Free ebooks ==> www.Ebook777.

com

Simulations of
Physics

Butikov
Simulations of Oscillatory Systems: with Award-Winning Software,

Oscillatory Systems
Physics of Oscillations provides a hands-on way of visualizing and un-
derstanding the fundamental concepts of the physics of oscillations. Both
the book and software are designed as exploration-oriented supplements for
courses in general physics and the theory of oscillations.

Simulations of Oscillatory Systems

The book is conveniently structured according to mathematical complexity. with Award-Winning Software,
Each chapter in Part I contains activities, questions, exercises, and problems
of varying levels of difficulty, from straightforward to quite challenging. Physics of Oscillations
Part II presents more sophisticated, highly mathematical material that
delves into the serious theoretical background for the computer-aided study
of oscillations.

The software package allows you to observe the motion of linear and
nonlinear mechanical oscillatory systems and to obtain plots of the
variables that describe the systems along with phase diagrams and plots
of energy transformations. These computer simulations provide clear, vivid
illustrations of oscillations in various physical systems, bringing to life many
abstract concepts, developing your physical intuition, and complementing
the analytical study of the subject.

Features
• Provides interactive software that serves as a desktop laboratory for
exploring simulated systems and replicating the experiments
• Enables you to perform interesting mini-research projects involving
the physics of oscillations
• Develops your physical intuition and theoretical foundations
• Includes various examples of the behavior of simulated systems
• Requires no previous knowledge of algorithmic languages or
programming

K24904

Eugene I. Butikov
6000 Broken Sound Parkway, NW
Suite 300, Boca Raton, FL 33487 ISBN: 978-1-4987-0768-8
711 Third Avenue 90000
New York, NY 10017
an informa business 2 Park Square, Milton Park
www.crcpress.com Abingdon, Oxon OX14 4RN, UK
9 781498 707688
w w w.crcpress.com
www.Ebook777.com

K24904 cvr mech.indd 1 1/19/15 11:32 AM

 i i

i i

i i

i i
 i i

i i
Free ebooks ==> www.Ebook777.com

Simulations of
Oscillatory Systems

with Award-Winning Software,

Physics of Oscillations

i
www.Ebook777.com i

i i
 i i

i i

i i

i i
 i i

i i
Free ebooks ==> www.Ebook777.com

Simulations of
Oscillatory Systems

with Award-Winning Software,

Physics of Oscillations

Eugene I. Butikov
Department of Physics
Saint Petersburg State University
Saint Petersburg, Russia

i
www.Ebook777.com i

i i
CRC Press
Taylor & Francis Group
6000 Broken Sound Parkway NW, Suite 300
Boca Raton, FL 33487-2742
© 2015 by Taylor & Francis Group, LLC
CRC Press is an imprint of Taylor & Francis Group, an Informa business

No claim to original U.S. Government works

Version Date: 20150121

International Standard Book Number-13: 978-1-4987-0770-1 (eBook - PDF)

This book contains information obtained from authentic and highly regarded sources. Reasonable
efforts have been made to publish reliable data and information, but the author and publisher cannot
assume responsibility for the validity of all materials or the consequences of their use. The authors and
publishers have attempted to trace the copyright holders of all material reproduced in this publication
and apologize to copyright holders if permission to publish in this form has not been obtained. If any
copyright material has not been acknowledged please write and let us know so we may rectify in any
future reprint.

Except as permitted under U.S. Copyright Law, no part of this book may be reprinted, reproduced,
transmitted, or utilized in any form by any electronic, mechanical, or other means, now known or
hereafter invented, including photocopying, microfilming, and recording, or in any information stor-
age or retrieval system, without written permission from the publishers.

For permission to photocopy or use material electronically from this work, please access www.copy-
right.com (http://www.copyright.com/) or contact the Copyright Clearance Center, Inc. (CCC), 222
Rosewood Drive, Danvers, MA 01923, 978-750-8400. CCC is a not-for-profit organization that pro-
vides licenses and registration for a variety of users. For organizations that have been granted a photo-
copy license by the CCC, a separate system of payment has been arranged.

Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and are
used only for identification and explanation without intent to infringe.
Visit the Taylor & Francis Web site at
http://www.taylorandfrancis.com
and the CRC Press Web site at
http://www.crcpress.com
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

Contents

Preface xi

Introduction xiii
Classification of Oscillations . . . . . . . . . . . . . . . . . . . . . . . xiv
Simulated Physical Systems . . . . . . . . . . . . . . . . . . . . . . . xvi
How to Use the Software . . . . . . . . . . . . . . . . . . . . . . . . . xvii
Notes to the Instructor . . . . . . . . . . . . . . . . . . . . . . . . . . . xxii

I Oscillations in Simple Systems 1

1 Free Oscillations of a Linear Oscillator 3
1.1 Summary of the Theory . . . . . . . . . . . . . . . . . . . . . . 3
1.1.1 General Concepts . . . . . . . . . . . . . . . . . . . . . 3
1.1.2 Differential Equation of a Linear Torsion Oscillator . . . 4
1.1.3 The Time of Damping and the Quality Factor Q . . . . . 6
1.1.4 The Phase Diagram of a Linear Oscillator . . . . . . . . 9
1.1.5 Energy Transformations . . . . . . . . . . . . . . . . . 11
1.1.6 The Computer Simulation of a Linear Oscillator . . . . . 13
1.2 Review of the Principal Formulas . . . . . . . . . . . . . . . . . 13
1.3 Questions, Problems, Suggestions . . . . . . . . . . . . . . . . . 15
1.3.1 Free Undamped Oscillations . . . . . . . . . . . . . . . 15
1.3.2 Damped Free Oscillations . . . . . . . . . . . . . . . . 16
1.3.3 Non-Oscillatory Motion of the System . . . . . . . . . . 17

2 Torsion Spring Oscillator with Dry Friction 19

2.1 Summary of the Theory . . . . . . . . . . . . . . . . . . . . . . 19
2.1.1 General Concepts . . . . . . . . . . . . . . . . . . . . . 19
2.1.2 The Physical System . . . . . . . . . . . . . . . . . . . 19
2.1.3 The Differential Equation of the Oscillator . . . . . . . . 22
2.1.4 Damping Caused by Dry Friction . . . . . . . . . . . . . 23
2.1.5 The Phase Trajectory . . . . . . . . . . . . . . . . . . . 24
2.1.6 Energy Transformations . . . . . . . . . . . . . . . . . 25
2.1.7 The Role of Viscous Friction . . . . . . . . . . . . . . . 26

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

vi CONTENTS

2.2 Review of the Principal Formulas . . . . . . . . . . . . . . . . . 28

2.3 Questions, Problems, Suggestions . . . . . . . . . . . . . . . . . 28
2.3.1 Damping Caused by Dry Friction . . . . . . . . . . . . . 29
2.3.2 Influence of Viscous Friction . . . . . . . . . . . . . . . 30

3 Forced Oscillations in a Linear System 33

3.1 Summary of the Theory . . . . . . . . . . . . . . . . . . . . . . 33
3.1.1 Basic Concepts . . . . . . . . . . . . . . . . . . . . . . 34
3.1.2 Discussion of the Physical System . . . . . . . . . . . . 34
3.1.3 The Differential Equation for Forced Oscillations . . . . 36
3.1.4 The Principle of Superposition . . . . . . . . . . . . . . 37
3.2 Steady-State Forced Oscillations . . . . . . . . . . . . . . . . . 38
3.2.1 Forced Oscillations in the Absence of Friction . . . . . . 38
3.2.2 The Resonance Curve . . . . . . . . . . . . . . . . . . . 40
3.2.3 Resonance of the Angular Velocity . . . . . . . . . . . . 42
3.2.4 Energy Transformations . . . . . . . . . . . . . . . . . 42
3.3 Transient Processes . . . . . . . . . . . . . . . . . . . . . . . . 45
3.3.1 Initial Conditions That Eliminate a Transient . . . . . . 46
3.3.2 Forced Oscillations from Rest at Resonance . . . . . . . 47
3.3.3 Mechanical Analogue of the Stimulated Emission of
Radiation . . . . . . . . . . . . . . . . . . . . . . . . . 50
3.3.4 Transient Processes Near Resonance . . . . . . . . . . . 53
3.3.5 Transient Processes Far from Resonance . . . . . . . . . 54
3.3.6 Transient Processes and the Phase Trajectory . . . . . . 55
3.4 Review of the Principal Formulas . . . . . . . . . . . . . . . . . 57
3.5 Questions, Problems, Suggestions . . . . . . . . . . . . . . . . . 58
3.5.1 Steady-State Forced Oscillations . . . . . . . . . . . . . 58
3.5.2 Transient Processes . . . . . . . . . . . . . . . . . . . . 60

4 Square-Wave Excitation of a Linear Oscillator 65

4.1 Theoretical Background . . . . . . . . . . . . . . . . . . . . . . 65
4.1.1 Model of the Physical System . . . . . . . . . . . . . . 65
4.1.2 The Differential Equation of Forced Oscillations . . . . 67
4.2 Steady-State Forced Oscillations under the Square-Wave Torque 68
4.2.1 Harmonics of the Driving Force and of the Steady-State
Response . . . . . . . . . . . . . . . . . . . . . . . . . 68
4.2.2 Forced Oscillations as Natural Oscillations about the
Alternating Equilibrium Positions . . . . . . . . . . . . 73
4.3 Transient Processes under the Square-Wave External Torque . . . 75
4.4 Estimation of the Amplitude of Steady-State Oscillations . . . . 77
4.4.1 Resonant Amplitude of Steady-State Oscillations . . . . 78
4.4.2 Amplitude of Steady Oscillations at Strong Friction . . . 80
4.4.3 Amplitude of Steady Oscillations at T = 2nT0 . . . . . 81
4.4.4 Steady-State Oscillations at High Frequencies of the
Square-Wave Torque . . . . . . . . . . . . . . . . . . . 82

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

CONTENTS vii

4.5 Energy Transformations . . . . . . . . . . . . . . . . . . . . . . 83

4.6 The Electromagnetic Analogue of the Mechanical System . . . . 85
4.7 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . 86
4.8 Review of the Principal Formulas . . . . . . . . . . . . . . . . . 87
4.9 Questions, Problems, Suggestions . . . . . . . . . . . . . . . . . 88
4.9.1 Swinging of the Oscillator at Resonance . . . . . . . . . 88
4.9.2 Non-Resonant Forced Oscillations . . . . . . . . . . . . 89

5 Parametric Excitation of Oscillations 93

5.1 Summary of the Theory. General Concepts . . . . . . . . . . . . 93
5.1.1 Classification of Oscillations . . . . . . . . . . . . . . . 93
5.1.2 The Simulated Physical System . . . . . . . . . . . . . 94
5.1.3 Electromagnetic Analogue of the Mechanical System . . 97
5.1.4 Conditions for Parametric Resonance . . . . . . . . . . 97
5.1.5 The Threshold of Parametric Excitation . . . . . . . . . 98
5.1.6 Differential Equation for Parametric Oscillations . . . . 101
5.1.7 The Mean Natural Period at Large Modulation . . . . . . 102
5.2 Frequency Ranges of Parametric Excitation . . . . . . . . . . . . 103
5.2.1 Main Interval of Parametric Excitation . . . . . . . . . . 104
5.2.2 Third-Order Interval of Parametric Instability . . . . . . 107
5.2.3 Frequency Ranges for Resonances of Even Orders . . . . 109
5.2.4 Intersections of the Boundaries at Large Modulation . . 112
5.2.5 Intervals of Excitation in the Presence of Friction . . . . 113
5.3 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . 118
5.4 Questions, Problems, Suggestions . . . . . . . . . . . . . . . . . 118
5.4.1 Principal Parametric Resonance . . . . . . . . . . . . . 118
5.4.2 Manual Control of the Parameter . . . . . . . . . . . . . 121
5.4.3 Parametric Resonances of High Orders . . . . . . . . . . 122

6 Sinusoidal Modulation of the Parameter 125

6.1 Summary of the Theory: Basic Concepts . . . . . . . . . . . . . 125
6.1.1 The Physical System . . . . . . . . . . . . . . . . . . . 125
6.1.2 Physical Reasons for Parametric Excitation at Smooth
Modulation . . . . . . . . . . . . . . . . . . . . . . . . 126
6.1.3 Conditions of Parametric Resonance . . . . . . . . . . . 126
6.1.4 Energy Transformations at Parametric Excitation . . . . 128
6.1.5 The Threshold of Parametric Excitation . . . . . . . . . 131
6.1.6 Differential Equation for Sinusoidal Motion of the
Weights along the Rod . . . . . . . . . . . . . . . . . . 132
6.2 The Intervals of Parametric Instability . . . . . . . . . . . . . . 134
6.2.1 The Principal Interval of Instability . . . . . . . . . . . . 134
6.2.2 Resonance of the Second Order . . . . . . . . . . . . . 139
6.2.3 Resonances of the Third and Higher Orders . . . . . . . 144
6.3 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . 147
6.4 Questions, Problems, Suggestions . . . . . . . . . . . . . . . . . 148

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

viii CONTENTS

6.4.1 Principal Parametric Resonance . . . . . . . . . . . . . 148

6.4.2 The Principal Interval of Parametric Resonance . . . . . 150
6.4.3 The Second Parametric Resonance . . . . . . . . . . . . 151

II Nonlinear Oscillations 153

7 Free Oscillations of the Rigid Pendulum 155
7.1 Summary of the Theory . . . . . . . . . . . . . . . . . . . . . . 155
7.1.1 The Physical System . . . . . . . . . . . . . . . . . . . 155
7.1.2 The Differential Equation of Motion for a Pendulum . . 157
7.1.3 Dependence of the Period on the Amplitude . . . . . . . 158
7.1.4 The Phase Portrait of the Pendulum . . . . . . . . . . . 159
7.1.5 The Phase Portrait in the Simulation Program . . . . . . 164
7.1.6 The Limiting Motion along the Separatrix . . . . . . . . 165
7.2 Oscillations of the Pendulum with Extremely Large
Amplitudes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
7.2.1 Oscillations with Amplitudes Approaching 180◦ . . . . 169
7.2.2 Another Derivation of the Expression for the Period
of Large Oscillations . . . . . . . . . . . . . . . . . . . 173
7.3 Period of Revolutions and Large Oscillations . . . . . . . . . . . 174
7.3.1 The Period of Fast Revolutions . . . . . . . . . . . . . . 174
7.3.2 Relationship between the Periods of Revolutions
and Large Oscillations . . . . . . . . . . . . . . . . . . 174
7.3.3 Mean Values of the Potential and Kinetic Energies . . . 178
7.4 The Influence of Friction . . . . . . . . . . . . . . . . . . . . . 179
7.4.1 The Phase Portrait of the Pendulum in the Presence
of Friction . . . . . . . . . . . . . . . . . . . . . . . . . 179
7.4.2 Revolutions Followed by Oscillations . . . . . . . . . . 180
7.5 Review of the Principal Formulas . . . . . . . . . . . . . . . . . 181
7.6 Questions, Problems, Suggestions . . . . . . . . . . . . . . . . . 182
7.6.1 Small Oscillations of the Pendulum . . . . . . . . . . . 182
7.6.2 Oscillations with Large Amplitudes . . . . . . . . . . . 184
7.6.3 The Rotating Pendulum . . . . . . . . . . . . . . . . . . 187

8 Rigid Planar Pendulum under Sinusoidal Forcing 189

8.1 Regular Response of a Harmonically Driven Rigid Pendulum . . 189
8.1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . 189
8.1.2 The Physical Model . . . . . . . . . . . . . . . . . . . . 190
8.1.3 Behavior of the Pendulum under the Slow Varying
Sinusoidal Torque Whose Amplitude is Close to 1 . . . . 191
8.2 Steady-State Response-Frequency Curves . . . . . . . . . . . . 196
8.2.1 Approximate Theoretical Resonance Curve . . . . . . . 196
8.2.2 Autoresonance, Hysteresis, and Bistability . . . . . . . . 198
8.2.3 Nonlinear Resonance and a “Bell-Ringer Mode” . . . . 199

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

CONTENTS ix

8.2.4 Symmetry-Breaking and Period-Doubling Bifurcations,

Chaos and the Crisis . . . . . . . . . . . . . . . . . . . 201
8.3 Subharmonic and Superharmonic Resonances . . . . . . . . . . 203
8.4 Other Extraordinary Regular Forced Oscillations . . . . . . . . . 206
8.5 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . 211

9 Pendulum with a Square-Wave Modulated Length 213

9.1 The Investigated Physical System . . . . . . . . . . . . . . . . . 213
9.1.1 The Square-Wave Modulation of the Pendulum Length . 214
9.1.2 Conditions and Peculiarities of Parametric Resonance . . 217
9.2 The Threshold of Parametric Excitation . . . . . . . . . . . . . . 218
9.2.1 The Energy Supplied by the Square-Wave Modulation . 218
9.2.2 Regime of Parametric Regeneration . . . . . . . . . . . 219
9.2.3 Transients over the Threshold . . . . . . . . . . . . . . 221
9.2.4 Parametric Swinging with the Feedback . . . . . . . . . 222
9.3 Autoresonance, Bifurcations, Multistability . . . . . . . . . . . . 223
9.3.1 Autoresonance . . . . . . . . . . . . . . . . . . . . . . 223
9.3.2 Bifurcations of Symmetry Braking and Period Doubling 223
9.4 Quantitative Theory of Parametric Excitation . . . . . . . . . . . 225
9.4.1 Differential Equation for Parametric Oscillations . . . . 225
9.4.2 The Mean Natural Period at Large Depth of Modulation 226
9.5 Frequency Ranges for Parametric Resonance . . . . . . . . . . . 227
9.5.1 Resonances of Odd Orders . . . . . . . . . . . . . . . . 227
9.5.2 Main Interval of Parametric Instability . . . . . . . . . . 228
9.5.3 Third-Order Interval of Parametric Instability . . . . . . 232
9.5.4 Parametric Resonances of Even Orders . . . . . . . . . . 233
9.5.5 Intersections of the Boundaries at Large Modulation . . 236
9.6 Intervals of Parametric Excitation in the Presence of Friction . . 237
9.6.1 Boundaries of Instability for Resonances of Odd Orders . 238
9.6.2 Resonances of Even Orders . . . . . . . . . . . . . . . . 242
9.7 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . 243

10 Rigid Pendulum with Oscillating Pivot 245

10.1 Introductory Notes . . . . . . . . . . . . . . . . . . . . . . . . . 245
10.2 Kapitza’s Pendulum — Dynamic Stabilization . . . . . . . . . . 246
10.3 The Physical Model of the Investigated System . . . . . . . . . . 248
10.4 Parametric Resonance . . . . . . . . . . . . . . . . . . . . . . . 249
10.5 Physical Reasons for Stability of the Inverted Pendulum . . . . . 251
10.6 An Approximate Quantitative Theory of the Inverted Pendulum . 253
10.7 Exact Differential Equation for Pendulum with Oscillating Pivot 257
10.8 Effective Potential Function for a Pendulum . . . . . . . . . . . 258
10.9 Subharmonic Resonances of High Orders . . . . . . . . . . . . . 262
10.9.1 Multiple-Nodding Oscillations of the Parametrically
Driven Pendulum . . . . . . . . . . . . . . . . . . . . . 263
10.9.2 Spectrum of Small-Amplitude n-Periodic Oscillations . . 267

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

x CONTENTS

10.9.3 Lower Boundary of the Dynamic Stabilization . . . . . . 269

10.9.4 Subharmonic Resonances of Fractional Orders . . . . . 273
10.9.5 Coexistence of Subharmonic Resonances of Different
Orders n . . . . . . . . . . . . . . . . . . . . . . . . . . 275
10.10 Upper Boundary of Dynamic Stability . . . . . . . . . . . . . . 277
10.10.1 The “Flutter” Mode and Ordinary Parametric Resonance 278
10.10.2 Boundaries of the Second-Order Subharmonic
Resonance . . . . . . . . . . . . . . . . . . . . . . . . . 281
10.10.3 The Influence of Friction . . . . . . . . . . . . . . . . . 283
10.11 Enhanced Criterion for Kapitza’s Pendulum Stability . . . . . . 286
10.11.1 Subharmonic Resonances at Arbitrary Frequencies
and Amplitudes of the Pivot Vibration . . . . . . . . . . 287
10.11.2 Improved Lower Boundary of Dynamic Stabilization . . 290
10.11.3 Improved Upper Boundary of Dynamic Stabilization . . 294
10.12 Complicated Regular Motions of the Parametrically Driven
Pendulum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 298
10.13 Chaotic Motions of the Pendulum . . . . . . . . . . . . . . . . 300
10.14 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . 303

11 Torsion Pendulum with Dry and Viscous Damping 307

11.1 Basics of the Theory . . . . . . . . . . . . . . . . . . . . . . . . 307
11.1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . 307
11.1.2 The Physical System . . . . . . . . . . . . . . . . . . . 309
11.1.3 The Differential Equation of the Oscillator . . . . . . . . 311
11.1.4 Damping of Free Oscillations under Dry Friction . . . . 312
11.2 Sinusoidally Driven Oscillator with Dry Friction . . . . . . . . . 313
11.2.1 Resonance in the Oscillator with Dry Friction under
Sinusoidal Excitation . . . . . . . . . . . . . . . . . . . 313
11.2.2 Analytical Solution for the Second Half-Cycle of the
Resonant Excitation . . . . . . . . . . . . . . . . . . . . 314
11.2.3 The Threshold of the Resonant Growth . . . . . . . . . 316
11.2.4 Resonance in the Presence of Dry and Viscous Friction . 317
11.2.5 Non-Resonant Forced Oscillations . . . . . . . . . . . . 319
11.3 Excitation at Sub-Resonant Frequencies . . . . . . . . . . . . . 322
11.4 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . 330

Bibliography 333

Index 339

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

Preface

This textbook, Simulations of Oscillatory Systems, is accompanied by the

award-winning educational software package, Physics of Oscillations. The text-
book and the software are developed as exploration-oriented supplements to var-
ious courses in general physics and the theory of oscillations, and can be helpful
to a wide range of students and researches. Successful usage of the software is
possible without any knowledge of algorithmic languages or programming. The
software, which is an essential part of the textbook, is designed to be a desktop
laboratory that can serve as an electronic training course for individual, interactive
work on a computer.
The software package, Physics of Oscillations, includes several highly interac-
tive computer programs presenting an extensive set of computer-simulated exper-
iments. Contemporary interactive media provides students with a powerful means
to visualize and to explore the fundamental concepts of the physics of oscillations.
With the textbook and the programs, students and their instructors get a powerful
tool that enables them to investigate phenomena that are difficult to imagine and
study in an abstract conventional manner. The simulation programs make visi-
ble the magnificence of mathematics in its application to the wonderful world of
oscillations.
Students can work at a pace they can enjoy, varying parameters of the sim-
ulated systems and repeating the most interesting experiments several times on
their own. The experience based on a student’s own actions results in deeper un-
derstanding than the mere reception of someone else’s knowledge. No doubt, for
a great majority of human beings, a visual impression is much more intensive and
permanent than a heard or read one. With some of the suggested programs, stu-
dents have an opportunity to perform interesting mini-research projects in physics.
The structure of the textbook and programs allows students to study the sub-
ject at different levels of difficulty, depending on the time available and on the
mathematical complexity of the course. The structuring of levels by degree of
difficulty and mathematical complexity is especially convenient. At an introduc-
tory level, the underlying concepts and physical laws are discussed without much
mathematics: The software suggests experiments that demonstrate typical exam-
ples of behavior of the simulated systems, and helps to develop physical intuition.
At an intermediate level, a more detailed though rather elementary description is
available. In Part I of the textbook each chapter is supplied with a section contain-

xi

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

xii PREFACE

ing suggested activity, questions, exercises, and thought-out problems with a wide
range of difficulty from straightforward to quite challenging. For those students
who’d like an in-depth investigation, Part II of the textbook and the corresponding
programs provide much more sophisticated, highly mathematical material, which
delves into the serious theoretical background for the computer-aided study of
oscillations.
Computer simulations provide very clear and impressively vivid illustrations
of oscillations in various physical systems. The screen displays subtle details that
usually escape us in direct observation. It is possible to change time scales and
to widely vary parameters and other experimental conditions. We can investigate
interesting situations that are inaccessible in a real experiment. The graphic repre-
sentation of experimental results allows us to easily collect and understand large
amounts of information.
The software package, Physics of Oscillations, allows users to observe the mo-
tion of various linear and nonlinear mechanical oscillatory systems directly on the
computer screen and to obtain plots of the variables that describe the system along
with phase diagrams and plots of energy transformations. The plots and phase di-
agrams appear on the screen simultaneously with the display of the motion. These
simulations bring to life many abstract concepts of the physics of oscillations. The
simulations aid greatly in developing our physical intuition and complement the
analytical study of the subject in a manner that is mutually reinforcing.
The package, Physics of Oscillations, runs under the Windows operating sys-
tem. To use the software, you should install it on your machine. To install the
first part of the package called, “Simple Systems,” download from the Web the
archive file butikov.faculty.ifmo.ru/MasterDiskOsc.zip and unpack it to a tem-
porary folder. Then, from Windows Explorer click on autorun.exe in the “Auto”
subfolder, or on setup.exe in the “Software” subfolder, and follow instructions on
the screen. The setup program will create an entry in the Start menu, and a shortcut
(Physics of Oscillations) on your desktop. To start Physics of Oscillations, click
the icon on the desktop, or open the Physics of Oscillations in the “Programs”
menu, and choose the subprogram you need. To install the second part, Nonlinear
Systems, download the archive file butikov.faculty.ifmo.ru/Nonlinear.zip.

Eugene I. Butikov
St. Petersburg State University, Russia

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

Introduction

Welcome to the wonderful world of oscillations! Oscillations are everywhere. The

Earth, like a giant bell, vibrates under our feet. Oscillating electric and magnetic
fields carry light to our eyes. Vibrating air carries sound to our ears. Through
electromagnetic waves and acoustic vibrations, we receive the major part of infor-
mation about the world surrounding us.
Oscillations in various physical systems may differ in physical nature, but they
also have much in common. A branch of science called the theory of oscillations
deals with the analysis of laws common to all oscillations.
It is easier to understand common laws of oscillation processes if we ana-
lyze them in the most plain and obvious examples; e.g., in mechanical systems
that are accessible to direct visual observation. For this purpose, the simulation
experiments in the package Physics of Oscillations deal with commonly known
mechanical systems such as the spring harmonic oscillator and the simple pendu-
lum.
A linear (or harmonic) oscillator is a system in which a displacement from
the equilibrium position causes a restoring force to appear that is proportional to
the displacement. The time dependence of the state of the system is described
by linear differential equations. Mechanical examples of such systems include a
weighted spring (spring–mass system) and a torsion pendulum. The latter could
be a flywheel that is attached to an elastic spiral spring. The spring twists when
the flywheel is turned around its axis, much like the balance device of a mechan-
ical watch. The vibrations of such a linear torsion pendulum are simulated in the
suggested software.
A common example of a nonlinear mechanical system is an ordinary pendu-
lum in the gravitational field. When the pendulum is moved from the equilibrium
position, the restoring torque of gravity is proportional to the sine of the deflection
angle. Therefore, for small oscillations, the pendulum may be considered a linear
oscillator, but at large angular displacements, its behavior differs greatly from that
of the linear oscillator.
In addition to natural (or free) oscillations, which are excited by some initial
action on an isolated system that is then left to itself, the software simulates forced
oscillations occurring under some periodic external action. It also treats parametric
oscillations occurring during the periodic change of some parameter of the system
to which the motion of the system is sensitive.

xiii

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

xiv INTRODUCTION

The simulation experiments allow the user to observe oscillations of various

physical systems directly on the screen and at the same time to draw the time-
dependent plots of the variables that characterize the system. The phase diagram
as well as the plots of potential, kinetic, and total energy are also available for a
detailed investigation of the system being analyzed. The experiments have been
designed to be plain and obvious. They provide the capability of observing re-
peatedly and thoroughly the fine details of phenomena that usually escape notice
during direct observation. The user can widely modify parameters of the physical
system and the initial conditions. The graphic presentation of the results allows
students to see and easily understand large amounts of information.

Classification of Oscillations
1. According to the physical nature of the phenomena involved, oscillations
in various systems are divided into mechanical oscillations and electromag-
netic ones. Mechanical oscillations are characterized by alternating conver-
sions of the kinetic energy into one (or several) kinds of potential energy
and back. In electromagnetic oscillations alternating conversions occur be-
tween the electric field energy (which is analogous to the potential energy
in mechanical systems) and the magnetic field energy (the analogue of the
kinetic energy). Sometimes oscillations have a combined mechanical and
electromagnetic nature, e.g., oscillations in plasma. Oscillations of different
physical nature obey common laws. These laws common to all oscillations
are studied by the theory of oscillations.
2. According to kinematics, i.e., the character of time dependence of some
physical quantity x(t) that characterizes the physical system, oscillations
are classified as periodic and non-periodic. An important variety of periodic
oscillations are sinusoidal or harmonic oscillations, when the time depen-
dence of x(t) is described by a sine (or cosine) function. The most important
kinds of non-periodic oscillations are almost sinusoidal ones. Such oscilla-
tions can be treated approximately as sinusoidal with a slowly changing
amplitude. Nearly sinusoidal oscillations with slowly varying amplitude—
modulated oscillations—are widely used in radio communication: A bear-
ing electromagnetic wave of a high frequency is modulated by oscillations
of a low (e.g., acoustic) frequency.
3. According to the means of excitation, oscillations are divided into four main
groups:

• Free or natural oscillations occurring in a system that is left to itself

after some initial excitation. Such oscillations take place about the po-
sition of stable equilibrium. In a conservative system the energy con-
versions during free oscillations are reversible, and (in a system with
one degree of freedom) the oscillations are exactly periodic. In a linear

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

CLASSIFICATION OF OSCILLATIONS xv

system, in which a restoring force is proportional to the displacement

from the equilibrium position, free oscillations without friction give
an example of simple harmonic motion. Their period is independent
of the amplitude. This property of a linear oscillator is called isochro-
nism.
In real systems, conversions of the energy from one kind to the other
and back are partially irreversible. Dissipation of the energy due to
friction (or electric resistance) causes damping of free oscillations.
• Forced oscillations occurring in a system (that can execute free oscil-
lations) under the influence of some periodic external action. Gener-
ally, the oscillations are called forced if the periodic external action
on the oscillatory system can be expressed by a separate term in the
differential equation describing the system. This term must be a given
periodic function of time.
On the expiry of some time after the periodic external force begins
to operate, forced oscillations become exactly periodic: They have a
constant amplitude and acquire the period of the external force. The
phase and the amplitude of these steady-state forced oscillations are
independent of the initial conditions. The dependence of the ampli-
tude of steady-state oscillations on the frequency ω of the external
force has a resonant character—the amplitude grows appreciably as ω
approaches the natural (resonant) frequency ω0 of the oscillator.
Initial conditions influence a transient process—the process of estab-
lishing steady-state oscillations. In a linear system the transient pro-
cess can be represented as a superposition of steady-state oscillations
whose frequency equals that of the external sinusoidal force and fad-
ing free oscillations that have the natural frequency. Usually a transient
process lasts as long a time as is necessary for natural oscillations in
the system to damp away.
• Parametric oscillations occurring in an oscillatory system excited by
a periodic variation of some parameter of the system. The most famil-
iar example of parametric oscillations is given by an ordinary child’s
swing excited by periodic changes of its length.
In the case of parametric oscillations the system is subjected to non-
stationary forces that depend not only on time (explicitly) but also on
coordinates. This case is more complicated for investigation. Let, for
example, a restoring force F = −kx arise when the system is dis-
placed from the equilibrium position, but in contrast to the stationary
case the parameter k changes with time due to some periodic influ-
ence: k = k(t).
In the differential equation of the system the coefficient of x is not
constant; it depends on time explicitly. Oscillations in such a system
differ essentially both from natural free oscillations, described by a

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

xvi INTRODUCTION

differential equation with constant coefficients, and from forced oscil-

lations caused by an external force that depends only on time.
In the case of periodic changes of the parameter k, when k(t + T ) =
k(t), oscillations in a system are called parametrically excited or sim-
ply parametric. When the process of oscillations caused by the pe-
riodic modulation of some parameter acquires an increasing charac-
ter, the phenomenon is called parametric resonance. In the case of
parametric resonance the state of equilibrium becomes unstable: The
system leaves it executing oscillations with progressively increasing
amplitude.
Parametric resonance differs considerably in some characteristics
from ordinary resonance caused by a periodic external force exerted
on the oscillatory system. Parametric resonance occurs when different
relationships between the frequency of modulation of a parameter and
the natural frequency of the system are fulfilled. To excite resonance,
the amplitude of modulation must be large enough: The depth of mod-
ulation must exceed some threshold value in order to cause parametric
resonance.
• Self-excited or self-sustained oscillations (or auto-oscillations) occur-
ring without any periodic influence. They are possible in oscillatory
systems that have a constant (non-periodic) source of energy and a
positive feedback. Examples of self-sustained oscillations are given
by mechanical watches or by electromagnetic oscillators. These non-
linear systems can regulate energy supply from a constant source to
compensate losses caused by friction or resistance.

4. According to complexity, oscillatory systems are divided into simple sys-

tems (systems with only one degree of freedom) characterized by a single
natural frequency; systems with lumped (concentrated) parameters with a
finite number (or countable infinite number) of degrees of freedom, hav-
ing as many natural frequencies (frequencies of normal oscillations) as the
number of degrees of freedom; systems with distributed parameters having
a continuous infinite number of degrees of freedom and consequently an
infinite number of natural frequencies.

Simulated Physical Systems

Part I of the textbook deals with the following types of oscillations in mechanical
systems:

• Free oscillations of a linear torsion spring pendulum

• The torsion spring oscillator with dry and viscous friction
• Forced oscillations in a linear system under a sinusoidal force

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

HOW TO USE THE SOFTWARE xvii

• Non-sinusoidal (square-wave) external force in a linear system

• Parametric square-wave excitation of a linear torsion pendulum
• Parametric resonance at sinusoidal excitation a linear pendulum

Part II of the textbook covers more advanced and sophisticated topics of the
physics of oscillations. Nonlinear systems that demonstrate chaotic behavior are
considered. In particular, the package contains the following simulations:

• Free oscillations and rotations of a nonlinear rigid pendulum

• Forced swinging of a rigid planar pendulum and nonlinear resonance
• Parametric excitation of a rigid pendulum by periodic changes of its length
• Parametrically driven pendulum with vertical oscillations of the pivot
• Dynamic stabilization of inverted pendulum by oscillations of the pivot
• Torsion spring oscillator with dry and viscous friction under sinusoidal ex-
ternal force

How to Use the Software

Each chapter of the textbook is related to one of the simulation programs of
the package Physics of Oscillations. A chapter begins with a description of the
simulated physical system and includes brief but important information on the
theory of the phenomenon under consideration. In part I, much of this informa-
tion may be familiar to you from lectures or traditional textbooks. If you feel sure
you know the material well, you can skip it and begin with the part entitled “Ques-
tions, Problems, Suggestions.” If you want to learn something new while using the
software, this section of each chapter is very important. Try to solve the problems
given therein beforehand. You will then be able to verify your solutions in a com-
putation experiment. This will make your work on the computer much more in-
teresting. If you find that the experimental results disagree with your predictions,
try to find the reason for the differences. Much real research in physics amounts
to the analysis of discrepancies between theoretical predictions and experimental
results.
Below is a brief overview of the basic control functions carried out with the
help of the command buttons and menu items.

• Menu Bar at the top of each window displays the commands used to op-
erate the program. Also right-clicking the mouse anywhere in the program
window invokes a pop-up menu with several commands.

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

xviii INTRODUCTION

• Command Buttons under the menu bar provide quick access to several
commonly used commands. You click a command button once to carry out
the action represented by that button.
• Sliders with the labels “Speed Up” and “Slow Down” allow you to vary the
speed of animation for a convenient observation by changing the time scale
in which the motion is simulated and displayed.

Command Buttons common to most of the programs execute the following ac-
tions:

• Start, Pause, Go – starts the simulation, makes a pause in the simulation,

continues the interrupted simulation;
• Restart – restores the initial conditions and repeats the simulation from the
beginning;
• Erase – clears the windows (erases all graphs and rescales), and continues
the simulation;
• Skip Over – skips the simulation over several periods of the external action;
• Erase – removes the old plots (and rescales) before going on;
• Change – calls up the control panel to change properties;
• Options – calls up the control panel to choose options (see details below);
• Exit – returns to the starting page of the program;
• Time Scale – changes the number of periods displayed on the plots;
Animation – speeds up or slows down the animation.

Menu items common to all the programs carry out the following functions:

• File:

– Exit – closes all windows and terminates the program;

– Print – opens the panel to make the printer settings;
• Input – offers to choose a parameter for input or opens the panel for en-
tering several parameters of the simulated system (and the initial conditions
for the simulation);
• Options:
If you click the button Options or the menu item “Options,” a control panel
appears from which you can choose a mode of simulation and a mode of
displaying information that best fits your requirements:

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

HOW TO USE THE SOFTWARE xix

– You can choose one of three option buttons in plotting functions: 1)

the program is to pause in the simulation when all the space for the
plots is used and is to wait for your next command, or 2) it should
continue drawing new plots over the old ones, or 3) it should erase
them before continuing plotting.
– You can choose the thickness of lines for drawing the plots, and the
background color (dark or white) by checking the switches;
– You can make the image of the system more realistic (with the massive
bobs on the rotor) by choosing “two-sided” from the option “needle
image;”
– If you wish to suppress the automatic choice of scale and to choose
your own scale, select the option ”Custom Scale” and indicate the
maximal values of the angular displacement and angular velocity to
be displayed on the plots.

• Examples – offers a list of predefined examples to choose from or opens

the panel with a set of examples provided;
• Zoom – uses all the space available on the screen for drawing the plots and
phase diagram.

Other controls specific for separate programs are described in the corresponding
sections of the textbook. For more information on controls, search “How To . . . ”
under the menu “Help” from within the programs.
Context-sensitive physical explanations can be called by pressing the F1 key
or by clicking the menu item “Help on physics.” Contents of the displayed help
reference depend on the place where the program was interrupted. Each simulation
program has a section entitled “Information about the Physical System.” It might
be a good idea to begin your work with a program by looking through this section.
You will find a brief description of the physical system and some dynamic illus-
trations of its behavior, and gradually you will get used to its conventional image
on the display screen. This image is intentionally made to appear very schematic
to remind you that in the simulation experiment you are dealing not with a real
physical system but with some idealized model of it.
It may prove helpful to look through the sections concerning an experiment
in a survey mode (by choosing various examples from the menu item Examples)
before you proceed to a detailed examination. In this mode, you need not enter
data or choose options. The sets of available examples vary when you change
from one screen configuration to another by clicking the “View” item. You can
thereby get an idea of the contents and of the amount of work to be done. All
quantities are then assigned suitable default values.
Clicking the menu item “View” allows you to choose one of the four screen
configurations, which differ by the amount of information displayed:

1. Image of the physical system (only);

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

xx INTRODUCTION

2. . . . plus the plots of time dependence of the angular displacement and of the
angular velocity;
3. . . . plus the phase diagram;
4. . . . plus the plots of energy transformations.

Command buttons in the Tutorial perform the following functions:

Go, Pause – begin the simulation, pause, resume;
<< Back – return to the preceding item of the Tutorial;
More >> – go to the next page of the Tutorial;
Contents – call up the contents of the Tutorial;
Exit – return to the starting page of the program;
Animation – speed up or slow down the animation.
To change properties of the physical system (such as the quality factor, the
dead zone, the frequency and amplitude of the external action, etc.—depending
on the system under consideration) or the initial conditions (values of the angular
displacement and angular velocity at t = 0), click on the button Change or the
item “Settings” (or “Input”) in the menu. A control panel then appears in which
you can type in the new values in corresponding dialog boxes, or you can change
the values by dragging the slides on the scroll-bars. When all chosen values are
specified, confirm your choice by clicking the OK button. Or click the Cancel
button, if you wish to delete the changes and return to the previous values.
Instead of entering data, you can look for a suitable example of the case you
are interested in by clicking the menu item “Examples.” In each example all nec-
essary properties and initial conditions are already specified. The set of available
examples depends on the screen configuration you have chosen from the menu
item “View.”
The software allows you to make a hard copy of the plots with the help of your
system printer. The plots will have the best resolution your printer can provide. All
the graphs in this textbook are obtained in this way. To print the plots, perform the
following steps:

• Input, if necessary, the values of parameters and initial conditions (the plot
will correspond to the current values of the system parameters);
• Choose the menu item “Print.” A printer control panel will appear;
• In the panel, choose the plot you wish to print (angular displacement or an-
gular velocity time dependence, phase diagram, or energy transformations)
by clicking the corresponding option button;
• Input the plot dimensions (width and height) in the units you are used to
(inches, centimeters, pixels, or points). The program will round out the val-
ues in order to get the best result on your printer;

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

HOW TO USE THE SOFTWARE xxi

• Select the time interval for the plot: Indicate the number of the natural peri-
ods (or the external force periods, or the periods of modulation, depending
on the program) to be reproduced (time scale), and the initial instant of the
time interval;
• If you wish, you can add a caption to the plot (type its text into the corre-
sponding box), and a header (for the whole page where you can print several
plots);
• Choose whether or not to print the legend and the values of parameters
(under the plot);

• Click the Go command button after all the necessary settings have been
made.
• After the image of the plot is stored in the computer’s memory (the label
“Wait...” will change to the label “Ready”), you can choose several possi-
bilities to continue:

1. Get a hard copy of the plot by clicking the Eject command button.
Only one plot will be printed out;
2. If you click the More command button, then the next part of the
process (one more interval ordered in the box “Time Scale”) will be
placed over the same plot (together with the previous part);
3. You can store another plot image in order to print it below on the
same sheet of paper. To do this, repeat the previous steps and make
the settings for the second plot. The program will warn you if free
space on the sheet is insufficient for the next plot;

• Click on the Eject command button to send all the stored images to the
printer.

If you check the box “Write to ASCII file,” all the data of the simulation ex-
periment will be placed in a file (without any output to your printer). You can
use this file afterwards, managing the data with a third-party software. The first
row of the sheet identifies the values of the corresponding columns (time, angular
displacement, angular velocity, energy, etc.). If you store plotting data in a file,
you need not indicate the dimensions of the plot or choose other options (what
to print or what captions or headers to make). The only significant parameter you
need to specify is the number of periods of the simulation experiment that should
be performed and stored. The stored data will apply to the current values of the
properties of the physical system and the initial conditions.
On-line instructions concerning the operation are available from within the
programs under the menu item “Help,” “Help on Controls.”

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

xxii INTRODUCTION

Notes to the Instructor

This textbook, Simulations of Oscillatory Systems, with the software package,
Physics of Oscillations, gives the instructor several choices on how to use it in
physics courses. Certainly, it can be used in lecture demonstrations to illustrate
various aspects of free, forced, and parametric oscillations in mechanical systems
and in oscillatory circuits. The teacher can save much time and effort in avoiding
long and dull explanations by using the dynamic simulations offered by these pro-
grams. But surely there is a much more efficient way to exploit the clear and plain
character of the simulations.
A more important use of the textbook and the software is as a desktop labo-
ratory equipped with a set of ready and well-adjusted experimental devices with
which students may easily interact. When our aim is to learn physics, we should
avoid wasting time on auxiliary technical details concerning the apparatus, in spite
of their importance to the success of a real experiment. The same is true for the
computation experiment in which we deal with idealized mathematical models
rather than with real physical systems. If we are not interested in the inner work-
ings of the simulation program itself, an evident advantage of ready and debugged
software is that it allows us to concentrate entirely on the physical principles in-
volved.
The simulation experiments in the Physics of Oscillations software involve
mechanical systems primarily because the motion of such systems is easily rep-
resented on the computer screen. Such visualization makes the simulation exper-
iments very convincing and easy to understand. When we begin with systems
whose behavior is familiar (a simple torsion spring oscillator or a rigid pendu-
lum) and see that the simulation shows us just what we would expect from a real
system, the subsequent simulations of nonlinear and more complex systems (the
driven pendulum or the parametrically excited oscillator) become more convinc-
ing. These are cases when our intuition based on common everyday experience
often fails to work.
The most satisfactory pedagogical effect can be achieved if the learning pro-
cess is organized so that students study the subject matter in the textbook before-
hand and solve the problems and exercises assigned by the instructor. They are
then more likely to know what to do when working with the program. It is also
helpful for students to study the derivations of principal formulas and to attempt
some of the derivations omitted in the textbook. However, especially important
are the problems and the exercises.
Stars (asterisks) are added to problem numbers to grade them by difficulty.
Simple problems have no stars. One star means that the problem requires thought.
Two- and especially three-star problems are challenging in some way. The best
learning will take place while a student is struggling with the tough problems.
Most of the exercises ask students to calculate some value they should input
in order to get a desired result. Students thus have an opportunity to verify their
answers (and thus the understanding of the subject as a whole) not by peeping in
the back of the textbook, but rather by a real experiment with the mathematical

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

NOTES TO THE INSTRUCTOR xxiii

model of the simulated physical system. The students’ work on the computer is
thus made much more interesting and fruitful.
The textbook and the software are designed for use at a wide range of levels
and can be adapted to meet the needs of many physics courses. The structure
of the textbook and the software makes them accessible to some extent even to
introductory physics students who are unfamiliar with the calculus and are unable
to follow all the derivations given in the textbook. Nevertheless, such students
can acquire a qualitative understanding of the phenomena and can improve their
physical intuition through experimental work on the computer. Animation and the
simultaneous drawing of plots of different variables make it easy to understand a
large amount of information.
Advanced undergraduate and graduate students can find in the textbook and
the software a lot of both interesting and useful optional material. With it they
may perform small research projects and thereby come to appreciate the beauty of
oscillatory phenomena.
The choice of subjects to be studied depends mainly on the content of the
physics course in question and on the classroom hours available. In any case,
certain material ought to be considered obligatory: The program “Free Oscilla-
tions in a Linear System” (including damping caused by viscous friction); the
program “Forced Oscillations of Linear Torsion Spring Pendulum” (which may be
restricted to steady-state oscillations, with only a notion of transients—especially
at the resonance).
Material that can be optional includes the program “Torsion Spring Oscillator
with Dry Friction” (interesting because it illustrates (1) the physical cause for ran-
dom errors arising in measuring instruments using a needle, and (2) the method
of joining the solutions of linear differential equations describing the motion dur-
ing successive time intervals); the program “Non-Sinusoidal External Force in a
Linear System”—an impressive example of the spectral decomposition of a peri-
odic process, and of the concept of how the spectrum of a signal is transformed
in a linear system (an explanation is given of how a square-wave input voltage
is distorted by an oscillatory circuit and transformed into the output voltage of a
different shape); the programs “Parametric Excitation of Linear Oscillator” and
“Sinusoidal Modulation of the Parameter.”
The physical systems that are simulated in these latter two programs may seem
exotic and even somewhat ridiculous. However, they give a very clear example
of the parametric excitation of a linear mechanical system. All peculiarities of
parametric resonance can be exhaustively investigated in this case, and its physi-
cal properties are completely explained. The possibility of manual control of the
moment of inertia aids a great deal in understanding the phenomenon. This sim-
ulation provides a good background for the study of more complicated nonlinear
parametric systems like a pendulum whose length is periodically changed (model
of a swing), or a pendulum with the suspension point driven periodically in the
vertical direction. (These systems are included in Part II of the textbook).
In Part II of the textbook several nonlinear physical systems are investigated
with the help of simulation programs dealing with more advanced and sophisti-

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

xxiv INTRODUCTION

cated problems of the physics of oscillations. In particular, the following physical

systems are considered:

• Free oscillations and rotations of a planar rigid pendulum. Dependence of

the period on the amplitude. Phase portrait. Limiting motion along the sep-
aratrix. Oscillations with extremely large amplitudes and revolutions.
• Forced swinging and revolutions of the rigid planar pendulum by the peri-
odic external torque. Nonlinear resonance, phase locking, hysteresis at slow
scanning of the external force frequency, chaotic behavior of a simple dy-
namical system, strange attractor.
• Parametric excitation of the rigid planar pendulum by square-wave modu-
lation of its length (a simplified model of the playground swing). Intervals
of parametric instability.
• Constrained vertical oscillations of the suspension point of a rigid pendu-
lum. Dynamic stabilization of the parametrically driven inverted pendulum
at vertical oscillations (Kapitza’s pendulum).
• Forced oscillations in a mechanical system with dry (Coulomb) and viscous
friction.

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

Part I

Oscillations in Simple Systems

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

Chapter 1

Free Oscillations of a Linear

Oscillator

Annotation. In Chapter 1 free (unforced) oscillations in a simple mechanical sys-

tem and its electromagnetic analogue—series LCR-circuit—are investigated both
analytically and with the help of computer simulations. Basic general concepts
of the theory of oscillations are introduced and discussed. Chapter 1 includes a
description of the simulated physical system and a summary of the relevant theo-
retical material for students as a prerequisite for the virtual lab “Free Oscillations
of Linear Torsion Pendulum” from the software package Physics of Oscillations
(Part I, Simple Systems). A set of theoretical and experimental problems to be
solved by students on their own is included, as well as various assignments that
the instructor can offer students for possible individual work.

1.1 Summary of the Theory

1.1.1 General Concepts
Equilibrium of a physical system is called stable if under any disturbance a restor-
ing force (or a restoring torque) arises that tends to return the system to the equi-
librium position. If the system is disturbed from this state by some external initial
action and then left to itself, the system subsequently oscillates about its equilib-
rium position. Such unforced oscillations are called free or natural, because they
are not driven by an external force. Any system that is able to execute free oscilla-
tions in the vicinity of its position of stable equilibrium—a repetitive motion back
and forth around the equilibrium position—is called an oscillator.
In the absence of friction, energy transmitted to the system at the initial ex-
citation is conserved, that is, remains constant during subsequent oscillations. In
such an idealized conservative system with one degree of freedom, the motion is
strictly periodic and continues indefinitely without damping. If there is friction in

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

4 CHAPTER 1. FREE OSCILLATIONS OF A LINEAR OSCILLATOR

L
j (t)

Figure 1.1: Schematic image of the torsion spring oscillator simulated in the pro-
gram and its electromagnetic analog—LCR-circuit.

the system, free oscillations are damped: they gradually fade away because of the
dissipation of energy, and the system eventually comes to rest in the equilibrium
position.
If the restoring force that tends to return a disturbed conservative system to its
equilibrium position is proportional to the displacement from this position, oscil-
lations of the system are harmonic. Simple harmonic motion occurs when there
is a linear restoring force. The simplest example is a mass on a spring (a spring–
mass system). If there is no friction or other dissipation, or when the damping
force is proportional to the velocity and oppositely directed (viscous friction), the
differential equation describing the motion of the system is linear because the dis-
placement and its time derivatives are to the first power. Such a physical system is
called a linear oscillator.

1.1.2 Differential Equation of a Linear Torsion Oscillator

The linear oscillator simulated in the suggested computer program is a balanced
flywheel (rotor) whose center of mass lies on the axis of rotation. Such a flywheel
may consist, for example, of a rigid rod with two equal masses, as shown in the
left-hand panel of Figure 1.1.
The rod of the flywheel can rotate about an axis that passes through its center.
A spiral spring with one end fixed and the other attached to the flywheel flexes
when the flywheel is turned, and creates a restoring torque N which tends to return
the flywheel to the equilibrium position. In our model we assume that the spring
obeys Hooke’s law: The torque N (ϕ) is proportional to the angular displacement
ϕ from the equilibrium position:

N = −Dϕ. (1.1)

Here D is a constant of proportionality called the torsion spring constant. Mea-

sured in units of torque per radian, its value depends on the strength of the spring.
The angular displacement ϕ(t) of the flywheel from its equilibrium position ϕ = 0

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

1.1. SUMMARY OF THE THEORY 5

(vertical in Figure 1.1) is measured by a needle attached to the flywheel, and a

fixed dial.
The right-hand panel of Figure 1.1 shows an LCR-oscillatory circuit that can
be regarded as an electromagnetic analogy of the mechanical device. Both sys-
tems are described by identical differential equations and thus are dynamically
isomorphic. However, the mechanical system has a definite didactic advantage for
exploration of oscillations because it allows us to observe a direct visualization of
motion.
The law of rotation of a flywheel whose moment of inertia about its axis of
symmetry is J, and which is acted upon by a torque N in the absence of friction,
gives the following differential equation:

J ϕ̈ = −Dϕ or ϕ̈ + ω02 ϕ = 0, (1.2)

where we have introduced the notation ω02 = D/J for the coefficient of ϕ. The
general solution of Eq. (1.2) can be written as a superposition of two harmonic
oscillations with the same frequency ω0 , one cos-like and the other sin-like, with
arbitrary amplitudes C and S, respectively:

ϕ(t) = C cos ω0 t + S sin ω0 t. (1.3)

For a particular solution, the values of C and S depend on the initial conditions.
Another equivalent form of the general solution (with two other arbitrary constants
A0 and δ0 ) represents a simple harmonic oscillation:

ϕ(t) = A0 cos(ω0 t + δ0 ). (1.4)

The two arbitrary constants A0 and δ0 in the general solution (1.4) have the physi-
cal sense of the amplitude and the initial phase, respectively. Their values, as well
as the values of C and S in Eq. (1.3), depend on the initial conditions (that is, on
the angular displacement ϕ(0) and the angular velocity ϕ̇(0) at the initial moment
t = 0). In other words, these characteristics of the motion depend on the way in
which such free oscillations are excited.
The subsequent natural oscillations occur with the frequency ω0 , the squared
value of which is proportional to the spring constant D and inversely proportional
to the moment of inertia J of the flywheel. The frequency ω0 and the correspond-
ing period T0 = 2π/ω0 , unlike the amplitude and initial phase, do not depend
on the initial conditions – they are entirely determined by the properties of the
system, i.e., by the values of the physical parameters D and J. Free oscillations
of the system always occur with the same natural frequency ω0 at arbitrary initial
conditions, that is, independently of the mode of excitation.
When the flywheel is also acted upon by a force of viscous friction, which is
proportional to and oppositely directed to the angular velocity ϕ̇, the differential
equation of motion has the form:

ϕ̈ + 2γ ϕ̇ + ω02 ϕ = 0, (1.5)

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

6 CHAPTER 1. FREE OSCILLATIONS OF A LINEAR OSCILLATOR

where the decay constant (or damping constant) γ characterizes the strength of
viscous friction in the system.
Equations (1.2) and (1.5) are linear differential equations because the depen-
dent variable, ϕ(t), and its time derivatives occur only to the first power. We shall
therefore refer to this system as a linear oscillator. These equations are also ho-
mogeneous, because ϕ(t), or its derivatives, appear to the same power in every
term of the equation. The homogeneity of Eqs. (1.2) and (1.5) implies that an
external driving force, independent of ϕ(t) or its time derivatives, is not present.
We call the unforced oscillations described by Eqs. (1.2) and (1.5) free or natural
oscillations. They occur when there is no external driving force.
As for any homogeneous equation, Eqs. (1.2) and (1.5) have the trivial solu-
tion, ϕ(t) = 0 and ϕ̇(t) = 0. This solution describes a system that is always at
rest in its equilibrium position. Since the solutions to this second-order differential
equation are completely determined by the values of ϕ and ϕ̇ at some particular
moment, it is clear that if the initial values of ϕ and ϕ̇ are zero, they must remain
zero forever. Hence, if non-zero values ϕ and ϕ̇ are to be found, ϕ(0) and ϕ̇(0)
cannot both be zero. Oscillations of the system are produced only when there is
some initial excitation of the system.
When friction is sufficiently weak, so that γ < ω0 , the general solution of
Eq. (1.5) can be written in the form:

ϕ(t) = A0 exp(−γt) cos(ω1 t + δ0 ). (1.6)

This solution describes damped oscillations whose amplitude A0 exp(−γt) de-

creases exponentially with time. The amplitude constant A0 (the initial ampli-
tude) and the initial phase δ0 depend on the initial conditions. The frequency ω1
appearing in the cosine term in Eq. (1.6) is given by
q p
ω1 = ω02 − γ 2 = ω0 1 − (γ/ω0 )2 . (1.7)

In the case of relatively weak damping, when constant γ is small compared to the
natural frequency ω0 (γ/ω0 ≪ 1), frequency ω1 is very close to ω0 :

ω1 ≈ ω0 − γ 2 /(2ω0 ). (1.8)

The fractional difference (ω0 − ω1 )/ω0 of these frequencies is proportional to the

square of the small parameter γ/ω0 .
Graphs of the deflection angle and of the angular velocity for oscillations
damped by the force of viscous friction are shown in Figure 1.2.

1.1.3 The Time of Damping and the Quality Factor Q

In the cases of weak and moderate damping, in Eq. (1.6) the time-dependent fac-
tor A0 exp(−γt) can be treated as a slowly decreasing amplitude of diminishing
oscillations. After an interval τ = 1/γ, the amplitude is e ≈ 2.72 times smaller
than its initial value. The time τ is called the decay time or the time of damping.

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

1.1. SUMMARY OF THE THEORY 7

o
120 .
ϕ(t ) ϕ(t )

o
-120
0 1 2 3 4 5 6 7 8T 0

Figure 1.2: Plots of the deflection angle ϕ(t) and of the angular velocity ϕ̇(t) for
damped natural oscillations.

When γ ≪ ω0 , or τ ≫ T0 = 2π/ω0 (condition of weak damping), the

oscillator executes a large number N of oscillations during the decay time τ :
N = τ /T0 ≫ 1. Consecutive maximal deflections from the equilibrium position
diminish in a geometric progression. Letting ϕn be the maximal angular displace-
ment of the n-th oscillation, we have

ϕn+1 /ϕn ≈ exp(−γT0 ) ≈ 1 − γT0 . (1.9)

That is, the ratio of successive terms in this infinite geometric progression is less
than unity by the small value γT0 = T0 /τ ≪ 1.
The strength of viscous friction in the system is usually characterized either by
the damping constant γ, which, as can be seen from Eq. (1.5), has the dimension of
frequency, or by a dimensionless quantity Q, called the quality factor. The quality
factor is defined by:
ω0 τ
Q= =π . (1.10)
2γ T0
The number of cycles during which the amplitude of oscillations decreases by a
factor e ≈ 2.72 is given by Q/π, and the number of cycles N1/2 during which the
amplitude is halved is given by:

N1/2 = (ln 2/π)Q = 0.22 Q = Q/4.53. (1.11)

It follows from Eq. (1.11) that N1/2 = 4 if Q = 18.13, that is, for this value of the
quality factor the amplitude of natural oscillations halves under the viscous fric-
tion after each 4 cycles. The graphs in Figure 1.2 obtained in computer simulation
of natural oscillations at Q = 18.13 show clearly that indeed the amplitude halves
after the first four cycles, and halves again after the next four cycles.
When γ ≥ ω0 (condition of strong damping), a disturbed oscillator returns to
the equilibrium position without oscillating. In this motion, the oscillator either
approaches the equilibrium position asymptotically from one side or overshoots

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

8 CHAPTER 1. FREE OSCILLATIONS OF A LINEAR OSCILLATOR

o
105
ϕ(t )

.
ϕ(t )
0

o
-105
0 1 2T0

Figure 1.3: Plots of the deflection angle ϕ(t) and of the angular velocity ϕ̇(t) for
the case of critical damping (γ = ω0 , ϕ(0) = 0 and ϕ̇(0) = 5ω0 ).

the equilibrium position only once and then asymptotically reapproaches it from
the other side. This latter case occurs only when the initial angular velocity of the
oscillator is directed toward the equilibrium position, and its magnitude is large
enough.
When γ = ω0 , the system is said to be critically damped. The general solution
to the differential equation of motion, Eq. (1.5), for the critically damped system
takes the form:
ϕ(t) = (C1 t + C2 ) exp(−γt), (1.12)
where C1 and C2 are constants defined by the initial conditions. For example, if
the system is given an initial velocity Ω0 at the equilibrium position, that is, if
ϕ(0) = 0 and ϕ̇(0) = Ω0 , then C1 = Ω0 , C2 = 0, and the motion of the system
is described by the function:

ϕ(t) = Ω0 t exp(−γt). (1.13)

The graphs of the deflection angle ϕ(t) and of the angular velocity ϕ̇(t) for this
case of excitation at critical damping are shown in Figure 1.3.
An interesting feature of the critically damped system is that, after an initial
disturbance, it returns to rest in the equilibrium position usually sooner than it does
in any other case (i.e., than it does for any other value of the damping constant γ
for a given value of ω0 ). It is seen from Eq. (1.10) that the value of the quality
factor that corresponds to critical damping (γ = ω0 ) is Q = 0.5.
Non-oscillatory motion at strong friction, when γ > ω0 , can be represented as
a superposition of two exponential functions, which have different time constants
τ1 = −1/α1 and τ2 = −1/α2 :
q
ϕ(t) = C1 eα1 t + C2 eα2 t , where α1,2 = −γ ± γ 2 − ω02 . (1.14)

Values of C1 and C2 are determined by the initial conditions.

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

1.1. SUMMARY OF THE THEORY 9

In measuring instruments such as moving-coil galvanometers, damping is in-

troduced intentionally in order to overcome the difficulty of taking a reading from
an oscillating needle. A measuring instrument is said to be critically damped if
the needle just fails to oscillate and comes to rest in the shortest possible time.
If the instrument is underdamped (Q > 0.5), the needle oscillates repeatedly
before coming to rest. If the instrument is overdamped (Q < 0.5), the needle does
not oscillate, though generally it takes longer to come to rest than it does when the
instrument is critically damped.

1.1.4 The Phase Diagram of a Linear Oscillator

The mechanical state of a torsion pendulum at any instant is determined by the
two quantities: the angular displacement ϕ and the angular velocity ϕ̇ (or instead
of ϕ̇ by the angular momentum J ϕ̇). The evolution of the mechanical state of the
system during its entire motion can be graphically demonstrated very clearly in a
phase diagram. This is a graph that plots the angular velocity ϕ̇ (or the angular
momentum J ϕ̇) versus the angular displacement ϕ.
The mechanical state of the system at any instant is represented by a point,
called the representative point, in the phase plane. If the motion of the physical
system is periodic, the representative point, moving clockwise, generates a closed
path in the phase plane. The phase trajectory of periodic motion is closed, because
the system returns to the same mechanical state after a full cycle.
The phase diagram for harmonic oscillations (e.g., for oscillations in a linear
system without friction) is an ellipse (or a circle at the appropriate choice of the
scales). The points of intersection of the phase curve with the ϕ-axis (the extreme,
or turning points) correspond to maximal deflections of the oscillator from the
equilibrium position. At these points, the sign of the angular velocity ϕ̇ changes,
and the tangent to the phase curve is perpendicular to the abscissa axis.
As noted above, the period of harmonic oscillations is determined entirely by
the parameters of the physical system, specifically by the values of the spring con-
stant D and the moment of inertia J. Unlike the amplitude and the initial phase,
the period does not depend on the initial conditions, that is, on the way oscilla-
tions are excited. This property of the harmonic oscillator is called isochronism.
Because of this property, the representative point generates ellipses of different
sizes (which correspond to various amplitudes of oscillations in the same system)
during the same time T0 .
In the presence of relatively weak viscous friction (γ < ω0 ) the extreme dis-
placements, as well as the extreme values of the angular velocity, gradually dimin-
ish with each subsequent cycle of oscillation. Consequently the phase trajectory
for free oscillations is transformed from a closed curve (an ellipse or a circle) into
a shrinking spiral that winds around a focal point located at the origin of the phase
plane. The lower panel of Figure 1.4 shows the phase diagram of damped oscilla-
tions. In the upper panel the parabolic potential well Epot (ϕ) of the linear oscilla-
tor is shown together with the graphs Etot (ϕ) of the total energy and Ekin (ϕ) of
the kinetic energy. During the oscillations, the point representing the total energy

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

10 CHAPTER 1. FREE OSCILLATIONS OF A LINEAR OSCILLATOR

E
E tot

E pot

E kin

.
ϕ ϕ

Figure 1.4: The parabolic potential well (upper panel) and the phase diagram
(lower panel) of damped oscillations.

Etot travels in this well from one slope to the other descending gradually to the
bottom of the well.
The family of phase trajectories, which correspond to the same values of the
parameters of the system but to different initial conditions, forms a phase portrait
of the system. This phase portrait gives a clear graphic representation of all possi-
ble motions of the system. The phase portrait of a conservative linear oscillator is
formed by a set of similar ellipses with a common center at the origin of the phase
plane. The center represents a state of rest in the equilibrium position.
When friction is relatively weak (γ < ω0 ), this center becomes an attractor
of the phase trajectories called the focal point. That is, all phase trajectories of
damped oscillations spiral in toward the origin, forming an infinite number of
gradually shrinking loops, as in Figure 1.5a.
When friction is relatively strong (γ > ω0 ), the attractor of the phase trajec-
tories becomes a node: all phase trajectories of non-oscillatory motion approach
this node directly, without spiraling.
The phase portrait of an overdamped system is shown in Figure 1.5b. The
phase curves asymptotically approach theporigin, where they have a common
tangent ϕ̇ = α1 ϕ, where α1 = −γ + γ 2 − ω02 . At specific initial condi-
tions, when ϕ̇(0) = α1 ϕ(0), the representative point moves towards the node
directly alongpthis tangent. The other rectilinear phase trajectory ϕ̇ = α2 ϕ
(α2 = −γ − γ 2 − ω02 ) occurs at initial conditions of the type ϕ̇(0) = α2 ϕ(0).

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

1.1. SUMMARY OF THE THEORY 11

. .
ϕ ϕ

ϕ ϕ

a b

Figure 1.5: Phase portrait of (a) an underdamped (γ = 0.1ω0 ) and of (b) an over-
damped (γ = 1.1ω0 ) linear oscillator with viscous friction.

In general, only one phase trajectory passes through a given point of the phase
plane. Indeed, if we consider this arbitrary point as an initial state of the system,
the further motion of the system is defined uniquely. This motion is represented
by a single phase trajectory, passing through the point. However, there may be
exceptions in that either no phase trajectory passes through a phase point, or there
are several trajectories at once. Phase points of this kind are called singular.
For a linear oscillator, there is only one singular point: the origin of the phase
plane. It corresponds to the state of rest in the equilibrium position, where both ϕ
and ϕ̇ are zero. When γ = 0, this point is a center, and no phase trajectory passes
through it. When 0 < γ < ω0 , or when γ > ω0 , it is respectively a focus or a
node, to which all phase trajectories are attracted. For nonlinear systems such as
a planar rigid pendulum, there exists another kind of singular points, namely the
saddle points.

1.1.5 Energy Transformations

The total energy E of the torsion spring pendulum is the sum of the elastic poten-
tial energy Epot of the strained spring and the kinetic energy Ekin of the rotating
flywheel:
1 1
E = Epot + Ekin = Dϕ2 + J ϕ̇2 . (1.15)
2 2
Free oscillations in the absence of friction are characterized by the exchange
of energy between its potential and kinetic forms. At the points of maximum dis-
placement from the equilibrium position, the kinetic energy is zero, and the total
energy of the oscillator is the potential energy of the strained spring. A quarter of
a period later, the oscillator passes through the equilibrium point, where the po-
tential energy is zero and the total energy of the oscillator is the kinetic energy of

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

12 CHAPTER 1. FREE OSCILLATIONS OF A LINEAR OSCILLATOR

E kin
Epot
Etotal

0 1 2 3 4 T0

Figure 1.6: Energy transformations during damped natural oscillations.

the flywheel. During the next quarter period, the reverse exchange of energy oc-
curs: the kinetic energy is transformed into potential energy. Such transformations
happen twice during one period. That is, oscillations of the two kinds of energy,
180◦ out of phase with one another, are executed between zero and a maximal
value 12 DA20 with the frequency 2ω0 , i.e., with double the natural frequency ω0 of
the system.
The exchanges between potential and kinetic energy described above are char-
acteristic of a conservative system, in which such transformations are reversible.
The sum of the kinetic and potential energy, i.e., the total mechanical energy E of
the oscillator, is the same at every instant and is equal to the maximum values of
both kinetic energy and potential energy:
1 1
E= DA20 = Jω02 A20 . (1.16)
2 2
The total energy of the system is proportional to the square of the amplitude
A0 . The values of the two forms of energy, averaged over a period, are each equal
to one half of the total energy:
1 1 1
hEpot i = hEkin i = E = DA20 = Jω02 A20 . (1.17)
2 4 4

In the presence of friction, the exchanges between kinetic and potential en-
ergy are partially irreversible because of the dissipation of mechanical energy.
This dissipation occurs nonuniformly during a complete cycle: Its instantaneous
rate, −dE/dt, is zero when the flywheel, in a given cycle, is at the extremes of its
motion and its angular velocity, ϕ̇, is zero. This is clearly seen in Figure 1.6, which
shows the transformations of energy occurring in damped oscillations. The rate of
dissipation is greatest when the flywheel moves in the vicinity of the equilibrium
position, where its angular velocity is maximal. Indeed, the rate of energy dissi-
pation −dE/dt = −Nfr ϕ̇ caused by viscous friction is proportional to the square
of the angular velocity and hence to the momentary value of the kinetic energy

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

1.2. REVIEW OF THE PRINCIPAL FORMULAS 13

J ϕ̇2 /2 of the oscillator. The statement, frequently encountered in textbooks, that

the energy of a damped oscillator decays exponentially, applies only to values of
the total energy averaged over a period.
The graphs in Figure 1.6 correspond to the same value of the quality factor
(Q = 18.13) as the graphs of ϕ(t) and ϕ̇(t) in Figure 1.2. We see clearly that
for this quality factor the amplitude of oscillations halves after each four cycles
of natural oscillations (Figure 1.2), while the total energy halves after each two
cycles (Figure 1.6).

1.1.6 The Computer Simulation of a Linear Oscillator

The computer simulation of the torsion spring pendulum in the suggested simu-
lation program of the package PHYSICS OF OSCILLATIONS is based on nu-
merical integration of the differential equation, Eq. (1.5). Although this equation
can be integrated analytically, the analytic solution is not used in those parts of
the computer program that demonstrate the evolution of the system with time.
The analytic solution is used only to determine the maximal values of the angu-
lar displacement and of the angular velocity in order to establish suitable scales
for the corresponding plots. The agreement between the results of the numerical
integration in the computer simulation and the analytic predictions can serve as a
confirmation of the quality of the algorithm used (the fourth order Runge – Kutta
method). This verification gives us confidence in the reliability of the the com-
puter simulations of nonlinear systems in subsequent programs of this software
package since the simulations are based on the same numerical method.
In this numerical simulation of the linear oscillator we adopt a unit of time
that is appropriate for the system under consideration, namely the period T0 =
2π/ω0 of free oscillations in the absence of friction. Thus the simulated oscillator
may be characterized by only one physical parameter: either by the dimensionless
ratio of the damping constant to the natural frequency γ/ω0 , or by the equivalent
dimensionless quantity – the quality factor Q = ω0 /(2γ), inversely proportional
to γ/ω0 .
The angular displacement, ϕ, is expressed in radians in the program, though
for convenience of observation, the dial on the screen and the plots involving the
angle of deflection are graduated in degrees. The angular velocity, ϕ̇, is measured
in units of the natural frequency, ω0 . When the initial conditions are set in a simu-
lation computer experiment, the initial angular velocity also must be expressed in
units of ω0 .

1.2 Review of the Principal Formulas

The differential equation of a free (unforced) linear torsion oscillator:

ϕ̈ + 2γ ϕ̇ + ω02 ϕ = 0. (1.18)

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

14 CHAPTER 1. FREE OSCILLATIONS OF A LINEAR OSCILLATOR

The frequency and the period of free oscillations without friction (at γ ≪ ω0 ):
r
D 2π
ω0 = , T0 = . (1.19)
J ω0
An oscillatory solution (valid at γ < ω0 ):
ϕ(t) = A0 e−γt cos(ω1 t + δ0 ), (1.20)
where the constants A0 and δ0 are determined by the initial conditions ϕ(0), ϕ̇(0).
The frequency ω1 of damped oscillations
q
ω1 = ω02 − γ 2 . (1.21)
An equivalent form of the general solution:
ϕ(t) = e−γt (C cos ω1 t + S sin ω1 t), (1.22)
where the constants C and S are determined by the initial conditions. They are
related to A0 and δ0 :
p
A0 = C 2 + S 2 , tan δ0 = −S/C. (1.23)
In the case of weak damping (γ ≪ ω0 )
ω1 ≈ ω0 − γ 2 /(2ω0 ). (1.24)
The decay time (during which the amplitude is reduced by the factor e ≈ 2.72):
τ = 1/γ. (1.25)
A non-oscillatory motion at γ = ω0 :
ϕ(t) = (C1 t + C2 )e−γt . (1.26)
The quality factor Q of an oscillator:
τ ω0
Q=π = . (1.27)
T0 2γ
The number of oscillations, during which the amplitude is halved:
ln 2 Q
N1/2 = Q = 0.22 Q = . (1.28)
π 4.53
The total mechanical energy of the oscillator consists of elastic potential energy
of the strained spring and kinetic energy of the flywheel:
1 1
E = Epot + Ekin = Dϕ2 + J ϕ̇2 . (1.29)
2 2
The values of the potential energy and kinetic energy of the oscillator, averaged
over a cycle, equal one another, each of them constituting one half the total energy:
1 1 1
hEpot i = hEkin i = E = DA20 = Jω02 A20 . (1.30)
2 4 4

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

1.3. QUESTIONS, PROBLEMS, SUGGESTIONS 15

1.3 Questions, Problems, Suggestions

1.3.1 Free Undamped Oscillations
1.3.1.1 The Initial Conditions and the Shape of the Plots. In the absence
of friction a linear oscillator executes simple harmonic motion, which is charac-
terized by purely sinusoidal time dependence of the angular displacement and of
the angular velocity.
(a) What initial conditions give rise to oscillations of cosine time dependence,
and of sine time dependence? Suppose that you want to get oscillations with the
angular amplitude of 90◦ . What initial angular displacement ϕ(0) = ϕ0 at zero
initial angular velocity ϕ̇(0) = 0 ensures the desired amplitude?
(b) What initial angular velocity ϕ̇(0) = Ω ought you to impart to the oscil-
lator, at rest in the equilibrium position, in order to obtain the same amplitude of
90◦ ? Remember that the initial angular velocity Ω must be expressed for input in
units of the natural frequency ω0 . Verify your answer with a computer experiment,
using the appropriate initial conditions.
1.3.1.2 Maximal Deflection and Conservation of Energy. Imagine exciting
an oscillator initially at rest in the equilibrium position by a push that produces an
initial angular velocity Ω = 2ω0 .
(a) Calculate the angle ϕmax of maximal deflection using the law of the con-
servation of energy.
(b) Verify your result experimentally. Note that the simulation program per-
forms the numerical integration of the differential equation independently of con-
servation laws, such as the conservation of energy. That is, these laws are not used
in the program.
1.3.1.3 The Phase Trajectory and the Initial Conditions. Compare the mo-
tion of the representative point along the phase trajectory of a conservative oscil-
lator with the time-dependent plots of the angle of deflection and of the angular
velocity.
(a) How is the phase trajectory changed if you change the initial conditions?
(b) Does the direction of the motion of the representative point along the phase
trajectory depend on the initial conditions?
(c) Is it possible that phase trajectories for different initial conditions coincide?
If so, formulate the requirements for the coincidence.
1.3.1.4 Elliptical and Circular Shape of the Phase Trajectory.
(a) Prove analytically that the phase trajectory of a conservative linear oscil-
lator is an ellipse with its center at the origin of the phase plane. Use the general
solution of Eq. (1.2), expressed by Eq. (1.4). What are the semiaxes of the ellipse?
(b) Show that the elliptical shape of the phase diagram of a conservative linear
oscillator follows immediately from the law of the conservation of the energy.
(c) What scale on the axis of the ordinate (the angular velocity axis) of the
phase plane produces a circular phase trajectory?

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

16 CHAPTER 1. FREE OSCILLATIONS OF A LINEAR OSCILLATOR

(d) Does the time interval during which the representative point passes along
one loop of the phase trajectory depend on the initial conditions?
1.3.1.5 The Phase Diagram and Energy Transformations. Compare the
phase trajectory with the plot of potential energy versus the angle of deflection.
The positioning of plots on the display screen (if you open the window “Phase
diagram”) is convenient for such comparison. Pay special attention to the positions
of the extreme points (turnarounds) on the phase trajectory and in the parabolic
potential well. For the initial conditions ϕ(0) = ϕ0 , ϕ̇(0) = Ω, what are the
values of the potential energy and the kinetic energy at the extreme points and at
the equilibrium position?
What are the extreme deflection ϕmax and the maximal angular velocity ωmax
of the flywheel?
1.3.1.6 The Shape and the Frequency of Energy Oscillations. Consider the
plots of the time dependence of kinetic energy and potential energy.
(a) What can you say about their maximal and average values? Compare these
plots with the plots of the angular displacement and the angular velocity.
(b) At what frequency do the oscillations of each kind of energy occur? What
are the limits (the extreme values) and the mean (averaged over a period) values
of each kind of energy in these oscillations?
1.3.1.7 The Phase Trajectories with the Same Energy. Consider the oscil-
lations of a conservative oscillator at different initial conditions but with the same
total energy. What differences do you observe in the plots and the phase trajecto-
ries in these cases?

1.3.2 Damped Free Oscillations

1.3.2.1 The Sequence of Maximal Deflections. Under the action of a weak
force of viscous friction, the sequence of maximal deflections of a free, damped
linear oscillator forms a decreasing geometric progression: Each consecutive max-
imal deflection is smaller than the preceding one by the same factor, exp(−γT0 ) ≈
1 − γT0 [see Eq. (1.10)].
(a) Calculate the value of the quality factor Q at which the amplitude halves
during every two complete oscillations.
(b) Input this value in a computer experiment and verify the theoretically pre-
dicted constant ratio of successive maximal deflections. Note that this ratio does
not depend on the initial conditions.
(c) Evaluate the increment of the period of oscillations at this value of the
quality factor with respect to the period T0 in the absence of friction (in percent).
Can you detect the increment in the simulation experiment? The marks on the time
axis correspond to integer numbers of periods T0 = 2π/ω0 without friction.
1.3.2.2∗ Maximal Deflection after an Initial Push. Imagine that we excite
oscillations with an initial push which imparts an initial angular velocity of 2ω0
to the flywheel in its equilibrium position.

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

1.3. QUESTIONS, PROBLEMS, SUGGESTIONS 17

(a) Calculate the first maximal deflection of the flywheel after the excitation
for the quality factor Q = 5.
(b) What will be the value of the subsequent extreme deflection that occurs in
the direction opposite to the first? Verify your answers using the simulation.
1.3.2.3∗∗ Complex Initial Conditions.
(a) Let the initial deflection of the torsion pendulum be 155 degrees, and the
initial angular velocity be 2ω0 . The quality factor Q = 5. Calculate the maximal
deflection of the flywheel.
(b) With the same initial deflection (155 degrees) and the same quality fac-
tor Q = 5 as in the preceding item (a), calculate the maximal deflection of the
flywheel, if the initial angular velocity equals −2ω0 .
(c) Let the initial deflection of the torsion pendulum be −155 degrees. What
initial angular velocity would ensure the maximal deflection of 155 degrees (to
the opposite side), if the quality factor Q = 20?
1.3.2.4∗ The Phase Trajectory of Damped Oscillations. The phase trajec-
tory of damped free oscillations for Q > 0.5 is a spiral that makes an infinite
number of gradually shrinking loops around the focus located at the origin of the
phase plane. This focus corresponds to the state of rest in the equilibrium position,
and the phase trajectory approaches it asymptotically.
(a) How does the size of these loops change while the curve approaches the
focus?
(b) Does the time interval during which the representative point makes one
revolution along the spiral change as the loops of the curve shrink?
1.3.2.5∗ The Dissipation of Energy. Compare the transformation of potential
energy into kinetic energy (and vice versa) for free undamped oscillations in the
absence of friction with that for free damped oscillations in the presence of viscous
friction.
(a) Show, using a simulation experiment, that if Q = 18.1, the amplitude is
halved during four complete oscillations and the total energy is halved during two
complete oscillations.
(b) Why is the dissipation of mechanical energy nonuniform during one cycle
of oscillations? At what instants during a cycle is the time-rate of energy dissipa-
tion greatest and at what instants is it smallest?

1.3.3 Non-Oscillatory Motion of the System

When viscous friction is strong (Q ≤ 0.5), a disturbed system returns to the
equilibrium position without oscillating. In the computer simulation, the needle
asymptotically approaches the zero point from one side.
1.3.3.1∗ Non-Oscillatory Motion at Critical Damping. Consider the case of
critical damping, γ = ω0 .
(a) Why is critical damping preferable in measuring instruments using a needle
as an indicator? How might your answer apply to the suspension system in an
automobile?

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

18 CHAPTER 1. FREE OSCILLATIONS OF A LINEAR OSCILLATOR

Show that the value of Q in the case of critical damping is 0.5.

(b) Calculate the maximal angle of deflection if the system, with Q = 0.5,
receives an initial velocity Ω = 5ω0 in the equilibrium position.
(c) In what lapse of time does the needle move towards this extreme point?
Verify your answers by simulating the experiment on the computer. Note that
the needle approaches the equilibrium position from one side – it does not cross
the zero point of the dial.
1.3.3.2 Critical Damping.
(a) Prove that the value Q = 0.5 (γ = ω0 ) is really critical. Do so by showing
that at slightly greater values of Q, the needle of a perturbed oscillator executes
heavily damped oscillations, slowly moving to and fro across the zero point of the
dial.
(b) For a critically damped system, express the constants C1 and C2 in the gen-
eral solution ϕ(t) = (C1 t + C2 ) exp(−γt) of the differential equation, Eq. (1.5),
in terms of the initial displacement ϕ(0) = ϕ0 and the initial angular velocity
ϕ̇(0) = Ω0 .
(c) Is it possible for a critically damped system to move after an initial distur-
bance according to a pure exponential law? If so, what initial conditions give rise
to such motion? What is the phase trajectory of this motion? Prove your answers
experimentally.
(d) At what initial conditions will the flywheel of a disturbed critically damped
system cross the equilibrium position? For a given initial displacement ϕ0 , what
initial angular velocity Ω should you impart to the flywheel of the critically
damped oscillator in order to have it cross the equilibrium position after a lapse of
time t = 3T0 , where T0 = 2π/ω0 is the natural period (the period of oscillations
in the absence of friction)?
1.3.3.3∗ Motion of an Overdamped System.
(a) For arbitrary initial conditions (ϕ(0) = ϕ0 , ϕ̇(0) = Ω0 ), express the values
of C1 and C2 in the general solution (1.14) of the differential equation for an
overdamped system in terms of ϕ0 and Ω0 .
(b) At what initial conditions will the motion of an overdamped system be
described by a monoexponential function of time? What are the phase trajectories
that correspond to such motions?
(c) Explain why, at arbitrary initial conditions, non-oscillatory motion of the
flywheel towards the equilibrium position occurs more slowly and requires more
time than at critical damping. Is it possible for an overdamped system to return
to the equilibrium position faster than for the critically damped system with the
same ω0 ? If so, what conditions of excitation ensure the motion?
(d) What is the principal difference between the phase trajectories correspond-
ing to a non-oscillatory motion and those corresponding to damped oscillations?
(e) Is it possible for an overdamped system (γ > ω0 ) to cross the equilibrium
position after excitation? If so, what initial conditions give rise to such motion? Is
it possible for the oscillator to cross the equilibrium position more than once?

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

Chapter 2

Torsion Spring Oscillator with

Dry Friction

Annotation. Chapter 2 deals with natural oscillations in a simple mechanical sys-

tem with dry (Coulomb) friction. It includes a description of the simulated phys-
ical system and a summary of the relevant theoretical material for students as a
prerequisite for the virtual lab “Torsion Spring Oscillator with Dry Friction” from
the software package Physics of Oscillations (Part I, Simple Systems). Peculiari-
ties of damping of oscillations under dry friction are discussed in detail. Chapter
2 also includes a set of theoretical and experimental problems to be solved by
students, as well as various assignments that the instructor can offer students for
possible individual work on their own.

2.1 Summary of the Theory

2.1.1 General Concepts
Chapter 2 and the relevant simulation program are aimed at investigation of free
oscillations of a torsion spring pendulum damped by dry (Coulomb) friction.
An idealized mathematical model of dry friction described by the so-called z-
characteristic is assumed. In this model, the force of kinetic friction does not de-
pend on speed and equals the limiting force of static friction. The physical system
modeled here allows us to understand the origin of accidental errors in reading
some measuring instruments.

2.1.2 The Physical System

The rotating component of the torsion spring oscillator is a balanced flywheel (a
rigid rod with two equal weights) whose center of mass lies on the axis of rotation.
Figure 2.1 shows a schematic image of the simulated system. A spiral spring, one

19

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

20 CHAPTER 2. SPRING OSCILLATOR WITH DRY FRICTION

- jm +j
m

j (t)

Figure 2.1: The image of the torsion spring oscillator with dry friction.

end fixed and the other end attached to the flywheel, flexes when the flywheel
is turned. The system is in equilibrium when the rod of the flywheel is vertical
(ϕ = 0). The spring provides a restoring torque whose magnitude is proportional
to the angular displacement ϕ of the flywheel from the equilibrium position.
The dynamical behavior of such a system under the influence of viscous fric-
tion (for which the torque is proportional to the angular velocity) is discussed in
Chapter 1, “Free oscillations of a linear torsion pendulum.” When friction is vis-
cous, free oscillations of a spring pendulum are described by a linear differential
equation. The amplitude of such oscillations decreases exponentially with time.
That is, the consecutive maximal deflections of the oscillator from its equilibrium
position are in a diminishing geometric progression because their ratio is constant.
In principle such oscillations continue indefinitely, their amplitude asymptotically
approaching zero. However, it is convenient to characterize the duration of expo-
nential damping by a decay time τ . This conventional time of damping τ is the
lapse of time during which the amplitude of free oscillations decreases by a factor
of e ≈ 2.72.
The exponential character of damping caused by viscous friction follows from
the proportionality of friction to velocity. Some other relationship between friction
and velocity produces damping with different characteristics.
The case of dry or Coulomb friction has important practical applications. In
this case, as long as the system is moving, the magnitude of dry friction is very
nearly constant and its direction is opposite that of the velocity. An idealized sim-
plified characteristic of dry friction (called the z-characteristic) is shown in Fig-
ure 2.2. The graph shows dependence of the frictional torque N on the angular
velocity ϕ̇ of rotation. Here the magnitude of friction is constant, but its direc-
tion changes each time the direction of the velocity changes. When the system is
at rest, the torque of static dry friction takes on any value from −Nmax to Nmax .
The actual value depends on the friction needed to balance the other forces exerted

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

2.1. SUMMARY OF THE THEORY 21

.
N (j)

Nmax

.
0 j
- Nmax

Figure 2.2: An idealized characteristic of dry friction (z-characteristic).

on the system. The magnitude of the torque of kinetic dry friction is assumed in
this model to be equal to the limiting torque Nmax of static friction.
In real physical systems dry friction is characterized by a more complicated
dependence on velocity. The limiting force of static friction is usually greater than
the force of kinetic friction. When the speed of a system increases from zero, ki-
netic friction at first decreases, reaches a minimum at some speed, and then grad-
ually increases with a further increase in speed. These peculiarities are ignored in
the idealized z-characteristic of dry friction. Nevertheless, this idealization helps
us to understand many essential properties of oscillatory processes in real physical
systems.
Because the magnitude of the torque of static friction can assume any value up
to Nmax , there is a range of values of displacement called the stagnation interval
or dead zone in which static friction can balance the restoring elastic force of the
strained spring. The stagnation interval extends equally to either side of the point
at which the spring is unstrained. The stronger the dry friction in the system, the
more extended the stagnation interval. The boundaries ±ϕm of the interval are
determined by the limiting torque Nmax of static friction: Nmax = Dϕm . If the
velocity becomes zero at some point of the dead zone, the system remains at rest
there. The boundaries ±ϕm of the dead zone are indicated in Figure 2.1. At any
point within the stagnation interval the system can be at rest in a state of neutral
equilibrium, in contrast to a single position of stable equilibrium provided by the
spring in the case of viscous friction.
An important feature of oscillations damped by dry friction is that motion
ceases after a finite number of cycles. As the system oscillates, the sign of its
velocity changes periodically, and each subsequent change occurs at a smaller
displacement from the mid-point of the stagnation interval. Eventually the turning
point of the motion occurs within the stagnation interval, in which static friction
can balance the restoring force of the spring, and so the motion abruptly stops.
The exact position in the stagnation interval at which this event occurs depends on
the initial conditions, which may vary from one situation to the next.
These characteristics are typical of various mechanical systems with dry fric-

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

22 CHAPTER 2. SPRING OSCILLATOR WITH DRY FRICTION

tion. For example, dry friction may be encountered in measuring instruments, such
as a moving-coil galvanometer, in which readings are taken with a needle. In the
galvanometer, a light coil of wire is pivoted between the poles of a magnet. When
a current flows through the coil, it turns against a spiral return spring. If the coil
axis is fixed in unlubricated bearings and hence experiences dry friction, the nee-
dle of the coil may come to rest and show at any point of the stagnation interval on
either side of the dial point, which gives the true value of the measured quantity.
So we can now understand one of the reasons that random errors inevitably occur
in the readings of moving-coil measuring instruments. The larger the dry friction,
the larger the errors of measurement.

2.1.3 The Differential Equation of the Oscillator

The rotating flywheel of the torsion oscillator is simultaneously subjected to the
action of the restoring torque −Dϕ produced by the spring, and of the torque
Nfr of kinetic dry friction. The differential equation describing the motion of the
flywheel, whose moment of inertia is J, is thus

J ϕ̈ = −Dϕ + Nfr . (2.1)

According to the idealized z-characteristic of dry friction, the torque Nfr is

directed oppositely to the angular velocity ϕ̇, and is constant in magnitude while
the flywheel is moving, but may have any value in the interval from −Nmax up to
Nmax while the flywheel is at rest:

−Nmax for ϕ̇ > 0,
Nfr (ϕ̇) = (2.2)
Nmax for ϕ̇ < 0,

or Nfr = −Nmax sign(ϕ̇). Here Nmax is the limiting value of the static frictional
torque. It is convenient to express the value Nmax in terms of the maximal possible
deflection angle ϕm of the flywheel at rest:

Nmax = Dϕm . (2.3)

The angle ϕm corresponds to the boundary of the stagnation interval.

The differential equation, Eq. (2.1), in the general case of an oscillator with
dry friction, is nonlinear because the torque Nfr (ϕ̇) abruptly changes when the
sign of ϕ̇ changes at the extreme points of oscillation, and because when the fly-
wheel moves, the torque is usually not constant. But in the idealized case of the
z-characteristic we may consider the following two linear equations instead of
Eq. (2.1):
J ϕ̈ = −D(ϕ + ϕm ) for ϕ̇ > 0, (2.4)
J ϕ̈ = −D(ϕ − ϕm ) for ϕ̇ < 0. (2.5)
Whenever the sign of the angular velocity ϕ̇ changes, the pertinent equation of
motion also changes. The nonlinear character of the problem reveals itself in al-
ternate transitions from one of the linear equations, Eqs. (2.4)–(2.5), to the other.

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

2.1. SUMMARY OF THE THEORY 23

o
2.84 175 .
ϕ( t) ϕ (t )

0 0

o
-2.84 -175
o o
-175 0 175 0 1 2 3 4 T0
Dead zone 12 degrees, initial deflection 175 degrees, initial angular velocity 0.0

Figure 2.3: Phase trajectory (left) and graphs of ϕ(t) and ϕ̇(t) (right) for oscilla-
tions whose damping occurs under dry friction.

The solution to Eqs. (2.4)–(2.5), which corresponds to a set of given initial

conditions, can be found by using the method of the stage-by-stage integration
of each of the linear equations for the half-cycle during which the direction of
motion is unchanged. These solutions are then joined together at the instants of
transition from one equation to the other in such a way that the displacement at
the end point of one half-cycle becomes the initial displacement at the beginning
of the next half-cycle. This array of solutions continues until the end point of a
half-cycle lies within the dead zone.
If in addition to dry friction the oscillator also experiences viscous friction, we
must add to the equations of motion, Eqs. (2.4)–(2.5), one more term proportional
to the angular velocity ϕ̇:

ϕ̈ = −ω02 (ϕ + ϕm ) − 2γ ϕ̇ for ϕ̇ > 0, (2.6)

ϕ̈ = −ω02 (ϕ − ϕm ) − 2γ ϕ̇ for ϕ̇ < 0. (2.7)

Here ω02 = D/J is the squared natural frequency of the oscillator (the frequency
of oscillations in the absence of friction), and γ is the damping constant. It is
convenient to characterize viscous friction by the dimensionless quality factor,
Q = ω0 /2γ.

2.1.4 Damping Caused by Dry Friction

In order to discover the basic characteristics of oscillations that are damped under
the action of dry friction, we shall first assume that viscous friction is absent, that
is, the damping constant γ = 0.
At the initial instant t = 0, let the flywheel be displaced to the right (clock-
wise) from the equilibrium position so that ϕ(0) > 0. If this displacement ex-
ceeds the boundary of the stagnation interval, i.e., if ϕ(0) > ϕm , the flywheel,
being released without a push, begins moving to the left (ϕ̇ < 0), and its motion is
described by Eq. (2.5). The solution to Eq. (2.5) with the given initial conditions
(ϕ(0) = ϕ0 , ϕ̇(0) = 0) is simple harmonic motion whose frequency is ω0 . The

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

24 CHAPTER 2. SPRING OSCILLATOR WITH DRY FRICTION

midpoint of the motion is ϕm . This point coincides with the right-hand bound-
ary of the stagnation interval. The displacement ϕm of the midpoint from zero is
caused by the constant torque of kinetic friction. This torque is directed to the right
side (clockwise) while the flywheel is moving to the left. The amplitude of this os-
cillation about the midpoint ϕm equals ϕ0 − ϕm . The first segment of the graph
in the right-hand panel of Figure 2.3 (the first half-cycle of the sine curve, whose
midpoint is at a height of ϕm above the abscissa axis) is a plot of this portion of
the motion. Since the amplitude of the first half-cycle is ϕ0 − ϕm , the extreme
left position of the flywheel at the end of the half-cycle is ϕ(0) − 2ϕm . When
the flywheel reaches this position, its velocity is momentarily zero, and it starts to
move to the right. Since its angular velocity ϕ̇ is subsequently positive, we must
now consider Eq. (2.4). The values of ϕ and ϕ̇ at the end of the preceding half-
cycle are taken as the initial conditions for this half-cycle. Thus the subsequent
motion is again a half-cycle of harmonic oscillation with the same frequency ω0
as before but with the midpoint −ϕm displaced to the left, i.e., with the midpoint
at the left-hand boundary of the stagnation interval. This displacement is caused
by the constant torque of kinetic friction, whose direction was reversed when the
direction of motion was reversed. The amplitude of the corresponding segment of
the sine curve is ϕ0 − 3ϕm .
Continuing this analysis half-cycle by half-cycle, we see that the flywheel ex-
ecutes harmonic oscillations about the midpoints alternately located at ϕm and
−ϕm . The frequency of each cycle is the natural frequency ω0 , and so the dura-
tion of each full cycle equals the period T0 = 2π/ω0 of free oscillations in the
absence of friction.
The joining together of these sinusoidal segments, whose midpoints alternate
between the boundaries of the stagnation interval, produces the curve that de-
scribes oscillatory motion damped by dry friction (Figure 2.3). The maximal de-
flection decreases after each full-cycle of these oscillations by a constant value
equal to the doubled width of the stagnation interval (i.e., by the value 4ϕm ). The
oscillation continues until the end point of some next-in-turn segment of the sine
curve occurs within the dead zone (−ϕm , ϕm ).
Thus, in the case of dry friction, consecutive maximal deflections diminish
linearly in a decreasing arithmetic progression, and the motion stops after a final
number of cycles, in contrast to the case of viscous friction, for which the maximal
displacements decrease exponentially in a geometric progression, and for which
the motion continues indefinitely.

2.1.5 The Phase Trajectory

The character of oscillations in the presence of dry friction is given clearly by the
phase trajectory shown in the left-hand panel of Figure 2.3. The system is initially
at rest (ϕ̇(0) = 0) and displaced to the right (ϕ(0) = ϕ0 > ϕm ). This initial
state is represented by the point on the curve that lies to the extreme right on
the horizontal axis, the ϕ-axis. The portion of the phase trajectory lying below the
horizontal axis represents the motion during the first half-cycle, when the flywheel

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

2.1. SUMMARY OF THE THEORY 25

is moving to the left. This curve is the lower half of an ellipse (or of a circle
if the scales have been chosen appropriately) whose center is at the point ϕm
on the horizontal axis. This point corresponds to the right-hand boundary of the
stagnation interval.
The second half-cycle, when the flywheel is moving to the right, is represented
by half an ellipse lying above the ϕ-axis, where angular velocities are positive. The
center of this second semi-ellipse is at the point −ϕm , on the ϕ-axis. The complete
phase trajectory is formed by such increasingly smaller semi-ellipses, alternately
centered at ϕm and −ϕm . The diameters of these consecutive semi-ellipses lie
along the ϕ-axis and decrease each half-cycle by 2ϕm . The phase trajectory ter-
minates on the ϕ-axis at the point at which the curve meets the ϕ-axis inside the
dead zone (the portion of the ϕ-axis lying between ϕm and −ϕm ).
This phase trajectory is to be compared with that of the oscillator acted upon
by viscous friction (see Figure 1.4 of Chapter 1). In the latter case, the curve spirals
around a focal point located at the origin of the phase plane. The curve consists
of an infinite number of turns, which gradually become smaller and approach the
focus asymptotically. In the present case of dry friction, the loops of the phase
curve are equidistant. The phase trajectory consists of a finite number of cycles
and terminates at the point at which it meets the dead zone – the segment of the
ϕ-axis between the points −ϕm and ϕm .
If dry friction in the system is accompanied by a rather weak viscous friction
(γ < ω0 ), the semi-ellipses become distorted and their axes shrink during the
motion. The loops of the phase trajectory are no longer equidistant. Nevertheless
their shrinking does not last indefinitely: the phase trajectory in this case also
terminates after some finite number of turns around the origin when it reaches the
stagnation interval on the ϕ-axis.

2.1.6 Energy Transformations

While the flywheel is rotating in one direction, the torque Nmax of kinetic fric-
tion, independent of the velocity, is constant, and the total energy of the oscillator
decreases linearly with the angular displacement, ϕ, of the flywheel. This linear
dependence of the total energy on ϕ is clearly indicated in the left-hand panel of
Figure 2.4, where the parabolic potential well of the elastic spring is shown. The
representing point whose ordinate gives the total energy E(ϕ) and whose abscissa
gives the angular displacement of the flywheel travels in the course of time be-
tween the slopes of this well, gradually descending to the bottom of the well. The
trajectory of this point consists of rectilinear segments lying between the sides
of the well. These segments are straight because the negative work done by the
force of dry friction is proportional to the angle of rotation, ∆ϕ. The amount of
this work |Nmax ∆ϕ| equals the decrease −∆E of total energy. However, the de-
pendence of total energy on time, E(t), is not linear because the rotation of the
flywheel is nonuniform. The time rate of dissipation of the total energy, −dE/dt,
is proportional to the magnitude of the angular velocity, |ϕ̇(t)|. Thus, the greatest
rate of dissipation of mechanical energy through friction occurs when the mag-

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

26 CHAPTER 2. SPRING OSCILLATOR WITH DRY FRICTION

E tot E tot
E pot
E kin

E pot

- j0 0 j0 0 1 2 3 4 T0

Figure 2.4: Energy transformations in oscillations damped by dry friction.

nitude of the angular velocity, ϕ̇, is greatest, that is, when the flywheel crosses
the boundaries of the dead zone. Near the points of extreme deflection, where the
angular velocity is near zero, the time rate of dissipation of mechanical energy is
smallest. Typical plots of energy transformation are shown in the right-hand panel
of Figure 2.4.
Unlike the case of viscous friction, the oscillator with dry friction may retain
some mechanical energy Ef at the termination of the motion. Such occurs if the
final angular displacement (within the dead zone) is not at the midpoint of the stag-
nation interval. Then the spring remains strained, and its elastic potential energy is
not zero. The remaining energy does not exceed the value Dϕ2m /2 = Nmax ϕm /2.
When the initial excitation is large enough, that is, when the initial energy is
much greater than Dϕ2m /2, the oscillator executes a large number of cycles before
the oscillations cease. In this case it is reasonable to consider the total energy
averaged over the period of an oscillation, hE(t)i. The decrease of hE(t)i during
a large number of cycles depends quadratically on the lapse of time because the
amplitude of oscillation decreases linearly with time and because the averaged
total energy is proportional to the square of the amplitude.
If we let tf be the final moment when oscillations cease, then at the time t the
averaged total energy hE(t)i is proportional to (t − tf )2 . This statement (which
clearly applies only for t < tf ) is exactly true only when the flywheel comes to
rest at the center of the stagnation interval. However, even if such is not the case
and there is a residual potential energy stored in the spring after the motion ceases,
the statement is approximately true.

2.1.7 The Role of Viscous Friction

In real systems, dry friction is always accompanied to some extent by viscous
friction. The damping of oscillations in this case can also be investigated by the
above-described method, namely by the stage-by-stage solving of Eqs. (2.6) and
(2.7) and by using the final mechanical state (the angular displacement and ve-
locity) of every half-cycle as the initial conditions for the next half-cycle. That
is, the solutions are joined by equating their angular displacements and angular

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

2.1. SUMMARY OF THE THEORY 27

velocities (always zero) at the boundaries.

The clearest representation of the mechanical evolution of the system experi-
encing both dry and viscous friction is given by a phase diagram. Unlike the case
of pure dry friction, the path in phase space is no longer a series of diminishing
semi-ellipses (or semicircles) with alternating centers. Instead the phase trajectory
consists of the shrinking alternating halves of spiral loops that are characteristic
of a linear damped oscillator. The focal points of these spirals alternate between
the boundaries of the stagnation interval.
To compare the relative importance of viscous versus dry friction, we con-
sider below the decrease in amplitude caused by each of these effects during one
complete cycle.
It was established above that under the action of dry friction this decrease
equals the constant value of the doubled width of the stagnation interval 4ϕm . On
the other hand, viscous friction decreases the amplitude of the oscillation during
a complete cycle by an amount that is not constant but rather is proportional to
the amplitude. Indeed, for γT0 ≪ 1, i.e., for rather large values of the quality
factor Q, expression for the decrease ∆a during one period T0 in the momentary
amplitude a due to viscous friction can be expanded in a series:
2π πa
∆a = a(1 − e−γT0 ) ≈ aγT0 = aγ = . (2.8)
ω0 Q
Equating ∆a to the doubled width 4ϕm of the stagnation interval, we find the
amplitude ã, which delimits the predominance of one type of friction over the
other:
4ϕm 4
ã = = ϕm Q ≈ ϕm Q. (2.9)
γT0 π
If the actual amplitude is greater than ã, the effect of viscous friction domi-
nates. Conversely, if the actual amplitude is less than ã, the effect of dry friction
dominates.
When the initial excitation of the oscillator is great enough, the amplitude may
exceed the value ã ≈ Qϕm . In this instance, the initial damping of the oscillations
is influenced mainly by viscous friction. This case may be illustrated in the phase
diagram. The decrement in the width of several initial loops of the phase trajectory
(caused by viscous friction) is greater than the separation of the centers of adjoin-
ing half-loops (i.e., the decrement exceeds the width of the stagnation interval). It
is clear that in this case the shrinking of the spiral caused by viscous friction is
more influential in showing the effects of damping than is the alternation of the
centers of half-loops caused by dry friction.
When the value of a falls below that of ã (when a < ã = Qϕm ), the effects
of dry friction dominate. In the phase plane this dominance produces a trajectory
of consecutive half-loops whose centers alternately jump between the ends of the
stagnation interval, −ϕm and ϕm , until the phase trajectory reaches the segment
of the ϕ-axis in the stagnation interval.
When viscous friction is strong, that is, when values of the quality factor Q
are less than the critical value of 0.5 (when γ > ω0 ), and when the initial displace-

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

28 CHAPTER 2. SPRING OSCILLATOR WITH DRY FRICTION

ment of the flywheel ϕ(0) lies beyond the boundaries of the stagnation interval,
|ϕ(0)| > ϕm , the needle of the released flywheel moves without oscillating toward
the point of the dial that corresponds to the nearest boundary of the stagnation in-
terval. At this point the flywheel stops turning.

2.2 Review of the Principal Formulas

The differential equation of motion of an oscillator acted upon by dry friction:
J ϕ̈ = −D(ϕ + ϕm ) for ϕ̇ > 0, (2.10)
J ϕ̈ = −D(ϕ − ϕm ) for ϕ̇ < 0, (2.11)
where ϕm is the angle corresponding to the boundaries of the dead zone. If in
addition, viscous friction is present, a term proportional to the angular velocity is
also present:
ϕ̈ = −ω02 (ϕ + ϕm ) − 2γ ϕ̇ for ϕ̇ > 0, (2.12)
ϕ̈ = −ω02 (ϕ − ϕm ) − 2γ ϕ̇ for ϕ̇ < 0, (2.13)
where ω0 is the natural frequency of oscillations in the absence of friction:
D
ω02 = . (2.14)
J
The damping factor γ that characterizes the viscous friction is related to the quality
factor Q by the equation:
ω0
Q= . (2.15)
2γ
The boundary value of the amplitude that delimits the two cases in which the
effects either of viscous friction or of dry friction predominate:
4ϕm 4
a= = ϕm Q ≈ ϕm Q. (2.16)
γT π

2.3 Questions, Problems, Suggestions

The preceding analysis of the behavior of the oscillator under the influence of dry
friction is based upon the method of the stage-by-stage analytic integration of the
differential equations that describe the system. These equations are linear for the
time intervals occurring between consecutive extreme deflections. These intervals
are bounded by the instants at which the angular velocity is zero. The complete
solution is obtained by joining together these half-cycle solutions for consecutive
time intervals. On the other hand, the computer simulation of the torsion pendu-
lum in this software package is based on the numerical integration of differential
equations (the Runge – Kutta method to fourth order). To answer the questions
below, you may apply the analysis described above. Then you can verify your
analytic results by simulating the experiment on the computer.

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

2.3. QUESTIONS, PROBLEMS, SUGGESTIONS 29

2.3.1 Damping Caused by Dry Friction

The strength of dry friction in the system is characterized by the width of the dead
zone. This interval is defined in the program when you input the value of the angle
ϕm , which sets the limits of the dead zone on both sides of the middle position at
which the spring is unstrained. Total width of this dead zone is 2ϕm . The value of
ϕm must be expressed in degrees.
2.3.1.1 Oscillations without Dry Friction. Begin with the value ϕm = 0
corresponding to the absence of dry friction. Show that in this case the system dis-
plays the familiar behavior of a linear oscillator, i.e., simple harmonic oscillations
with a constant amplitude in the absence of friction and with an exponentially
decaying amplitude in the presence of viscous friction. The strength of viscous
friction is characterized by the quality factor Q.
2.3.1.2 Dry Friction after an Initial Displacement. To display the role of
dry friction clearly, choose a large value of the angle ϕm that determines the lim-
its of the dead zone (say, 15 to 20 degrees), and let viscous friction be zero. Such
conditions are somewhat unrealistic. They are far unlike the situation characteris-
tic of measuring instruments using a needle, such as moving-coil galvanometers.
These instruments are constructed so that the dead zone is as small as possible,
and critical viscous damping is deliberately introduced in order to avoid taking a
reading from an oscillating needle. When an instrument is critically damped, its
moving system just fails to oscillate, and it comes to rest in the shortest possible
time. If the dead zone is narrow, the needle stops at a position very close to the
dial point, which gives the true value of the measured quantity. Here, on the other
hand, conditions are chosen to clarify the role of dry friction.
(a) What can you say about the succession of maximal deflections if damping
is caused only by dry friction with the ideal z-characteristic? What is the law
of their diminishing? How is the difference of consecutive maximal deflections
related to the half-width of the dead zone?
(b) Let the angle ϕm that defines the boundaries of the stagnation zone be,
say, 15◦ , let the initial angle of deflection ϕ0 be 160◦ , and let the initial angular
velocity be zero. Calculate the point of the dial at which the needle eventually
comes to rest. How many semi-ellipses form the phase trajectory of this motion,
from its initial point to the point at which the motion stops? Verify your predictions
by simulating the motion on the computer.
(c) In the graph of the time dependence of the deflection angle, where are
the midpoints of the half-cycles of the sinusoidal oscillations located? Note how
these individual segments of the sine curves are joined to form a continuous plot
of damped oscillations.
(d) In the graph of the angular velocity versus time, note the abrupt bends in
the curve at the instants at which the midpoints abruptly replace one another. What
is the reason for these bends? Prove that these instants are separated by half the
period of harmonic oscillations in the absence of dry friction. (Note that points on
the time scale of the graphs correspond to integral multiples of the period.)

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

30 CHAPTER 2. SPRING OSCILLATOR WITH DRY FRICTION

2.3.1.3∗ Dry Friction after an Initial Push. Choose different initial condi-
tions: Let the initial deflection be zero, and the initial angular velocity be, say, 2ω0
(where ω0 is the natural frequency of oscillations). Use the same value ϕm = 15◦
as above.
(a) Calculate the maximal deflection of the needle.
(b) To what position on the dial does the needle point when oscillations cease?
How many turns are present in the complete phase trajectory of this motion? Ver-
ify your answer using a simulation experiment on the computer.
2.3.1.4∗ Damping by Dry Friction at Various Initial Conditions. Assuming
the same width of the dead zone as above, calculate the maximal angle of deflec-
tion and the final position on the dial to which the needle points when oscillations
cease, for the more complicated initial conditions:
(a) The initial deflection angle ϕ(0) = 135◦ , and the initial angular velocity
ϕ̇(0) = 1.5ω0 (ω0 is the natural frequency of the oscillator).
(b) The initial deflection angle ϕ(0) = −135◦, and the initial angular velocity
ϕ̇(0) = 1.5ω0 .
Verify your calculated values by simulating an experiment on the computer.
2.3.1.5∗ Energy Dissipation at Dry Friction.
(a) The graph of the total mechanical energy versus the angle of deflection
consists of rectilinear segments joining the slopes of the parabolic potential well
(when you work in the section “Energy transformations” of the computer pro-
gram). Suggest an explanation.
(b) Letting the initial angular velocity ϕ̇(0) = 2ω0 , where ω0 is the natural
frequency, and using energy considerations, calculate the entire angular path of
the flywheel, excited from the midpoint of the dead zone by an initial push if the
half-width of the dead zone ϕm = 10◦ .
2.3.1.6 Oscillations in the Case of a Narrow Dead Zone. Choose a small
value for the angle ϕm (less than 5◦ ), and set the initial angular displacement to
be many times the width of the dead zone, 2ϕm .
(a) How many cycles does the flywheel execute before stopping?
(b) When the number of cycles is large, the plots clearly demonstrate the linear
decay of the amplitude and the equidistant character of the loops in the phase di-
agram. What can you say about the time dependence of the total energy, averaged
over a cycle?

2.3.2 Influence of Viscous Friction

2.3.2.1∗ Transition of the Main Role from Viscous to Dry Friction. When
damping is caused both by dry and viscous friction, it is interesting to observe
the change in the character of damping when the main contribution passes from
viscous to dry friction.
Let the angle ϕm that determines the width of the dead zone be about 1◦ and let
the quality factor Q, which characterizes the strength of viscous friction, be about

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

2.3. QUESTIONS, PROBLEMS, SUGGESTIONS 31

30. Let the initial angular deflection be 120◦ and let the initial angular velocity be
zero.
(a) Does dry or does viscous friction determine the initial damping effects?
(b) At what value of the amplitude does the character of damping change?
How does this change manifest itself on the plots of time dependence of the angle
of deflection and of the angular velocity? On the phase trajectory?
2.3.2.2∗ Both Viscous and Dry Friction. Let the boundaries of the stagnation
interval be at ϕm = 10◦ and the quality factor Q = 5. Let the initial velocity be
2ω0 and let the initial deflection be zero.
(a) Calculate the maximal angular deflection of the needle at these initial con-
ditions. Verify your answer experimentally.
(b) What kind of friction, dry or viscous, initially dominates the damping of
oscillations?
(c)∗∗ Let the boundaries of the stagnation zone be determined by the angle
ϕm = 10◦ . Let the quality factor Q be 3, the initial deflection 65◦ , and the initial
angular velocity −2ω0 . Calculate the maximal angular deflection of the needle in
the direction opposite the initial deflection. Verify your answer experimentally.
2.3.2.3 Dry Friction and Critical Viscous Damping.
(a) Choose the quality factor Q to be near the critical value 0.5 and investigate
the character of damping experimentally. Where within the limits of the dead zone
is the needle most likely to stop if the quality Q is slightly greater than the critical
value? Give some physical explanation of your observations.
(b) Where would the needle stop if the quality factor Q is less than 0.5 (that is,
if the system is overdamped)? Does the answer depend on the initial conditions?

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

Chapter 3

Forced Oscillations in a
Linear System

Annotation. Chapter 3 deals with a linear torsion pendulum driven by a sinusoidal

external torque and its electromagnetic analogue – LCR-circuit with applied alter-
nate voltage. The chapter includes a description of the simulated physical system
and a summary of the relevant theoretical material for students as a prerequisite
for the virtual lab “Forced Oscillations of Linear Torsion Pendulum” from the
software package Physics of Oscillations (Part I, Simple Systems).
A preliminary study of natural oscillations (Chapter 1, “Free Oscillations of
a Linear Torsion Pendulum”) is strongly recommended. In Chapter 3 steady-state
forced oscillations and transient processes at different driving frequencies are in-
vestigated. This chapter also includes a set of theoretical and experimental prob-
lems to be solved by students on their own, as well as various assignments that the
instructor can offer students for possible individual mini-research projects.

3.1 Summary of the Theory

This chapter is concerned with forced oscillations of a torsion spring pendulum
driven by a periodic external force with sinusoidal dependence on time. In the
model of the physical system adopted here, a kinematic mode of excitation is used:
One part of the system (a driving rod) is constrained to execute simple harmonic
motion.
We consider both steady-state sinusoidal forced oscillations, which take place
later than a sufficiently large time interval elapsed after the driving force started to
operate, and transient processes, the latter depending on the initial conditions. The
decomposition of a transient process onto the sum of sinusoidal oscillations of a
constant amplitude, which have the frequency of the external force, and damped
free oscillations with the natural frequency, are examined.

33

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

34 CHAPTER 3. FORCED OSCILLATIONS IN A LINEAR SYSTEM

3.1.1 Basic Concepts

In the conventional classification of oscillations by their mode of excitation, oscil-
lations are called forced if an oscillator is subjected to an external periodic influ-
ence whose effect on the system can be expressed by a separate term, a periodic
function of the time, in the differential equation of motion. We are interested in
the response of the system to the periodic external force.
The behavior of oscillatory systems under periodic external forces is one of the
most important topics in the theory of oscillations. A noteworthy distinctive char-
acteristic of forced oscillations is the phenomenon of resonance, in which a small
periodic disturbing force can produce an extraordinarily large response in the os-
cillator. Resonance is found everywhere in physics and so a basic understanding
of this fundamental problem has wide and various applications. The phenomenon
of resonance depends upon the whole functional form of the driving force and
occurs over an extended interval of time rather than at some particular instant.
In the case of unforced (free, or natural) oscillations of an isolated system,
motion is initiated by an external influence acting before a particular instant. This
influence determines the mechanical state of the system, that is, the displacement
and the velocity of the oscillator, at the initial instant. These in turn determine
the amplitude and phase of subsequent free oscillations. Frequency and damping
of such oscillations are determined by the physical properties of the system. On
the other hand, the characteristics of forced oscillations generated by a periodic
external influence depend not only on the initial conditions and physical properties
of the oscillator but also on the nature of the external disturbance, that is, on its
amplitude and (primarily) on frequency.

3.1.2 Discussion of the Physical System

To study forced oscillations in a linear system excited by a sinusoidal external
force, we consider here the same torsion spring pendulum described in Chapter
1 (which is devoted to free oscillations), namely, a balanced flywheel attached to
one end of a spiral spring. The flywheel turns about its axis of rotation under the
restoring torque of the spring, much like the devices used in mechanical watches.
However, unlike the situation of free oscillations in which the other end of the
spring is fixed, now this end is attached to an exciter, which is a rod that can
be turned back and forth about an axis common with the axis of rotation of the
flywheel. A schematic diagram of the driven torsion oscillator is shown in the
left-hand panel of Figure 3.1.
The right-hand panel of Figure 3.1 shows an oscillatory LCR-circuit with al-
ternate input voltage. This circuit can be regarded as an electromagnetic analog of
the mechanical device. Both systems are described by identical differential equa-
tions and thus are dynamically isomorphic. However, the mechanical system has a
definite didactic advantage for exploration of forced oscillations because it allows
us to observe a direct visualization of motion.
Indeed, a mechanical system such as this one is ideal for the study of resonance

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

3.1. SUMMARY OF THE THEORY 35

- q0 +q 0
q (t)
L
j (t)
Vin C Vout

Figure 3.1: The torsion spring oscillator excited by a given sinusoidal motion of
the driving rod attached to the spiral spring (left), and its electromagnetic analogue
– LCR-circuit under sinusoidal input voltage (right).

because it is possible to see directly what is happening. When the driving rod (the
exciter) is turned through a given angle θ, the equilibrium position of the flywheel
is displaced through the same angle, alongside the rod. The flywheel can execute
free damped oscillations about this displaced position. For weak and for moderate
friction the angular frequency of these oscillations is close to the natural frequency
ω0 of the flywheel. This frequency depends onp the torsion spring constant D and
the moment of inertia J of the flywheel: ω0 = D/J.
If the rod is forced to execute a periodic oscillatory motion, the flywheel is
subjected to the action of a periodic external torque. This action is an example
of the kinematic excitation of forced oscillations. This method of excitation is
characterized by a given periodic motion of some part of the system. The kine-
matic mode of excitation is chosen here for the computer simulations of forced
oscillations because the motion of the exciting rod can be displayed directly on
the computer screen. Computer experiments with the system can show clearly,
among other things, the phase shift between the exciter and the flywheel, and the
ratio of their amplitudes.
Another possible mode of excitation of forced oscillations is characterized by
a given periodic external force whose value does not depend on the position and
velocity of the excited oscillator. This mode of excitation is called dynamic. Such
excitation is difficult to display on the screen because it does not arise from the
mechanical motion of the external source. Moreover, this mode of excitation of
a mechanical system is not easy to realize experimentally. Nevertheless, in most
textbooks forced oscillations are treated under the assumption that an oscillatory
system is excited by a given periodic force.
The differential equations describing forced oscillations are the same for both
modes of excitation. The physical differences appear primarily in the character of
energy transformations. When the excitation is kinematic, the equilibrium position
of the flywheel moves alongside the moving rod. The corresponding parabolic

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

36 CHAPTER 3. FORCED OSCILLATIONS IN A LINEAR SYSTEM

potential well then moves also as a whole, to and fro, alongside the rod. On the
other hand, when the excitation is dynamic, the potential well is stationary. We
discuss these differences below (see Section 3.2.4, p. 42).

3.1.3 The Differential Equation for Forced Oscillations

We assume that the exciting rod is constrained by some external source to execute
simple harmonic motion about a middle position (the vertical in Figure 3.1). The
amplitude of the rod is θ0 and the angular frequency is ω. The angular displace-
ment of the rod θ(t) varies with time t sinusoidally:

θ(t) = θ0 sin ωt. (3.1)

If at some instant t the flywheel is displaced through an angle ϕ(t) from the
central position (which is the origin of the scale in Figure 3.1), and the rod of the
exciter is simultaneously displaced through an angle θ, the spring exerts a torque
−D(ϕ − θ) = −Dϕ + Dθ0 sin ωt on the flywheel because the spring is strained
through the angle ϕ − θ. (Compare this torque with the torque −Dϕ for the case
of free oscillations.) Hence, in the absence of friction, the differential equation of
rotation of the flywheel with the moment of inertia J is:

J ϕ̈ = −Dϕ + Dθ0 sin ωt. (3.2)

This equation is also the differential equation of forced oscillations excited by

a given external sinusoidally varying torque Dθ0 sin ωt whose constant amplitude
is Dθ0 . That is, the equation of motion is the same for forced oscillations excited
by these two modes (an oscillating motion of the rod and a sinusoidal external
torque).
p by the moment of inertia J of the flywheel
Dividing both sides of this equation
and introducing the notation ω0 = D/J for the natural frequency (ω02 = D/J),
we rewrite the above equation in a canonical form:

ϕ̈ + ω02 ϕ = ω02 θ0 sin ωt. (3.3)

In the presence of viscous friction whose torque is proportional to the angular

velocity ϕ̇ of the flywheel, we must add the appropriate frictional term to the
differential equation of motion describing forced oscillations:

ϕ̈ + 2γ ϕ̇ + ω02 ϕ = ω02 θ0 sin ωt. (3.4)

The damping constant γ characterizes the strength of viscous friction. As in

the case of free oscillations, it is related to the dimensionless quality factor Q by
the expression 2γ/ω0 = 1/Q.
Forced oscillations of the electric charge q stored in a capacitor of a resonant
series LCR-circuit (see the right-hand panel of Figure 3.1) excited by a sinusoidal

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

3.1. SUMMARY OF THE THEORY 37

input voltage Vin (t) obey the same differential equation as does the forced oscil-
lation of a mechanical torsion spring oscillator excited by sinusoidal variations in
position of the driving rod:

q̈ + 2γ q̇ + ω02 q = ω02 CVin (t). (3.5)

In this equation ω0 is the natural frequency of oscillations of charge in the circuit

in the absence of resistance. It depends on √ the capacitance C of the capacitor
and the inductance L of the coil: ω0 = 1/ LC. The damping constant γ =
R/(2L) characterizes the dissipation of electromagnetic energy occurring in a
resistor whose resistance is R.
Because of this similarity, the mechanical system described above enables us
to give a very clear explanation for transformation of the input voltage Vin (t) =
V0 sin ωt into the output voltage Vout (t) = VC (t) = q/C (voltage across the
capacitor C). The output voltage VC (t) is analogous to the deflection angle ϕ(t)
of the rotor, while the alternating electric current I(t) = q̇(t) in the circuit is
analogous to the rotor angular velocity ϕ̇(t) of the mechanical model. However,
some caution is necessary in interpreting the analogy between the mechanical
oscillator and the electric LCR-circuit with respect to the energy transformations.
The momentary position of the modeled physical system at any time instant is
determined by two angular coordinates, namely, by the angles ϕ(t) and θ(t). But
the coordinate θ(t) is entirely determined by external conditions. (The physical
meaning of θ(t) is the instantaneous position of the rod that executes a given
motion.) The angle θ it is not a “free” coordinate, and so the system does not
actually have a second degree of freedom. The only “free” coordinate (that is, the
coordinate whose functional dependence on time is yet to be determined) is the
angle ϕ, which gives the deflection of the flywheel from the central position. To
find this unknown function ϕ(t), we need only one differential equation, Eq. (3.4).
The differential equation corresponding to the second coordinate θ can be used to
find the external torque that must be exerted on the rod in order to provide its given
sinusoidal motion. The external source of such torque inevitably experiences the
reaction of the oscillator.

3.1.4 The Principle of Superposition

Our investigation of forced oscillations in a system whose differential equation
of motion is linear is facilitated by the principle of superposition. This principle
states that if several external forces act simultaneously on a linear system, the
forced oscillations caused by each force acting separately are to be added together
(superimposed) to get the complete solution. In other words, in linear systems
there is no interaction (no mutual influence) of the individual oscillations excited
by several external forces acting simultaneously.
It follows from the principle of superposition that, in addition to the forced
oscillations caused by a given external force, a linear oscillator can simultane-
ously execute free damped oscillations. These free (or natural) oscillations may
be thought of as arising from a null external force on the right-hand side of the

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

38 CHAPTER 3. FORCED OSCILLATIONS IN A LINEAR SYSTEM

differential equation of forced oscillations, Eq. (3.4). We can always infer that
along with a given driving force a null force is also present. These natural oscilla-
tions are excited when an external force is switched on (or when its amplitude or
initial phase is changed).
A superposition of such damped natural oscillations and driven forced oscil-
lations of constant amplitude occurs during a transient process, when forced os-
cillations, over a period of time, acquire the frequency of the external force and a
constant amplitude. The duration τ of this transient process of establishing forced
steady-state oscillations equals (in the general case) the duration of damping of
free oscillations: τ = 1/γ.

3.2 Steady-State Forced Oscillations

During some time after the external force has been activated (after the rod has be-
gun its given periodic motion), the transient natural oscillations inevitably damp
out. Since only these oscillations depend on the initial conditions, we can say figu-
ratively that the oscillator eventually “forgets” its initial state, and its forced oscil-
lations become steady: The flywheel executes harmonic oscillations of a constant
amplitude with the frequency of the external driving force. These steady-state os-
cillations are described by the periodic particular solution of the inhomogeneous
differential equation of motion, Eq. (3.4):

ϕ(t) = a sin(ωt + δ). (3.6)

The steady-state oscillations are characterized by definite values of amplitude

a and phase lag δ. The phase lag δ is the angular difference between the instants
at which the flywheel and the driving rod cross the zero point of the dial (or reach
their maximal displacements). Both a and δ depend on ω (the frequency of the
external action), ω0 (the natural frequency), and γ (the damping constant of the
oscillator).
The dependencies of a and δ on the external driving frequency, ω, are called
the amplitude—frequency and the phase—frequency characteristics of the oscil-
lator. When friction is relatively weak (γ ≤ ω0 or Q ≥ 1), the dependence of the
amplitude on the frequency has a resonance character: the amplitude increases
sharply as ω approaches ω0 .
The graph of the dependence of the steady-state amplitude on the frequency
ω is called the resonance curve (see Figure 3.2). The greater the quality factor
Q, the sharper the peak of the resonance curve, that is, the more pronounced the
resonance in the system.

3.2.1 Forced Oscillations in the Absence of Friction

When the driving frequency is sufficiently far from the resonant frequency, we
may neglect the influence of friction on the amplitude a and the phase lag δ of
steady-state oscillations. That is, we may employ here the idealized frictionless

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

3.2. STEADY-STATE FORCED OSCILLATIONS 39

model to describe the behavior of a real system in which there is some friction.
(Note that the applicability of a physical model to a real system depends not only
on the properties of the system, but also on the problem that we are solving.) Thus,
in order to describe steady-state oscillations in the case when |ω − ω0 | ≫ γ, we
can use Eq. (3.3), which is valid for a forced linear oscillator in the absence of
friction.
As a particular periodic solution describing steady-state oscillations, we try
the expression:
ϕ(t) = a sin ωt. (3.7)
Substituting this expression into Eq. (3.3), we find that Eq. (3.7) actually gives
a solution of Eq. (3.3) if the amplitude a(ω) as a function of frequency ω is:

ω02 θ0
a(ω) = . (3.8)
ω02 − ω 2

If the driving frequency ω is set equal to zero, Eq. (3.8) yields a = θ0 : The
flywheel is at rest in the displaced equilibrium position. If ω ≪ ω0 , a ≈ θ0 : In
the case of a very slow motion of the driving rod, the flywheel follows the rod
quasistatically. That is, the flywheel remains in the equilibrium position, which
itself moves slowly alongside the slowly moving rod. At very low frequencies
of the external action, kinematically excited steady-state forced oscillations of
the flywheel occur with almost the same amplitude and the same phase as the
compelled motion of the driving rod.
Equation (3.8) shows that as the driving frequency ω is increased, the ampli-
tude of forced oscillations of the flywheel becomes greater. For ω → ω0 the value
a of the amplitude tends to infinity. Consequently, it is inadmissible to ignore fric-
tion in the vicinity of resonance (at ω ≈ ω0 ). This case is considered below.
We note that according to Eq. (3.8), the value of a becomes negative if ω >
ω0 . The negative sign of a means here that for ω > ω0 , steady-state oscillations
occur with a phase opposite that of the external force: When the rod turns in one
direction, the flywheel turns in the other, both reaching their opposite extreme
deflections simultaneously.
We can write the solution for ω > ω0 in the form of Eq. (3.6), retaining the
positive amplitude a for all frequencies, if we assume a to be equal to the absolute
value of the right-hand side of Eq. (3.8), and the phase shift δ to be equal to −π.
When ω < ω0 , the phase shift δ in the absence of friction (and also for rela-
tively weak friction, as we shall see later) is zero, so that the flywheel and the rod
oscillate in phase. That is, moving in the same direction, they pass the mid-point at
the same time, and reach the extremes in their deflections simultaneously. How-
ever, as indicated by Eq. (3.8), the extreme deflection of the flywheel is greater
than that of the rod and increases infinitely when the driving frequency approaches
the natural frequency (when ω → ω0 ).

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

40 CHAPTER 3. FORCED OSCILLATIONS IN A LINEAR SYSTEM

a
5 q0 Q=5

4
Q = 3.5
3
Q = 2.5
2

0 w
1 2 3 w
d 0

− π /2 w
w0

−π

Figure 3.2: Resonance curves of a linear oscillator.

3.2.2 The Resonance Curve

In the vicinity of resonance (at driving frequencies ω that satisfy the condition
|ω − ω0 | ≤ γ), it is necessary to take friction into account in the differential
equation of forced oscillations. That is, we need to solve Eq. (3.4). Steady-state
forced oscillations are described by its particular periodic solution. We can write
this solution in the form ϕ(t) = a sin(ωt + δ) (see Eq. (3.6)). Substituting that
solution into Eq. (3.4), we can search for the values of a and δ for which the
function a sin(ωt + δ) satisfies Eq. (3.4). Leaving the derivations as an exercise,
we give here the final expressions for the amplitude a and the phase shift δ:

ω02 θ0 2γω
a(ω) = p 2 , tan δ = − . (3.9)
(ω0 − ω 2 )2 + 4γ 2 ω 2 ω02− ω2

The graphs of the dependence of the amplitude on the frequency a(ω) (the
resonance curves) and the phase lag on the frequency δ(ω) for different values of
the quality factor Q are shown in upper and lower parts of Figure 3.2, respectively.
The frequency ω of excitation is measured in this figure in units of the natural fre-
quency ω0 , and the amplitude a of the flywheel – in units of the exciter amplitude
θ0 . The maximum value of the amplitude of steady oscillations amax occurs at the
resonant frequency ωres : q
ωres = ω02 − 2γ 2 . (3.10)

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

3.2. STEADY-STATE FORCED OSCILLATIONS 41

√
This expression for ωres is valid if friction is not too large, that is, if ω0 > 2 γ.
When friction is sufficiently small, that is, if γ ≪ ω0 or Q ≫ 1, we can use
instead of Eq. (3.10) the following approximate expression:
s
2γ 2 γ2
   
1
ωres = ω0 1 − 2 ≈ ω0 1 − 2 = ω0 1 − . (3.11)
ω0 ω0 4Q2

That is, the resonant frequency nearly coincides with the natural frequency ω0 :
The value ωres differs from ω0 only by a term of the second order in the small pa-
rameter γ/ω0 . For example, if Q = 10 (moderate friction), the resonant frequency
differs from the natural frequency only by 0.25%.
The amplitude of steady-state oscillations at resonance amax is determined by
the following expression:
ω2θ ω 0 θ0
amax = p0 0 ≈ = Qθ0 . (3.12)
2γ ω02 − γ 2 2γ

We see from Eq. (3.12) that the amplitude amax of steady-state oscillations at
resonance is approximately Q times greater than the amplitude θ0 of the driving
rod (provided that the quality factor Q is not too low). In other words, the am-
plitude amax of steady-state oscillations at resonance is Q times greater than the
amplitude a(0) of steady-state oscillations at a very low driving frequency ω (at
slow oscillations of the rod). We note that the resonant properties of a linear oscil-
lator under forced oscillations and the damping of its natural free oscillations are
characterized by the same quantity, the quality factor Q.
When there is no friction, Eq. (3.9) shows that the amplitude of the flywheel
during steady-state forced oscillations is greater than the amplitude
√ θ0 of the rod
at all frequencies ω between zero and the boundary value 2 ω0 . When the fre- √
quency ω of the driving force exceeds the natural frequency ω0 by more than 2
times, the amplitude of steady-state forced oscillations is smaller than θ0 and ap-
proaches zero as the frequency ω increases further. In this range of frequencies
the dynamic effect of an external driving force is less than the static effect of a
constant force of the same magnitude. The physical cause of such behavior is the
inertia of the flywheel: When the driving frequency of the rod is considerably
greater than the natural frequency of the flywheel, the massive flywheel cannot
follow the rapid motion of the rod. The same is true also in the presence of√mod-
erate friction, except that the bounding frequency is slightly smaller than 2 ω0 ,
as can be seen from the graphs of a(ω) plotted in Figure 3.2 using Eq. (3.9).
Equation (3.9) for the phase shift δ and the corresponding graphs in the lower
panel of Figure 3.2 show that steady-state forced motion always lags behind the
driving force since δ is always negative. Far from resonance at ω < ω0 this lag is
nearly zero, and the flywheel oscillates nearly in phase with the exciting rod. When
ω = ω0 , steady-state oscillations of the flywheel lag in phase behind oscillations
of the exciting rod by a quarter of the period (δ = −π/2) for all values of friction.
In this case the displacement of the flywheel is greatest when the displacement of
the rod is zero, and vice versa. When ω is much greater than ω0 , the phase shift δ

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

42 CHAPTER 3. FORCED OSCILLATIONS IN A LINEAR SYSTEM

approaches −π. That is, the lag is nearly 180◦ , which means that in this case the
flywheel and the exciting rod always rotate in opposite directions.
If there is no friction, Eq. (3.9) indicates that the phase lag is either 0 (for ω <
ω0 ) or 180◦ (for ω > ω0 ). That is, when ω = ω0 , there is an abrupt change from
motion of the flywheel exactly in phase with that of the rod to motion in which they
oscillate exactly in opposite phase. (In the absence of friction, the amplitude of
the flywheel at the transition is infinite.) In the presence of friction, the transition
from in-phase steady-state oscillations of the flywheel and the rod to opposite-
phase steady-state oscillations takes place gradually over a range of frequencies
centered about ω0 . The width of this range, as can be seen from Figure 3.2, is
proportional to the damping constant γ.

3.2.3 Resonance of the Angular Velocity

In steady-state oscillations under a sinusoidal force, the angular velocity of the
flywheel ϕ̇ = aω cos(ωt + δ) changes with time harmonically with the frequency
ω of the external driving force. The expression for the amplitude of the angular
velocity Ω = ϕ̇ differs from Eq. (3.9) for the amplitude a(ω) of these oscillations
by an additional factor ω:
ω02 θ0
Ω(ω) = ωa(ω) = p 2 . (3.13)
(ω0 /ω − ω)2 + 4γ 2
Dependence of the velocity amplitude on the frequency is shown in the upper
panel of Figure 3.3.
As we can see from Eq. (3.13), the maximum of the resonance curve for the
angular velocity is located at ω = ω0 independently of the damping factor γ.
Therefore resonance of the angular velocity occurs at the value of the driving fre-
quency ω, which exactly equals the natural frequency ω0 of the oscillator for both
weak and strong friction.
p On the other hand, resonance of the angular displace-
ment occurs at ωres = ω02 − 2γ 2 .
The lower panel of Figure 3.3 shows the dependence of the phase shift between
the driving rod and the angular velocity. At resonance (at ω = ω0 ) this phase shift
equals zero: The driving rod oscillates in phase with the velocity. This means that
at resonance the energy is transmitted in one direction – from the exciter to the
oscillator – during the whole period.

3.2.4 Energy Transformations

Though the amplitude is constant in steady-state forced oscillations, the total en-
ergy of the oscillator is constant only on the average. During one quarter of a
cycle, energy is transmitted from the driving rod to the oscillator, and during the
next quarter cycle, energy is transferred back from the oscillator to the external
source driving the rod. In contrast to free oscillations, not only do the kinetic and
potential energies oscillate, but so does their sum, the total mechanical energy.
The total energy oscillates with double the frequency of the external force.

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

3.2. STEADY-STATE FORCED OSCILLATIONS 43

5 Q=5

4
Q = 3.5
3
Q = 2.5
2

π /2 w
1 2 3 w
0

0 w
w0

− π /2

Figure 3.3: Resonance curves for the velocity of a linear oscillator.

In our discussion of steady-state energy transformations, we pay special at-

tention to the distinguishing characteristics of the energy transfer for kinematic
excitation of forced oscillations modeled in these computer simulations, and for
dynamic excitation, for which the external torque is a given time-dependent quan-
tity. We have already mentioned these distinctions above. Here we discuss them
in detail.
When the dynamic mode of excitation is used, an external torque is applied
directly to the flywheel. With one end of the spiral spring fixed, the deformation
of the spring (the amount of twisting from its unstrained state) is determined by the
angular displacement ϕ of the flywheel from the midpoint (from the equilibrium
position). Thus the potential energy of the spring is given by
1 1 1
Epot = Dϕ2 (t) = Da2 sin2 (ωt + δ) = Jω02 a2 [1 − cos 2(ωt + δ)]. (3.14)
2 2 4
The kinetic energy of the oscillating flywheel is independent of the mode of
excitation and is given by the following expression
1 2 1 1
Ekin = J ϕ̇ (t) = Jω 2 a2 cos2 (ωt+δ) = Jω 2 a2 [1+cos 2(ωt+δ)]. (3.15)
2 2 4
It is seen that, while the oscillator executes steady-state forced oscillations
with the frequency ω, the values of its potential and kinetic energies are oscillat-

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

44 CHAPTER 3. FORCED OSCILLATIONS IN A LINEAR SYSTEM

ing harmonically with the frequency 2ω in opposite phases with respect to one
another. The ratio of their maximal (and average) values is equal to the squared
ratio of the natural frequency to the driving frequency:

hEpot i ω2
= 02 . (3.16)
hEkin i ω

Hence, when ω < ω0 , the potential energy predominates on the average over the
kinetic energy. In particular, when ω ≪ ω0 , the spring is twisted quasistatically,
and nearly all the energy of the oscillator is the elastic potential energy of the
strained spring. On the other hand, for external frequencies that exceed the natural
one (ω > ω0 ), the kinetic energy predominates over the potential energy.
The peculiarities of energy transformations for the kinematic excitation of os-
cillations are related to the fact that the equilibrium position of the flywheel (and
its potential well as a whole) is displaced when the driving rod is turned. The de-
formation of the spring in this case is determined by the difference in the angles
ϕ(t) and θ(t). The expression for its potential energy then takes the form:
1
Epot = D(ϕ − θ)2 . (3.17)
2
When the frequency of the rod is much less than the resonant frequency, the
flywheel moves as though it were attached to the slowly moving rod. The flywheel
remains close to its equilibrium position, which is displaced by the driving rod.
The spring remains nearly unstrained and its potential energy is nearly zero. In
other words, the oscillator is always located near the bottom of its potential well,
which slowly oscillates alongside the rod. Therefore, at low frequencies of the
driving mechanism, the kinetic energy predominates over the potential energy, in
direct contrast to the case of dynamic excitation.
When the driving frequency is large compared to the natural frequency, the
inertia of the flywheel diminishes the response the flywheel is able to make to
the displacement of the equilibrium position by the external source, and so the
flywheel oscillates with a relatively small amplitude. At high driving frequencies,
the amplitude of the steady-state deflections of the flywheel from its central po-
sition is much smaller than the amplitude θ0 of the angular displacement of the
driving rod. The spring is twisted back and forth through approximately the an-
gle θ0 , while the angular velocity of the flywheel remains relatively small. Hence
the elastic potential energy arising from the deformation of the spring predomi-
nates over the kinetic energy of the flywheel, again in direct contrast to the case
of dynamic excitation.
When friction is small (γ ≪ ω0 ), the ratio of the average value of the potential
energy to the average value of the kinetic energy for kinematic excitation can be
found from Eq. (3.17) and Eq. (3.6) for ϕ(t) and Eq. (3.1) for θ(t):

hEpot i ω2
= 2. (3.18)
hEkin i ω0

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

3.3. TRANSIENT PROCESSES 45

Comparing Eq. (3.18) with Eq. (3.16), we see that the ratios of the average
value of the potential energy to the average value of the kinetic energy for dy-
namic and kinematic modes of excitation of forced oscillations are inverses of one
another, in agreement with the general discussion above.

3.3 Transient Processes

The amplitude and the phase of steady-state oscillations do not depend on the
initial conditions. Loosely speaking, during the transient process the oscillator
eventually “forgets” them. We should keep in mind that steady-state oscillations
are described by Eq. (3.6), which is the periodic particular solution to the in-
homogeneous differential equation, Eq. (3.4). The graphs of the amplitude and
phase versus the driving frequency (the amplitude—frequency and the phase—
frequency characteristics of a linear oscillator), displayed in Figure 3.2, refer to
this particular solution and are valid only for steady-state oscillations.
The initial conditions, namely the initial angle of deflection ϕ(0) and the initial
angular velocity ϕ̇(0), are influential only during the transient process. During the
transition, natural damped oscillations are superimposed on the steady-state forced
oscillations. The effects of the natural oscillations disappear once the steady-state
oscillations have been established.
Mathematically a transient process is represented by the general solution to
the inhomogeneous equation, Eq. (3.4). This complete solution is given by:

ϕ(t) = a sin(ωt + δ) + Ce−γt cos(ω1 t + α). (3.19)

The first term on the right is the periodic particular solution, Eq. (3.6), to the
inhomogeneous equation, Eq. (3.4). The second term on the right, called the tran-
sient term, is the general solution to the corresponding homogeneous equation,
namely Eq. (3.4) in which the right-hand side is zero. This solution of the homo-
geneous equation is the contribution of damped natural oscillations to the transient
process. The frequency ω1 of this term nearly equals the natural frequency ω0 pro-
vided the friction is not too large:
s
γ2 γ2
   
1
q
ω1 = ω02 − γ 2 = ω0 1 − 2 ≈ ω0 1 − = ω 0 1 − . (3.20)
ω0 2ω02 8Q2

The fractional difference (ω0 − ω1 )/ω0 in most cases of practical importance

is so small that we can neglect it and assume that ω1 = ω0 . Indeed, if Q = 5, the
fractional difference is only 0.5%: (ω0 − ω1 )/ω0 = 0.005.
The transient term of the general solution contains two arbitrary constants,
C and α. Their values depend on the initial conditions, namely on the angular
displacement and the angular velocity of the flywheel at the instant the external
force begins to act.
Thus the transient process is described by a superposition of two oscillations:
A sinusoidal steady oscillation with a constant amplitude a and a frequency ω of

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

46 CHAPTER 3. FORCED OSCILLATIONS IN A LINEAR SYSTEM

o
41.7 ϕ( t )

o
-41.7
0.53 .
ϕ( t)

-0.53
0 1 2 3 4T

Figure 3.4: Decomposition of a transient process onto the sum of steady-state

forced oscillations and damped natural oscillations (time-dependent graphs of the
angular displacement and angular velocity) at ω = 0.5ω0 , ϕ(0) = 0, ϕ̇(0) = 0.

the driving force, and a damped natural oscillation with the frequency ω1 ≈ ω0
and a decaying amplitude. In principle, the transient process continues indefi-
nitely, but in practice it is considered completed after the transient term essentially
dies out, that is, after about Q cycles of natural oscillations. (We recall that Q is
the dimensionless quality factor, given by Q = ω0 /2γ.) Generally, the smaller the
friction, the longer the transient process lasts. However, as we shall see in the next
section, it is possible to choose initial conditions such that there is no transient
term and no transient process.
There is an option in the suggested simulation program that allows the user
to display the plots of the two simple oscillations, which constitute the transient
process, while simultaneously plotting their superposition. An example of the de-
composition of a transient process onto the two separate simple components—the
steady-state sinusoidal oscillation with a constant amplitude and the frequency of
the external driving action (periodic particular solution of the differential equa-
tion), and the decaying natural transient oscillation (solution of the correspond-
ing homogeneous equation)—is shown in Figure 3.4. The graphs correspond to
ω = 0.5ω0 (the driving frequency equals one half the natural one), and zero initial
conditions (ϕ(0) = 0, ϕ̇(0) = 0).

3.3.1 Initial Conditions That Eliminate a Transient

As noted above, it is possible to choose the initial conditions in such a way that
there is no transient process, that is, so that steady-state oscillations appear imme-
diately upon activation of the external force. To find these conditions, we observe
that if we take as the angular displacement and the angular velocity at t = 0

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

3.3. TRANSIENT PROCESSES 47

the corresponding values that characterize the steady-state oscillations, the initial
conditions are satisfied by the steady-state itself, without the addition of natural
oscillations.
It follows from Eq. (3.6) that the required initial angular deflection ϕ(0) is
a sin δ, and the required initial angular velocity ϕ̇(0) is aω cos δ, where a and δ
are the amplitude and the phase of steady-state oscillations given by Eq. (3.9).
Since the steady-state term in Eq. (3.19) satisfies the initial conditions by itself,
the transient term vanishes, leaving only the steady-state, so that C in Eq. (3.19)
must be zero. In other words, if ϕ(0) = a sin δ and ϕ̇(0) = aω cos δ, no transient
natural oscillation arises when the external force begins to act, and there is no
contribution to the motion from the homogeneous equation.
Transient processes make the phenomenon of forced oscillations much more
complicated than are simple harmonic steady-state oscillations. In many cases
these transient processes are important and interesting in themselves. They are
worth considerable attention in your work with the simulation computer program.

3.3.2 Forced Oscillations from Rest at Resonance

We next restrict our study of transient processes to the case in which the initial
conditions are zero. That is, we consider forced oscillations that correspond to the
flywheel at rest in the equilibrium position at the instant the driving force begins
to operate:
ϕ(0) = 0, ϕ̇(0) = 0. (3.21)
At t = 0, the driving rod starts moving according to:

θ(t) = θ0 sin ωt. (3.22)

Let us first consider the case of an oscillator damped by relatively weak fric-
tion (γ ≪ ω0 ). Because there is little friction, the resonant frequency is very
nearly the natural frequency ω0 . For the case in which the driving frequency ω is
set equal to ω0 , it follows from Eqs. (3.9) that the periodic particular solution that
describes the steady-state oscillations is given by:
ω0  π
ϕ(t) ≈ θ0 sin ω0 t − = −Qθ0 cos ω0 t. (3.23)
2γ 2
In this case the amplitude of oscillation of the flywheel is greater than the ampli-
tude of oscillation of the rod by the factor Q, and the phase lag is −π/2. That is,
the oscillations of the flywheel are one quarter of a cycle behind the oscillations
of the driving rod.
The transient term in Eq. (3.19) with ω1 = ω0 is next added to Eq. (3.23) to
obtain a complete solution:

ϕ(t) = −Qθ0 cos ω0 t + Ce−γt cos(ω0 t + α). (3.24)

The arbitrary constants C and α are determined by satisfying the initial condi-
tions, Eqs. (3.21). In the case of weak friction, when γ ≪ ω0 , the exponential

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

48 CHAPTER 3. FORCED OSCILLATIONS IN A LINEAR SYSTEM

factor e−γt in Eq. (3.24) changes little over many oscillations and when taking
the time derivative of Eq. (3.24), can be considered approximately constant for
long durations:

ϕ̇(t) ≈ Qθ0 ω0 sin ω0 t − Ce−γt ω0 sin(ω0 t + α). (3.25)

Requiring that ϕ̇(0) be zero implies that α = 0, and requiring that ϕ(0) be
zero implies that C = Qθ0 . Hence, if Q ≫ 1 and ω = ω0 , the solution of
differential equation of motion (3.6), satisfying the zero initial conditions defined
by Eq. (3.21), is:

ϕ(t) = −Qθ0 (1 − e−γt ) cos ω0 t = −b(t) cos ω0 t, (3.26)

where
b(t) = Qθ0 (1 − e−γt ). (3.27)
This superposition of forced and slowly decaying natural oscillations, each
with the frequency ω0 , can be considered as a single nearly harmonic oscillation
with the frequency ω0 and an amplitude b(t), which slowly increases with time,
asymptotically approaching the steady-state value, Qθ0 . The graph of ϕ(t) for
such a transient process at resonance in the case of zero initial conditions is shown
in the upper panel of Figure 3.5 (the angular displacement of the flywheel ϕ(t)
together with the angular displacement θ(t) of the exciter). The lower panel of
this figure shows decomposition of this process onto the sum of steady-state forced
oscillations of constant amplitude and damped natural oscillations.
o
100
ϕ (t)

0
θ (t )
o
-100
100
o 1 2 3 4 5 6 7 8 9 10 11 12T

o
-100
0 1 2 3 4 5 6 7 8 9 10 11 12T

Figure 3.5: The transient process of resonant excitation from the state of rest in
the equilibrium position (upper panel), and decomposition of this process onto the
sum of steady-state forced oscillations and damped natural oscillations (ω = ω0 ,
θ0 = 10◦ , Q = 10, initial conditions ϕ(0) = 0, ϕ̇(0) = 0).

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

3.3. TRANSIENT PROCESSES 49

By convention, the duration of this transient process is assumed to be equal

to the time of damping τ (τ = 1/γ), which is the time during which natural free
oscillations of the transient term fade away.
This gradual growth of the amplitude b(t) and its gradual asymptotic approach
to a maximum value during the transient process at the resonant frequency can be
easily explained on the basis of energy transformations. At resonance a definite
phase relation establishes between the oscillation of the flywheel and that of the
rod. Namely, the rod oscillates in phase with the angular velocity of the flywheel.
This phase relation provides the conditions that are favorable for the transfer of
energy from the rod to the oscillator. For large values of the quality factor Q, the
amplitude of the flywheel eventually increases to the value Qθ0 , which consider-
ably exceeds (by a factor of Q) the amplitude θ0 of the rod. The greater the quality
factor Q, the greater the energy eventually stored by the oscillator and the greater
the number of cycles required to transfer this energy to the oscillator by a weak
driving force. The growth of the amplitude decreases to zero when the velocity-
dependent friction dissipates any further addition of energy from the driving rod.
During steady-state oscillations, the energy dissipated over a cycle equals the en-
ergy transmitted to the oscillator from the external source that drives the exciting
rod.
In the case of weak friction the duration of the transient process is large com-
pared with the period of oscillations: τ ≫ T0 . The amplitude of oscillation of the
flywheel, initially at rest, increases over many cycles. The early growth of the am-
plitude occurs almost linearly with time. This behavior is evident from Eq. (3.27),
in which we let γt ≪ 1. Expanding the exponential in a power series and keeping
only the linear term, we have:
ω0 1
b(t) = Qθ0 (1 − e−γt ) = θ0 (1 − e−γt ) ≈ θ0 ω0 t. (3.28)
2γ 2
Cancelation of the damping constant γ in the latter expression means that
the linear growth of the amplitude during the early stage of the transient pro-
cess (while γt ≪ 1, or t ≪ QT0 ) occurs just as though friction were absent. In
the idealized case of the complete absence of friction, such linear growth of the
amplitude would continue indefinitely.
Thus, steady-state oscillations are impossible for a frictionless oscillator at the
resonant frequency. In this case, the particular solution of the inhomogeneous dif-
ferential equation of motion (Eq. (3.4), with γ = 0 and ω = ω0 ), rather than
being periodic, increases without limit: ϕ(t) = 21 θ0 ω0 t cos ω0 t. To obtain the
general solution containing two arbitrary constants, we must also include the gen-
eral solution of the corresponding homogeneous equation, which in this case de-
scribes natural oscillations of constant amplitude. Determining the constants from
the zero initial conditions, we find:
1
ϕ(t) = θ0 (ω0 t cos ω0 t − sin ω0 t). (3.29)
2
The amplitude of this oscillation with the frequency ω0 is 21 θ0 (ω0 t)2 + 1 ≈
p
1
2 θ0 ω0 t. Thus, during the resonant transient process in the absence of friction the

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

50 CHAPTER 3. FORCED OSCILLATIONS IN A LINEAR SYSTEM

amplitude of the flywheel, initially at rest, increases steadily at almost a constant

rate. This rate is the same as the early rate in the presence of weak friction, because
during the early stage of the transient process the growth of mechanical energy is
almost unimpeded by the velocity-dependant friction.
The unlimited growth of the amplitude at resonance predicted by Eq. (3.29)
means, as noted above, that in this frictionless system steady-state oscillations
are not possible when ω = ω0 . In fact this growth means that in conditions of
resonance the idealized frictionless model does not work; that is, we cannot use
the model for description of resonance in a real system no matter how small the
friction.
In a real system, at sufficiently large amplitudes, either friction dissipates the
added energy (so that in the model of the system it is necessary to take friction
into account) or the amplitude of the oscillation increases beyond the linear limit
of the restoring force of the spring so that Hooke’s law fails. In the latter case, the
nonlinear dependence of the restoring force on the angle of deflection changes the
period of natural oscillations as the amplitude grows, and resonance is destroyed.
For a nonlinear system, the growth of the amplitude is restricted even in the ab-
sence of friction: The resonant conditions become violated at large amplitudes.
To say which of these reasons (friction or nonlinearity) restricts the growth of the
amplitude in a real physical system, we need to know more about the properties
of the system.

3.3.3 Mechanical Analogue of the Stimulated Emission of

Radiation
As we have seen above, in conditions of exact tuning to resonance, that is, when
the external frequency is equal to the natural frequency of the oscillator, the am-
plitude of forced oscillations grows monotonically. In the absence of friction, as
follows from Eq. (3.29), the amplitude grows indefinitely almost linearly with the
time t: b(t) ≈ 21 θ0 ω0 t. The growth of energy is provided by the external source
that drives the rod of the exciter. Now let us think about the following question:
Is the backward transfer of energy from the oscillator to the source of external
excitation possible in conditions of exact tuning to resonance?
The process of resonant growth of the amplitude from the state of rest in the
equilibrium position occurs at definite relations between the phases of the exciter
and the flywheel. These relations are favorable for the transfer of energy from the
external source to the oscillator: As follows from Eq. (3.29), the angular velocity
of the flywheel depends on time according to ϕ̇(t) ≈ 12 θ0 ω02 t sin ω0 t almost from
the very beginning, that is, the velocity changes in the same phase with the external
torque Dθ0 sin ω0 t in the right-hand side of Eq. (3.4). This means that the torque
is always directed in the same way as the angular velocity, pushing the flywheel in
the direction of its rotation. We note that such phase relationships are characteristic
for zero initial conditions.
But what will happen if at the moment at which the external force is switched
on, the oscillator is already excited, i. e., it already has energy and is executing

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

3.3. TRANSIENT PROCESSES 51

natural oscillations? It is clear that the subsequent process depends on the phase
relations between the existing natural oscillations of the flywheel and oscillations
of the exciter. If the angular velocity varies in the same phase with the external
torque, the energy again will be transferred from the external source to the oscilla-
tor, and the amplitude of existing oscillations will be increasing immediately after
the external source is switched on. On the other hand, if the angular velocity of
the flywheel varies in the opposite phase with respect to the external torque, that
is, according to ϕ̇(t) ∼ − sin ω0 t, the external torque will be directed all the time
against the velocity of the flywheel, and will slow down its rotation. This means
that the energy will be transferred from the oscillator to the external source. Cer-
tainly, this process will continue only until the amplitude becomes zero, that is,
until all the energy of the oscillator is transferred to the external source. After this
moment all will happen in the way that corresponds to zero initial conditions: The
amplitude will grow because the phase relations become favorable for the transfer
of energy from the external source to the oscillator.
We can easily see that if the exciting rod moves according to θ(t) = θ0 sin ω0 t,
the phase of the flywheel’s natural oscillation that is favorable for the transfer of
energy from the oscillator to the external source will be provided, for example, if
at the initial moment the flywheel is at rest and is displaced from the equilibrium
position in the positive direction, that is, at the initial conditions of the type ϕ(0) =
ϕ0 (where ϕ0 > 0), ϕ̇(0) = 0. The solution to differential equation of motion,
Eq. (3.4), which corresponds to such initial conditions, has the form:
1 1
ϕ(t) = (ϕ0 − θ0 ω0 t) cos ω0 t + θ0 sin ω0 t
2 2
1
≈ (ϕ0 − θ0 ω0 t) cos ω0 t. (3.30)
2
The approximate expression in (3.30) is valid when it is possible to ignore the con-
tribution of natural oscillations into ϕ(t), i. e., when the constant amplitude 21 θ0 of
the sin ω0 t term is small compared to the time-dependent amplitude |ϕ0 − 12 θ0 ω0 t|
of the cos ω0 t term in Eq. (3.30). It follows from Eq. (3.30) that the amplitude
first diminishes linearly with time to almost the zero value during the interval
t = 2ϕ0 /(ω0 θ0 ) = (ϕ0 /πθ0 )T0 . Then the time-dependent term of the amplitude
in Eq. (3.30) changes its sign. This means that the phase relation between the fly-
wheel and the exciting rod becomes the same as in the above considered case of
resonant swinging from rest; that is, it becomes favorable for the transfer of en-
ergy to the oscillator from the external source. As a result, the amplitude grows
indefinitely.
The time-dependent graph of the angle ϕ(t) and angular velocity ϕ̇(t) for this
process is shown in Figure 3.6. We note that during the initial stage of the process
the external torque that is proportional to θ(t) varies in the opposite phase with
respect to ϕ̇(t) and thus slows down the flywheel.
During the second stage the torque θ(t) varies in phase with the angular ve-
locity ϕ̇(t) and thus accelerates the flywheel – the energy is transferred from the
external source to the flywheel.

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

52 CHAPTER 3. FORCED OSCILLATIONS IN A LINEAR SYSTEM

o
100 .
ϕ (t ) ϕ (t )
θ ( t)
0

o
-100
0 1 2 3 4 5 6 7 8 9 10 T0

Figure 3.6: The transfer of energy from the excited oscillator to the external source
and back in conditions of exact tuning to resonance.

We can see here an analogue between the above considered processes of reso-
nant interaction of the mechanical oscillator with the periodic external source, and
the familiar optical phenomena of resonant absorption and stimulated emission of
radiation by atoms. The mechanical oscillator can be considered as an atom, or,
more precisely, as an optical electron linked to the atom by a quasi elastic force.
Such an electron can execute natural oscillations at a definite frequency. The ex-
ternal source that excites forced oscillations of the mechanical oscillator can be
regarded as an analog of the electromagnetic field of the light wave that interacts
with the atom. The energy can be transferred either to the atom from the elec-
tromagnetic field whose frequency equals the natural frequency of the electron
in the atom (more precisely, with the frequency corresponding to a transition of
the optical electron between atomic energy levels), or from the excited atom to
the electromagnetic wave which interacts with the atom. The first possibility cor-
responds to the resonant absorption of light by the atom; the second possibility
corresponds to the stimulated emission of radiation by the atom. Which one of the
two possibilities actually takes place depends on the phase relationships between
oscillations of the electromagnetic field in the light wave and oscillations of the
optical electron in the atom.
When the energy is transferred from the atom to the electromagnetic field, the
amplitude of the light wave increases with conservation of all other characteris-
tics of the wave (including its phase). This explains the coherent character of the
stimulated radiation emitted by a lot of excited atoms in the field of one and the
same electromagnetic wave. By virtue of this high coherence of the stimulated ra-
diation, lasers have their remarkable properties: The possibility to concentrate the
energy in the spectrum of radiation (to get a highly monochromatic emitted light);
the possibility to concentrate the emitted energy in space and in the direction of
propagation (to focus the radiation); and the possibility to concentrate the energy
in time (to generate extremely short light impulses).
This analogue is very useful for understanding optical phenomena, but we
should not take it too literally. The processes of absorption and emission of light
by atoms obey quantum laws, which can predict only the probabilities of different

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

3.3. TRANSIENT PROCESSES 53

o
140
ϕ (t )

o θ (t )
-140
0 1 2 3 4 5 6 7 8 9 10 11 12T

Figure 3.7: Gradually fading beats during a transient process near the resonant
frequency (ω = 0.8 ω0 ). The graphs show time dependencies for angular displace-
ments ϕ(t) of the flywheel and θ(t) of the exciter at the zero initial conditions.

results of interaction of light with an atom. Therefore the deterministic processes

in a mechanical oscillator can be compared to those in a statistical group of atoms
interacting with radiation, rather than to those in an isolated atom.

3.3.4 Transient Processes Near Resonance

If the driving frequency ω is near the frequency ω1 of damped natural oscilla-
tions, then during the transient process, while the natural oscillations have not yet
damped away, we observe the addition of two oscillations with slightly different
frequencies, ω and ω1 . As already mentioned, the frequency ω1 nearly equals the
natural frequency ω0 provided the friction is not too large (see Eq. (3.20), p. 45),
so that we need not distinguish between them here.
The oscillation resulting from this addition is modulated: its amplitude slowly
alternately increases and decreases with a beat frequency equal to the difference
|ω−ω0| between the driving and natural frequencies. The external force first drives
the oscillator to amplitudes that exceed the steady-state value; then the accumu-
lated phase shift between the oscillations of the flywheel and those of the driving
rod causes energy to flow back from the oscillator to the source of the external ac-
tion, and the amplitude of the flywheel decreases. These cycles of transient beats
(of slow variations of the amplitude) are repeated over and over until the damped
natural oscillations die out.
In the presence of viscous friction, the modulation of the amplitude is gradu-
ally diminished as damping decreases the contribution of the transient oscillations
with the natural frequency.
Figure 3.7 displays graphs of such fading transient beats for the case in which
the flywheel is initially at rest. In this example, four cycles of the driving force at
the frequency ω = 0.8 ω0 occur during five cycles of natural oscillations at the
frequency ω0 . (On the graph, the vertical hatch marks correspond to the external
period T .) Hence one period, Tb = 2π/|ω − ω0 |, of the beat cycle occurs during
four periods, T = 2π/ω, of the driven (steady-state) cycle and during five periods,
T0 = 2π/ω0 , of natural oscillations.

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

54 CHAPTER 3. FORCED OSCILLATIONS IN A LINEAR SYSTEM

32.1 o
ϕ (t )

0
θ (t )

o
-32.1
0.19
.
ϕ (t )

.
θ (t )
-0.19
0 1 2 3 4T

Figure 3.8: The angular displacement and angular velocity during a transient pro-
cess at a low driving frequency (ω = 0.2ω0 , θ0 = 30◦ , Q = 40, initial conditions
ϕ(0) = 0, ϕ̇(0) = 0).

In the absence of friction, steady-state oscillations with the frequency ω occur

in phase with the external force at ω < ω0 , or 180◦ out of phase at ω > ω0 . Their
amplitude a is determined by Eq. (3.8). The other contribution in the transient
process is given by oscillations with the natural frequency ω0 . At γ = 0 these nat-
ural oscillations are not damped. Thus, in this case an addition of two harmonic
oscillations with slightly different frequencies ω and ω0 and constant amplitudes
occurs after the external force is activated. For the zero initial conditions the am-
plitude of the natural oscillation equals −(ω/ω0 )a. The resulting beat oscillation
is pure in the sense that the beats do not decay. We can consider the beat oscilla-
tion as an oscillation with a mean frequency (ω + ω0 )/2 and a slowly periodically
varying amplitude. The envelope of the oscillations changes sinusoidally, period-
ically equaling zero. The transient process lasts indefinitely, and so there is no
steady-state oscillation in the absence of friction (at γ = 0).

3.3.5 Transient Processes Far from Resonance

Here we consider non-resonant cases in which the external frequency ω is much
different from the natural frequency ω0 .
If the external driving frequency is much less than the natural frequency of the
oscillator (ω ≪ ω0 ), the equilibrium position of the flywheel (in which the spring
is unstrained) slowly moves back and forth alongside the rod. Simultaneously,
the flywheel executes relatively rapid damped oscillations at its natural frequency
about this slowly moving equilibrium position. As a result, these gradually fading
rapid natural oscillations, superimposed on the slow steady-state forced oscilla-
tions of a constant amplitude, produce a pattern of motion like that shown by the

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

3.3. TRANSIENT PROCESSES 55

o
12.4
ϕ (t )

o
-12.4
0.38 .
ϕ (t )

-0.38
0 1 2 3 4 5 6 7 8 9 10 11 12T

Figure 3.9: The angular displacement and angular velocity during a transient pro-
cess at a high driving frequency (ω = 5.0ω0 , θ0 = 60◦ , Q = 10, initial conditions
ϕ(0) = 0, ϕ̇(0) = 0).

graph in Figure 3.8. When these natural oscillations die out, the plot evolves into a
pure, undistorted sine wave corresponding to steady-state oscillations whose fre-
quency is the slow driving frequency ω.
In the opposite case in which the exciting rod oscillates with a high frequency
(ω ≫ ω0 ), relatively rapid forced oscillations at the frequency ω and a constant
amplitude occur about a middle position, which, during the transient process,
slowly executes damped oscillations at the natural frequency ω0 . After these slow
oscillations have died out, only the rapid forced oscillations of a constant ampli-
tude remain. These rapid steady-state oscillations occur symmetrically about the
value θ = 0, i.e., about the central position of the driving rod. This case is il-
lustrated in Figure 3.9. Together with the graphs of ϕ(t) and ϕ̇(t), the graphs of
damped slow natural oscillations and of steady-state rapid oscillations that consti-
tute the transient are also shown. We note that the steady-state oscillations have
opposite phase with respect to oscillations of the driving rod.

3.3.6 Transient Processes and the Phase Trajectory

The equation of motion describing forced oscillations, Eq. (3.4), is explicitly time-
dependent. In this equation a given function of time, θ(t) = θ0 sin ωt, describes
the forcing periodic motion of the driving rod. Thus the mechanical state of the
system under consideration is determined by the three quantities: ϕ, ϕ̇, and t.
In order to display all of the characteristics of the mechanical state of an os-
cillator acted upon by a given time-dependent external force, we add a temporal
dimension to the phase plane (ϕ, ϕ̇). This temporal third dimension is introduced
by erecting a time axis perpendicular to the phase plane. In plotting the solution

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

56 CHAPTER 3. FORCED OSCILLATIONS IN A LINEAR SYSTEM

Figure 3.10: Phase trajectories with Poincaré sections for the transient at reso-
nance swinging of the oscillator (left), and for the gradually fading transient beats
at ω = 0.8 ω0 (right).

to Eq. (3.4), the computer simulation program traces the projection onto the plane
(ϕ, ϕ̇) of a twisted three-dimensional phase trajectory arising from forced oscil-
lations.
To obtain a clear graphic representation of the entire transient process, we
mark positions of the representative point on this phase trajectory at equal time
intervals, at the moments when the rod, moving from left to right, crosses the zero
point of the dial.
These points on the phase trajectory show the mechanical state of the system
at times equal to integer multiples of the period of oscillation, T = 2π/ω, of the
external force. Such points are called Poincaré sections. The simulation program
displays the phase trajectory with Poincaré sections (Figure 3.10) when you open
the window “Phase diagram” by using a relevant button on the control panel of
the program.
Since steady-state oscillations have the period of the external force and a con-
stant amplitude, the corresponding three-dimensional phase trajectory intersects
all the planes t = T, 2T, . . . , nT at the same values of ϕ and ϕ̇. Thus the pro-
jections of Poincaré sections onto the plane (ϕ, ϕ̇) at the times tn = nT coincide.
However, for a transient process, projected Poincaré sections form a set of
points in the plane (ϕ, ϕ̇) that condense gradually to the point ϕ = a sin δ, ϕ̇ =
aω cos δ, the projected Poincaré section for the steady-state oscillations.
At the resonant frequency (ω = ω0 ), the oscillation of the flywheel lags be-
hind that of the rod by a quarter cycle (δ = π/2), and the coordinates of the
limiting condensing point of the Poincaré sections for the transient process are
ϕ = −a, ϕ̇ = 0. If the initial angular deflection and angular velocity of the fly-
wheel are zero, the rod starts to move when the flywheel is at rest in the equilib-
rium position. Then all the projections of the Poincaré sections lie on the abscissa
axis of the phase plane, starting at the origin and gradually approaching the above-
mentioned condensing point, ϕ = −a, ϕ̇ = 0 (left-hand panel of Figure 3.10).
The right-hand panel of this figure corresponds to the process of transient beats at
ω = 0.8 ω0 , whose graph is shown in Figure 3.7, p. 53.

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

3.4. REVIEW OF THE PRINCIPAL FORMULAS 57

3.4 Review of the Principal Formulas

The differential equation of forced oscillations for the kinematic excitation:

ϕ̈ + 2γ ϕ̇ + ω02 ϕ = ω02 θ0 sin ωt. (3.31)

Steady-state forced oscillations are described by the particular periodic solution

of this inhomogeneous differential equation:

ϕ(t) = a sin(ωt + δ). (3.32)

The amplitude a and the phase shift δ of steady-state forced oscillations:

ω02 θ0 2γω
a(ω) = p 2 , tan δ = − . (3.33)
(ω0 − ω 2 )2 + 4γ 2 ω 2 ω02 − ω 2

The driving frequency ωres at which the amplitude of steady-state forced oscilla-
tions is a maximum is given by:

γ2
   
1
q
ωres = ω02 − 2γ 2 ≈ ω0 1 − 2 = ω0 1 − . (3.34)
ω0 4Q2
√
The amplitude of steady-state oscillations at resonance (for γ < 2 ω0 ) and its
approximate value for γ ≪ ω0 are given by:

ω 2 θ0 ω 0 θ0
amax = p0 ≈ = Qθ0 . (3.35)
2γ ω02 − γ 2 2γ

Here the quality factor Q is the same dimensionless quantity, Q = ω0 /(2γ), that
characterizes the damping of free oscillations.
The amplitude Ω of the angular velocity of steady-state oscillations:

ω02 θ0
Ω = ωa(ω) = p 2 . (3.36)
(ω0 /ω − ω)2 + 4γ 2
The maximal amplitude Ω of the angular velocity of steady-state oscillations at
resonance (for which ω = ω0 ) is given by:

ω02 θ0
Ωmax = ωa(ω0 ) = = ω0 Qθ0 . (3.37)
2γ
The initial conditions that eliminate the transient process are:

ϕ0 = a sin δ, ϕ̇(0) = aω cos δ, (3.38)

where a and δ are the amplitude and the phase of steady-state oscillations.
The transient process at the resonant frequency and zero initial conditions:

ϕ(t) = −Qθ0 (1 − e−γt ) cos ω0 t. (3.39)

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

58 CHAPTER 3. FORCED OSCILLATIONS IN A LINEAR SYSTEM

3.5 Questions, Problems, Suggestions

The simulation of forced oscillations of a torsion spring pendulum in the com-
puter program is based on the numerical integration of the differential equation
of motion, Eq. (3.4), by means of the Runge–Kutta method of the fourth order.
Although this linear differential equation can be solved analytically, the analytic
solution is not used in this simulation program. The agreement between the ana-
lytic theoretical predictions and the observed behavior of the linear oscillator in
these simulations is an indicator of the reliability of the numerical method. Thus
we get confidence in the reliability of the simulations of nonlinear systems in other
programs in the package “Physics of Oscillations,” because these simulations are
based on the same numerical method. This confidence is not without value since
analytic solutions are unavailable for nonlinear systems whose behavior in simu-
lations is often hard to reconcile with common sense.

3.5.1 Steady-State Forced Oscillations

In order to display steady-state oscillations without a preliminary transient pro-
cess (when the option “Show steady-state” is chosen), the simulation program
automatically sets the initial conditions to be ϕ(0) = a sin δ and ϕ̇(0) = aω cos δ
(overriding any initial conditions you may have entered). These initial conditions
provide steady-state oscillations immediately after the external force is activated.
The values of the amplitude a and the phase δ are calculated in the program from
Eq. (3.9) using the values of the external frequency ω and the quality factor Q you
have entered.
3.5.1.1 Steady-State Forced Oscillations without Friction.
(a) Setting properties of the system, choose full absence of friction. In this
case, the transient process lasts indefinitely, so that steady-state oscillations do
not establish. How can you explain the physical sense of the periodic analytical
solution that describes steady-state forced oscillations in the absence of friction?
Is this solution applicable to a real system? If so, what conditions must be satisfied
in order that it be possible in a real system to observe the motion described by this
analytical solution?
(b) Convince yourself that for driving frequencies less than the natural fre-
quency (ω < ω0 ), the steady-state sinusoidal oscillations of the flywheel occur
exactly in phase with the oscillations of the driving rod. At what frequency is
the amplitude of the flywheel twice that of the rod? Calculate this frequency and
verify your answer with a simulation experiment.
(c) Convince yourself that for driving frequencies greater than the natural fre-
quency (ω > ω0 ), the phase of the steady-state oscillations of the flywheel is
exactly opposite that of the rod. At what value of the driving frequency (ω > ω0 )
is the amplitude of the flywheel again twice that of the rod? At what driving fre-
quency are these amplitudes equal? At what frequency is the amplitude of the
flywheel one half that of the rod?

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

3.5. QUESTIONS, PROBLEMS, SUGGESTIONS 59

3.5.1.2∗ Transformations of Energy for Steady-State Oscillations.

(a) Using the plots of potential, kinetic, and total mechanical energy, find out
during which parts of the driving cycle is energy transmitted from the rod to the
oscillator. Give a physical explanation of this direction of energy transfer. Dur-
ing which parts of the cycle is energy transferred back from the oscillator to the
external source?
(b) For the kinematic excitation of forced oscillations, what is the ratio of the
average value of the potential energy to the average value of the kinetic energy for
each of those values of frequency for which the amplitude of the flywheel is twice
that of the rod? For the case when the amplitudes are equal? For the case when
the amplitude of the flywheel equals one half that of the rod? Compare the values
observed on the experimental plots with your calculated values.
3.5.1.3 The Amplitude and Phase of Steady-State Oscillations.
Examine steady-state oscillations in the presence of friction. Input some mod-
erate value of the quality factor, say, Q = 5.
(a) Evaluate the percentage shift of the resonant frequency from the natural
frequency.
(b) What is the ratio of the amplitude of steady-state oscillation of the flywheel
to the amplitude of the exciting rod at resonance?
(c) What is the phase lag of the oscillation of the flywheel relative to the phase
of the rod at the resonant frequency and at a driving frequency equal to 0.8 of the
resonant value?
Answer the same questions for Q = 20.
3.5.1.4∗∗ Peculiarities of the Kinematic Excitation. In the case of the dy-
namic excitation of oscillations by a given force whose value is independent of the
position of the flywheel, the ratio of the average potential energy to the average
kinetic energy equals (ω0 /ω)2 , so that for low frequencies the potential energy
predominates. For kinematic excitation, the ratio of the average energies is differ-
ent.
(a) Analyze the variations with time of both kinds of energy, and of the total
energy for the kinematic mode of excitation. Give a reasonable physical expla-
nation for these energy variations. Calculate the ratio of the average values of
potential energy to kinetic energy in this case.
(b) At what frequency of the external torque are the average values of the
potential and kinetic energy equal to each other?
(c) In the case of dynamic excitation, mean values of both kinds of energy
at resonance are equal to one another, and their changes occur exactly in op-
posite phase, so that total mechanical energy remains constant. However, in the
kinematic mode of excitation of forced oscillations, total mechanical energy is
subjected to variations even at resonance. Explain these variations. Calculate how
much the maximal and minimal values of the total energy differ from its average
value (in percent).
3.5.1.5∗∗ Steady-State Oscillations at Various Frequencies.
(a) Let the driving frequency ω of the rod be a little less than the natural fre-

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

60 CHAPTER 3. FORCED OSCILLATIONS IN A LINEAR SYSTEM

quency ω0 , say, ω = 0.9 ω0 , and let the value of Q be 5. What is the ratio of
the amplitude of the steady-state oscillations to the resonant amplitude? What is
the phase lag of the oscillation of the flywheel relative to the phase of the rod (in
fractions of a cycle)?
(b) At what values of the driving frequency (on either side of the resonant
frequency) is the amplitude of steady-state oscillations one half of the resonant
amplitude? What is the corresponding phase lag in each of these cases? What
kind of energy (averaged over a cycle) predominates in each of these cases?
(c) At what driving frequency ω are the amplitudes of the flywheel and of the
rod equal? What is the phase lag of the flywheel relative to the phase of the rod in
this case?
3.5.1.6∗∗ Half-Width of the Resonance Curve.
(a) Examine analytically and experimentally the dependence of the resonant
amplitude of steady-state oscillations on the value of the quality factor Q. How
does the half-width of the resonance curve depend on the quality factor Q? (The
half-width is the interval of driving frequencies within the limits of which the
amplitude of steady-state oscillations exceeds one half of its maximal value.) In
other words, how does the sharpness of the resonance peak change when damping
is increased?
Make the necessary calculations and then verify your answers by simulating
the appropriate experiments on the computer.
(b) How does the position of the resonance peak change as damping is in-
creased? At what value of Q is this maximum shifted to a frequency of zero? (In
this case the effect of a static external force applied to the system exceeds that of
a driving force oscillating at any frequency.)
3.5.1.7∗∗ Power Absorbed and Dissipated. Lorentzian.
(a) Prove analytically that for steady-state forced oscillations the power re-
ceived by the oscillator from the external source, averaged over a period, equals
the averaged value of energy dissipated by friction. Use the values of amplitude a
and phase δ for steady-state oscillations, expressed by Eq. (3.9).
(b) Show that the spectral distribution of power absorbed by the oscillator with
weak friction for steady-state oscillations is described by the function
1
F (ω) = ,
1 + (ω − ω0 )2 τ 2
where τ = 1/γ. (This function is encountered in various problems of physics. It
is called Lorentzian.)

3.5.2 Transient Processes

3.5.2.1∗ Initial Conditions That Eliminate a Transient.
(a) Under certain initial conditions there is no transient term. That is, imme-
diately after the external driving force is activated, the oscillator executes steady-
state oscillations with a constant amplitude at the driving frequency. What are

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

3.5. QUESTIONS, PROBLEMS, SUGGESTIONS 61

these initial conditions? Express the angle of deflection and the angular velocity
corresponding to these initial conditions in terms of the quality factor Q of the
oscillator and the parameters of the external action – the driving frequency ω and
the amplitude θ0 of oscillations of the rod.
(b) Letting Q = 5, ω = ω0 , and θ0 = 15◦ , calculate the initial angle of
deflection and the initial angular velocity for which there is no transient. Enter
your calculated values and verify experimentally that the forced oscillations are
purely sinusoidal from the beginning of the motion. Repeat the same procedure
for different values of the system parameters: Q = 25, ω = 0.5 ω0 , θ0 = 15◦ , and
say, Q = 25, ω = 1.5 ω0 , θ0 = 15◦ .
3.5.2.2∗ Transient Processes at Resonance.
Examine transient processes at resonance experimentally. Enter the follow-
ing values for the parameters: Q = 5, ω = ω0 , θ0 = 15◦ , and enter the initial
conditions ϕ(0) = 0, ϕ̇(0) = 0.
(a) Calculate the lapse of time, measured in units of the period, during which
the amplitude reaches 90% of its steady-state value. Verify your answer experi-
mentally. Note the monotonic growth of the amplitude and its exponential asymp-
totic approach to its steady-state value.
Analyze the character of energy conversions using the graphs of the kinetic,
potential, and total energy. At what instants of time is the growth in the total energy
of the oscillator a maximum?
(b) Carefully examine the graphs of the decomposition of the resonant tran-
sient process into its simple component parts (the transient term and the steady-
state oscillation). Note especially the exponential damping of the transient term.
Why does the initial value of the amplitude of this natural oscillation equal the
amplitude of the steady-state oscillation?
(c)∗ Taking into account the analytic expression, Eq. (3.26) for ϕ(t) corre-
sponding to the resonant case (ω = ω0 ), predict the behavior of the Poincaré
sections in the phase plane. Verify your prediction by a simulation experiment.
(d)∗ Consider a transient process at resonance in the absence of friction. How
does the amplitude of oscillations increase with time during the transient process
that begins from the state of rest in the equilibrium position?
(e)∗∗ During a transient process, is it possible for the amplitude to decrease if
the frequency of the external force is exactly the resonant frequency? Give physi-
cal arguments for your answer. Can you prove your answer experimentally?
3.5.2.3∗ Transient Processes Near Resonance.
Explore transient processes near resonance. Let the frequency of the external
force ω be equal to, say, 1.2 ω0 (ω0 is the natural frequency of the oscillator).
Consider first of all behavior of the system in the absence of friction when the
initial conditions are zero. The left-hand panel of Figure 3.11 shows the phase
diagram with Poincaré sections, and the right-hand panel shows the graphs of
ϕ̇(t) and ϕ(t).
(a)∗ Calculate the amplitude of the transient term, that is, the amplitude of
oscillations with the natural frequency contributing to the transient process. Also

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

62 CHAPTER 3. FORCED OSCILLATIONS IN A LINEAR SYSTEM

1.43
.
ϕ (t )
θ ( t)
0

-1.43 o o
-74.2 0 74.2
ϕ (t )
θ ( t)
0

o
- 74.2
0 2 4 6 8 10 12T

Figure 3.11: Beats in the absence of friction during forced oscillations at driving
frequency ω = 1.2 ω0 .

calculate the amplitude of the angular velocity of these oscillations. What is the
ratio of the amplitude of the transient term to the amplitude of steady-state forced
oscillations? Verify your answers experimentally using the option “Decomposi-
tion of the Transient Process.”
(b) Through how many cycles does the rod oscillate before the amplitude of
oscillation of the flywheel reaches its maximal value? What is the lapse of time
between successive moments at which the amplitude is zero? In other words, what
is the beat period?
(c)∗∗ Calculate the maximal values of the angular velocity and of the angular
displacement during the beats. What is the ratio of the maximal amplitude of the
flywheel to the amplitude of the rod?
(d)∗∗ Note the distribution of the Poincaré sections of the phase trajectory for
this process of beats in the absence of friction (Figure 3.12). Explain this distribu-
tion.
(e)∗∗ Consider the effect of friction by entering a moderate value of the quality
factor Q (20−25), while keeping the previous values of the remaining parameters.
What changes in the behavior of the system do you expect? What is the corre-
sponding distribution of the Poincaré sections in the phase plane? (Several initial
loops of the phase trajectory are shown in Figure 3.12.) Follow behavior of the
Poincaré sections for as long a time as is needed for the steady-state oscillations
to establish. Explain the distribution of the Poincaré sections displayed.
(f)∗∗ Change the driving frequency by a small amount. For example, let ω be
1.19 ω0 or 1.21 ω0 instead of 1.20 ω0 . What changes in the transient process does
this change in the frequency cause? What are the corresponding distributions of
the Poincaré sections in the phase plane if there is no friction and if there is weak
friction? What are the reasons for the differences in appearance of the display

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

3.5. QUESTIONS, PROBLEMS, SUGGESTIONS 63

1.32
.
ϕ (t ) θ ( t)
0

-1.32 o
-68.7 o 0 68.7
ϕ (t ) θ ( t)
0

o
- 68.7
0 2 4 6 8 10 12 T

Figure 3.12: Fading of beats caused by weak friction (Q = 50) during forced
oscillations at ω = 1.2 ω0 .

from the case in which ω = 1.2 ω0 ?

(g)∗ Predict the distribution of Poincaré sections for ω = 1.25ω0 . Verify your
prediction by the simulation.
3.5.2.4∗ Transient Processes Far from Resonance.
Investigate transient processes far from resonance. Let the frequency of the
driving rod be, say, four times smaller than the natural frequency of the oscillator:
ω = 0.25 ω0 . Set the two initial conditions to zero.
(a) What are the oscillations like during the transient process if there is no
friction? Calculate the amplitudes of the two superimposed oscillations, one with
the natural frequency ω0 and the other with the driving frequency ω. What are the
the two amplitudes of the corresponding angular velocities?
(b) Observe and explain the shape of the phase trajectory and the distribution
of Poincaré sections.
(c) Next introduce moderate friction by setting Q to be approximately 15,
keeping the values of the other parameters. Note the gradual fading of the contri-
bution of the natural oscillations. (Use the option “Decomposition of the Transient
Process” to see how the time dependencies of these contributing simple oscilla-
tions are plotted.) Observe how the complex phase trajectory plotted during the
early stage of the transient process is transformed into the ellipse corresponding
to steady-state oscillations. Explain the shape of the phase trajectory and the dis-
tribution of Poincaré sections (Figure 3.13a).
(d)∗∗ Consider the opposite case of a driving force with a high frequency. For
example, let the driving frequency be, say, four times greater than the natural fre-
quency of the oscillator: ω = 4 ω0 . Set the two initial conditions to zero, and
let friction be zero. What is the ratio of the amplitudes of the two superimposed
oscillations with the frequencies ω and ω0 ? By how many times does the maxi-

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

64 CHAPTER 3. FORCED OSCILLATIONS IN A LINEAR SYSTEM

a b

Figure 3.13: Several loops of the phase trajectory and the succession of Poincaré
sections for transients at ω = 0.25 ω0 (a) and ω = 4 ω0 (b). The thick lines show
the ellipses which correspond to steady-state oscillations.

mal deflection during the transient process exceed the amplitude of steady-state
oscillations, contributing to the transient?
(e)∗∗ Compare the shapes of the phase trajectories and the distributions of the
Poincaré sections for the two cases, ω = 0.25 ω0 and ω = 4 ω0 , in the absence
of friction. How can you explain the similarity in the phase trajectories for these
cases?
(f)∗ Introduce weak friction for the case in which ω = 4 ω0 . Observe the
way in which frequent steady-state forced oscillations are established while the
contribution of slow transient natural oscillation gradually fades away. How does
the phase trajectory evolve in this case? What is the behavior of the Poincaré
sections during the corresponding transient process?
(g)∗∗ Repeat the simulation experiments for other values of the driving fre-
quency, say ω = 0.125 ω0, ω = 0.5 ω0 , ω = 1.5 ω0 , ω = 2 ω0 , ω = 6 ω0 ).
Explain the peculiarities of the transient processes for these cases.

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

Chapter 4

Square-Wave Excitation of a
Linear Oscillator

Annotation. Chapter 4 deals with forced oscillations in a linear system driven by

a non-sinusoidal external force, namely by a force with square-wave time depen-
dence. It includes a description of the simulated mechanical physical system and
its electromagnetic analogue, and a summary of the relevant theoretical material
for students as a prerequisite for the virtual lab “Square-wave Excitation of Linear
Oscillator.” Chapter 4 also includes a set of theoretical and experimental prob-
lems to be solved by students on their own, as well as various assignments that the
instructor can offer students for possible individual mini-research projects.

4.1 Theoretical Background

This chapter deals with forced oscillations of a torsion spring pendulum excited
by an external square-wave driving torque. Two different ways of determining
the steady-state response of the oscillator to a non-harmonic driving force are
described and compared. Behavior of this familiar mechanical system can help
a student to better understand why and how an electromagnetic oscillatory LCR-
circuit transfers the square-wave voltage from input to output with a distortion of
its shape.

4.1.1 Model of the Physical System

Forced oscillations in a linear system under a sinusoidal driving force are con-
sidered in Chapter 3. Most textbooks on general physics treat this case rather ex-
tensively. The general case of a periodic but non-sinusoidal excitation of a linear
oscillator is usually only mentioned with reference to the principle of superposi-
tion and an expansion of an arbitrary periodic force as the Fourier series of sine

65

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

66 CHAPTER 4. SQUARE-WAVE EXCITATION OF AN OSCILLATOR

- q0 +q 0
q (t)
L
j (t)
Vin C Vout

Figure 4.1: Schematic image of the torsion spring oscillator (left) and its elec-
tromagnetic analogue—LCR-oscillatory circuit excited by the square-wave input
voltage (right).

and cosine functions. In this chapter an alternative approach to the problem of

forced oscillations is suggested and compared with the traditional treatment.
To study forced oscillations caused in a linear system by a non-sinusoidal pe-
riodic external influence, we employ a simplified model of a torsion spring oscil-
lator. Its schematic image is shown in the left panel of Figure 4.1. The oscillator is
similar to the balance devices of ordinary mechanical watches—a balanced mas-
sive rotor (flywheel) attached to one end of an elastic spiral spring. A mechanical
system such as this one is ideal for the study of resonance because it is possible
to see directly what is happening. The spring provides a restoring torque propor-
tional to the angular displacement of the flywheel from the equilibrium position.
To provide an external excitation, the other end of the spiral spring is attached to a
driving rod (exciter) that can be turned about the axis common with the axis of the
flywheel. When the rod is constrained to move periodically to and fro about some
middle position, an additional periodic torque is exerted on the flywheel. This
mode of excitation is called kinematical because it is characterized by a given
motion of some part of the system rather than by a given external torque.
When the driving rod is turned, the equilibrium position of the flywheel is
displaced alongside the rod through the same angle. The flywheel can execute
damped natural oscillations about this new displaced equilibrium position. An ex-
ternal piecewise constant torque whose shape is that of a periodic square-wave can
be realized by abruptly displacing the driving rod alternately in opposite directions
through the same angle in equal time intervals. We suppose that the displacements
of the rod and thus of the equilibrium position of the flywheel occur so quickly
that there is no significant change in either the angular position or velocity of the
flywheel during the abrupt displacement of the driving rod.
The right-hand panel of Figure 4.1 shows an LCR-oscillatory circuit that can
be regarded as an electromagnetic analogue of the mechanical device. Both sys-
tems are described by identical differential equations and thus are dynamically

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

4.1. THEORETICAL BACKGROUND 67

isomorphic. However, the mechanical system has a definite didactic advantage for
exploration of forced oscillations because it allows us to observe a direct visual-
ization of motion.

4.1.2 The Differential Equation of Forced Oscillations

When the system is at rest, we let the needle attached to the flywheel be parallel to
the driving rod. In other words, the spring is not strained when the needle points
in the same direction as does the rod. The zero point of the dial indicates the
central position of the exciting rod (the vertical position in Figure 4.1). The angle
of deflection ϕ of the needle from this zero point indicates the position of the
flywheel. When the rod is deflected from the vertical position through an angle θ,
the spiral spring is twisted from its unstrained state through the angle ϕ − θ. The
spring then exerts a torque −D(ϕ − θ) on the flywheel, where D is the torsion
spring constant. Thus, the differential equation of rotation of the flywheel, whose
moment of inertia about the axis of rotation is J, is given by
J ϕ̈ = −D(ϕ − θ). (4.1)
We transfer −Dϕ (the part of the elastic torque that is proportional to ϕ) to the
p of Eq. (4.1), divide the resulting equation by J, and introduce the value
left side
ω0 = D/J, whose physical meaning is the frequency of natural oscillations in
the absence of friction. Thus we obtain:
ϕ̈ + ω02 ϕ = ω02 θ. (4.2)
The right-hand side ω02 θ of this equation can be treated as an external torque
(divided by J ) caused by the displacement of the rod from its central position
through an angle θ. We let the instantaneous displacements of the rod occur alter-
nately to the right and to the left after the lapse of equal time intervals T /2, so that
during the interval (0, T /2) the equilibrium position of the flywheel is displaced
to the right through a fixed angle θ0 , and during the next interval (T /2, T ) the
equilibrium position is displaced to the left through the same angle (the angle θ
in Eq. (4.2) equals −θ0 ). Thus, T is the full period of the external non-sinusoidal
action, repeated indefinitely. In the presence of viscous friction, a term 2γ ϕ̇ pro-
portional to the angular velocity ϕ̇ should be added to Eq. (4.2), in which the
damping constant γ characterizes the strength of viscous friction in the system:
ω02 θ0 ,

2 (0, T /2),
ϕ̈ + 2γ ϕ̇ + ω0 ϕ = (4.3)
−ω02 θ0 , (T /2, T ).
Forced oscillations of the electric charge q stored in a capacitor of a resonant
series LCR-circuit (see the right-hand panel of Figure 4.1) excited by a square-
wave input voltage Vin (t) obey the same differential equation as does the forced
oscillation of a mechanical torsion spring oscillator excited by periodic abrupt
changes of position of the driving rod:
q̈ + 2γ q̇ + ω02 q = ω02 CVin (t). (4.4)

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

68 CHAPTER 4. SQUARE-WAVE EXCITATION OF AN OSCILLATOR

In this equation, ω0 is the natural frequency of oscillations of charge in the circuit

in the absence of resistance. It depends on √ the capacitance C of the capacitor
and the inductance L of the coil: ω0 = 1/ LC. The damping constant γ =
R/(2L) characterizes the dissipation of electromagnetic energy occurring in a
resistor whose resistance is R.
Because of this similarity, the mechanical system described above enables us
to give a clear explanation for transformation of the square-wave input voltage
Vin (t) = ±V0 into the output voltage Vout (t) = VC (t) = q/C (voltage across
the capacitor C), whose time dependence differs considerably from the piecewise
constant input voltage. The output voltage VC (t) is analogous to the deflection
angle ϕ(t) of the rotor, while the alternating electric current I(t) = q̇(t) in the
circuit is analogous to the rotor angular velocity ϕ̇(t) of the mechanical model.
However, some caution is necessary in interpreting the analogy between the me-
chanical oscillator and the electric LCR-circuit with respect to the energy trans-
formations. We note that in using the mentioned analogy it would be incorrect
to exactly associate the electric potential energy of a charged capacitor with the
elastic potential energy of a strained spring because the latter depends directly on
the angle (ϕ ± θ0 ), while the energy of a capacitor depends directly on the charge
q or on the corresponding voltage VC = q/C (not on the voltage (VC ± V0 )). In
contrast to the spring oscillator, for which a jump in the position of the rod causes
an abrupt change in the elastic potential energy of the spring, a jump of the in-
put voltage across an electric circuit does not abruptly change the charge and the
energy of a capacitor.

4.2 Steady-State Forced Oscillations under the

Square-Wave Torque
Because of friction, natural oscillations of the flywheel gradually damp out, and
a while after the external force began to act, a steady-state periodic motion of
the flywheel is eventually established with a period equal to the period T of the
driving force. The greater the decay time, τ = 1/γ, of natural oscillations, the
longer the duration of this transient process.
In the case of a sinusoidal driving torque, the steady-state oscillations of the
flywheel acquire not only the period of the external action but also the sinusoidal
time dependence whose frequency equals the driving frequency. However, a peri-
odic driving force, whose time dependence is something other than a pure sinu-
soid, produces a steady-state response that has the same period but whose time
dependence differs from that of the driving force.

4.2.1 Harmonics of the Driving Force and of the Steady-State

Response
We consider below two different ways of determining the steady-state response
of the oscillator to a non-harmonic driving force, such as the square-wave time-

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

4.2. STEADY-STATE FORCED OSCILLATIONS 69

dependent force discussed above. One (traditional) way is based on expansion of

the time dependence of the external force in a Fourier series, i.e., on the repre-
sentation of this force as a superposition of sinusoidal components (harmonics).
Because the differential equation of motion for the spring oscillator is linear, the
influence of each of these harmonic components of the external force can be con-
sidered independently. Each sinusoidal component of the driving torque produces
its own sinusoidal response of the same frequency in the motion of the flywheel.
The amplitude and phase of each sinusoidal response can be calculated separately.
The corresponding formulas are the same as for the familiar case of monohar-
monic excitation.
The net steady-state forced motion of the flywheel can be found as the su-
perposition of these individual responses. Thus, to each sinusoidal component of
the periodic driving force (to the input harmonic), there corresponds a sinusoidal
component of the same frequency in the steady-state motion of the responding
oscillator (we can call it the output harmonic). Since the relative contributions
of harmonic components to this response differ from the corresponding contri-
butions to the driving force, the graph of the net motion of the flywheel has a
different shape than does the graph of the motion of the driving rod.
In particular, it may occur that one of the input harmonics with relatively small
amplitude induces especially large amplitude in the output oscillations. Such is the
case when the frequency of this harmonic is close to the natural frequency ω0 of
the oscillator since forced oscillations caused by this sinusoidal force occur under
conditions of resonance. On the other hand, the relative contributions of the input
harmonics whose frequencies lie far from the maximum of the resonance curve
are considerably attenuated in the output oscillations. In the output steady-state
oscillations such harmonics are appreciably suppressed. The oscillator responds
selectively to sinusoidal external forces of different frequencies. The phenomenon
of resonance depends upon the whole functional form of the driving force and
occurs only if its spectrum contains the component whose frequency is close to
the natural frequency of the oscillator.
Differences between the time dependence of output steady oscillations and
that of the input driving force (distortions of the signal from input to output) are
caused not only by changes in the relative amplitudes of different harmonics but
also by changes in their phases from input to output. In the case of weak damping
the resonance curve (amplitude versus frequency) is very sharp, and the depen-
dence of phase on frequency is nearly a step-function. Specifically, all harmonic
components whose frequencies ωk = kω = (2π/T )k (T – driving period) are
lower than the natural frequency ω0 contribute to the output oscillations of the
flywheel nearly in the same phases as they do to the input driving force. But har-
monics whose frequencies ωk are higher than the natural frequency contribute to
the output oscillations with nearly inverted phases: in the Fourier expansion of
the output steady-state oscillations, their phases are almost opposite to the phases
of the corresponding harmonics of the driving force. If there is a sinusoidal com-
ponent in the driving force whose frequency lies close to ω0 , this harmonic pro-
duces a significantly increased relative contribution to the output oscillations of

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

70 CHAPTER 4. SQUARE-WAVE EXCITATION OF AN OSCILLATOR

q (t)
q0

j (t)

q0
t

Figure 4.2: Transformation of the spectrum of the input square-wave external

torque into the spectrum of steady-state output oscillations (see text for detail).

the flywheel. This output harmonic component lags in phase by π/2 behind the
corresponding input harmonic in the spectrum of the driving force.
The analytic expression for Eq. (4.3) for which the square-wave shape of its
right-hand side has been Fourier decomposed has the following form:
∞
X 4θ0 ω02
ϕ̈ + 2γ ϕ̇ + ω02 ϕ = sin ωk t. (4.5)
πk
k=1, 3, 5...

The Fourier series of the piecewise-constant external force in Eq. (4.3) con-
tains only odd-number harmonics with frequencies ωk = kω (k = 1, 3, 5, . . . ),
where ω = 2π/T , the frequency of the driving force. We note that the amplitudes
of harmonics of the square-wave function decrease rather slowly, as 1/k, with the
increase of their index k and their frequency ωk . This case is a good example of a
multi-harmonic external excitation of the oscillator since the frequency spectrum
of the square-wave driving force is rich in harmonics.
Figure 4.2 illustrates the transformation of the input spectrum of an external
square-wave periodic force into the output spectrum of the steady-state response
of the oscillator. The upper panel of Figure 4.2 shows the square-wave graph of
θ(t) that corresponds to the exciter’s motion, and harmonics of θ(t) from the first
up to the eleventh. The graph of the sum of all these harmonics, which approxi-
mates the square-wave motion of the exciter, is also shown.

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

4.2. STEADY-STATE FORCED OSCILLATIONS 71

T = 3T0
4
Q=5
3

0 2 4 6 8 10 w
w0

Figure 4.3: The resonance curve, which shows how the amplitudes of separate
odd-numbered harmonics of the input square-wave motion of the exciter (whose
period T equals 3T0 ) are transformed by the oscillator with Q = 5 into the am-
plitudes of the output harmonics of the flywheel steady-state oscillations (see text
for details).

The lower panel of Figure 4.2 shows the same harmonics in the output steady-
state motion of the flywheel ϕ(t). For each sinusoidal term (each input harmonic)
in the right-hand side of Eq. (4.5), the amplitude and phase of the periodic si-
nusoidal particular solution (of the output harmonic) are given by the commonly
known expression (see Eq. (3.9), p. 40 in Chapter 3). Figure 4.3 illustrates how the
amplitudes of separate odd-numbered harmonics of the input square-wave motion
of the exciter (whose period T equals 3T0 ) are transformed by the oscillator with
Q = 5 into the the amplitudes of the output harmonics of the flywheel steady-state
oscillations. The resonance curve shows how the oscillator with certain given pa-
rameters responds to the individual harmonic components of the external force.
Since for the case shown in Figure 4.3 the driving period equals three natural peri-
ods, the third harmonic of the square-wave force occurs under the maximum of the
resonance curve, and this harmonic component dominates the output spectrum.
Adding solutions that correspond to all input harmonics, we get the follow-
ing time dependence of the angular displacement, ϕ(t), for steady-state forced
oscillations under the square-wave excitation:
∞
X 4θ0 ω02
ϕ(t) = sin(ωk t + αk ), (4.6)
πk (ω02 − ωk2 )2 + 4γ 2 ωk2
p
k=1, 3, 5...

where the phases αk of the individual harmonics are determined by

2γωk
tan αk = . (4.7)
ωk2 − ω02
Equations (4.6) and (4.7) display clearly the above-discussed peculiarities of
the oscillator response to the square-wave driving action of the exciting rod. A

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

72 CHAPTER 4. SQUARE-WAVE EXCITATION OF AN OSCILLATOR

resonant response from the oscillator occurs each time the denominator in one of
the terms of the sum in Eq. (4.6) is minimal, that is, when the frequency ωk of one
of the harmonics of the external force is equal to the resonant frequency ωres of
the oscillator:
γ2
q  
ωres = ω02 − 2γ 2 ≈ ω0 1 − 2 .
ω0
The latter approximate expression for ωres is valid for a weakly damped oscillator
(γ ≪ ω0 ), whose quality factor Q is large (Q = ω0 /2γ ≫ 1). Since the frac-
tional difference between ωres and ω0 is of the second order in the small parameter
γ/ω0 = 1/(2Q), in most cases of practical importance we need not distinguish
the resonant frequency from the natural one and can assume that ωres = ω0 . For
ωk < ω0 Eq. (4.7) yields αk ≈ 0, which means that the corresponding harmonic
contributes to the output oscillations in the same phase as to the input square-wave
force. Conversely, for ωk > ω0 Eq. (4.7) yields αk ≈ −π, and this harmonic com-
ponent enters into the output oscillations with the inverted phase.
When the frequency of the sinusoidal external force is slowly varied, the res-
onant steady-state response of the oscillator can occur at only one value of the
driving frequency ω, namely ω = ωres , the resonant frequency of the oscillator. In
other words, in the case of sinusoidal excitation there is only one resonance, and
it occurs when the driving period T equals the natural period T0 of the oscilla-
tor. However, in the case of the square-wave excitation, resonance occurs not only
when the periods are equal, but each time the driving period T is an odd-number
multiple of the natural period T0 of the oscillator, that is, when T = (2n + 1)T0 ,
where n = 0, 1, 2, . . . Resonances, for which n ≥ 1, occur when the fre-
quency of one of the odd harmonics of the square-wave driving torque approaches
the resonant frequency of the oscillator. Each resonance corresponds to a definite
harmonic in the input spectrum (spectrum of the square-wave force).
Generalizing, we note that a linear oscillator with a sharp resonance curve
(and given resonant frequency) appreciably responds only to a certain single har-
monic component of an arbitrarily complex external force. In this respect such an
oscillator can be regarded as a spectral instrument, which selects a definite spec-
tral component of an external action. That is, if we cause to “sweep” the natural
frequency of an oscillator through a range of frequencies, such an oscillator re-
sponds resonantly to the complex external input each time its natural frequency
coincides with one of the harmonic frequencies in the Fourier expansion of the
external force. In other words, a sweep-frequency oscillator with a large quality
factor provides us with a means by which a complex periodic input can be physi-
cally decomposed into its Fourier components.
The mathematical representation of the square-wave function by a series on
the right-hand side of Eq. (4.3) is not unique. The function can be represented as
a sum of other functions in many different ways. That is, it is possible to express
the external action either as a Fourier series of sine and cosine functions or as a
series of other complete sets of functions. From the mathematical point of view,
all such expansions are equally valid. The usefulness of the Fourier expansion in
the case under consideration is associated with physics. It is related to the capa-

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

4.2. STEADY-STATE FORCED OSCILLATIONS 73

bility of a linear harmonic oscillator to perform this expansion physically. When

the phenomenon of resonance is used as a means of experimental investigation,
only the Fourier representation of the analyzed complex process is adequate and
expedient.

4.2.2 Forced Oscillations as Natural Oscillations about the

Alternating Equilibrium Positions
Another way to obtain an analytic solution to the differential equation of mo-
tion (4.3) for steady-state oscillations forced by the square-wave external torque is
based on viewing the steady-state motion as a sequence of free oscillations, which
take place about an equilibrium position that periodically alternates between +θ0
and −θ0 . During the first half-cycle, from t = 0 to T /2, the equilibrium position
is located at ϕ = +θ0 . For this half-cycle the general form of the dependence of
ϕ(t) on t can be written as:

ϕ(t) = θ0 + Ae−γt cos(ω1 t + α), (0, T /2), (4.8)

p
where ω1 = ω02 − γ 2 is the frequency of damped natural oscillations, A and
α are arbitrary constants of integration determined by conditions at the beginning
of the half-cycle. The graphs of ϕ̇(t) and ϕ(t) for this stage of motion (from
t = 0 to T /2) are shown in the right-hand panel of Figure 4.4 by segments of
curves between points marked 1 and 2. The corresponding segment of the phase
trajectory is shown in the left-hand panel of this figure between points 1 and 2.
The graphs correspond to the case for which T = 3T0 (that is, when the period T
of the driving force is set equal to three natural periods T0 ).

1.47 .
ϕ (t ) θ ( t)

0 13 2 1 2 3

-1.47 o
o 101
-101 0 ϕ (t )
2 θ ( t)

o 1 3
-101
0 1 2T

Figure 4.4: Phase diagram, graphs of the time dependence of the angular velocity
ϕ̇(t) and the deflection angle ϕ(t) at resonant steady-state oscillations for T =
3T0 , Q = 5, θ0 = 30◦ .

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

74 CHAPTER 4. SQUARE-WAVE EXCITATION OF AN OSCILLATOR

During the next half-cycle (T /2, T ), damped natural oscillations occur about
an equilibrium position displaced through the same angle θ0 but in the opposite
direction. For this half-cycle the time dependence of ϕ(t) has the form:

ϕ(t) = −θ0 − Ae−γ(t−T /2) cos(ω1 (t − T /2) + α), (T /2, T ), (4.9)

where the constants A and α have the same values as they do in Eq. (4.8). These
values follow from the fact that, in steady-state oscillations, the graph of time
dependence during the second half-cycle (when the driving rod is displaced to
the left) must be the mirror image of the graph for the first half-cycle, shifted by
T /2 along the time axis. This relationship can clearly be seen from Figure 4.4, in
which the graphs of ϕ̇(t) and ϕ(t) for this stage of motion (from t = T /2 to T )
are shown in the right-hand panel by segments of curves between points marked
2 and 3. The corresponding segment of the phase trajectory is shown in the left-
hand panel of this figure between points 2 and 3 (the latter point coincides with
the initial point 1).
The constants A and α for any given values of T , θ0 , and γ can be calculated
from the condition that during the instantaneous change in the positions of the
driving rod at t = T /2, from one equilibrium position to the other, the angular
deflection and the angular velocity of the flywheel do not change. In other words,
we should equate the right-hand sides of Eqs. (4.8) and (4.9) and their time deriva-
tives at t = T /2. These conditions give us two simultaneous equations for A and
α. Solving the equations we find:

e−γT /2 [ω1 sin(ω1 T /2) + γ cos(ω1 T /2)] + γ

tan α = − (4.10)
e−γT /2 [ω1 cos(ω1 T /2) − γ sin(ω1 T /2)] + ω1
and
2θ0
A=− . (4.11)
e−γT /2 cos(ω1 T /2 + θ) + cos α
Equations (4.8)–(4.11) describe the steady-state motion only during the time
interval from 0 to T . That is, if we substitute a value of t greater than T into these
equations, they do not give the correct value for ϕ(t). Nevertheless, we can find
the value of ϕ(t) for an arbitrary t by taking into account that ϕ(t) is a periodic
function of t: ϕ(t + T ) = ϕ(t). Thus, having obtained the graph of ϕ(t) for
the time interval [0, T ], we can simply translate the graph to the adjacent time
intervals [T, 2T ], [2T, 3T ], and so on.
The treatment of forced oscillations as natural oscillations about alternating
equilibrium positions provides especially clear explanation of a rather complex
behavior of the oscillator under the square-wave force whose period is consider-
ably longer than the natural period. Figure 4.5 shows the phase diagram (left),
and the plots of ϕ̇(t) and ϕ(t) (right) for the steady-state forced oscillations at
T = 7T0 and relatively strong friction (Q = 3). The segments of the phase tra-
jectory are marked in the left-hand panel of this figure by points 1, 2, and 3 (the
latter point coincides with the initial point 1), which correspond to the segments
of the graphs of ϕ̇(t) and ϕ(t) on the right-hand side, marked by the same figures.

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

4.3. TRANSIENT PROCESSES 75

0.85 .
ϕ (t ) θ ( t)

0 2
1 2 3
1 3

-0.85 o o
-66.1 0 66.1 ϕ (t ) θ ( t)
2
0

1 3
o
- 66.1
0 1 2T

Figure 4.5: Damped oscillations about alternating displaced equilibrium positions

at resonant steady-state oscillations for T = 7T0 , Q = 3, θ0 = 30◦ .

We see clearly how after each, in turn, abrupt displacement of the driving rod, the
flywheel makes several natural oscillations of gradually diminishing amplitude
about the new equilibrium position. These natural oscillations replace both abrupt
fronts of each rectangular impulse, thus distorting its shape from input to output.

4.3 Transient Processes under the Square-Wave

External Torque
The above treatment of forced oscillations excited by a square-wave external
torque as natural oscillations about alternating equilibrium positions helps us to
understand many characteristics of both steady-state oscillations and transient pro-
cesses. In particular, it enables us to clearly understand the physical reason for the
resonant growth of the amplitude when the period of the driving force equals the
natural period of the oscillator or when it equals some odd-number multiple of
that period.
Suppose that before the external square-wave torque is applied, the oscillator
has been at rest in its equilibrium position, ϕ = 0. When, at t = 0, the driving
rod abruptly turns into a new position, θ0 , the flywheel, initially at rest, begins to
execute damped natural oscillations about the new equilibrium position at θ0 with
the frequency ω1 ≈ ω0 . This oscillation begins with an initial velocity of zero. As
long as the rod remains at θ0 , the time dependence of the angular displacement of
the flywheel, ϕ(t), is given by the following function:

ϕ(t) = θ0 − θ0 exp(−γt) cos ω1 t.

That is, the flywheel, starting out with ϕ = 0 at t = 0, passes through the displaced

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

76 CHAPTER 4. SQUARE-WAVE EXCITATION OF AN OSCILLATOR

equilibrium position ϕ = θ0 when ω0 t = π/2, and reaches its extreme deflection

of nearly ϕ = 2θ0 at ω0 t = π. (Damping prevents it from quite reaching ϕ = 2θ0 .)
If the period T of the driving rod equals T0 , the flywheel arrives at the extreme
point ϕ ≈ 2θ0 and its angular velocity becomes zero just at the moment t = T /2,
when one half of the driving period has elapsed. At this moment the rod instantly
moves to the new position −θ0 , which becomes the new equilibrium position of
the flywheel for the next time interval (T0 /2, T ). Hence, the next half-cycle of
its natural oscillation starts again with an angular velocity of zero, but its initial
angular displacement from the new central point is nearly 3θ0 . This value is nearly
2θ0 greater than its value in the preceding half-cycle. It would be exactly 2θ0
greater in the absence of friction, and the amplitude of oscillation would increase
by the same value 4θ0 during each full cycle of the external force, provided the
driving period equals the natural period of the oscillator (or some odd-number
multiple of that period).
In a real system such an unlimited growth of the amplitude linearly with time
is impossible because of friction. The growth of the amplitude is approximately
linear during the initial stage of the transient process. This resonant growth grad-
ually decreases, and steady-state oscillations are eventually established, during
which the increment of the amplitude occurring at every instantaneous displace-
ment of the equilibrium position (at any jump of the driving rod) is nullified by an
equal decrement caused by viscous friction during the intervals between succes-
sive jumps.
Such a process of gradual growth of the amplitude, which eventually results
in oscillations of a constant amplitude, is depicted very clearly by the phase tra-
jectory shown in the left-hand upper corner of Figure 4.6. Since the oscillator is at
rest at the moment the external force is activated, the phase trajectory originates
at the origin of the phase plane (point 1). Its first section up to point 2 is a portion
of a spiral that winds around a focus located at the point (+θ0 , 0). This focus
corresponds to the equilibrium position displaced to the right. The next section
between points 2 and 3, a continuation of the phase trajectory, describes damped
natural oscillation about the other equilibrium position after the driving rod has
jumped to the left. This is a segment of a similar spiral that winds around the
symmetrical point (−θ0 , 0) of the phase plane.
If the period of the square-wave external action equals an odd-numbered mul-
tiple of the natural period, the transition of the representative point from one spi-
ral, centered say at (+θ0 , 0), to the adjoining spiral centered at the other focus
(−θ0 , 0), occurs at a point of the ϕ-axis to the right of θ0 , at a maximal dis-
tance from the new focus. As a result, the new loop of the phase trajectory turns
out to be larger than the preceding one. Such untwisting of the phase trajectory
continues at a decreasing rate until the expansion of loops due to the alternation
of the foci is nullified by their contraction caused by viscous friction. Eventually
a closed phase trajectory is formed that corresponds to steady-state oscillations.
This curve has a central symmetry about the origin of the phase plane. It consists
of two branches, each representing damped natural oscillations about one of the
two alternating symmetrical equilibrium positions. For T = 7T0 such a closed

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

4.4. AMPLITUDE OF STEADY-STATE OSCILLATIONS 77

1.47 .
ϕ (t ) θ ( t)

2
0 3 1 01 2 3

-1.47 o o
-102 0 102 ϕ (t ) θ ( t)
2
0 1
3
o
-102
0 1 2T

Figure 4.6: The phase diagram (left), graphs of the time dependence of the angular
velocity and position angle (right) for the transient process of excitation from
equilibrium for resonance occurring at T = 3T0 .

phase trajectory is shown in Figure 4.5 (left-hand upper corner).

Any transient process in a linear system can be represented as a superposition
of the periodic solution to Eq. (4.3) that describes the steady-state oscillations, and
a solution of the corresponding homogeneous equation (with the right-hand side
equal to zero) that describes the damped natural oscillations. The simulation pro-
gram displays such a decomposition of the transient process if the corresponding
option is chosen. Figure 4.6 shows the above-discussed graphs for the transient
process of swinging from the state of rest in the equilibrium position that takes
place under the square-wave torque with a period T = 3T0 . A decomposition
of this transient process onto periodic steady-state oscillations and fading natural
oscillations is given by Figure 4.7, in which the graphs of damped natural oscil-
lations with exponentially decreasing amplitude are singled out especially clearly.
Curves 1 show the time-dependent graphs of the transient; curves 2 correspond
to the contribution of natural oscillations, curves 3 show the periodic steady-state
oscillations.

4.4 Estimation of the Amplitude of Steady-State

Oscillations
Next we evaluate the maximal angular deflection, ϕm , attained in the steady-state
oscillations when the oscillator is driven by a square-wave torque. The value ϕm
certainly can be found from Eqs. (4.8)–(4.11). However, such a calculation is
rather complicated. Using the simple arguments suggested in the previous sec-
tions, we can avoid tedious calculations, at least for some special cases.

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

78 CHAPTER 4. SQUARE-WAVE EXCITATION OF AN OSCILLATOR

1.47 .
ϕ (t )
1
0

2 3
-1.47o
102 ϕ (t )
θ ( t)
1
0

3
o
2
-102
0 1 2T

Figure 4.7: Graphs of the angular velocity (upper panel) and the angular position
(lower panel) that show the decomposition of the transient (curves 1) onto the
damping natural oscillations (curves 2) and the periodic steady-state oscillations
(curves 3).

4.4.1 Resonant Amplitude of Steady-State Oscillations

We consider first the main resonance in which the driving period equals the natural
period: T = T0 . The estimation of ϕm can be made in the following way. The
closed phase trajectory for the steady-state oscillation consists in this case of a
single loop intersecting the ϕ-axis at the points −ϕm and ϕm , which are the points
of extreme angular displacements of the oscillator. The angular separation of these
points from the equilibrium position at θ0 are ϕm + θ0 (on the left side of θ0 ) and
ϕm − θ0 (on the right side of θ0 ).
The upper part of the phase trajectory is a half-loop of a spiral whose focus is
at the point +θ0 , displaced to the right from the origin. While the representative
point passes along this upper half-loop from −ϕm to ϕm , the oscillator executes
one half of a period of damped natural oscillation about the equilibrium position
θ0 , displaced to the right side: the flywheel passes from the extreme deflection
|ϕm + θ0 | on the left side (measured from θ0 ) to the extreme deflection ϕm − θ0
on the right side.
When the oscillator reaches this extreme point, the equilibrium position in-
stantly switches to the focus −θ0 , and the representative point then passes along
the lower half-loop, thus closing the phase trajectory of the steady-state motion.
The relative decrease of the amplitude because of the viscous friction during
one half of the natural period (t = T0 /2) of oscillation equals exp(−γT0 /2). So
the left and right separations for the upper half-loop are related to one another
through this exponential factor giving the frictional decay for a half-cycle:

(ϕm + θ0 ) exp(−γT0 /2) = ϕm − θ0 . (4.12)

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

4.4. AMPLITUDE OF STEADY-STATE OSCILLATIONS 79

1.47 0.85

ϕ0 θ0
-ϕ0
-ϕm ϕm -ϕm ϕm
-θ0 θ0
-θ0

-1.47 o -0.85 o o
-102 0 102 -66.1 0 66.1

Figure 4.8: Phase trajectories for the resonance at T = 3T0 , Q = 5, θ0 = 30◦

(left) and at T = 7T0 , Q = 3, θ0 = 30◦ (right).

For the case in which γ ≪ ω0 , that is, γT0 ≪ 1 (oscillator with relatively
weak friction), we can assume exp(−γT0 /2) ≈ 1 − γT0 /2. Using this approxi-
mation in Eq. (4.12) and solving for ϕm , we obtain the desired estimate:
2 4
ϕm ≈ θ0 = Q θ0 . (4.13)
γT0 /2 π

The product of the damping constant γ and the natural period T0 is expressed here
in terms of the quality factor Q = ω0 /2γ.
Equation (4.13) shows that for resonance induced by the fundamental har-
monic of the square-wave external torque (T = T0 ) the amplitude of steady-state
oscillation is Q times greater than the amplitude (4/π)θ0 of this harmonic com-
ponent in the square-wave motion of the exciting rod. (See Eq. (4.5).) The same
conclusion can be reached from a spectral approach to the treatment of stationary
forced oscillations.
Through a similar (though more complicated) calculation we can obtain an
estimate of the maximal displacement, ϕm , attained in steady-state oscillations for
any of the higher resonances when the period of the square-wave external torque
is an odd multiple of the natural period. For example, for the resonance occurring
at T = 3T0 (three natural periods during one cycle of the exciting rod) we can
consider a half of the closed phase trajectory (left-hand panel of Figure 4.8), which
consists of three shrinking half-loops spiraling to the right-hand focus (θ0 , 0).
At the beginning of each cycle, when the exciter abruptly turns from left to
right, the representative point occurs on the abscissa axis of the phase plane at
some point −ϕ0 (see the left-hand panel of Figure 4.8) to the left of the new equi-
librium position θ0 . After a half-period of natural oscillation about θ0 (after T0 /2)
the flywheel reaches its maximal deflection ϕm to the right-hand side, which we
are going to evaluate. Then during the next period T0 of natural oscillation the
representative point makes a full revolution about the focus (θ0 , 0), and occurs
again on the abscissa axis at the point ϕ0 , which is displaced to the right from
the focus (θ0 , 0). At this moment the exciting rod turns abruptly to the left, and
the remaining part of the closed phase trajectory is formed by the one-and-a-half
loops spiraling toward the left focus (−θ0 , 0).

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

80 CHAPTER 4. SQUARE-WAVE EXCITATION OF AN OSCILLATOR

After each half-period T0 /2 of natural oscillations the maximal angular elon-

gation of the flywheel from the equilibrium position diminishes by the same factor
q ≈ exp(−γT0 /2) = exp(−π/2Q). At the beginning of the cycle the repre-
sentative point is displaced to the left from the equilibrium position θ0 through
|ϕ0 + θ0 —. After T0 /2 later this point reaches its maximal elongation ϕm − θ0 to
the right of θ0 . Therefore

(ϕ0 + θ0 )q = ϕm − θ0 . (4.14)

During the next period T0 the representative point traces a whole loop around
θ0 , and its maximal elongation is multiplied by q 2 . Since now its displacement
from the origin equals ϕ0 , this point is displaced from the equilibrium position
through ϕ0 − θ0 . Therefore we can write:

(ϕm − θ0 )q 2 = ϕ0 − θ0 . (4.15)

Solving the system of Eqs. (4.14)–(4.15), we get the following expression for the
desired maximal deflection ϕm :
 
2q
ϕm = θ0 1 + . (4.16)
1 − q3
For example, at Q = 5 the factor q = exp(−π/2Q) equals 0.73, and Eq. (4.16)
yields ϕm = 3.39 θ0 . For θ0 = 30◦ we get ϕm = 102◦ (see Figure 4.4).
We note that for this case it is not so simple to evaluate the maximal elongation
of the flywheel on the basis of the spectral approach. Indeed, the shape of output
oscillations now depends not only on the third harmonic of the input square-wave
oscillation that is in resonance with the oscillator, but also on the first (fundamen-
tal) harmonic that produces in the output a contribution of the same order of mag-
nitude. The shape of output oscillations in this case due to this contribution differs
considerably from a pure sinusoid. This is clearly seen from the phase trajectory
and the plots in Figure 4.4 of resonant steady-state oscillations for T = 3T0 ,
Q = 5, θ0 = 30◦ .

4.4.2 Amplitude of Steady Oscillations at Strong Friction

For the case of strong friction, the maximum elongation of the flywheel even in
resonant conditions is only slightly greater than the amplitude of the exciting rod.
The shape of the output pulses differs from the square-wave input impulses not
so drastically as at weak friction. The principal distortion of the pulses’ shape
reveals itself in some smoothing of the abrupt leading front of input impulses.
This is clearly seen from the graphs of ϕ(t) in Figure 4.5 for T = 7T0 , Q = 3.
In order to evaluate the maximum elongation of the flywheel at steady-state os-
cillations in conditions of strong friction (quality factor Q of the order 2 – 3), we
note that for T /T0 > Q the natural oscillations about a displaced equilibrium po-
sition almost damp out before the rod of the exciter makes its next-in-turn abrupt
rotation to the new position. In these conditions the phase trajectory that spirals

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

4.4. AMPLITUDE OF STEADY-STATE OSCILLATIONS 81

in towards the focus (θ0 , 0) approaches very close to this point before the transi-
tion to spiraling towards the other focus (see the right-hand panel of Figure 4.8).
This means that at the beginning of each in-turn half-cycle of steady-state oscil-
lations the representative point starts almost from one of the foci. This occurs at
T /T0 > Q independently of the exact value of T /T0 : it does not matter whether
the period of excitation has a resonant value (that is, equals an odd number of
natural periods) or not.
Next we find the maximum elongation in such conditions. At abrupt rotation
of the exciting rod from left to right, the representing point starts to move from the
left focus (−θ0 , 0) (see the right-hand panel of Figure 4.8) along a shrinking spi-
ral, winding around the right focus (θ0 , 0). This spiral intersects the abscissa axis
at the point ϕm , which we are interested in after a half-period T1 /2 ≈ T0 /2 of
damped natural oscillations about the displaced equilibrium position θ0 . The ini-
tial angular displacement from this position equals approximately 2θ0 , and after a
half-period the displacement equals ϕm −θ0 . These initial and final displacements
differ by a factor q ≈ exp(−γT0 /2) = exp(−π/2Q); therefore 2θ0 q = ϕm − θ0 .
This yields the following final expression for the maximum elongation of the fly-
wheel:
ϕm ≈ θ0 (1 + 2q) ≈ θ0 (1 + 2e−π/2Q ). (4.17)
For example, at Q = 3 Eq. (4.17) gives for ϕm the value 2.18 θ0 , i. e., the max-
imum elongation of the flywheel is more than twice the displacement θ0 of the
exciting rod from its middle position. For θ = 30◦ we have ϕm ≈ 66◦ (see Fig-
ure 4.5).
This approximate estimate for ϕm is almost independent of the period of exci-
tation. However, it is valid only under assumption that during a half of this period
natural oscillations nearly fade away. To find the precision of the above estimate,
we can consider the residual displacement of the flywheel from the focal point
θ0 . For example, at T = 7T0 this residual displacement equals approximately
2θ0 exp(−7π/2Q), which for Q = 3 yields 0.03 θ0. This means that at the begin-
ning of each half-cycle the representing point starts not exactly from ±θ0 as we
assumed above, but rather from some point slightly displaced from ±θ0 to one or
the other side depending on the exact value of T /T0 .

4.4.3 Amplitude of Steady Oscillations at T = 2nT0

When the driving period is an even-numbered multiple of the natural period, the
maximal deflection of the flywheel ϕm attained in steady-state forced oscillations
at weak friction is close to 2θ0 , that is, about twice the amplitude of square-wave
motion of the exciter. We can easily see this result from the shape of the corre-
sponding phase trajectory: each of its two symmetrical halves consists of an inte-
gral number of shrinking loops of a spiral winding around one of the foci θ0 and
−θ0 . Figure 4.9 shows this kind of the phase trajectory and the time-dependent
graphs for a special case in which T = 4T0 .
For T = 2T0 , one complete cycle of the natural oscillation occurs while the
equilibrium position is displaced to one side. In the absence of friction both closed

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

82 CHAPTER 4. SQUARE-WAVE EXCITATION OF AN OSCILLATOR

0.67 .
ϕ (t ) θ ( t)

-0.67 o
o
-62.6 0 62.6 θ ( t)
ϕ (t )

o
-62.6
0 1 2T

Figure 4.9: The phase trajectory (left) and graphs of the time dependence of the
angular velocity and of the angular position for steady-state forced oscillations at
T = 4T0 .

loops of the steady-state phase trajectory meet at the origin of the phase plane, and
the magnitude ϕm of the maximal displacement on each side of the zero point just
equals 2θ0 . Friction causes the loops to shrink, and the maximal displacement ϕm
of the flywheel becomes slightly smaller than 2θ0 .

4.4.4 Steady-State Oscillations at High Frequencies of the

Square-Wave Torque
For high frequencies of the external force, when the square-wave period T of the
driving rod is very short compared to the natural period T0 of the oscillator, in
steady-state motion the flywheel executes only small vibrations about the mid-
point ϕ = 0. The period of these vibrations is the same as that of the driving rod.
However, they occur in the opposite phase with respect to the exciter motion, and
their amplitude is small compared to the amplitude θ0 of the exciter.
Since the flywheel moves little while the position of the rod is fixed at either θ0
or −θ0 , we can consider the torque of the spring exerted on the flywheel as nearly
constant during the intervals between successive jumps of the rod. Therefore the
graph of the angular velocity in steady-state short-period oscillations consists of
nearly rectilinear segments (Figure 4.10). They correspond to the rotation of the
flywheel with a uniform angular acceleration ω02 θ0 caused by the constant torque
of the strained spring. Such rotation continues in one direction during time inter-
vals between abrupt turns of the rod. After each succeeding turn the acceleration
changes sign, remaining nearly the same in magnitude. In the graph of the angular
velocity, the straight segments join to form a saw-toothed pattern of isosceles tri-
angles. The corresponding graph of the angular deflection is formed by adjoining

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

4.5. ENERGY TRANSFORMATIONS 83

0.96
.
ϕ (t )

-0.96 o o
-6.18 0 6.18
ϕ (t )
0

o
-6.18
0 1 2T

Figure 4.10: Steady-state oscillations at T ≪ T0 (T = 0.25 T0 , Q = 20, θ0 =

75◦ )—phase diagram (left) and time dependent graphs (rectilinear segments of the
saw-teeth graph of the angular velocity ϕ̇(t) and parabolic segments of the angle
of deflection ϕ(t)).

parabolic segments alternating after each half of the driving period.

We can easily find the height of these segments, that is, the maximal displace-
ment ϕm of the rotor in the case under consideration, by calculating the angu-
lar path of the rotor as it moves with a constant angular acceleration ω02 θ0 =
4π 2 θ0 /T02 from the zero point of the dial to maximal elongation ϕm during a
quarter driving period T /4:
2
π2

T
ϕm ≈ θ0 . (4.18)
8 T0

4.5 Energy Transformations

The exchange of mechanical energy between the oscillator and the source of the
external driving force occurs in the investigated system only at the instants when
the driving rod turns abruptly from one position to the other. During the intervals
between such instant turns, while the oscillator executes damped natural oscilla-
tions about one of the two displaced equilibrium positions, only an alternating
partial conversion between the elastic potential energy of the strained spring and
the kinetic energy of the flywheel occurs, accompanied by the gradual dissipation
of mechanical energy because of friction.
To get a general idea of these energy transformations, we can refer to the
section “Energy Transformations” in the corresponding simulation program, and
consider the motion of the point representing the total energy in the graph of po-
tential energy versus the angle of deflection.

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

84 CHAPTER 4. SQUARE-WAVE EXCITATION OF AN OSCILLATOR

E total
Epot
E kin

0 T/2 T

Figure 4.11: Energy transformations at steady-state oscillations with T = 3T0 ,

Q = 8. Curve Epot – potential energy, curve Ekin – kinetic energy, curve Etotal
– total mechanical energy.

A parabolic potential well corresponds to each of the two equilibrium positions

of the oscillator. When the flywheel is located at the angle ϕ from its central
position, the corresponding potential energy of the spring is given by one of the
two quadratic functions:
1
U (ϕ) = k(ϕ ∓ θ0 )2 . (4.19)
2
We must take the upper sign in Eq. (4.19) if the equilibrium position is displaced
to the right (to the point +θ0 ), and the lower sign if the equilibrium position is
displaced to the left.
At an instantaneous turn of the driving rod from one position to the other
the angular velocity of the massive flywheel and hence its kinetic energy do not
change. An abrupt change occurs only in the value of the elastic potential energy
of the spring. This causes the representative point to make an abrupt vertical tran-
sition from one of the parabolic potential wells to the other at a fixed value of the
angle ϕ.
During the interval before the next jump, while the oscillator executes damped
natural oscillations about a displaced equilibrium position, the point that repre-
sents the total energy travels back and forth between the walls of the corresponding
potential well, descending gradually toward the bottom because of energy losses
caused by friction.
The time-dependent graphs of the total mechanical energy Etotal (t), kinetic
energy Ekin (t), and potential energy Epot (t) during one period of steady-state
oscillations are shown in Figure 4.11 for T = 3T0 , Q = 8.
It is important to note that in this simplified model of the physical system the
deformation of the spiral spring is assumed to be quasistatic. In other words, we
ignore the possibility that the spring vibrates as a system with distributed parame-
ters, whose each portion has both elastic and inertial properties. For a light spring
(attached to a comparatively massive flywheel) these vibrations are characterized
by much higher frequencies than the frequencies of the torsional oscillations of
the flywheel. Our simplified model is applicable to such a physical system be-
cause rapid vibrations of the spring quickly damp out.

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

4.6. THE ELECTROMAGNETIC ANALOGUE 85

4.6 The Electromagnetic Analogue of the

Mechanical System
Forced oscillations of the electric charge q stored in a capacitor in a resonant
series LCR-circuit excited by a square-wave input voltage V (t) (see the right-
hand panel of Figure 4.1) obey the same differential equation as does the forced
oscillation of a mechanical torsion spring oscillator excited by periodic abrupt
changes of position of the driving rod:

q̈ + 2γ q̇ + ω02 q = ω02 CV (t). (4.20)

In this equation ω0 is the natural frequency of oscillations of charge in the circuit

in the absence of resistance. It depends on √ the capacitance C of the capacitor
and the inductance L of the coil: ω0 = 1/ LC. The damping constant γ =
R/(2L) characterizes the dissipation of electromagnetic energy occurring in a
resistor whose resistance is R.
Because of this similarity, the mechanical system simulated in the computer
program enables us to give a very clear treatment of transformation of the square-
wave input voltage V (t) = ±V0 into the output voltage VC (t) = q/C (voltage
across the capacitor C). The output voltage VC (t) is analogous to the deflection
angle ϕ(t) of the rotor. The alternating electric current I(t) = q̇(t) in the circuit
is analogous to the angular velocity ϕ̇(t) of the mechanical model.
The assumption concerning quasistatic character of the spring deformation in
the mechanical model, i.e., the assumption of possibility to neglect rapid vibra-
tions of the spring as a distributed parameter system, corresponds to the ordinary
implicit assumption of quasistationary current in the circuit. According to this as-
sumption, the momentary value of the current is the same along the whole circuit.
The assumption is valid if the inductance and the capacitance of the wires are
negligible compared with the inductance of the coil and the capacitance of the ca-
pacitor, respectively. In this case the oscillatory circuit can be treated as a system
with lumped parameters (as a system with one degree of freedom), in which all the
capacitance is concentrated in the capacitor and all the inductance is concentrated
in the coil.
Some caution is necessary in interpreting the analogy between the mechanical
oscillator and the electric LCR-circuit with respect to energy transformations. It
is incorrect to identify exactly the electric potential energy of a charged capacitor
with the elastic potential energy of a strained spring because the latter depends
directly on the angle (ϕ ± φ0 ), while the energy of a capacitor depends directly
on the charge q or on the corresponding voltage VC = q/C (not on the voltage
(VC ± V0 )). In contrast to the spring oscillator, for which a jump in the position
of the rod causes an abrupt change of the elastic potential energy of the spring, a
jump in the input voltage across an electric circuit does not abruptly change the
charge and the energy of a capacitor.
The spectral treatment of the transformation of a square-wave driving force (an
input) into the steady-state oscillations of the mechanical spring oscillator (the

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

86 CHAPTER 4. SQUARE-WAVE EXCITATION OF AN OSCILLATOR

output) is equally valid for the transformation of an input square-wave voltage

into the output oscillations of charge in the analogous electric circuit. Only those
harmonics of the input signal that are near the resonant frequency of the circuit
are noticeably present in the output voltage across the capacitor.
A resonant circuit selectively transmits to the output signal those harmonic
components of the input signal whose frequencies are near the resonant frequency
of the circuit. The greater the quality factor Q, the sharper the resonance curve
and the finer the selectivity of the oscillatory circuit. √
It is possible to vary the natural frequency ω0 = 1/ LC of a resonant circuit
by varying either the capacitance C or the inductance L. Such tunable resonant
circuits with high selectivity can serve as spectral instruments that are physically
able to Fourier analyze a complex input signal.
The assumption concerning the quasistatic character of the spring deforma-
tion in the mechanical model, i.e., the assumption of possibility to neglect rapid
vibrations of the spring as a distributed parameter system, corresponds to the or-
dinary implicit assumption of quasistationary current in the analogous oscillatory
circuit. According to this assumption, the momentary value of the current is the
same along the whole circuit. The assumption is valid if the inductance and ca-
pacitance of the wires are negligible compared with the inductance of the coil and
the capacitance of the capacitor, respectively. In this case the oscillatory circuit
can be treated as a system with lumped parameters (as a system with one degree
of freedom), in which all the capacitance is concentrated in the capacitor and all
the inductance is concentrated in the coil.

4.7 Concluding Remarks

We have considered in this chapter two different ways of determining the steady-
state response of the linear oscillator to a non-harmonic driving force. The tradi-
tional approach based on the Fourier expansion of the input driving force is cer-
tainly quite general because it is applicable to an arbitrary periodic excitation. The
second method based on representing forced oscillations as some sequence of nat-
ural oscillations about displaced equilibrium positions can be used only for piece-
wise constant excitations. Nevertheless, this approach is physically much more
obvious, and also allows us to understand transient processes. Combining both
approaches in teaching students, we can hope to give them better understanding
of this important topic.
The obvious, intuitive treatment of the transformation of a square-wave driv-
ing force (an input) into the steady-state oscillations of the mechanical spring
oscillator (the output) described in this chapter is equally valid for the transforma-
tion of an input square-wave voltage into the output oscillations of charge in the
analogous electric circuit. Therefore, behavior of this familiar mechanical system
can help a student to better understand why and how an electromagnetic oscil-
latory LCR-circuit transfers the square-wave voltage from input to output with a
distortion of its shape. Mechanical analogies allow a direct visualization and thus

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

4.8. REVIEW OF THE PRINCIPAL FORMULAS 87

can be very useful in gaining an intuitive understanding of complex phenomena.

This simple mechanical system helps us to understand both the complex shape
of the output oscillations and also their spectral composition. Only those harmon-
ics of the input signal, which are close to the resonant frequency of the circuit,
are noticeably present in the output voltage across the capacitor. In other words,
such a resonant circuit selectively responds to different harmonic components of
the input signal. The greater the quality factor Q, the sharper the resonance curve,
and the finer the selectivity of the oscillatory circuit. √
It is possible to easily vary the natural frequency ω0 = 1/ LC of a resonant
circuit by varying either the capacitance C or the inductance L. Such a tunable
resonant circuit with high selectivity can serve as a spectral instrument that is
able to accomplish the mathematical task of Fourier expansion of a complex input
signal onto separate harmonic components on a physical level. The mechanical
system described in this chapter provides a clear and plain way to understand this
possibility.

4.8 Review of the Principal Formulas

The differential equation of motion for a torsion spring oscillator driven by a
square-wave external force:

ω02 θ0 ,

2 (0, T /2),
ϕ̈ + 2γ ϕ̇ + ω0 ϕ = (4.21)
−ω02 θ0 , (T /2, T ).

The same equation in which the square-wave shaped right-hand side is represented
as a Fourier series:
∞
X 4θ0 ω02
ϕ̈ + 2γ ϕ̇ + ω02 ϕ = sin ωk t. (4.22)
πk
k=1, 3, 5...

The particular periodic solution of the equation (describing steady-state oscilla-

tions):
∞
X 4θ0 ω02
ϕ(t) = sin(ωk t + αk ), (4.23)
πk (ω02 − ωk2 )2 + 4γ 2 ωk2
p
k=1, 3, 5...

where the phases αk of the individual harmonics are determined by:

2γωk
tan αk = . (4.24)
ωk2 − ω02

The time dependence of ϕ(t) during the interval 0 ≤ t ≤ T /2, when the equilib-
rium position is located at ϕ = θ0 :

ϕ(t) = θ0 + Ae−γt cos(ω1 t + θ), (0, T /2), (4.25)

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

88 CHAPTER 4. SQUARE-WAVE EXCITATION OF AN OSCILLATOR

p
where ω1 = ω02 − γ 2 , the frequency of natural damped oscillations, and A and
θ are some constants.
The time dependence of ϕ(t) during the interval T /2 ≤ t ≤ T , when damped
natural oscillations occur about the equilibrium position located at −θ0 :

ϕ(t) = −θ0 − Ae−γ(t−T /2) cos(ω1 (t − T /2) + θ), (T /2, T ). (4.26)

For a steady-state process, the constants A and θ here have the same values as they
do for the interval (0, T /2).

4.9 Questions, Problems, Suggestions

4.9.1 Swinging of the Oscillator at Resonance
4.9.1.1 The Principal Resonance in the Absence of Friction. Select the
idealized case of no friction. Enter the period T of the external force, letting this
period be the period T0 of natural oscillations. Choose null initial conditions. That
is, let the oscillator be at rest in the equilibrium position at the moment the external
force is activated.
(a) What must be the value θ0 of the angular amplitude of the square-wave
motion of the driving rod in order that the amplitude reach 180◦ after the first 10
cycles? Verify your answer by a simulation experiment.
(b) What regularity does the growth of the amplitude exhibit? Explain the form
of the phase trajectory displayed. How does the energy of the oscillator grow with
time?
(c) If the oscillator is exactly tuned to resonance, is it possible for the ampli-
tude to diminish? Give some physical justification for your answer. Can you test
your answer by a simulation experiment?
4.9.1.2 High Resonances in the Absence of Friction. Examine the resonant
excitation of the oscillator initially at rest in the equilibrium position when the
period of the external square-wave force is three times longer than the natural
period: T = 3T0 .
(a) At what value θ0 of the amplitude of a square-wave oscillation of the driv-
ing rod does the oscillator reach its maximal deflection of 180◦ in 10 first cycles
of the external action? Verify your prediction experimentally.
(b) What are the differences in the graphs and the phase trajectories between
this case and the previous case (Problem 1.1) for which T = T0 ?
4.9.1.3∗ Transient Process and Steady-State Oscillations at the Principal
Resonance. Letting the period of the square-wave motion of the driving rod be
at the principal resonance, T = T0 , and letting the flywheel be initially at rest
in the equilibrium position, examine the transient process and the steady-state
oscillations in the presence of friction:
(a) Calculate the amplitude of steady-state oscillations for the values θ0 = 10◦
and Q = 10. Verify your result experimentally.

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

4.9. QUESTIONS, PROBLEMS, SUGGESTIONS 89

(b) What regularity does the growth of the amplitude exhibit in this case?
Explain the peculiarities of the phase trajectory.
(c) What is the initial amplitude of damped natural oscillations constituting
the transient process for θ0 = 10◦ and Q = 10?
(d) What initial conditions cause steady-state forced oscillations to appear
from the moment the external force begins to act, thus eliminating the transient
process? Verify your answer experimentally.
(e) Examine the spectrum of steady-state oscillations (the output) in this case.
Why are these oscillations nearly purely harmonic in spite of the non-sinusoidal,
square-wave shape of the input?
4.9.1.4∗ Steady-State Oscillations at High Resonances.
(a) Calculate the amplitude of steady-state oscillations for θ0 = 25◦ , Q =
5, and T = 3T0 . Explain the shape of the graphs displayed and of the phase
trajectory.
(b) What energy transformations take place during steady-state oscillations?
Compare the graphs of the time dependence of the kinetic, potential, and total
energy with the corresponding graphs of the angular deflection and the angular
velocity of the flywheel. Explain the shape of the graph of the total energy versus
the angle of deflection, and explain its relationship to the parabolic potential wells
shown in the same diagram.
(c) Which harmonic components determine the shape of output steady-state
oscillations in this case? Why, in spite of the exact tuning of the oscillator to the
frequency of the third harmonic of the input external force, does the first harmonic
component of this force appreciably influence the shape of the output oscillations?
How does this harmonic exhibit itself in the pattern of the output oscillations?
(d) Examine the influence of friction on the shape and on the spectral compo-
sition of steady-state oscillations at T = 3T0 . Note the relative reduction in the
contribution of the first and fifth harmonics as the quality factor of the oscillator
is increased.
(e) Explore the resonant oscillations of the flywheel when the frequency of the
fifth or the seventh harmonic of the external square-wave driving force coincides
with the natural frequency of the oscillator. Observe the transformation of the
spectrum from input to output, and the dependence of the spectrum of steady-
state oscillations on the quality factor of the oscillator. What is the shape of the
phase trajectory in these cases? How can you estimate the value of the maximal
displacement of the flywheel from the mid-point of its oscillations (from the zero
point of the dial) when friction is large (when Q ranges say from 1 to 3)?

4.9.2 Non-Resonant Forced Oscillations

4.9.2.1∗ Conditions That Eliminate a Transient at T = 2T0 .
(a) Predict the shape of the graphs of the angular deflection, angular veloc-
ity, and the phase trajectory of forced oscillations in the absence of friction, for
T = 2T0 . Under what initial conditions is the steady-state oscillation established

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

90 CHAPTER 4. SQUARE-WAVE EXCITATION OF AN OSCILLATOR

immediately after the force is activated? Verify your predictions in a simulation

experiment.
(b) Why does the total energy remain constant in these oscillations? When the
driving rod executes a jump, why does the oscillator neither gain energy from nor
give back energy to the external source?
(c) Examine the spectral composition of steady-state oscillations. Note the
contribution of the third harmonic and the influence of its alteration in phase on
the shape of the output oscillations: The frequency of the third harmonic in this
case is higher than the natural frequency of the oscillator. Therefore, its phase in
the output oscillations is inverted. As a result, this harmonic component, super-
imposed on the fundamental, produces a time-dependent graph with bulges at the
positions of the flat parts of the square-wave input graph.
4.9.2.2 Steady-State Oscillations for T = 2T0 .
(a) Consider forced oscillations for the case in which T = 2T0 in the presence
of friction, by setting Q ≈ 5 − 10. How do the time-dependent graphs and the
phase trajectory differ from the preceding case (Problem 2.1), in which friction is
absent? Calculate the maximal deflection of the flywheel attained in these steady-
state oscillations. Note changes in the energy transformations.
(b) Why does friction not noticeably influence the spectral composition of
steady-state oscillations in this case, in contrast to the case in which T = 3T0 ?
4.9.2.3 Steady-State Oscillations for a Large External Period.
(a) Examine forced oscillations for a case in which the natural frequency of
the oscillator lies somewhere between the frequencies of two consecutive high
odd harmonics of the external action (e.g., let 5T0 < T < 7T0 ). Which harmonic
components dominate in the output steady-state oscillations? Compare the shape
of the output steady-state oscillations with the shape of the input square-wave
impulses. What is the main difference between the patterns of input and output
oscillations?
(b) Investigate the influence of friction on the character of steady-state os-
cillations. Why are the distortions of the output less prominent the stronger the
damping? That is, why is the shape of the output curve for large friction nearly
rectangular?
(c) Explain the energy transformations in these oscillations using the graph
of the total energy versus the angle of deflection. What is the relationship of this
graph with the parabolic potential wells shown in the same diagram?
4.9.2.4 Steady-State Oscillations Forced by Short-Period Impulses.
(a) Choose a value for the period T of the square-wave external force to be a
small fraction (say 0.2 − 0.3) of the natural period T0 of the oscillator. The graph
of the angular velocity versus time for steady-state output oscillations has a saw-
toothed pattern with teeth that are nearly rectilinear isosceles triangles. Suggest
an explanation. What is the difference between the graph of the angular deflection
versus time for this case and a sine curve?
(b) Evaluate theoretically the height of a tooth of the angular velocity graph.
Also evaluate the maximal deflection angle for these non-sinusoidal steady-state

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

4.9. QUESTIONS, PROBLEMS, SUGGESTIONS 91

oscillations. Consider the case of weak or moderate friction. Let, for example,
the period T be T0 /4 and the angle θ0 describing the instantaneous deflections
of the driving rod be 30◦ . What spectral composition is characteristic of such
oscillations?

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

Chapter 5

Parametric Excitation of a
Linear Oscillator

Annotation. Chapter 5 is concerned with the phenomenon of parametric reso-

nance arising from a periodic square-wave modulation of the moment of inertia
of a linear torsion spring oscillator. Computer simulations of oscillations show
the motion of the system and are accompanied by plotting the time dependencies
of the angle of deflection and the angular velocity. Phase trajectories and energy
transformations for different values of the depth of modulation are analyzed, and
conditions of parametric regeneration and of parametric resonance are discussed.
Ranges of frequencies where parametric excitation is possible are determined.
Stationary oscillations on the boundaries of these ranges are investigated both an-
alytically and by computer simulation.

5.1 Summary of the Theory. General Concepts

5.1.1 Classification of Oscillations
In the conventional classification of oscillations by their method of excitation,
oscillations are called forced if an oscillator is subjected to an external periodic
influence whose effect on the system can be expressed by a separate term, a pe-
riodic function of the time, in the differential equation of motion describing the
system. Forced oscillations are discussed in Chapter 3 and Chapter 4.
The investigation of non-stationary, position-dependent forces, i.e., the forces
that are explicitly determined by both temporal and spatial coordinates, is more
complicated. For example, let a restoring force F = −kx arise when the system is
displaced through some distance x from the equilibrium position. But in contrast
to the stationary case, the parameter k changes with time because of some periodic
influence: k = k(t). In the differential equation of motion for the system,

mẍ = −k(t)x, (5.1)

93

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

94 CHAPTER 5. PARAMETRIC EXCITATION OF OSCILLATIONS

the coefficient of x is not constant: It explicitly depends on time. Oscillations in

such a system are essentially different from both free oscillations, which occur
when k is constant, and forced oscillations, which occur when k is constant and
an additional time-dependent forcing term is added to the right-hand side of the
equation of motion, Eq. (5.1).
In the case of periodic changes of the parameter k, when k(t + T ) = k(t),
where T is the period, the corresponding differential equation, Eq. (5.1), is called
Hill’s equation. Oscillations in a system described by Hill’s equation are called
parametrically excited or simply parametric oscillations. When the amplitude
of oscillation caused by the periodic modulation of some parameter increases
steadily, we describe the phenomenon as parametric resonance. In parametric res-
onance, equilibrium becomes unstable and the system performs oscillations whose
amplitude increases exponentially.
The characteristics and causes of parametric resonance are considerably dif-
ferent from those of the resonance occurring when the oscillator responds to a pe-
riodic external force. Specifically, the resonant relationship between the frequency
of modulation of the parameter and the mean natural frequency of oscillation of
the system is different from the relationship between the driving frequency and
the natural frequency for the usual resonance in forced oscillations. And if there
is friction, the amplitude of modulation of the parameter must exceed a certain
threshold value in order to cause parametric resonance.

5.1.2 The Simulated Physical System

A physical system undergoes a parametric forcing if one of its parameters is mod-
ulated periodically with time. A common familiar example of parametric excita-
tion of oscillations is given by the playground swing on which most people have
played in childhood. The swing can be treated as a physical pendulum whose re-
duced length changes periodically as the child squats at the extreme points, and
straightens when the swing passes through the equilibrium position.
It is easy to illustrate this phenomenon by the following simple experiment.
Let a thread with a bob hanging on its end pass through a little ring fixed immov-
ably in a support. The other end of the thread that you are holding in your hand you
can pull by some small length each time when the swinging bob passes through
the middle position and release the thread to its previous length each time the bob
reaches the utmost deflection. These periodic variations of the pendulum length
with the frequency twice the frequency of natural oscillation cause the amplitude
to increase progressively. Another canonical example of parametric pumping is
given by a pendulum whose support oscillates vertically.
However, such systems do not perfectly suit the initial acquaintance with the
parametric excitation because the ordinary pendulum is a nonlinear physical sys-
tem: The restoring torque of the gravitational force is proportional to the sine of
the deflection angle.
That is why we suggest here to study the basics of parametric resonance by
using the simplest linear mechanical system in which the phenomenon is possible,

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

5.1. SUMMARY OF THE THEORY 95

L(t)

j (t)

C
J (t)
R

Figure 5.1: Schematic image of the torsion spring oscillator with a rotor whose
moment of inertia is forced to vary periodically (left), and an analogous LCR-
circuit with a coil whose inductance is modulated by moving periodically an iron
core in and out of the coil (right).

namely, the torsion spring oscillator, similar to the balance device of a mechanical
watch.
The suggested computer program simulates a simple physical system that
perfectly suits the initial acquaintance with the basics of parametric resonance,
namely, a torsion spring oscillator (Figure 5.1) similar to the balance device of
a mechanical watch. It consists of a rigid rod that can rotate about an axis that
passes through its center. Two identical weights are balanced on the rod. An elas-
tic spiral spring is attached to the rod. The other end of the spring is fixed. When
the rod is turned about its axis, the spring flexes. The restoring torque −Dϕ of the
spring is proportional to the angular displacement ϕ of the rotor from the equilib-
rium position. After a disturbance, the rotor executes natural harmonic torsional
oscillations.
We assume that the weights can be shifted simultaneously along the rod in
opposite directions into other symmetrical positions so that the rotor as a whole
remains balanced. However, its moment of inertia is changed by such displace-
ments of the weights.
When the weights are shifted toward or away from the axis, the moment of
inertia decreases or increases respectively. As the moment of inertia of the rotor
is changed, so also is the natural frequency of its oscillation. Thus the moment of
inertia of the rotor is the parameter to be modulated in this system.
This physical system is ideal for the study of parametric resonance and has
several advantages in an educational context because it gives a very clear example
of the phenomenon in a linear mechanical system. All peculiarities of paramet-
ric excitation in this linear system can be completely explained and exhaustively
investigated by modest means even quantitatively.
Another similar mode of the parametric modulation – a smooth periodic vari-
ation of the moment of inertia by sinusoidal motion of the weights along the rod
– is considered in Chapter 6.

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

96 CHAPTER 5. PARAMETRIC EXCITATION OF OSCILLATIONS

In the case of the square-wave modulation, abrupt, almost instantaneous in-

crements and decrements of the moment of inertia occur sequentially, separated
by equal time intervals. We denote these intervals by T /2, so that T equals the
period of the variation in the moment of inertia (the period of modulation).
The square-wave variation of a parameter can produce considerable oscillation
of the rotor if the period of modulation is chosen properly. For example, suppose
that the weights are drawn closer to each other at the instant at which the rotor
passes through the equilibrium position, when its angular velocity is almost max-
imal. While the weights are being moved, the angular momentum of the system
remains constant since no torque is needed to effect this displacement. Thus the
resulting reduction in the moment of inertia is accompanied by an increment in the
angular velocity, and the rotor acquires additional energy. The greater the angular
velocity, the greater the increment in energy. This additional energy is supplied by
the source that moves the weights along the rod.
On the other hand, if the weights are instantly moved apart along the rotating
rod, the angular velocity and the energy of the rotor diminish. The decrease in
energy is transmitted back to the source.
In order that increments in energy occur regularly and exceed the amounts of
energy returned, i.e., in order that, as a whole, the modulation of the moment of
inertia regularly feed the oscillator with energy, the period of modulation must
satisfy certain conditions.
For instance, let the weights be drawn closer to and moved apart from one an-
other twice during one mean period of the natural oscillation. The angular velocity
increases at the moment the weights come together, and vice versa. Furthermore,
let the weights be drawn closer at the instant of maximum angular velocity, so
that the rotor gains as much energy as possible. Then, after a quarter period of the
natural rotary oscillation, the weights are moved apart, and this occurs almost at
the instant of extreme deflection, when the angular velocity is nearly zero. There-
fore this particular motion causes almost no change in the angular velocity and
kinetic energy of the rotor. Thus, modulating the moment of inertia at a frequency
twice the mean natural frequency generates the greatest growth of the amplitude,
provided that the phase of the modulation is chosen in the way described above.
It is evident that the energy of the oscillator is increased most greatly not
only when two full cycles of variation in the parameter occur during one natural
period of oscillation, but also when two cycles occur during three, five, or any odd
number of natural periods. We shall see later that the delivery of energy, though
less efficient, is also possible if two cycles of modulation occur during an even
number of natural periods.
If the changes of a parameter are produced with the above-mentioned periodic-
ity but not abruptly, the influence of these changes on the oscillator is qualitatively
quite similar, though the efficiency of the parametric delivery of energy (at the
same amplitude of the parametric modulation) is a maximum for the square-wave
time dependence, because this form of modulation provides optimal conditions
for the transfer of energy to the oscillating system. The case of the sinusoidal
modulation of some parameter is important for practical applications.

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

5.1. SUMMARY OF THE THEORY 97

5.1.3 Electromagnetic Analogue of the Mechanical System

Parametric excitation is possible in various oscillatory systems. Electromagnetic
oscillations in a series LCR-circuit containing a capacitor, an inductor (a coil), and
a resistor can be excited by periodic changes of the capacitance if we periodically
move the plates closer together and farther apart, or by changes of the inductance
of the coil if we periodically move an iron core in and out of the coil, as shown
in the right-hand panel of Figure 5.1. Such periodic changes of the inductance are
quite similar to the changes of the moment of inertia in the mechanical system
considered above. The strongest oscillations are excited when the cycle of such
changes is repeated twice during one period of natural electromagnetic oscillations
in the circuit, i.e., when the frequency of a parametric modulation is twice the
natural frequency of the system. It is evident that parametric excitation can occur
only if at least weak natural oscillations already exist in the system.
Parametric excitation is possible only with the modulation of one of the
energy-consuming parameters, C or L (D or J in the mechanical system). Modu-
lation of the resistance R (or of the damping constant γ in the mechanical system)
can affect only the character of the damping of oscillations. It cannot generate an
increase in their amplitude.
The mechanical system simulated in the relevant computer program has cer-
tain spectacular didactic advantages primarily because its motion is easily repre-
sented on the computer screen, and it is possible to see directly what is happening.
Such visualization makes the simulation experiments very convincing and easy to
understand, aiding a great deal in developing our physical intuition.

5.1.4 Conditions for Parametric Resonance

There are several important differences that distinguish parametric resonance from
the ordinary resonance caused by an external force acting directly on the system.
The growth of the amplitude and hence of the energy of oscillations during para-
metric excitation is provided by the work of forces that cause the modulation, that
is, by the work of forces that periodically change the parameter.
The most efficient energy transfer to the oscillatory system occurs when the
parameter is changed twice during one period of the excited natural oscillations.
But the delivery of energy, though less efficient, is possible when the parameter
changes once during one period, twice during three periods, and so on. That is,
resonance is possible when one of the following conditions for the frequency ω
(or for the period T ) of a parameter modulation is fulfilled:

ω = 2ω0 /n, T = nT0 /2, n = 1, 2, . . . (5.2)

For a given amplitude of parametric modulation, the higher the order n of

resonance, the less (in general) the amount of energy delivered to the oscillating
system during one period.
One of the most interesting characteristics of parametric resonance is the pos-
sibility of exciting increasing oscillations not only at the frequencies ωn given in

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

98 CHAPTER 5. PARAMETRIC EXCITATION OF OSCILLATIONS

Eq. (5.2), but also in a range of frequencies ω lying on either side of the values ωn
(in the ranges of instability.) These intervals become wider as the degree (depth)
of modulation is increased, that is, as the range of parametric variation is extended.
By this range we mean, in the case of the rotor, the difference in the maximal and
minimal values of its moment of inertia, and in the oscillating electrical circuit,
the differences in the inductance of the coil.
An important difference between parametric excitation and forced oscillations
is related to the dependence of the growth of energy on the energy already stored
in the system. While for forced excitation the increment of energy during one
period is proportional to the amplitude of oscillations, i.e., to the square root of
the energy, at parametric resonance the increment of energy is proportional to the
energy stored in the system.
Energy losses caused by friction (unavoidable in any real system) are also pro-
portional to the energy already stored. In the case of direct forced excitation, an
arbitrarily small external force gives rise to resonance. However, energy losses re-
strict the growth of the amplitude because these losses grow faster with the energy
than does the investment of energy arising from the work done by the external
force.
In the case of parametric resonance, both the investment of energy caused by
the modulation of a parameter and the frictional losses are proportional to the
energy stored (to the square of the amplitude), and so their ratio does not depend
on the amplitude.
This means that parametric resonance is possible only when a threshold is
exceeded, that is, when the increment of energy during a period (caused by the
parametric variation) is larger than the amount of energy dissipated during the
same time. To satisfy this requirement, the range of the parametric variation (the
depth of modulation) must exceed some critical value.
The critical (threshold) value of the modulation depth depends on friction.
However, if the threshold is exceeded, the frictional losses of energy cannot re-
strict the growth of the amplitude. In a linear system the amplitude of parametri-
cally excited oscillations must grow infinitely.
In a nonlinear system the natural period depends on the amplitude of oscilla-
tions. If conditions for parametric resonance are fulfilled at small oscillations and
the amplitude begins to grow, the conditions of resonance become violated at large
amplitudes. In a real system the growth of the amplitude is restricted by nonlinear
effects.

5.1.5 The Threshold of Parametric Excitation

We can use arguments employing the conservation laws to evaluate the modulation
depth that corresponds to the threshold of parametric excitation of the torsion
oscillator.
Let the changes in the moment of inertia J of the rotor occur between maximal
J1 and minimal J2 values that equal J0 (1 + m) and J0 (1 − m) respectively,
where J0 is a mean value of the moment of inertia, and m is the dimensionless

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

5.1. SUMMARY OF THE THEORY 99

quantity called the depth of modulation, which equals fractional increments and
decrements of the modulated parameter.
During abrupt radial displacements of the weights along the rod, the angular
momentum L = Jω of the rotor is conserved. Therefore it is convenient to use the
expression Ekin = L2 /(2J), which gives the kinetic energy of the rotor in terms
of L and J.
For the increment ∆E in the rotor kinetic energy that occurs during an abrupt
shift of the weights toward the axis, when the moment of inertia decreases from
the value J1 = J0 (1 + m) to the value J2 = J0 (1 − m), we can write:
L2 L2
 
1 1
∆E = − ≈ 2m ≈ 2mEkin . (5.3)
2J0 1 − m 1 + m 2J0
The approximate expressions in (5.3) are valid for small values of the modulation
depth (m ≪ 1). If the event occurs near the equilibrium position of the rotor,
when the total energy E of the pendulum is approximately its kinetic energy Ekin ,
Eq. (5.3) shows that the fractional increment in the total energy ∆E/E approxi-
mately equals twice the value of the modulation depth m: ∆E/E ≈ 2m.
When the frequencies and phases have those values that are favorable for the
most effective delivery of energy, the abrupt displacement of the weights toward
the ends of the rod occurs at the instant when the rotor attains its greatest deflection
(more precisely, when the rotor is very near it). At this instant the angular velocity
of the rotor is almost zero, and so this radial displacement of the weights into their
previous positions causes nearly no decrement in the energy.
For the principal resonance (n = 1) the investment in energy occurs twice
during the natural period T0 of oscillations. That is, the fractional increment in
energy ∆E/E during one period approximately equals 4m.
A process in which the increment in energy ∆E during a period is propor-
tional to the energy stored E (in the case under consideration ∆E ≈ 4mE) is
characterized on the average by the exponential growth of the energy with time:

E(t) = E0 exp(αt). (5.4)

In this case the index of growth α is proportional to the depth of modulation m of
the moment of inertia: α = 4m/T0 . When the modulation is exactly tuned to the
principal resonance (T = T0 /2), the decrease of energy is caused almost only by
friction.
Dissipation of energy due to viscous friction during an integral number of
natural cycles (for t = nT0 ) is described by the following expression:
E(t) = E0 exp(−2γt). (5.5)
Comparing Eqs. (5.4) and (5.5), we obtain the following estimate for the
threshold (minimal) value mmin of the depth of modulation corresponding to the
excitation of the principal parametric resonance:
1 π
mmin = γT0 = . (5.6)
2 2Q

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

100 CHAPTER 5. PARAMETRIC EXCITATION OF OSCILLATIONS

0.44
∆ J( t) .
ϕ (t )
0

-0.44 o o
-20.7 0 20.7
ϕ (t ) ∆ J( t)
0

o
-20.7
0 1 2T
(period of modulation 0.5T0, depth of modulation 20.0%, quality factor 8.29,
o
initial deflection 20.0 , initial angular velocity -0.07 w0)

Figure 5.2: The phase trajectory (left) and the time-dependent graphs of ϕ̇(t) and
ϕ(t) (right) of stationary oscillations at the threshold condition m ≈ π/2Q for
T = T0 /2.

Here we introduced the dimensionless quality factor Q = ω0 /(2γ) to characterize

friction in the system.
The phase trajectory and the time-dependent plots of the angular velocity and
the angle of deflection for parametric oscillations occurring at the threshold condi-
tions, Eq. (5.6), are shown in Figure 5.2. This mode of steady oscillations (which
have a constant amplitude in spite of the dissipation of energy) is called paramet-
ric regeneration.
For the third resonance (T = 3T0 /2) the threshold value of the depth of
modulation is three times greater than its value for the principal resonance:
mmin = 3π/(2Q). In this instance two cycles of the parametric variation occur
during three full periods of natural oscillations. In conditions of the third para-
metric resonance, radial displacements of the weights again happen at the most
favorable moments for pumping energy to the oscillator just like in conditions of
the principal resonance. Therefore, at the same depth of modulation, the same in-
vestment in energy occurs during an interval that is three times longer than the
interval for the principal resonance.
When the depth of modulation exceeds the threshold value, the energy of os-
cillations increases exponentially with time. The growth of the energy again is
described by (5.4). However, now the index of growth α is determined by the
amount by which the energy delivered through parametric modulation exceeds
the simultaneous losses of energy caused by friction: α = 4m/T0 − 2γ. The
amplitude of parametrically excited oscillations also increases exponentially with
time (Figure 5.3): a(t) = a0 exp(βt). The index β in the growth of amplitude

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

5.1. SUMMARY OF THE THEORY 101

0.60 .
∆ J( t) ϕ (t )

-0.60 o o
-29.3 0 29.3
ϕ (t ) ∆ J( t)

o
-29.3
0 2 4 6 8 10T
(period of modulation 0.5T0, depth of modulation 20%, quality factor 18.0,
o
initial deflection 10 , initial angular velocity 0.0)

Figure 5.3: Exponential growth of the amplitude of oscillations at conditions of

the first order parametric resonance (n = 1).

is one half the index of the growth in energy. For the principal resonance, when
the investment in energy occurs twice during one natural period of oscillation, we
have β = 2m/T0 − γ = mω0 /π − γ.

5.1.6 Differential Equation for Parametric Oscillations

We next consider a more rigorous mathematical treatment of parametric resonance
under a square-wave modulation of the parameter. This treatment is based on the
differential equations governing the phenomenon.
During the time intervals (0, T /2) and (T /2, T ), the value of the moment
of inertia is constant, and the motion of the rotor can be considered as a free
oscillation described by a linear differential equation. However, the coefficients in
this equation are different for the adjacent time intervals (0, T /2) and (T /2, T ):
1
ϕ̈ = − (ω 2 ϕ + 2γ ϕ̇) for 0 < t < T /2, (5.7)
1+m 0
1
ϕ̈ = − (ω 2 ϕ + 2γ ϕ̇) for T /2 < t < T. (5.8)
1−m 0
p
Here ω0 = D/J0 is the natural frequency of the oscillator and γ is the damping
constant characterizing the strength of viscous friction. Both these quantities cor-
respond to the mean value J0 = 21 (J1 + J2 ) of the moment of inertia. For small
and moderate values of m, the moment of inertia equals J0 when the weights are
near the half-way point between their extreme positions on the rod. For large m
this is not the case because the moment of inertia depends on the square of the
distance of the weights from the axis of rotation.

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

102 CHAPTER 5. PARAMETRIC EXCITATION OF OSCILLATIONS

At each instant tn = nT /2 (n = 1, 2, . . .) of an abrupt change in the moment

of inertia we must make a transition from one of the linear Eqs. (5.7)–(5.8) to
the other. During each half-period T /2 the motion of the oscillator is a segment
of some harmonic (or damped) natural oscillation. An analytical investigation of
parametric excitation can be carried out by fitting to one another known solutions
to the linear Eqs. (5.7)–(5.8) for consecutive adjacent time intervals.
The initial conditions for each subsequent time interval are chosen accord-
ing to the physical model in the following way. Each initial value of the angular
displacement ϕ equals the value ϕ(t) reached by the oscillator at the end of the
preceding time interval. The initial value of the angular velocity ϕ̇ is related to the
angular velocity at the end of the preceding time interval by the law of conserva-
tion of the angular momentum:

(1 + m)ϕ̇1 = (1 − m)ϕ̇2 . (5.9)

In Eq. (5.9) ϕ̇1 is the angular velocity at the end of the preceding time interval,
when the moment of inertia of the rotor has the value J1 = J0 (1 + m), and ϕ̇2
is the initial value for the following time interval, during which the moment of
inertia is equal to J2 = J0 (1 − m). The change in the angular velocity at an
abrupt variation of the inertia moment from the value J2 to J1 can be found in the
same way.
That we may use the conservation of angular momentum, as expressed in
Eq. (5.9), is allowed because, at sufficiently rapid displacement of the weights
along the rotor, we can neglect the influence of the spring and consider the rotor
as if it were freely rotating about its axis. This assumption is valid provided the
duration of the displacement of the weights is a small portion of the natural period.
Considering conditions for which Eqs. (5.7)–(5.8) yield solutions with in-
creasing amplitudes, we can determine the ranges of frequency ω near the values
ωn = 2ω0 /n, within which the state of rest is unstable for a given modulation
depth m. In these ranges of instability an arbitrarily small deflection from equilib-
rium is sufficient for the progressive growth of small initial oscillations.

5.1.7 The Mean Natural Period at Large Depth of Modulation

The threshold for the parametric excitation of the torsion pendulum is determined
above for the resonant situations in which two cycles of the parametric modulation
occur during one natural period or during three natural periods of oscillation. The
estimate obtained, Eq. (5.6), is valid for small values of the modulation depth m.
For large values of the modulation depth m, the notion p of a natural period
needs a more precise definition. Let T0 = 2π/ω0 = 2π J0 /D be the period of
oscillation of the rotor when the weights are fixed in some middle positions, for
which the moment of inertia equals J0 . The √period is somewhat longer when the
weights are moved further apart: T1 = T0 1 + m ≈ T0 (1 + m/2). √ The period
is shorter when the weights are moved closer to one another: T2 = T0 1 − m ≈
T0 (1 − m/2).

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

5.2. FREQUENCY RANGES OF PARAMETRIC EXCITATION 103

It is convenient to define the average period Tav not as the arithmetic mean
1
2 (T1 + T2 ), but rather as the period that corresponds to the arithmetic mean fre-
quency ωav = 12 (ω1 + ω2 ), where ω1 = 2π/T1 and ω2 = 2π/T2 . So we define
Tav by the relation:
2π 2T1 T2
Tav = = . (5.10)
ωav (T1 + T2 )
The period T of the parametric modulation that is exactly tuned to any of the
parametric resonances is determined not only by the order n of the resonance, but
also by the depth of modulation m. In order to satisfy the resonant conditions, the
increment in the phase of natural oscillations during one cycle of modulation must
be equal to π, 2π, 3π, . . . , nπ, . . . During the first half-cycle the phase increases
by ω1 T /2, and during the second half-cycle by ω2 T /2. Consequently, instead of
the approximate condition expressed by Eq. (5.2), we obtain:
ω1 + ω2 π Tav
T = nπ, or T =n =n . (5.11)
2 ωav 2
Thus, for a parametric resonance of some definite order n, the condition for
exact tuning can be expressed in terms of the two natural periods, T1 and T2 . This
condition is T = nTav /2, where Tav is defined by Eq. (5.10). For moderate values
of m it is possible to use approximate expressions for the average frequency and
period:
 
ω0 1 1 3 3
ωav = √ +√ ≈ ω0 (1 + m2 ), Tav ≈ T0 (1 − m2 ).
2 1+m 1−m 8 8
(5.12)
The difference between Tav and T0 reveals itself in terms proportional to the
square of the depth of modulation m.

5.2 Frequency Ranges of Parametric Excitation

An infinite growth of the amplitude during parametric excitation is possible not
only at exact tuning to one of the resonances but in certain intervals of T -values.
These intervals, or the ranges of instability, surround the resonant values T =
Tav /2, T = Tav , T = 3Tav /2, . . . The width of the intervals increases with
the depth m of the parameter modulation.1 Outside the intervals, the equilibrium
position of a torsion pendulum is stable, and the amplitude of oscillations does not
grow.
In order to determine the boundaries of the frequency ranges of parametric
instability surrounding the resonant values T = Tav /2, T = Tav , T = 3Tav /2,
. . . , we can consider stationary oscillations that occur when the period of modu-
lation T corresponds to one of the boundaries. These stationary oscillations can
1 Strictly speaking, for high resonances (n ≥ 2) this statement is true only for small and moderate

values of the depth of modulation (see Figure 5.6 on p. 108 with the diagram of the ranges of parametric
resonance).

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

104 CHAPTER 5. PARAMETRIC EXCITATION OF OSCILLATIONS

.
∆ J( t) ϕ (t )

0 0
-T/2 T/2

-1 0 1 2 3T
Depth of modulation 30%, period of modulation 0.4363T0

Figure 5.4: Stationary parametric oscillations at the lower boundary of the princi-
pal interval of instability (near T = Tav /2).

be represented as an alternation of natural oscillations with the periods T1 and T2 .

In the absence of friction the graphs of such oscillations are formed by segments
of non-damped sine curves with the corresponding periods.

5.2.1 Main Interval of Parametric Excitation

We examine first the vicinity of the principal resonance occurring at T = Tav /2.
Suppose that the period T of the parametric variation is a little shorter than the
resonant value T = Tav /2, so that T corresponds to the left boundary of the inter-
val of instability. In this case a little less than a quarter of the mean natural period
Tav elapses between consecutive abrupt increases and decreases of the moment of
inertia. The graph of the angular velocity ϕ̇(t) for this periodic stationary process
has the characteristic pattern shown in Figure 5.4. The segments of the graphs of
free oscillations (which occur at time intervals during which the moment of inertia
is constant) are alternating parts of sine or cosine curves with the periods T1 and
T2 . These segments are symmetrically truncated on both sides.
To find conditions at which such stationary oscillations take place, we can
write the expressions for ϕ(t) and ϕ̇(t) during the adjacent intervals in which
the oscillator executes natural oscillations, and then fit these expressions to one
another at the boundaries. Such fitting must provide a periodic stationary process.
We let the origin of time, t = 0, be the instant when the weights are shifted
apart. The angular velocity is abruptly decreased in magnitude at this instant (see
Figure 5.4). Then during the interval (0, √ T /2) the graph describes a natural os-
cillation with the frequency ω1 = ω0 / 1 + m. It is convenient to represent this
motion as a superposition of sine and cosine waves whose constant amplitudes are
A1 and B1 , respectively:

ϕ1 (t) =A1 sin ω1 t + B1 cos ω1 t,

(5.13)
ϕ̇1 (t) =A1 ω1 cos ω1 t − B1 ω1 sin ω1 t.

Similarly, during the interval (−T /2, 0) the graph in Figure 5.4 is a segment of

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

5.2. FREQUENCY RANGES OF PARAMETRIC EXCITATION 105

√
natural oscillation with the frequency ω2 = ω0 / 1 − m:

ϕ2 (t) =A2 sin ω2 t + B2 cos ω2 t,

(5.14)
ϕ̇2 (t) =A2 ω2 cos ω2 t − B2 ω2 sin ω2 t.

To determine the values of constants A1 , B1 and A2 , B2 , we can use the con-

ditions that must be satisfied when the segments of the graph are joined together,
taking into account the periodicity of the stationary process.
At t = 0 the angle of deflection is the same for both ϕ1 and ϕ2 : ϕ1 (0) =
ϕ2 (0). From this condition we find that B1 = B2 . We later denote these equal
constants simply by B. The angular velocity at t = 0 undergoes a sudden change:

(1 + m)ϕ̇1 (0) = (1 − m)ϕ̇2 (0).

This condition gives us the following relation between A2 and A1 : A2 = kA1 =

kA, where we have denoted A1 simply by A and introduced a dimensionless quan-
tity k, which depends on the depth of modulation m:
r
1+m
k= . (5.15)
1−m
Equations for the constants A and B are determined by the conditions at the
instants −T /2 and T /2. For stationary periodic oscillations, corresponding to
the principal resonance (and to all resonances of odd numbers n = 1, 3, . . .
in Eq. (5.11)), these conditions are:

ϕ1 (T /2) = −ϕ2 (−T /2), (1 + m)ϕ̇1 (T /2) = −(1 − m)ϕ̇2 (−T /2). (5.16)

Substituting ϕ and ϕ̇ from Eq. (5.14) in Eq. (5.16), we obtain the system of ho-
mogeneous equations for the unknown quantities A and B:

(S1 − kS2 )A + (C1 + C2 )B = 0,

(5.17)
k(C1 + C2 )A − (kS1 − S2 )B = 0.

In Eqs. (5.17) the following notations are used:

C1 = cos(ω1 T /2), C2 = cos(ω2 T /2),

(5.18)
S1 = sin(ω1 T /2), S2 = sin(ω2 T /2).

The homogeneous system of equations for A and B, Eqs. (5.17), has a non-trivial
(non-zero) solution only if its determinant is zero:

2kC1 C2 − (1 + k 2 )S1 S2 + 2k = 0. (5.19)

This condition for the existence of a non-zero solution to Eqs. (5.17) gives us
an equation for the unknown variable T , which enters in Eq. (5.19) as the argu-
ments of sine and cosine functions in S1 , S2 and C1 , C2 . This equation determines

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

106 CHAPTER 5. PARAMETRIC EXCITATION OF OSCILLATIONS

the desired boundaries of the interval of instability for a given value of the mod-
ulation depth m, which enters in the parameter k defined by Eq. (5.15). These
boundaries T− and T+ are given by the roots of the equation, Eq. (5.19).
To find approximate solutions T to this transcendental equation, Eq. (5.19),
we transform it into a more convenient form. We first represent in Eq. (5.19) the
products C1 C2 and S1 S2 as follows:
1 ∆ωT 1 ∆ωT
C1 C2 = (cos + cos ωav T ), S1 S2 = (cos − cos ωav T ),
2 2 2 2
where ∆ω = ω2 − ω1 . Then, using the identity cos α = 2 cos2 (α/2) − 1, we
reduce Eq. (5.19) to the following form:
ωav T ∆ωT
(1 + k) cos = ±|1 − k| cos . (5.20)
2 4
For the boundaries of the instability interval that contains the principal reso-
nance n = 1, we search for a solution T of Eq. (5.20) in the vicinity of T = T0 /2.
For a given value of the depth of modulation m, Eq. (5.20) in the neighborhood of
T0 /2 ≈ Tav /2 has two solutions that correspond to the boundaries T− and T+ of
the instability interval. The phase diagram and the graph of the angular velocity
for the right boundary of the main interval (n = 1) are shown in Figure 5.5.
To find the boundaries T− and T+ of the instability interval, we replace T in
the argument of the cosine on the left-hand side of Eq. (5.20) by Tav /2 + ∆T ,
where ∆T ≪ T0 . Since ωav Tav = 2π, we can replace the cosine in Eq. (5.20)
with − sin(ωav ∆T /2). Then Eq. (5.20) becomes:
ωav ∆T |1 − k| ∆ω(Tav /2 + ∆T )
sin =∓ sin . (5.21)
2 1+k 4
This equation for ∆T can be solved numerically by iteration. We start with ∆T =
0 as an approximation of the zeroth order, substituting it into the right-hand side
of Eq. (5.21), taken, say, with the upper sign. Then the left-hand side of Eq. (5.21)
gives us the value of ∆T to the first order. We substitute this first-order value into
the right-hand side of Eq. (5.21), and on the left-hand side we obtain ∆T to the
second order. This procedure is iterated until a self-consistent value of ∆T for the
left boundary is obtained. To determine ∆T for the right boundary, we use the
same procedure, taking the lower sign on the right-hand side of Eq. (5.21).
After the substitution of one of the roots T− or T+ of Eq. (5.21) into
Eqs. (5.17) both equations for A and B become equivalent and permit us to find
only the ratio A/B. This limitation means that the amplitude of stationary os-
cillations at the boundary of the instability interval can be arbitrarily large. This
amplitude depends on the initial conditions. Nevertheless, these oscillations have a
definite shape that is determined by the ratio of the amplitudes A and B of the sine
and cosine functions whose segments form the pattern of the stationary parametric
oscillation (see Figures 5.4 and 5.5).
The periods of modulation T− and T+ , corresponding respectively to the left
and right boundaries of the instability interval that contains the principal paramet-
ric resonance n = 1, calculated numerically for different vales of the modulation

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

5.2. FREQUENCY RANGES OF PARAMETRIC EXCITATION 107

.
ϕ (t ) ∆ J( t)

0 0
-T/2 T/2

-1 0 1 2 3T
Depth of modulation 30%, period of modulation 0.5282T0

Figure 5.5: Stationary parametric oscillations at the upper boundary of the princi-
pal interval of instability (near T = Tav /2).

depth m with the help of the above-described procedure, are shown by the bor-
der lines of the first “tongue” in Figure 5.6. The central curve of this “tongue”
gives the period of modulation T = Tav /2 as a function of the modulation depth
m, which corresponds to exact tuning to the principal parametric resonance. The
other “tongues” in Figure 5.6 show the intervals of parametric instability of high
orders, discussed further on in this section.
To obtain an approximate analytic solution to Eq. (5.21) that is valid for small
values of the modulation depth m, we can simplify the expression on the right-
hand side by assuming k ≈ 1 + m, |1 − k| ≈ m. We may also assume the value
of the cosine to be 1. On the left-hand side of Eq. (5.21), the sine can be replaced
by its small argument, in which ωav = 2π/Tav . Thus we obtain the following
approximate expression that is valid up to terms to the second order in m:
1 m
T∓ = 1∓ Tav . (5.22)
2 π
p
Since the natural period T0 = 2π D/J0 is used in the relevant simulation
program as an appropriate time unit for the input of the period of modulation T ,
we express the values of T∓ given by Eq. (5.22) also in terms of T0 :

m 3m2
 
1
T∓ = 1∓ − T0 . (5.23)
2 π 8

5.2.2 Third-Order Interval of Parametric Instability

In a similar way we can determine the boundaries of the instability interval in the
vicinity of a resonance of the higher order n = 3. At the third order resonance
(n = 3) two cycles of variation of the inertia moment occur during approximately
three natural periods of oscillation (T ≈ 3Tav /2).
The phase trajectories and the time-dependent graphs of stationary oscillations
at the left and right boundaries of the third interval are shown in Figure 5.7.
The phase orbit of the periodic oscillation closes after two cycles of modu-
lation. This orbit is formed by two concentric ellipses that correspond to natural

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

108 CHAPTER 5. PARAMETRIC EXCITATION OF OSCILLATIONS

Depth of modulation, %

Period of modulation (natural periods)

Figure 5.6: Intervals of parametric instability at square-wave modulation of the

inertia moment in the absence of friction.

oscillations of the flywheel with frequencies ω1 and ω2 . The representative point

moves clockwise along this orbit, jumping from one ellipse to the other each time
the weights are shifted along the rod of the flywheel. The numbers in Figure 5.7
make it easier to follow how the representative point describes this orbit: Equiva-
lent points of the phase orbit and the graph of the angular velocity are marked by
equal numbers.
Considering conditions at which the graphs of natural oscillations with fre-
quencies ω1 and ω2 on the left boundary fit one another for adjacent time in-
tervals and produce the periodic process shown in Figure 5.7, we get the same
Eqs. (5.16)–(5.17) for A1 and A2 , as well as Eq. (5.22) for the period of modula-
tion. Actually, this is true for all intervals of parametric instability of odd orders.
Similarly, for the right boundary we get the same equations for B1 and B2 as
in case n = 1, and also Eq. (5.22) with the opposite sign for determination of
the corresponding period of modulation T . However, if we are interested in the
third interval, we should search for a solution to these equations in the vicinity of
T = 3Tav /2, as well as for any other interval of odd order n — in the vicinity of
T = nTav /2. The boundaries of intervals of the third and fifth orders, obtained by
a numerical solution, are also shown in Figure 5.6.
For small values of the depth of modulation m, we can find approximate an-
alytic expressions for the lower and the upper boundaries of the interval that are
valid up to quadratic terms in m:
 
3 m
T∓ = ∓ Tav . (5.24)
2 2π

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

5.2. FREQUENCY RANGES OF PARAMETRIC EXCITATION 109

. ∆ J( t)
8 5
ϕ (t )
5 8
7 6 6 7
2 0 3 2 3
1 4 1
1 4 -T/2 T/2
-1 0 1 2 3T
Depth of modulation 30%, period of modulation 1.4116T0
1 8 . 1 8 1
ϕ (t ) ∆ J( t)
2 7 2 7

0
3 6 -T/2 3 6
T/2
4 5 4 5
-1 0 1 2 3T
Depth of modulation 25%, period of modulation 1.4964T0

Figure 5.7: The phase trajectories (left) and the time-dependent graphs of the an-
gular velocity (right) of stationary parametric oscillations at the left (upper panel)
and right (lower panel) boundaries of the interval of instability near T = 3Tav /2.

In terms of the mean natural period T0 these boundaries are expressed as follows:
9m2
 
3 m
T∓ = ∓ − T0 . (5.25)
2 2π 16
In this approximation, the third interval has the same width (m/π)T0 as does
the interval of instability in the vicinity of the principal resonance. However, this
interval is distinguished by a greater asymmetry: Its central point is displaced to
the left of the value T = 3T0 /2 by (9/16)m2 T0 .

5.2.3 Frequency Ranges for Resonances of Even Orders

For small and moderate square-wave modulation of the moment of inertia, para-
metric resonance of the order n = 2 (one cycle of the parametric variation during
one natural period of oscillation) is relatively weak compared to the resonances
n = 1 and n = 3 considered above. In the case in which n = 2 the abrupt changes
of the moment of inertia induce both an increase and a decrease of the energy only
once during each natural period. The growth of oscillations occurs only if the in-
crease in energy at the instant when the weights are drawn closer is greater than
the decrease in energy when the weights are drawn apart. This is possible only if
the weights are shifted toward the axis when the angular velocity of the rotor is
greater in magnitude than it is when they are shifted apart. For T ≈ Tav , these
conditions can be fulfilled only because there is a small difference between the
natural periods T1 and T2 of the rotor, where T1 is the period with the weights

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

110 CHAPTER 5. PARAMETRIC EXCITATION OF OSCILLATIONS

.
∆ J ( t ) ϕ (t )

0 1 2 3 4 5 6T
Depth of modulation 30%, period of modulation 0.9651T0

Figure 5.8: The phase trajectory (left) and the graph of the angular velocity (right)
of oscillations at parametric resonance of the second order (T = Tav ).

shifted apart and T2 is the period with them shifted together. This difference is
proportional to m.
The growth of oscillations at parametric resonance of the second order is
shown in Figure 5.8. In this case, the investment of energy during a period is
proportional to the square of the depth of modulation m, while in the cases of res-
onances with n = 1 and n = 3 the investment of energy is proportional to the first
power of m. Therefore, for the same value of the damping constant γ (the same
quality factor Q), a considerably greater depth of modulation is required here to
exceed the threshold of parametric excitation.
The interval of instability in the vicinity of resonance with n = 2 is consid-
erably narrower compared to the corresponding intervals of the resonances with
n = 1 and n = 3. Its width is also proportional only to the square of m (for small
values of m).
To determine the boundaries of this interval of instability, we can consider,
as is done above for other resonances, stationary oscillations for T ≈ T0 formed
by alternating segments of free sinusoidal oscillations with the periods T1 and
T2 . The graph of the angular velocity and the phase trajectory of such stationary
periodic oscillations for one of the boundaries are shown in Figure 5.9. During os-
cillations occurring at the boundary of the instability interval, the abrupt increment
and decrement in the angular velocity exactly compensate each other.
To describe these stationary oscillations, we can use the same expressions for
ϕ(t) and ϕ̇(t) as we use in Eqs. (5.13)–(5.14). The conditions for joining the
graphs at t = 0 are also the same. However, differences begin with the equations
for the constants A and B. They are determined by the conditions of periodicity
at the instants −T /2 and T /2. For stationary periodic oscillations, corresponding
to resonance with n = 2 (and for all resonances of even orders n = 2, 4, . . . in
Eq. (5.11)), these conditions are:

ϕ1 (T /2) = ϕ2 (−T /2), (1 + m)ϕ̇1 (T /2) = (1 − m)ϕ̇2 (−T /2), (5.26)

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

5.2. FREQUENCY RANGES OF PARAMETRIC EXCITATION 111

.
∆ J( t) ϕ (t )

-1 0 1 2T
Depth of modulation 35%, period of modulation 0.9232T0

Figure 5.9: Stationary parametric oscillations at one of the boundaries of the in-
terval of instability of the second order (near T = Tav ≈ T0 ).

and we obtain the system of equations for the amplitudes A and B:

(S1 + kS2 )A + (C1 − C2 )B = 0,

(5.27)
k(C1 − C2 )A − (kS1 + S2 )B = 0,

where S1 , C1 and S2 , C2 are defined by the same Eqs. (5.18). The homogeneous
system of equations for A and B, Eqs. (5.27), has a non-trivial solution if its
determinant is zero:

2kC1 C2 − (1 + k 2 )S1 S2 − 2k = 0. (5.28)

In order to find the values T∓ = Tav + ∆T for the instability interval with n = 2
from Eq. (5.28), we transform the products C1 C2 and S1 S2 in Eq. (5.28) by using
the identity cos α = 1 − 2 sin2 (α/2):
ωav T ∆ωT
(1 + k) sin = ±|1 − k| sin . (5.29)
2 4
We next replace T in the argument of the sine on the left-hand side of Eq. (5.29)
by Tav + ∆T , where ∆T ≪ T0 . Since ωav Tav = 2π, we can write this sine as
− sin(ωav ∆T /2). Then Eq. (5.20) becomes:

ωav ∆T |1 − k| ∆ω(Tav + ∆T )
sin =∓ cos . (5.30)
2 1+k 4
This equation gives the left boundary T− of the instability interval when we
take the upper sign in its right-hand side, and the right boundary T+ when we take
the lower sign. Stationary oscillations, which correspond to the right boundary,
are shown in Figure 5.10.
Equation (5.30) for ∆T can be also solved numerically by iteration. Substi-
tuting T− or T+ obtained from (5.30) into one of the equations in (5.17), we get
the ratio of the amplitudes A and B that determines the pattern of stationary os-
cillations at the corresponding boundary of the instability interval. We note how
narrow the intervals of even resonances (n = 2, 4) are for small values of m. With

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

112 CHAPTER 5. PARAMETRIC EXCITATION OF OSCILLATIONS

.
∆ J( t) ϕ (t )

-1 0 1 2T
Depth of modulation 40%, period of modulation 0.9756T0

Figure 5.10: Stationary parametric oscillations at the other boundary of the inter-
val of instability of the second order (near T = Tav ≈ T0 ).

the growth of m the intervals expand and become comparable with the intervals
of odd orders.
For moderate values of the depth of modulation, it is possible to find an ap-
proximate analytical solution of Eq. (5.30):
 
1
T∓ = 1 ∓ m2 Tav . (5.31)
4

In terms of T0 the boundaries of the second interval are:

 
1 3
T∓ = T0 + ∓ − m2 T 0 , (5.32)
4 8

i.e., T− = T0 − (5/8)m2 T0 , T+ = T0 − (1/8)m2 T0 . As mentioned above, the

width T+ − T− = (m2 /2)T0 of this interval of instability is proportional to the
square of the modulation depth.
The intervals of instability for the first five parametric resonances at square-
wave modulation of the inertia moment in the absence of friction are shown by
the shaded “tongues” in Figure 5.6) for various values of the modulation depth
m. The diagram is obtained by numerical solution of the equations that are dis-
cussed above. We note how narrow the intervals of even resonances (n =2, 4) are
for small values of m. With the growth of m the intervals expand and become
comparable with the intervals of odd orders.

5.2.4 Intersections of the Boundaries at Large Modulation

Figure 5.6 shows that at some definite values of m both boundaries of intervals
with n > 2 coincide (we may consider that they intersect). Thus at these values of
m the corresponding intervals of parametric resonance disappear. These values of
m correspond to the natural periods T1 and T2 of oscillation (associated with the
weights far apart and close to each other), whose ratio is 2 : 1, 3 : 1, and 3 : 2. For
the corresponding values of the modulation depth m and the period of modulation
T , steady oscillations occur for arbitrary initial conditions.

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

5.2. FREQUENCY RANGES OF PARAMETRIC EXCITATION 113

.
∆ J( t) ϕ (t )

0 1 2 3 4T
Depth of modulation 60%, period of modulation 1.2649T0

Figure 5.11: The phase trajectory and the time-dependent graph of the angular
velocity for stationary oscillations at the intersection of both boundaries of the
third interval.

For the first intersection (ratio 2 : 1) exactly one half of the natural oscillation
with period T1 is completed during the first half of the modulation cycle (see
Figure 5.11). On the phase diagram, the representing point traces a half of the
smaller ellipse, and then abruptly jumps down to the larger ellipse. During the
second half of the modulation cycle the oscillator executes exactly a whole natural
oscillation with period T2 = T1 /2, so that the representing point passes in the
phase plane along the whole larger ellipse, and then jumps up to the smaller ellipse
along the same vertical segment.
During the next modulation cycle the representing point first generates the
other half of the smaller ellipse, and then again the whole larger ellipse. Therefore
during any two adjacent cycles of modulation the representing point passes once
along the closed smaller ellipse and twice along the larger one, finally returning
to the initial point of the phase plane. We see that such an oscillation is periodic
for arbitrary initial conditions. This means that for the corresponding values of the
modulation depth m and the period of modulation T the growth of amplitude is
impossible even in the absence of friction (the instability interval vanishes).
Similar explanations can be suggested for other cases in Figure 5.6 in which
the boundaries intersect.

5.2.5 Intervals of Excitation in the Presence of Friction

When there is friction in the system, the intervals of the period of modulation that
correspond to the parametric instability become narrower, and for strong enough
friction (below the threshold) the intervals disappear.
Next we show that above the threshold approximate values for the bound-
first interval are given by Eq. (5.22) provided we substitute for m the
aries of the p
expression m2 − m2min with the threshold value mmin = π/(2Q) defined by
Eq. (5.6). p
For the third interval, we can use Eq. (5.24), substituting m2 − m2min for
m, with mmin = 3π/(2Q). When m is equal to the threshold value mmin , the
corresponding interval of parametric resonance disappears.

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

114 CHAPTER 5. PARAMETRIC EXCITATION OF OSCILLATIONS

Figure 5.12: Intervals of parametric excitation at square-wave modulation of the

moment of inertia without friction, for Q = 20, and for Q = 10.

The boundaries of the second interval of parametric resonance in the presence

of friction are approximately
p given by Eq. (5.31) provided we substitute
p for m2
4 4
the expression m − mmin with the threshold value mmin = 2/Q, which
corresponds to the second parametric resonance.
The diagram in Figure 5.12 shows the boundaries of the first three intervals
of parametric resonance for Q = 20 and Q = 10 (and also in the absence of
friction). Note the “island” of parametric resonance for n = 3 and Q = 20. This
resonance disappears when the depth of modulation exceeds 45% and reappears
when m exceeds approximately 66%.
5.2.5.1 Main Interval of Instability in the Presence of Friction
Stationary oscillations occurring at the left boundary of the instability interval
in the vicinity of the principal parametric resonance in the presence of friction
are shown in Figure 5.13 (compare with Figure 5.4). Twice during the full cycle
the angular velocity abruptly increases, and twice it decreases. The increments
are greater than the decrements, so that as a whole the energy received by the
rotor exceeds the energy given away. This surplus compensates for the dissipation
of the energy that occurs at natural oscillation during the intervals between the
abrupt displacements of the weights along the rod of the flywheel.
To find conditions at which such stationary oscillations take place, we can
write the expressions for ϕ(t) and ϕ̇(t) during the adjacent intervals when the os-
cillator executes damped natural oscillations, and then fit these expressions to one
another at the boundaries. We choose as the time origin t = 0 the instant when
the weights are shifted apart, and the angular velocity is decreased in magnitude.
Then during the interval (0, T /2) the graph describes a damped natural oscilla-

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

5.2. FREQUENCY RANGES OF PARAMETRIC EXCITATION 115

.
∆ J( t) ϕ (t )

0 0
-T/2 T/2

-1 0 1 2 3T
Depth of modulation 35%, period of modulation 0.43T0 , quality 8.95

Figure 5.13: Stationary oscillations in the presence of friction at the left boundary
of the principal instability interval.

√
tion with the frequency ω1 = ω0 / 1 + m. Similarly to Eq. (5.13), p. 104, it is
convenient to represent this motion as a superposition of damped oscillations of
sine and cosine type with some constants A1 and B1 :

ϕ1 (t) = (A1 sin ω1 t + B1 cos ω1 t) e−γt ,

(5.33)
ϕ̇1 (t) ≈ (A1 ω1 cos ω1 t − B1 ω1 sin ω1 t) e−γt .

The latter expression for ϕ̇(t) is valid for relatively weak friction (γ ≪ ω0 ). To
obtain it, we differentiate ϕ(t) with respect to the time, considering the exponen-
tial factor e−γt to be approximately constant. Indeed, at weak damping the main
contribution to the time derivative originates from the oscillating factors sin ω1 t
and cos ω1 t in the expression for ϕ(t). Similarly, during the interval (−T /2, 0)
the graph in Figure 5.13 is a segment of damped natural oscillation with the fre-
quency ω2 :
ϕ2 (t) = (A2 sin ω2 t + B2 cos ω2 t) e−γt ,
(5.34)
ϕ̇2 (t) ≈ (A2 ω2 cos ω2 t − B2 ω2 sin ω2 t) e−γt .
Further calculations are similar to those leading from Eqs. (5.13) and (5.14) to
(5.19), but instead of Eq. (5.19) we get the following equation:

2kC1 C2 − (1 + k 2 )S1 S2 + k(p + 1/p) = 0, (5.35)

where the notation p = e−γT is used. This condition of existence of a non-zero

solution for constants A1 and B1 gives us an equation that determines the desirable
boundaries of the interval of instability. These boundaries are given by the values
of the unknown variable T (the roots of the equation), which enters in Eq. (5.35)
as the arguments of sine and cosine functions in S1 , S2 and C1 , C2 , and also as
the argument of the exponent in p = e−γT . To find approximate solutions T to
this transcendental equation, we transform it into a more convenient form in the
same way as in Section 5.2.1, p. 104:
r
ωav T ∆ωT
(1 + k) cos = ± (1 − k)2 cos2 − k (p + 1/p − 2). (5.36)
2 4

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

116 CHAPTER 5. PARAMETRIC EXCITATION OF OSCILLATIONS

To find the boundaries of the interval that contains the principal resonance, we
should search for a solution T of Eq. (5.36) in the vicinity of T = T0 /2 ≈ Tav /2.
If for a given value of the quality factor Q (Q enters in p = e−γT ) the depth
of modulation m exceeds the threshold value, Eq. (5.36) has two solutions that
correspond to the desirable boundaries T− and T+ of the instability interval. These
solutions exist if the expression under the radical sign in Eq. (5.36) is positive. Its
zero value corresponds to the threshold conditions:
(1 − k)2 ∆ωT
cos2 = p + 1/p − 2. (5.37)
k 4
To evaluate the threshold value of Q for small values of the modulation depth m,
we may assume here k ≈ 1 + m, and cos(∆ωT /4) ≈ 1. In the right-hand side
of Eq. (5.37), in p = e−γT , we can consider γT ≈ γT0 /2 = π/(2Q) ≪ 1, so
that p + 1/p − 2 ≈ (γT )2 = (π/2Q)2 . Thus, for the threshold of the principal
parametric resonance we obtain:
 
π  m π π π π
Qmin ≈ 1+ ≈ , mmin ≈ 1+ ≈ . (5.38)
2m 2 2m 2Q 4Q 2Q
At the threshold the expression under the radical sign in Eq. (5.37) is zero.
Both its roots (the boundaries of the instability interval) merge. This occurs when
the cosine in the left-hand side of Eq. (5.37) is zero, that is, when its argument
equals π/2:
T π π 1
ωav = , or T = = Tav ,
2 2 ωav 2
so that the threshold conditions, Eqs. (5.38), correspond to exact tuning to reso-
nance, when T = Tav /2.
To find the boundaries T− and T+ of the instability interval, we represent
T in the argument of the cosine function in the left-hand side of Eq. (5.36) as
Tav /2 + ∆T , where ∆T ≪ T0 . Since ωav Tav = 2π, we can write this cosine as
− sin(ωav ∆T /2). Then Eq. (5.36) becomes:
s
ωav ∆T 1 ∆ω( 12 Tav + ∆T ) (p − 1)2
sin =∓ (k − 1)2 cos2 −k . (5.39)
2 1+k 4 p
For zero friction p = 1, and Eq. (5.39) coincides with (5.21), p. 106. The
diagram in Figure 5.12 (p. 114) is obtained by numerically solving this equation
for ∆T by iteration.
To obtain an approximate solution of Eq. (5.39) that is valid for small values
of the modulation depth m up to terms to the second order of m, we can simplify
the expression under the radical sign in the right-hand side of Eq. (5.36), assuming
k ≈ 1 + m, (1 − k)2 ≈ m2 , and the value of the cosine function to be 1. The last
term of the radicand can be represented as (π/2Q)2 ≈ m2min . In the left-hand side
the sine can be replaced with its small argument, where ωav = 2π/Tav . Thus we
obtain:
 
∆T 1 Tav 1
q q
≈∓ 2 2
m − mmin , or T∓ = 1∓ 2 2
m − mmin . (5.40)
Tav 2π 2 π

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

5.2. FREQUENCY RANGES OF PARAMETRIC EXCITATION 117

For the case of zero friction mmin = 0, and these approximate expressions
for the boundaries of the instability interval reduce to Eq. (5.22), p. 107. For the
threshold conditions m = mmin , and both boundaries of the interval merge, that
is, the interval disappears.
5.2.5.2. The Second Interval of Instability in the Presence of Friction
When friction is taken into account, we arrive at, instead of Eq. (5.29), p. 111,
the following equation for the boundaries of the second interval of parametric
instability:
ωav T
q
(1 + k) sin = ± (1 − k)2 sin2 (∆ωT /4) − k(p + 1/p − 2). (5.41)
2
We should search for its solution T in the vicinity of T = T0 ≈ Tav . If for a given
value of the quality factor Q (Q enters in p = e−γT ) the depth of modulation m
exceeds the threshold value, Eq. (5.41) has two solutions that correspond to the
boundaries T− and T+ of the instability interval. These solutions exist if the ex-
pression under the radical sign in Eq. (5.41) is positive. Its zero value corresponds
to the threshold conditions:
(k − 1)2 (p − 1)2
sin2 (∆ωTav /4) = . (5.42)
k p
To estimate the threshold value of Q for small values of the modulation depth
m, we may assume here k ≈ 1 + m, and sin(∆ωT /4) ≈ ∆ωT /4. In the right-
hand side of Eq. (5.42), in p = e−γT , we can consider γT ≈ γT0 = π/Q ≪ 1,
so that p + 1/p − 2 = (p − 1)2 /p ≈ (γT )2 = (π/Q)2 . Thus, for the threshold of
the second parametric resonance we obtain:
r
2 2
Qmin ≈ 2 , mmin ≈ . (5.43)
m Q
The threshold conditions correspond to exact tuning to resonance, when T = Tav .
To find the boundaries T− and T+ of the instability interval, we represent T
in the argument of the sine function in the left-hand side of Eq. (5.41) as Tav +
∆T , where ∆T ≪ Tav ≈ T0 . Since ωav Tav = 2π, we can write this sine as
− sin(ωav ∆T /2). Then Eq. (5.41) becomes:
s
ωav ∆T 1 ∆ω(Tav + ∆T ) (p − 1)2
sin =∓ (k − 1)2 sin2 −k . (5.44)
2 1+k 4 p

This form of the equation is convenient for numerical solution by iteration.

For the zero friction, p = 1, and Eq. (5.44) coincides with Eq. (5.30), p. 111.
To obtain an approximate solution of Eq. (5.44), valid for small values of the
modulation depth m up to the terms of the second order of m, we can simplify the
expression under the radical sign in the right-hand side of Eq. (5.44), assuming
k ≈ 1 + m, (1 − k)2 ≈ m2 , and sin ∆ω(Tav + ∆T )/4 ≈ ∆ωTav . The last term

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

118 CHAPTER 5. PARAMETRIC EXCITATION OF OSCILLATIONS

of the radicand can be represented as (2/Q)2 ≈ m4min . In the left-hand side the
sine can be replaced by its small argument, in which ωav = 2π/Tav . Thus for the
boundaries of the second instability interval we obtain the following approximate
expressions:
 
∆T 1 1
q q
≈∓ m4 − m4min , or T∓ = 1 ∓ m4 − m4min Tav . (5.45)
Tav 4 4

In the presence of friction, for a given value m of the depth of square-wave

modulation, only several first intervals of parametric resonance can exist (the cor-
responding “tongues” of instability appear only if m exceeds the threshold value),
in contrast to the idealized case of zero friction (see the diagram in Figure 5.12,
p. 114).

5.3 Concluding Remarks

We have shown above that a linear torsion oscillator whose moment of inertia is
subjected to square-wave modulation by mass reconfiguration gives a very conve-
nient example in which the phenomenon of parametric resonance can be clearly
explained physically with all its peculiarities and even investigated quantitatively
by rather modest mathematical means.
In a linear system, if the threshold of parametric excitation is exceeded, the
amplitude of oscillations increases exponentially with time. In contrast to forced
oscillations, linear viscous friction is unable to restrict the growth of the amplitude
at parametric resonance. In real systems the growth of the amplitude is restricted
by nonlinear effects that cause the period to depend on the amplitude. During
parametric excitation the growth of the amplitude causes variation of the natural
period and thereby violates the conditions of resonance.
We note that even if the equilibrium of the system is unstable due to modu-
lation of the parameter (that is, if the conditions of parametric excitation are ful-
filled), when the initial values of ϕ and ϕ̇ are zero, they remain zero over the course
of time. This behavior is in contrast to that of resonance arising from forced oscil-
lations. In the latter instance, the amplitude increases with time even if the initial
conditions are zero. In other words, if parametric resonance is to be excited, the
system must already be oscillating, at least slightly, when the parametric variation
first occurs.

5.4 Questions, Problems, Suggestions

5.4.1 Principal Parametric Resonance
5.4.1.1∗ Principal Resonance (n = 1) in the Absence of Friction.
(a) Input a moderate value of the depth m of modulation of the moment of
inertia (about 10–15 percent). Choose the period of modulation T to be equal to

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

5.4. QUESTIONS, PROBLEMS, SUGGESTIONS 119

one half the mean natural period of the oscillator. What kind of initial conditions
ought you to enter in order to generate from the very beginning the fastest growth
of the amplitude? Remember that at the initial moment, t = 0, the weights are
suddenly moved away from one another, further from the axis of rotation.
(b) What initial conditions would lead at first to a fading away of oscillations
that are already present? Using the plots of the oscillations, explain the physical
reason for the increase or decrease in amplitude. Take into account the phase re-
lationship between the natural oscillation of the rotor and the periodic changes in
its moment of inertia. Why is it that some time later a phase relation is established
that generates a growth in the amplitude?
(c) Try to understand the reasons that determine the lapse of time between the
initial fading of the amplitude and its subsequent infinite growth.
5.4.1.2∗ The Growth of the Amplitude at the Principal Resonance without
Friction.
(a) If the modulation is to generate the principal parametric resonance, what
rule governs the growth of the amplitude when there is an initial deflecton and
an initial angular velocity of zero? Calculate the depth of modulation m that, in
the absence of friction, generates a doubling of the amplitude after 10 cycles of
the parametric modulation. Verify your result with a simulation experiment on the
computer.
(b) What difference do you find in your observations of part (a) if you set the
initial deflection to be opposite the deflection in part (a)?
5.4.1.3∗ The Threshold for the Principal Resonance.
(a) Choosing a moderate value for the modulation depth (say, m = 0.15),
estimate the threshold (minimal) value of the quality factor Qmin that corresponds
to stationary oscillations (i.e., to parametric regeneration) when the modulation is
tuned to the principal resonance (T = T0 /2).
(b) Make your calculated estimation of the threshold value Qmin more exact
by using an experiment on the computer. Describe the character of the plots and of
the phase trajectory under conditions of parametric regeneration and explain their
features.
(c) Is the mode of stationary oscillations at the threshold (for Q = Qmin)
stable with respect to small deviations in the properties of the system? Is the mode
stable with respect to small deviations in the initial conditions?
(d)∗∗ The threshold value of the quality factor for any given modulation depth
m is absolutely minimal when the modulation is exactly tuned to resonance. For
small values of m the principal resonance occurs when T = T0 /2. However,
when m increases, the resonant value of the modulation period T departs from
T0 /2. Find this resonant value of T for an arbitrarily large modulation depth m
and estimate values of T for m = 15% and m = 40%.
5.4.1.4∗ The Amplitude Growth over the Threshold.
(a) For the case in which T = T0 /2 and m = 15%, by what factor does
the amplitude of oscillation increase during 10 cycles of parametric oscillation

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

120 CHAPTER 5. PARAMETRIC EXCITATION OF OSCILLATIONS

if Q = 2Qmin? Does the answer depend on the initial conditions? Verify your
answer with a simulation experiment.
(b) What is the amplitude of oscillation after the next 10 cycles of modula-
tion? Why does friction not restrict the growth of the amplitude of parametrically
excited oscillations?
5.4.1.5∗∗ The Principal Interval of Parametric Resonance in the Absence
of Friction.
(a) Calculate the values of the period of modulation T corresponding to the
boundaries of the instability interval at a given modulation depth m (in the ap-
proximation m ≪ 1) for the case when friction is absent.
(b) How does the width of the interval depend on the depth of modulation? Do
the terms of second order influence the width of the interval?
5.4.1.6∗ The Initial Conditions for Steady Oscillations.
(a) Enter the value of the period of modulation corresponding to the left bound-
ary of the instability interval at a given value m of the modulation depth. Choose
the absence of friction. Input some initial deflection. What value of the initial
angular velocity ought you to enter for a given angular deflection in order that
stationary oscillations of a constant amplitude occur from the beginning of the
modulation?
(b) Verify your calculated approximate values of T for either boundary by
simulating an experiment, and find more precise values. Explain the appearance
of characteristic features of the plots and the phase trajectories of stationary oscil-
lations corresponding to each boundary of the instability interval.
(c) For a given value of the initial displacement ϕ0 , and for the calculated
value ϕ̇(0) of the initial angular velocity that provides stationary oscillations (at
each of the boundaries of the interval of instability), calculate the amplitude of
these oscillations. Verify the theoretical value by the experiment.
5.4.1.7∗∗ The Threshold of Excitation within the Instability Interval.
(a) Choose a value T of the period of modulation somewhere between the lim-
its of the interval of instability, e.g., approximately halfway between the resonant
value and one of the boundaries. Evaluate experimentally the growth of the ampli-
tude in the absence of friction, and from your observations, calculate the threshold
value of the quality factor Q = Qmin for parametric excitation at the given value
T of the modulation period.
(b) Verify your result experimentally and use the experiment to find a more
exact value of Qmin . Compare the observed plots of these stationary oscillations
with the plots of stationary (threshold) oscillations at exact tuning to resonance.
What are the differences between the plots (and the phase trajectories) of station-
ary oscillations at the threshold within the interval of parametric excitation with
friction, and the plots (and the phase trajectories) of stationary oscillations at the
boundaries of the instability interval without friction?
(c) If the threshold is exceeded, why does the amplitude continue to increase
indefinitely? In other words, why is friction unable to restrict the growth of the
amplitude of parametrically excited oscillations?

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

5.4. QUESTIONS, PROBLEMS, SUGGESTIONS 121

(d) For small values of the modulation depth m ≪ 1, calculate up to terms

of second order in m the threshold value Q = Qmin of the quality factor for the
period of modulation T lying somewhere within the interval of instability. Com-
pare your theoretical result with the value that you have obtained experimentally
in parts (a) and (b).
5.4.1.8∗∗ The Interval of Instability in the Presence of Friction.
(a) For some depth of modulation m, the frequency interval of parametric
excitation shrinks because of friction and disappears as the quality factor reaches
the threshold value. Let the quality factor Q be greater than the threshold value
Qmin . Find the values T− and T+ of the modulation period T that correspond to
the boundaries of the instability interval for a given m and Q (in the approximation
m ≪ 1). Express these values in terms of m and mmin , where mmin = π/(2Q)
is the approximate threshold value of the modulation depth m for a given quality
factor Q.
(b) In order to observe steady oscillations corresponding to these boundaries
as soon as the simulation begins, you need to set the initial conditions properly.
For a given value ϕ0 of the initial deflection, and for each of the boundaries of
the interval, what initial velocity produces steady oscillations from the very be-
ginning? Verify your answer by simulating the experiment.
5.4.1.9 Oscillations Outside the Interval of Parametric Resonance. For
a given value of m, enter a value T of the modulation period lying somewhere
outside the limits of the instability interval. Convince yourself that for any set of
initial conditions the oscillations eventually fade away, even if the friction is very
weak, and that the rotor comes to rest at the equilibrium position in spite of the
forced periodic changes in its moment of inertia.

5.4.2 Manual Control of the Parameter

There is an option in the relevant simulation program in which the abrupt changes
of the moment of inertia do not occur with a periodicity chosen beforehand, but
respond instead to signals that you send by the mouse or from the keyboard. First
you choose the depth m of modulation and the quality factor Q, input initial condi-
tions and start natural free oscillations. Then you can control the oscillator behav-
ior by changing manually its moment of inertia at moments you choose yourself.
In this way you can acquire both an understanding of and a feel for the physical
causes of parametric excitation.
In order to suppress the automatic modulation of the moment of inertia, mark
the check-box “Manual Control.” Then the button on the screen marked by ar-
rows becomes enabled, and you can produce abrupt shifts of the weights toward
and away from the axis by repeatedly clicking on this button or by toggling the
Spacebar.
5.4.2.1 Exciting Oscillations by Manual Control of Modulation.
(a) Enter some value m of the modulation depth (about 15–30 percent) and
choose a value of the quality factor exceeding the threshold (for the given m). By

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

122 CHAPTER 5. PARAMETRIC EXCITATION OF OSCILLATIONS

setting suitable initial conditions, you may excite natural oscillations of moderate
swing (10–20 degrees). Try to increase the swing by toggling the Spacebar (or by
repeatedly clicking on the button with arrows) regularly at appropriate moments.
At what instants ought you to shift the weights towards the axis, and at what
instants ought you to return them to their previous positions?
(b∗ ) Is it possible to increase the amplitude to a given value (say, 180◦ ) more
quickly by toggling the manual control than by the strictly periodic programmed
changes of the moment of inertia for the case of exact tuning to resonance (and
with the same values of the depth of modulation and the quality factor)? Give
convincing reasons for your answer.
5.4.2.2 Damping of Oscillations by Manual Control. After you have pro-
duced a large oscillation by appropriately changing the moment of inertia, try to
make the oscillations dampen out as fast as possible by changing the moment of
inertia. (You should choose a different phase for the changes.) At what time in-
stants ought you to shift the weights toward and away from the axis in order to
cause the quickest fading of oscillations?

5.4.3 Parametric Resonances of High Orders

5.4.3.1∗ The Third Parametric Resonance (n = 3) without Friction.
(a) Examine the parametric excitation of the rotor for abrupt changes of its
moment of inertia with the period T ≈ 3T0 /2 (approximately one and a half
times the natural period, or about three cycles of the parameter modulation during
two cycles of natural oscillations). What initial conditions ensure the growth of
the amplitude from the beginning of the modulation?
(b) What value m of the modulation depth in the absence of friction is nec-
essary in order to double the initial oscillation during 15 cycles of the parameter
modulation? After how many cycles does the amplitude double once more?
(c) For what initial conditions does the oscillation at first decrease? Why does
this fading inevitably change after a while into an increase in the amplitude?
5.4.3.2∗ The Threshold for the Third Resonance.
(a) For small values m of the modulation depth, calculate the threshold value
Qmin of the quality factor up to terms in the first order of m. How does this value
depend on m? Compare your answer with the principal resonance, n = 1 (see
Problem 5.4.1.3), and with the second resonance, n = 2 (see Problem 5.4.3.4).
What might be a qualitative explanation for the difference?
(b) For m ≈ 20% evaluate the minimal value Qmin of the quality factor for
which parametric resonance of the order n = 3 is possible. Improve your theo-
retical estimate by the simulation experiment. Explain the observed shape of the
angular velocity plot and the form of the phase trajectory of stationary oscillations
at Q = Qmin . The plots of oscillations occurring at the threshold of the third res-
onance are shown in Figure 5.14. What factor determines the amplitude of such
oscillations?

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

5.4. QUESTIONS, PROBLEMS, SUGGESTIONS 123

.
ϕ (t ) ∆ J( t)

0 1 2 3T
Depth of modulation 20%, period of modulation 1.4772T 0, quality 26.8

Figure 5.14: The phase trajectory and the time-dependent graph of the angular ve-
locity of stationary oscillations at the threshold of the third parametric resonance.

5.4.3.3∗∗ The Third Interval of Parametric Excitation.

(a) Calculate the values of the modulation period T which, in the absence of
friction, correspond to the boundaries of the third instability interval for a given
modulation depth m (in the approximation m ≪ 1). How does the width of the
interval depend on the depth of modulation? Do the terms of the second order
influence the width of the interval?
(b) What value of the initial angular velocity ought you to enter for a given
initial deflection ϕ0 in order to get stationary oscillations of a constant amplitude
from the very beginning of the modulation on each boundary of the instability
interval? Verify your calculated values experimentally.
What are the shapes of the phase trajectories that correspond to the left and
right boundaries of this interval?
(c) Explore the width of the third interval of parametric excitation without
friction at arbitrarily large values of the modulation depth m. Note how the interval
moves to the left and gradually shrinks as m becomes greater (see the diagram in
Figure 5.6, p. 108)).
At m = 60% both boundaries of the interval coincide. (You may also say that
they intersect at this value of m.) This coincidence means that for the correspond-
ing value of the modulation period T you get steady oscillations for arbitrary
initial conditions. What might be a physical explanation for this behavior?
Hint: What is the ratio of the natural periods for the maximum and minimum
values of the moment of inertia for this value of the modulation depth?
5.4.3.4∗∗ The Third Instability Interval with Friction.
(a) At small values of m the third parametric resonance occurs at T = 3T0 /2.
However, with the growth of m the resonant value of the modulation period T
departs from 3T0 /2. Find an analytical expression for this resonant value of T
(for an arbitrarily large modulation depth m) and make a numerical estimate for
m = 15% and m = 40%.
(b) How does friction reduce the width of the third interval of parametric ex-
citation? For a small depth of modulation m ≪ 1, calculate approximate values
of the period of modulation T that correspond to the boundaries of the interval for

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

124 CHAPTER 5. PARAMETRIC EXCITATION OF OSCILLATIONS

a given value of the quality factor Q. Express the results in terms of m and the
threshold value mmin = 3π/(2Q) (see Problem 5.4.3.2) for the given Q-value.
5.4.3.5∗∗ Parametric Resonance of the Second Order (n = 2).
(a) Choosing a moderate value of the modulation depth (m < 20%), excite
parametric resonance of the order n = 2 (for which the period of modulation is
approximately equal to the average natural period). Why does the growth in ampli-
tude occur much more slowly in this case than it does for the principal resonance,
and even more slowly than it does for the resonance of the order n = 3 (for the
same value m of the modulation depth)? Explain the observed shape of the phase
diagram for n = 2. Try to determine experimentally the threshold value of the
modulation depth for a given value of the quality factor (say, Q = 15).
(b) For small values of the modulation depth m ≪ 1, try to calculate the
threshold value of the quality factor Qmin. (You need to keep the terms of the sec-
ond order in m). How does the threshold value of Q depend on m? Compare your
calculated value with the threshold of principal resonance and of the third reso-
nance. Explain the difference qualitatively. Also compare the theoretical threshold
value with your experimental result of part (a).
5.4.3.6∗∗ The Second Interval of Parametric Excitation.
(a) For small values of the modulation depth m ≪ 1, calculate the width of
the interval (you need to keep the terms to the second order of m). How does the
width depend on m?
(b) Excite and experimentally examine stationary oscillations without friction
that correspond to the boundaries of the second instability interval (near the reso-
nance for n = 2). For small values of the modulation depth m ≪ 1, why is this
interval considerably narrower than the interval for resonance of a higher order
n = 3?
(c) Why do two different phase trajectories correspond to each boundary of
the interval? What is the difference between the two stationary oscillations that
correspond to the same boundary? How can each one of them be excited? What
initial conditions ensure steady oscillations from the beginning of modulation?
5.4.3.7∗∗∗ The Second Interval of Parametric Excitation with Friction.
How does friction influence the width of the second interval of parametric
excitation? For a small depth of modulation m ≪ 1, calculate approximate values
of the period of modulation T that correspond to the boundaries of the interval for
a given value of the quality factor. Write down the results in terms of m and the
threshold value Qmin (see Problem 3.5) for the given value of Q.

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

Chapter 6

Sinusoidal Modulation of the

Parameter

Annotation. The general concepts of the parametric excitation of a linear system

and the conditions necessary for parametric resonance are considered in Chapter 5,
in which we examine the square-wave modulation of a parameter. The present
chapter deals with another case of parametric excitation of a linear torsion spring
oscillator, namely that of excitation arising from a smooth, harmonic motion of
the weights along the rod of the flywheel.
To provide a growth of energy during a smooth periodic modulation of the
moment of inertia, the motion of the weights toward the axis of rotation must
occur while the angular velocity of the flywheel is greater in magnitude than its
angular velocity when the the weights are returned to the ends of the rod. To be
sure, a smooth parametric modulation is less efficient for the delivery of energy
to the flywheel than is a square-wave modulation. During a square-wave modula-
tion, the shifts of the weights toward and away from the axis occur instantly and
at the most favorable moments. On the other hand, during smooth modulations
these processes are protracted in time. Hence the threshold value of the depth of
modulation for smooth modulations is somewhat greater than it is for square-wave
modulation. Nevertheless, if the threshold of excitation is exceeded, the amplitude
during smooth modulation also grows indefinitely.

6.1 Summary of the Theory: Basic Concepts

6.1.1 The Physical System
The physical system investigated in the present chapter and simulated in the rele-
vant computer program is the same linear torsion spring oscillator that is used for
the discussion of square-wave modulation (Chapter 5). A flywheel (rotor), which
is a balanced rod with two identical weights, can rotate about an axis that passes

125

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

126 CHAPTER 6. SINUSOIDAL MODULATION OF THE PARAMETER

through its center (see Figure 5.1, p. 95). When in its equilibrium position, one end
of the rod points toward the zero of the dial. When the rod is displaced from the
equilibrium position, the spiral spring is twisted and produces a restoring torque
that is proportional to the angle of deflection.
To provide modulation of a system parameter, we assume that the weights can
be shifted simultaneously along the rod in opposite directions into other symmet-
rical positions so that the rotor as a whole remains balanced. However, its moment
of inertia J is changed by such displacements of the weights. When the weights
are shifted toward or away from the axis, the moment of inertia decreases or in-
creases respectively. Thus the moment of inertia of the rotor is the parameter to
be modulated in the investigated physical system.p As the moment of inertia J is
changed, so also is the natural frequency ω0 = D/J of the torsional oscillations
of the rotor. Periodic modulation of the moment of inertia can cause, under certain
conditions, a growth of (initially small) natural rotary oscillations of the rod.

6.1.2 Physical Reasons for Parametric Excitation at Smooth

Modulation
To understand how a change in the moment of inertia can increase or decrease
the angular velocity of the rotor, let us imagine for a while that the spiral spring
is absent. Then the angular momentum of the system would remain constant as
the weights are being moved along the rod. Thus the resulting reduction in the
moment of inertia is accompanied by an increment in the angular velocity, and the
rotor acquires additional energy. The system is similar in some sense to a spinning
figure skater, whose rotation accelerates as she moves her initially stretched arms
closer to her body.
The greater the initial angular velocity, the greater the increment in the velocity
and the energy. This additional energy is supplied to the rotor by the source that
moves the weights along the rod. On the other hand, if the weights are moved apart
along the rotating rod, the angular velocity and the energy of the rotor diminish.
The decrease in energy is transmitted back to the source.
In order that increments in energy occur regularly and exceed the amounts
of energy returned, i.e., in order that, as a whole, the modulation of the moment
of inertia regularly feeds the oscillator with energy, the period and the phase of
modulation must satisfy certain conditions.
An example of parametric excitation of the torsion oscillator during the sinu-
soidal motion of the weights along the rod with the period that equals one half of
the natural period is shown in Figure 6.1.

6.1.3 Conditions of Parametric Resonance

To provide a growth of energy by modulation of the moment of inertia, the motion
of the weights toward the axis of rotation must occur while the angular velocity
of the rotor is on the average greater in magnitude than it is when the weights are
moved apart to the ends of the rod. The graphs in Figure 6.1 correspond to this

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

6.1. SUMMARY OF THE THEORY: BASIC CONCEPTS 127

1.72
.
v ( t) ϕ (t )

-1.72 o o
-93.4 0 93.4
∆ l ( t) ϕ (t )

o
-93.4
0 2 4 6 8 10T
Period of modulation 0.5T0, depth of modulation 20%, quality factor 18.0

Figure 6.1: The phase trajectory, the graphs of the angular velocity and displace-
ment of the rotor, and of the weights motion in conditions of the principal para-
metric resonance.

case: We see clearly that during the intervals of negative values of v the angular
velocity ϕ̇ is greater in magnitude than during the intervals of positive v. Oth-
erwise the modulation of the moment of inertia aids the damping of the natural
oscillations.
The strongest parametric oscillations are excited when the cycle of modula-
tion is repeated twice during one period T0 of natural oscillations in the system,
i.e., when the frequency ω of parametric modulation is twice the natural frequency
ω0 of the system. But the delivery of energy is also possible when the parameter
changes once during one period, twice during three periods, and so on. That is,
parametric resonance is possible when one of the following conditions for the fre-
quency ω of modulation (or for the period of modulation T = 2π/ω) is fulfilled:

ω = 2ω0 /n, T = nT0 /2, (6.1)

where n = 1, 2, . . . . For a given amplitude of modulation of the parameter, the

higher the order n of parametric resonance, the less (in general) the amount of
energy delivered to the oscillating system during one period.
One of the most interesting characteristics of parametric resonance is the pos-
sibility of exciting increasing oscillations not only at the frequencies ωn given in
Eq. (6.1), but also in intervals of frequencies lying on either side of the values
ωn (in the ranges of instability). These intervals become wider as the range of
parametric variation is extended, that is, as the depth of modulation is increased.
An important distinction between parametric excitation and forced oscillations
is related to the dependence of the growth of energy on the energy already stored
in the system. While for a direct forced excitation the increment of energy during

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

128 CHAPTER 6. SINUSOIDAL MODULATION OF THE PARAMETER

one period is proportional to the amplitude of oscillations, i.e., to the square root
of the energy, at parametric resonance the increment of energy is proportional to
the energy stored in the system.
Energy losses caused by friction (unavoidable in any real system) are also pro-
portional to the energy already stored. In the case of direct forced excitation, an
arbitrarily small external force gives rise to resonance. However, energy losses re-
strict the growth of the amplitude because these losses grow with the energy faster
than does the investment of energy arising from the work done by the external
force.
In the case of parametric resonance, both the investment of energy caused by
the modulation of a parameter and the frictional losses are proportional to the en-
ergy stored (to the square of the amplitude), and so their ratio does not depend
on the amplitude. Therefore, parametric resonance is possible only when a thresh-
old is exceeded, that is, when the increment of energy during a period (caused
by the parametric variation) is larger than the amount of energy dissipated during
the same time. To satisfy this requirement, the range of the parametric variation
(the depth of modulation) must exceed some critical value. This threshold value
of the depth of modulation depends on friction (see Section 6.1.5). However, if the
threshold is exceeded, the frictional losses of energy cannot restrict the growth of
the amplitude. In a linear system the amplitude of parametrically excited oscilla-
tions must grow indefinitely.
In a nonlinear system the natural period depends on the amplitude of oscil-
lations. If conditions for parametric resonance are fulfilled at small oscillations
and the amplitude begins to grow, the conditions of resonance become violated at
large amplitudes. In a real system the growth of the amplitude over the threshold
is restricted by nonlinear effects.

6.1.4 Energy Transformations at Parametric Excitation

To understand why it is possible to excite torsional oscillations of the rotor by a
smooth (e.g., sinusoidal) motion of the weights along the rod, and what condi-
tions are necessary for parametric resonance, we make use of the conservation of
energy.
In the case of the parametric excitation of oscillations, additional energy must
be transmitted to the rotor by the source that makes the weights move periodically
along the rod. Therefore, we calculate the work done by the source during one
period of oscillation and find those conditions under which this work is positive.
To produce a parametric excitation of the oscillator, certain phase relationships
between the variations of its angular velocity and the changes in its moment of
inertia must hold. Namely, the weights must be drawn toward the axis when the
angular velocity of the flywheel is on the average greater in magnitude than when
the weights are moved apart. Otherwise the modulation of the moment of inertia
aids the damping of the natural oscillation.
In the model we let the forced motion of the weights along the rod be ex-
actly sinusoidal, and so their distance l from the axis of rotation varies with time

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

6.1. SUMMARY OF THE THEORY: BASIC CONCEPTS 129

according to the following expression:

l(t) = l0 (1 + m sin ωt). (6.2)

Here l0 is the mean distance of the weights from the axis of rotation, and m is the
dimensionless amplitude of their harmonic motion along the rod (m < 1). From
Eq. (6.2) we find that a weight moves along the rod with a velocity (relative to the
rod) that changes with time as cos ωt:

v(t) = dl/dt = ωl0 m cos ωt. (6.3)

The relative acceleration of the weight in its motion along the rod is:

ar (t) = dv/dt = −ω 2 l0 m sin ωt. (6.4)

In order to find the force F exerted on the weight by the device that makes it
move along the rod, we use a non-inertial reference frame rotating with the rod.
Using Newton’s second law applied to the motion of the weight in this rotating
frame of reference, we must take into account the pseudo centrifugal force of
inertia acting on the weight, M ϕ̇2 (t)l(t), where M is the mass of the weight and
ϕ̇(t) is the angular velocity of the rod:

M ar (t) = F (t) + M ϕ̇2 (t)l(t). (6.5)

We are interested in the work of this force F (t) done during one period of
oscillation. The amount of this work (for both weights) equals the change in the
energy of oscillations during one period. For the infinitesimal element of work
dW performed during a time interval dt, during which the weight is displaced
along the rod a distance dl = v(t)dt, we can write the following expression:

dW = F (t)dl = F (t)v(t)dt = [ M ar (t) − M ϕ̇2 (t)l(t) ]v(t)dt. (6.6)

As we see from Eq. (6.3), the radial velocity v(t) of the weight in Eq. (6.6) is
proportional to the dimensionless amplitude m of its forced motion along the rod.
If we restrict our calculations to the first order of the small parameter m, we need
keep only the second term in square brackets in Eq. (6.6), and we can substitute
for l(t) its mean value l0 in the equation:

dW ≈ −M ϕ̇2 (t)l0 v(t)dt = −M ϕ̇2 (t)l02 ωm cos ωt. (6.7)

As we noted above, the most favorable condition for the parametric excitation
of the rotor occurs if the weights execute two full cycles of the forced motion
during one mean period of natural oscillation. In other words, the frequency ω in
Eq. (6.2) and Eq. (6.7) must be approximately twice the mean natural frequency
ω0 = 2π/T0 of oscillation of the rotor. (Here ω0 is the frequency of free oscilla-
tions of the rotor with the weights fixed at their average distance l0 from the axis).
The frequency of modulation ω in Eq. (6.2) that is equal to the doubled value of

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

130 CHAPTER 6. SINUSOIDAL MODULATION OF THE PARAMETER

the mean natural frequency (ω = 2ω0 ) is exactly tuned to the principal resonance
(n = 1) for small values of the dimensionless amplitude m.
In addition, it is necessary that a certain phase relation between the forced mo-
tion of the weights and the torsional oscillations of the rotor be satisfied: Namely,
the weights must move with maximal relative velocity toward the axis of rota-
tion at moments when the oscillating rod moves with its greatest angular velocity
(which it does when it is near its equilibrium position). This phase relation is sat-
isfied for the motion of the weights described by Eqs. (6.2)–(6.3) provided we
assume the following time dependence for the torsional oscillations of the rotor:

ϕ(t) = ϕm cos ω0 t; ϕ̇(t) = −ϕm ω0 sin ω0 t. (6.8)

These are only approximate expressions because, strictly speaking, the torsional
oscillations of the rotor are not harmonic. Deviations from a sinusoidal oscillation
are caused by the motion of the weights since this motion influences the angular
velocity and the moment of inertia of the rotor.
After the substitution of ω = 2ω0 and ϕ̇(t) from Eq. (6.8) into Eq. (6.7) we
can integrate dW given by (6.7) over a period T0 = 2π/ω0 , taking into account
that Z T0
cos2 ω0 tdt = T0 /2.
0
Finally we find that (up to terms of the first order in the small value m), the work
W of the force F (t) done during a period T0 is given by the following expression:
1
W = M ϕ2m ω02 l02 · 2πm. (6.9)
2
The same relationship is valid for the second weight. And so as a whole the
forces exerted on the weights by the device that makes them move along the rod
perform positive work W > 0 during a period and increase the energy of the
oscillator by the amount:

∆E = 2W = M ϕ2m ω02 l02 · 2πm. (6.10)

For simplicity, we let the rod be very light compared to the weights so that we
can consider all kinetic energy of the rotor to be the kinetic energy of these massive
weights. The total energy E of the oscillator is equal to the maximal value of its
kinetic energy, which is attained at the instants when the oscillating rotor moves
near its equilibrium position and has its greatest angular velocity ω0 ϕm :

E = M ϕ2m ω02 l02 .

Comparing this expression with the right side of Eq. (6.10), we see the most es-
sential feature of parametric resonance, namely that the investment of energy ∆E
due to modulation of a parameter is proportional to the energy E already stored in
the oscillator:
∆E = 2πmE. (6.11)

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

6.1. SUMMARY OF THE THEORY: BASIC CONCEPTS 131

Eq. (6.11) means that at parametric resonance the total energy E of oscilla-
tions, averaged over a period T0 = 2π/ω0 of oscillation, grows exponentially
with time:
dE
= mω0 E, E(t) = E0 exp(2st), where 2s = mω0 . (6.12)
dt

6.1.5 The Threshold of Parametric Excitation

The exponential growth of energy in conditions of the principal parametric res-
onance described by Eq. (6.12) occurs in the absence of friction. Dissipation of
the mean energy E due to viscous friction is also described by an exponential
function:
dE
= −2γE, E(t) = E0 exp(−2γt). (6.13)
dt
The threshold of parametric resonance corresponds to the case for which these
energy losses are just compensated for by the delivery of energy arising from the
forced periodic motion of the weights. In this instance, γ = s. Thus we can find
the minimal value of m (for a given value of γ or of the quality factor Q) that
makes parametric excitation possible:
2γ 1
mmin = = . (6.14)
ω0 Q
Equivalently, the threshold condition can be expressed in terms of the maximal
value of the damping constant γ (or the minimal quality factor Q) for a given
value m of the amplitude in (6.2):
1 ω0 1
γmax = mω0 , Qmin = = . (6.15)
2 2γmax m
These results concerning the threshold of parametric excitation are approxi-
mate and are valid only for small values of the dimensionless amplitude m of the
forced motion of the weights along the rod. The simulation program performs nu-
merical integration of the differential equation of motion. This integration is not
restricted to small values of m. Thus the simulation allows us to determine the
threshold conditions experimentally with greater accuracy.
Steady oscillations occurring at the threshold are called parametric regenera-
tion. They are shown in Figure 6.2. These graphs should be compared with those
shown in Figure 6.1, which displays plots of resonant oscillations occurring above
the threshold, where the amplitude grows exponentially in spite of the friction.
When the depth of modulation exceeds the threshold value, the (averaged over
the period) energy of oscillations increases exponentially with time. The growth
of the energy is again described by Eq. (6.12), p. 131. However, now the index
of growth 2α is determined by the amount by which the energy delivered through
parametric modulation exceeds the simultaneous losses of energy caused by fric-
tion: 2α = mω0 − 2γ. The energy of oscillations is proportional to the square of

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

132 CHAPTER 6. SINUSOIDAL MODULATION OF THE PARAMETER

. ϕ (t )
ϕ (t )
v ( t)

0 1 2 3 4 5 6T
Period of modulation 0.4866T0, depth of modulation 20%, quality factor 5.12

Figure 6.2: The phase trajectory (left), the graphs of the angular velocity and of
radial velocity of the weights (right) for oscillations at the threshold condition
m ≈ 1/Q.

the amplitude. Therefore the amplitude of parametrically excited oscillations also

increases exponentially with time (see Figure 6.1): a(t) = a0 exp(αt) with the in-
dex α (one half the index 2α of the growth in energy). For the principal resonance
we have α = mω0 /2 − γ.
We can now compare the threshold values of the depth of modulation for the
case of square-wave modulation with that of harmonic modulation. For m ≪
1, as will be seen below from Eq. (6.18), the moment of inertia is modulated
nearly harmonically with the depth mJ ≈ 2m. (The depth of modulation of the
moment of inertia is approximately twice the depth of modulation of the distance
l between the axis and the weights because the moment of inertia is proportional
to the square of the distance l.) So for harmonic modulation, Eq. (6.14) gives
the threshold depth of the modulation of the moment of inertia: mJ = 2/Q.
This value is somewhat greater than mJ = π/(2Q) given by Eq. (5.6), p. 99 for
the case of square-wave modulation (in Eq. (5.6) the depth of modulation of the
moment of inertia mJ is denoted simply as m). As already mentioned, square-
wave modulation provides more favorable conditions for the transfer of energy to
the oscillator from the source that moves the weights along the rod of the flywheel.

6.1.6 Differential Equation for Sinusoidal Motion of the

Weights along the Rod
For simplicity we consider the rod itself to be very light, so that the moment of
inertia J of the rotor is due principally to the weights: J = 2M l2 (t). The angular
momentum J ϕ̇(t) changes with time according to the equation:

d
(J ϕ̇) = −Dϕ, (6.16)
dt
where −Dϕ is the restoring torque of the spring. Substituting into Eq. (6.16) l(t)
from Eq. (6.2) and taking into account the expression ω02 = D/J0 (J0 = 2M l02 is

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

6.1. SUMMARY OF THE THEORY: BASIC CONCEPTS 133

the moment of inertia with the weights in their mean positions), we finally obtain:
d 
(1 + m sin ωt)2 ϕ̇ = −ω02 ϕ − 2γ ϕ̇.

(6.17)
dt
We have also added the drag torque of viscous friction to the right-hand side
of Eq. (6.17). This equation is solved numerically in the relevant computer pro-
gram in real time during the simulation of oscillations at sinusoidal motion of the
weights.
We note that the harmonic motion of the weights along the rod described by
Eq. (6.2) does not mean that the moment of inertia is harmonically modulated.
Indeed, J is proportional to the square of the distance l(t) rather than to its first
power. The time dependence of J(t) includes the second harmonic of the fre-
quency ω. Only for small values of the amplitude m (when m ≪ 1) can we
consider the modulation of the moment of inertia to be approximately sinusoidal:

J(t) = 2M l2(t) = 2M l02(1 + m sin ωt)2 ≈

(6.18)
2M l02 (1 + 2m sin ωt) = J0 (1 + mJ sin ωt),

where J0 = 2M l02 is the mean value of the moment of inertia, and mJ = 2m is

the depth of its modulation. (We note that the value of mJ is twice the value of
m.) If we are interested only in an approximate solution valid up to terms of the
first order in the small parameter m, then instead of the exact differential equation
of motion, Eq. (6.17), we can solve the following approximate equation:

ϕ̈ + 2γ ϕ̇ + ω02 (1 − 2m sin ωt)ϕ = 0. (6.19)

We ignore here the modulation of the coefficient of ϕ̇ because for parametric res-
onance the variation of only those parameters that store energy (the moment of
inertia and the torsion spring constant) is essential. Modulation of the damping
constant γ cannot excite oscillations.
Equation (6.19) is a special case of Hill’s equation, Eq. (5.1), p. 93, with si-
nusoidal time dependence of the parameter. It is called Mathieu’s equation. The
theory of Mathieu’s equation has been fully developed, and all significant prop-
erties of its solutions are well known. A complete mathematical analysis of its
solutions is rather complicated and is usually restricted to the determination of the
frequency intervals in which the state of rest in the equilibrium position becomes
unstable: At arbitrarily small deviations from the state of rest, the amplitude of
incipient small oscillations begins to increase progressively with time. The bound-
aries of these intervals of instability depend on the depth of modulation 2m. It is
worth mentioning that even inside the intervals (when conditions for parametric
resonance are satisfied), if ϕ(0) and ϕ̇(0) are exactly zero simultaneously, they
remain zero. This property contrasts with the usual case of resonance in which the
system is acted upon by a periodic external force. In this case, the amplitude of
oscillations begins to grow from a state of rest in the equilibrium position.
We emphasize that the application of the theory of Mathieu’s equation to the
simulated system under consideration is restricted to the linear order in m. For fi-

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

134 CHAPTER 6. SINUSOIDAL MODULATION OF THE PARAMETER

nite values of the depth of modulation m, the resonant frequencies and the bound-
aries of the intervals of instability for the simulated system differ from those pre-
dicted by Mathieu’s equation. We shall see this point below, where we develop the
theory of our system up to the terms of the second order of m.

6.2 The Intervals of Parametric Instability

In this section we shall show how to use the differential equation of motion,
Eq. (6.17), in order to determine the intervals of the frequency ω of the forced
oscillations of the weights along the rod, in which the state of rest of the rotor in
the equilibrium position becomes unstable and parametric excitation of its oscil-
lations occurs at deviations from equilibrium no matter how small they are.

6.2.1 The Principal Interval of Instability

In the vicinity of the principal resonance the frequency of modulation is approxi-
mately twice the natural frequency (ω ≈ 2ω0 ), and we can express ω in the form
ω = 2ω0 + ε, where ε is a small detuning from resonance (|ε| ≪ ω0 ). We then
propose that an approximate solution to Eq. (6.17) represents a nearly harmonic
motion with the frequency ω̃ = ω/2 = ω0 + ε/2. We let the amplitude and phase
of ϕ(t) slowly vary with time:

ϕ(t) = p(t) cos ω̃t + q(t) sin ω̃t. (6.20)

Here p(t) and q(t) are functions of time that vary slowly relative to the oscillating
sine and cosine functions. In the exact solution to Eq. (6.17) there are also higher
harmonics with the frequencies 3ω̃, 5ω̃, . . . , but their contribution is proportional
to higher powers of the small parameter m ≪ 1. We do not include these higher
harmonics in the approximate solution expressed by Eq. (6.20).
The time variation of the amplitudes p(t) and q(t) is caused by the modula-
tion of the square of the natural frequency, and so the derivatives of the functions
p(t) and q(t) are also proportional to the small quantity m. Substituting ϕ from
Eq. (6.20) into the differential equation, Eq. (6.17), we can express the products
of the sine and cosine functions in the following way:

sin 2ω̃t cos ω̃t = (sin ω̃t + sin 3ω̃t)/2,

sin 2ω̃t sin ω̃t = (cos ω̃t − cos 3ω̃t)/2,

and omit in the equation the higher harmonics with the frequency 3ω̃. Thus for the
functions p(t) and q(t) we obtain the following system of differential equations
of the first order:
2ω̃ q̇ − (ω̃ 2 − ω02 ) p + (2γ ω̃ − mω02 ) q = 0,
(6.21)
− 2ω̃ ṗ − (2γ ω̃ + mω02 ) p − (ω̃ 2 − ω02 ) q = 0.

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

6.2. THE INTERVALS OF PARAMETRIC INSTABILITY 135

We have omitted here the terms 2γ ṗ and 2γ q̇ since parametric excitation is possi-
ble only if friction is small enough (from Eq. (6.15) we see that 2γ < mω0 ). The
contribution of these omitted terms to Eq. (6.21) is of the order m2 .
According to general rules, the solution to these equations can be searched
for in the form exp αt. The condition for the existence of a nontrivial (nonzero)
solution of this system of homogeneous equations gives the following expression
for α:
1p
α≈ (mω0 )2 − ε2 − γ. (6.22)
2
Here we have taken into account that ω̃ 2 ≈ ω02 + ω0 ε. If there is an exact tuning
to resonance, the deviation in frequency ε vanishes (ε = 0), and Eq. (6.22) gives
the following value for the index α that determines the exponential growth in the
amplitude of parametrically excited oscillations:

α ≈ mω0 /2 − γ. (6.23)

The amplitude of oscillation grows if α > 0. Therefore, for the threshold of

parametric resonance we obtain m = 2γ/ω0 = 1/Q. The same value for the
threshold of parametric excitation under conditions of exact tuning to resonance
is obtained above using the conservation of energy; see Eq. (6.14), p. 131. For
zero friction, the index of the exponential resonant growth in the amplitude is
proportional to the depth of modulation: α = mω0 /2.
For the case in which friction is absent (γ = 0), and for a given value m
of the depth of modulation, we find from Eq. (6.22) that increasing with time
solutions of the linearized differential equation, Eq. (6.17), exist in some interval
of frequencies that extends by ∆ω = mω0 on either side of the resonant value
ωres = 2ω0 . Thus, the half-width ∆ω of the interval of instability is given by:

∆ω = mω0 . (6.24)

For zero friction, the width 2∆ω of the interval within which parametric reso-
nance occurs is proportional to the amplitude m of the forced periodic motion of
the weights. For a value ω of the frequency of modulation lying somewhere within
the interval, the amplitude of parametrically excited oscillations grows exponen-
tially with time as exp(αt), where the index α of the growth is given by Eq. (6.22)
with γ = 0:
1p
α= (mω0 )2 − (ω − ωres )2 (6.25)
2
(for |ω − ωres | ≤ mω0 ). The value of α is zero at the boundaries ω± of the interval
of instability: ω± = ωres ± mω0 . At these boundaries, stationary oscillations of
constant amplitude are possible. An example of such oscillations is shown Fig-
ure 6.3.
The symmetric shape of these graphs shows that on average there is no energy
transfer to the frictionless oscillator: The energy gained during one half-cycle of
modulation is returned back during the next half-cycle of these oscillations on the
boundary of the instability interval.

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

136 CHAPTER 6. SINUSOIDAL MODULATION OF THE PARAMETER

.
ϕ (t ) ϕ (t )

0
v ( t)

0 1 2 3 4 5 6T
Period of modulation 0.4065T0, depth of modulation 30%, no friction

Figure 6.3: The phase trajectory of stationary oscillations occurring at the left
boundary of the principal instability interval (left), the time dependent graphs of
the angular velocity and displacement of the rotor, and of the radial velocity of the
weights (right).

The non-sinusoidal shape of the graphs (see Figure 6.3) shows clearly that the
spectrum of steady-state periodic oscillations on the boundaries of the instability,
besides the principal harmonic whose frequency ω̃ equals one half the frequency
of modulation ω = 2π/T , also contains harmonics of high orders.
Therefore, in order to obtain more precise values for the frequencies of mod-
ulation ω± that correspond to the boundaries of the instability interval, we need
to include these high harmonics in the approximate solution of Eq. (6.17). Their
frequency 3ω̃, 5ω̃, . . . is an odd-number multiple of the fundamental frequency
ω̃ ≈ ω0 /2. We look for a solution containing terms up to the second order in m.
We assume that this solution has the following form:

ϕ(t) = p0 cos ω̃t + q0 sin ω̃t + p1 cos 3ω̃t + q1 sin 3ω̃t. (6.26)

If we are interested only in the boundaries ω± of the interval of instability, where

oscillations are stationary and their amplitude does not vary with time, we can
assume the coefficients p0 , q0 , p1 , and q1 to be constant.
Substituting Eq. (6.26) in Eq. (6.17), we can omit the terms with the frequency
5ω̃. In the terms with the frequency ω̃ we need to keep quantities up to the first
and second order in m, while in the terms with the frequency 3ω̃ we need to keep
only the terms of the first order.
Finally we arrive at a system of homogeneous equations for p0 , q0 , and p1 ,
q1 . The condition for the existence of a nontrivial solution of the system gives us
the desired boundaries. These boundaries (for γ = 0) are given by the following
frequencies ω± :
11
ω± = 2ω0 ± mω0 + m2 ω0 . (6.27)
8
In the simulation computer program we are to enter not the frequency ω of the
parametric modulation, but rather the period T = 2π/ω. And so, as a convenient
reference here, we also indicate the boundaries of the instability interval near the

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

6.2. THE INTERVALS OF PARAMETRIC INSTABILITY 137

Figure 6.4: Principal instability interval (a) and the diagram showing the bound-
aries of the first three intervals (b). Thin curves that deviate slightly at large m
values from the boundaries of the principal interval are plotted according to the
approximate expression (6.28).

principal resonance in units of T0 :

 
T0 1 7
T = 1 ∓ m − m2 . (6.28)
2 2 16
The term of the second order in m has the same value for both boundaries of the
interval. It does not influence the width of the interval, shifting it as a whole by a
value proportional to m2 .
The structure of the principal interval of parametric instability is shown in
Figure 6.4a for the absence of friction (thick bounding curves), for Q = 20, and Q
= 10 (thin inner curves). It is more convenient to express the boundaries using not
the frequency ω of the parametric modulation, but rather the period T = 2π/ω.
This convention is usually used in presenting the stability map for Mathieu-type
systems by the so-called Incze–Strutt diagrams. We also use it in the simulation
program and for all figures in this chapter. The dashed regions in Figure 6.4b show
the first three intervals of parametric instability in one T – m diagram. 1
In the presence of viscous friction the principal interval shrinks. The interval
disappears if Q < 1/m: Its boundaries merge at the threshold. Equation (6.27)
gives for the threshold the value mmin = 1/Q, which has been found above, in
Eq. (6.14), from considerations based on the energy conservation.
An example of steady oscillations occurring at the left boundary of the princi-
pal instability interval is shown in Figure 6.3, p. 136. The upper panel of Figure 6.5
shows the phase diagram and the graphs of steady oscillations at the right bound-
ary of the interval in the absence of friction. We note the departure of the shape of
1 Actually the curves in Figure 6.4 are plotted with the help of somewhat more complicated formulas

than Eq. (6.28) (not cited here), which are obtained by holding several more harmonic components in
the trial function ϕ(t).

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

138 CHAPTER 6. SINUSOIDAL MODULATION OF THE PARAMETER

. ϕ (t )
ϕ (t )
v ( t)
0

0 1 2 3 4T
Period of modulation 0.5528T0 , depth of modulation 30%, no friction
. ϕ (t )
ϕ (t )

0
v ( t)

0 1 2 3 4T
Period of modulation 0.545T0, depth of modulation 30%, quality 7.0

Figure 6.5: The phase trajectory of stationary oscillations occurring at the right
boundary of the principal instability interval (left), the time-dependent graphs of
the angular velocity and angular deflection of the rotor, and of radial velocity of
the weights (right).

these graphs from a sine curve, which is caused by the contribution of higher har-
monics (mainly of the third harmonic with the frequency 3ω̃ = 23 ω). The ratio of
the amplitude of the third harmonic to the amplitude of the fundamental harmonic
is approximately the same for both boundaries (|C3 /C1 | ≈ 38 m). The difference
in the patterns of oscillations at the left and right boundaries (compare the graphs
in Figures 6.3 and 6.5) is explained by a different phase shift of the third harmonic
with respect to the fundamental one.
The lower panel of Figure 6.5 corresponds to the right boundary in the pres-
ence of friction. From the asymmetry of the graph it is clear that in this case the
energy received by the oscillator is greater than the energy returned back: During
the intervals of negative values of v (while the weights are moving toward the axis)
the angular velocity ϕ̇ is greater in magnitude. The energy excess compensates for
the frictional losses, providing the stationary oscillations. Outside the instability
interval, the modulation of the moment of inertia causes only a few changes in the
shape of those decaying natural oscillations that may have been excited.
The simulations show that stationary oscillations at the boundaries of the prin-
cipal resonance also include the fifth and even seventh harmonic components with
frequencies 25 ω and 72 ω respectively. To find the boundaries with greater precision,
we should include these high harmonics into the trial function ϕ(t), Eq. (6.26). For
the frictionless oscillator it is more convenient to choose the time origin in such
a way that the motion of the weights along the rod is described in Eq. (6.2) by
l(t) = l0 (1 + m cos ωt) instead of the sine function. In this case the sine and

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

6.2. THE INTERVALS OF PARAMETRIC INSTABILITY 139

cosine harmonics do not mix, that is, the stationary oscillations at the left bound-
ary of the interval include only harmonics of the cosine type, and at the right
boundary—of the sine type.
The final analytical expressions for the frequencies (and periods) of modula-
tion and for the relative contributions of high harmonics (as functions of m) at
the boundaries of the instability interval are complicated and hence not cited here.
However, they show a very good agreement with the simulations. We cite here the
calculated values for a certain modulation depth m = 0.3 (30%). The correspond-
ing experimental values (obtained in the simulation) are shown in the parenthesis:
Left (cosine-type) boundary: Period T /T0 = 0.4066 (0.4066);
C3 /C1 = −0.103 (−0.101); C5 /C1 = 0.015 (0.016); C7 /C1 = 0.002 (0.001).
Right (sine-type) boundary: Period T /T0 = 0.5528 (0.5528);
S3 /S1 = −0.129 (−0.129); S5 /S1 = 0.020 (0.020); S7 /S1 = 0.003 (0.003).
For arbitrary values of the modulation depth m, the boundaries of the principal
instability interval are shown by the first “tongue” of T – m diagram in Figure 6.4.
In the presence of viscous friction the interval of instability of the state of rest
in the equilibrium position shrinks. From Eq. (6.22) (with α = 0) we find that over
the threshold, when m > 1/Q, the following deviations ∆ω of the frequency ω on
both sides of its resonant value ωres correspond to the boundaries of the interval:
p p
∆ω = (mω0 )2 − 4γ 2 = m2 − (1/Q)2 ω0 . (6.29)

From this equation we see that at the threshold (when Q = 1/m) the boundaries
of the interval merge, and the interval of parametric instability disappears.
Outside the instability interval, the modulation of the moment of inertia causes
only a few changes in the shape of those decaying natural oscillations that may
have been excited.

6.2.2 Resonance of the Second Order

In contrast to the principal resonance, for which the energy supply due to the
parameter modulation occurs even if we assume the torsional oscillations to be
purely sinusoidal (see Eq. (6.8)), for the resonance of the second order a positive
net energy delivery is possible only by virtue of the asymmetric distortions in
the shape of the oscillations. These distortions are clearly seen in Figure 6.6. They
provide the motion of the weights toward the axis of rotation (v > 0) to happen on
average at a greater (in magnitude) angular velocity ϕ̇ than the backward motion.
The distortions can be described by the second harmonic component (frequency
2ω), whose contribution is proportional to the depth of modulation m. Hence the
amount of energy delivered by modulation in conditions of the second parametric
resonance is proportional not to m (as at the principal resonance, see Eq. (6.12),
p. 131), but only to m2 (for small values of the modulation depth m ≪ 1).
A system described by Mathieu’s equation, Eq. (6.19), with a harmonic time
dependence of the modulated parameter also has resonances of higher orders (n >

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

140 CHAPTER 6. SINUSOIDAL MODULATION OF THE PARAMETER

. ϕ (t )
ϕ (t )

v ( t)
0 0 1 2 3 4 5 6 7 8T
Period of modulation 0.9375T0, depth of modulation 40%, no friction

Figure 6.6: The phase trajectory of oscillations in conditions of the second para-
metric resonance (left), graphs of the angular velocity, angular deflection, and of
the radial velocity of the weights (right).

1) near the values ω = 2ω0 /n. The width ∆ω of these resonance bands (of the
intervals of instability for the state ϕ = 0, ϕ̇ = 0) diminishes very quickly as the
order n of resonance is increased—as mn . The index s of the rate of the amplitude
growth diminishes also as fast as does ∆ω with the increase in n. Both of these
properties make an experimental observation of parametric resonances of higher
orders n > 1 at moderate values of m very difficult.
In the case of sinusoidal motion of the weights, which is simulated in the rel-
evant computer program, parametric resonance of order n = 2 (with ω ≈ ω0 or
T ≈ T0 , that is, when the period of modulation is approximately equal to the natu-
ral period) is extremely weak and narrow for small values of the depth of modula-
tion. In order to find the boundaries of the second interval of parametric instability,
we look for a solution of Eq. (6.19) that describes stationary oscillations near the
value ω = ω0 . Considering terms up to the second order in the modulation depth
m, we should include in this approximate solution the sinusoidal oscillations with
the fundamental frequency2 ω = ω0 + ε (the frequency of modulation) and the
second harmonic with the frequency 2ω:

ϕ(t) = a0 cos ωt + b0 sin ωt + a1 cos 2ωt + b1 sin 2ωt. (6.30)

An example of such stationary oscillation, which corresponds to the left

boundary of the second instability interval in the absence of friction, is shown
in Figure 6.7. We note the asymmetry of this oscillation: The angular excursion of
the rotor on the right-hand side is smaller than on the left-hand side.
Obviously, for the same boundary there also exists one more asymmetric os-
cillation with greater excursion on the right-hand side. The phase orbits for these
two similar oscillations are the mirror images of one another with respect to the
ordinate axis.
2 However, it may be convenient to consider the fundamental frequency of parametrically excited

stationary oscillations to be always equal to one half of the frequency of modulation. Then the spectrum
of oscillations in the case of resonance of an odd order includes only odd harmonics. The spectrum of
stationary oscillations for resonance of an even order includes only even harmonics (the amplitude of
the fundamental harmonic is zero).

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

6.2. THE INTERVALS OF PARAMETRIC INSTABILITY 141

.
ϕ (t ) ϕ (t )

0
v ( t)

0 1 2 3T
Period of modulation 0.9502T0 , depth of modulation 30%, no friction

Figure 6.7: The phase trajectory of stationary oscillations occurring at the left
boundary of the second instability interval (left), graphs of the angular velocity
and angular displacement of the rotor, and of the radial velocity of the weights
(right).

Substituting ϕ(t) into Eq. (6.19), we transform there the products of sine and
cosine functions into sums, keeping the terms with the frequencies ω and 2ω.
Thus, for the coefficients C2 , S2 , and C4 , S4 we obtain the following system of
homogeneous equations:
ω2
 
3 2γ
1 − 02 C2 + m2 C2 + 2m S4 − S2 = 0,
ω 4 ω0
ω2
 
1 2γ (6.31)
1 − 02 S2 + m2 S2 + 2m C4 + C2 = 0,
ω 4 ω0
3 C4 − 2m S2 = 0, 3 S4 + 2m C2 = 0.

The last two equations of this system give us the expressions for the amplitudes
C4 and S4 of the second harmonic in ϕ(t) in terms of the depth of modulation m
and the amplitudes C2 and S2 of the principal harmonic:
2 2
C4 = m S2 , S4 = − m C2 . (6.32)
3 3
These relations mean essentially that the amplitude of the second harmonic in the
stationary oscillations equals 32 m times the amplitude of the principal harmonic.
The ratio of the amplitudes of these harmonics is the same for both boundaries of
the interval. However, for the left and right boundaries these harmonics add with
different relative phases, creating a different shape of the resulting oscillations.
Graphs of oscillations occurring at the right boundary of the second instability
interval are shown in Figure 6.8.
We note that for the right boundary there also exists an oscillation with the
opposite asymmetry of the angular velocity. The phase orbits for these two similar
oscillations are the mirror images of one another with respect to the abscissa axis
of the phase plane.
Substituting C4 and S4 from Eq. (6.32) into the first two equations of the
system, Eq. (6.31), and taking into account that ω 2 = (ω0 + ε)2 ≈ ω02 + 2ω0ε, we

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

142 CHAPTER 6. SINUSOIDAL MODULATION OF THE PARAMETER

.
ϕ (t ) ϕ (t )

0
v ( t)

0 1 2 3T
Period of modulation 0.9727T0, depth of modulation 30%, no friction

Figure 6.8: The phase trajectory of stationary oscillations occurring at the right
boundary of the second instability interval (left), graphs of the angular velocity
and angular displacement of the rotor, and of the radial velocity of the weights
(right).

obtain the system of two homogeneous equations for C2 and S2 :

 
2ε 7 2γ
− m2 C2 − S2 = 0,
ω0 12 ω0
  (6.33)
2γ 2ε 13
C2 + − m2 S2 = 0.
ω0 ω0 12

A nontrivial solution to this system exists if its determinant equals zero. This con-
dition determines the values of ε = ω − ω0 , which correspond to the boundaries
ω± of the second interval of instability:
 
5 2 1p 4 2
ω± = 1 + m ± m − (4/Q) ω0 . (6.34)
12 8

We note that even the lower boundary is displaced to a higher frequency from the
value ω0 . The boundaries of the interval merge at the threshold. From Eq. (6.34)
we find the threshold conditions for the second parametric resonance:
 
2 4 5
mmin = √ , Qmin = 2 , ωres = 1 + m2 ω0 . (6.35)
Q m 12

Stationary oscillations occurring at the threshold of the second parametric reso-

nance are illustrated by Figure 6.9.
In order to observe the mode of parametric regeneration (stationary oscilla-
tions at the threshold of the second parametric resonance) for a given modulation
depth m in the simulation experiment, we should choose the period of modula-
tion and the quality factor according to Eq. (6.35), and properly set the initial
conditions. For the threshold of the second resonance, Eqs. (6.33) give S2 = C2 .
Therefore,
2 4
ϕ(0) = C2 (1 + m), ϕ̇(0) = ω0 C2 (1 − m). (6.36)
3 3

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

6.2. THE INTERVALS OF PARAMETRIC INSTABILITY 143

.
ϕ (t ) ϕ (t )

0
v ( t)

0 1 2 3T
Period of modulation 0.9639T0, depth of modulation 30%, quality 47.5

Figure 6.9: The phase trajectory of stationary oscillations occurring at the thresh-
old of the second instability interval (left), graphs of the angular velocity, angular
displacement, and of the radial velocity of the weights (right).

To produce stationary oscillations, we can arbitrarily choose an initial angular

displacement ϕ(0), and enter an initial angular velocity ϕ̇(0) = ω0 ϕ(0)(1 − 2m),
as follows from Eq. (6.36). Or, equivalently, we can choose an arbitrary initial
velocity ϕ̇(0), and enter an initial displacement ϕ(0) = ϕ̇(0)(1 + 2m)/ω0 .
In the absence of friction the width of the second interval of instability is
proportional to the square of the depth of modulation: ω+ − ω− = m2 ω0 /4.
Equation (6.34) gives the following boundaries of the interval for zero friction:
   
13 2 7 2
ω+ = 1 + m ω0 , ω− = 1 + m ω0 . (6.37)
24 24
To find the frequencies corresponding to these boundaries with a greater pre-
cision, we should include more harmonics into the trial function ϕ(t), Eq. (6.26).
In the absence of friction it is more convenient to assume that the motion of the
weights along the rod is described in Eq.(5.7) by l(t) = l0 (1 + m cos ωt) instead
of the sine function. In this case the stationary oscillations at the left boundary of
the interval include only harmonics of the cosine type, and at the right boundary
of the sine type.
The final (rather complicated) expressions for the periods of modulation and
for the relative contributions of high harmonics at the boundaries of the instability
interval show a very good agreement with the simulations. Below we cite the
calculated values for the modulation depth m = 0.3 (30%). The corresponding
experimental values are shown in the parenthesis:
Left (cosine-type) boundary: Period T /T0 = 0.9502 (0.9502);
C4 /C2 = −0.203 (−0.202); C6 /C2 = 0.038 (0.039).
Right (sine-type) boundary: Period T /T0 = 0.9727 (0.9727);
S4 /S2 = −0.207 (−0.207); S6 /S2 = 0.039 (0.039).
For arbitrary values of the modulation depth m the calculated boundaries of
this instability interval are shown by the second “tongue” of the T – m diagram
in Figure 6.4.

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

144 CHAPTER 6. SINUSOIDAL MODULATION OF THE PARAMETER

.
ϕ (t ) ϕ (t )

v ( t)
0 1 2 3 4T
Period of modulation 1.4115T0, depth of modulation 35%, quality 300

Figure 6.10: The phase trajectory (left) and the time-dependent graphs (right) of
stationary oscillations at the threshold of the third parametric resonance.

In order to observe stationary oscillations in the simulation experiment for

the case when friction is zero, we should choose the period of modulation cor-
responding to one of these boundaries, and properly set the initial conditions. If
l(t) = l0 (1 + m cos ωt), for the left boundary we can choose arbitrarily an initial
angular displacement ϕ(0) and zero initial angular velocity. For the right bound-
ary, vice versa, we arbitrarily choose an initial angular velocity ϕ̇(0) and zero
initial displacement.

6.2.3 Resonances of the Third and Higher Orders

At parametric resonance of the third order, the flywheel makes three full oscilla-
tions during two cycles of modulation. Stationary oscillations at the threshold of
parametric resonance of the third order are shown in Figure 6.10.
In order to find the boundaries of the third interval of parametric instabil-
ity in the absence of friction, we assume that the weights move according to
l(t) = l0 (1 + m cos ωt), and use the trial function ϕ(t) that includes the funda-
mental harmonic of the frequency 12 ω and several high odd-numbered harmonics
of frequencies 23 ω, 52 ω, . . . .
Stationary oscillations at the left boundary comprise only harmonics of cosine
type, and at the right boundary only harmonics of sine type. After substituting the
trial function into the differential equation
 
d 2 d
(1 + m cos ωt) ϕ + ω02 ϕ = 0, (6.38)
dt dt
we equate to zero the coefficients of cosine (or sine) functions with frequencies
1 3 5
2 ω, 2 ω, 2 ω, . . . , and thus get a system of homogeneous equations for the coef-
ficients C1 , C3 , . . . (or S1 , S3 , . . . ) of harmonic components in the trial function.
The condition of existence of a nontrivial solution to this system yields an equa-
tion for the desired boundaries. This equation is the same as for the boundaries of
the principal instability interval, but this time we look for its approximate solution
in the vicinity of 23 T0 (instead of 12 T0 ). Third harmonic component (frequency
3
2 ω) dominates the spectrum.

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

6.2. THE INTERVALS OF PARAMETRIC INSTABILITY 145

.
ϕ (t ) ϕ (t )

v ( t)
0 1 2 3 4T
Period of modulation 1.4336T0, depth of modulation 30%, no friction

Figure 6.11: The phase trajectories and the time-dependent graphs of stationary
oscillations at the left boundary of the third interval of parametric instability.

ϕ (t )
.
ϕ (t )

v ( t)
0 1 2 3 4T
Period of modulation 1.4369T0, depth of modulation 30%, no friction

Figure 6.12: The phase trajectories and the time-dependent graphs of stationary
oscillations at the right boundary of the third interval of parametric instability.

To increase precision, more harmonics should be included into the trial func-
tion ϕ(t). We cite below the values of the period and of the relative contributions
of different harmonics at stationary oscillations for the modulation depth m = 0.3,
obtained by a calculation in which harmonics up to the thirteenth order were in-
cluded. Such calculations can be fulfilled, say, with the use of Mathematica pack-
age by Wolfram Research Inc. The corresponding experimental values are shown
in the parenthesis:
Left (cosine-type) boundary of the 3rd resonance: Period T /T0 = 1.4336 (1.4336);
C1 /C3 = 0.107 (0.110); C5 /C3 = −0.289 (−0.288); C7 /C3 = 0.065 (0.067).
Right (sine-type) boundary of the 3rd resonance: Period T /T0 = 1.4369 (1.4369);
S1 /S3 = 0.135 (0.136); S5 /S3 = −0.291 (−0.292); S7 /S3 = 0.066 (0.066).
Stationary oscillations at the left boundary of the third interval of parametric
instability are shown in Figure 6.11, and at the right boundary of this interval in
Figure 6.12.
Similar calculations allow us to find the periods of modulation at which reso-
nances of higher orders occur. Corresponding ranges of parametric instability are

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

146 CHAPTER 6. SINUSOIDAL MODULATION OF THE PARAMETER

very narrow, that is, both their boundaries nearly coincide. We cite below the cal-
culated values of the modulation periods and the spectral composition of station-
ary oscillations for the boundaries of the fourth and fifth resonances (at m = 0.3):
Left (cosine-type) boundary of fourth resonance: Period T /T0 = 1.9107 (1.9107);
C2 /C4 = 0.219 (0.220); C6 /C4 = −0.377 (−0.374);
C8 /C4 = 0.100 (0.102); C10 /C4 = −0.023 (−0.021).
Right (sine-type) boundary of fourth resonance: Period T /T0 = 1.9112 (1.9112);
S2 /S4 = 0.222 (0.222); S6 /S4 = −0.377 (−0.377);
S8 /S4 = 0.100 (0.100); S10 /S4 = −0.023 (−0.023).

Left (cosine-type) boundary of fifth resonance: Period T /T0 = 2.3872 (2.3872);

C1 /C5 = 0.017 (0.019); C3 /C5 = 0.319 (0.321);
C7 /C5 = −0.459 (−0.466); C9 /C5 = 0.124 (0.146).
Right (sine-type) boundary of fifth resonance: Period T /T0 = 2.3873 (2.3873);
S1 /S5 = 0.020 (0.020); S3 /S5 = 0.321 (0.321);
S7 /S5 = −0.468 (−0.468); S9 /S5 = 0.142 (0.144).
The spectral composition with the amplitudes of separate harmonics, phase
trajectories and time-dependent graphs of stationary oscillations at the boundaries
of the fourth and fifth intervals of parametric instability are shown in Figures 6.13
and 6.14, respectively.
Almost exact coincidence of both boundaries for the high-order intervals of
parametric instability means that at the period of modulation corresponding to
one of the intervals we can actually observe (in the absence of friction) not a
resonant growth but rather stationary (strongly non-harmonic) oscillations of a
constant (arbitrarily large) amplitude. From the graphs in Figures 6.13 and 6.14
we can conclude that at exact tuning to n-order resonance the oscillator completes
just the whole number n of natural oscillations (of varying period and amplitude)
during exactly two cycles of modulation. The process is periodic at arbitrary initial
conditions, in contrast to the boundaries of low orders, for which special initial
conditions are required to provide periodic oscillations.
This behavior can be explained in terms of the familiar phenomenon of fre-
quency modulation. For parametric resonances of high orders the weights move
along the rod of the exciter rather slowly: The period T of their motion is large
compared to the period T0 of natural torsional oscillations of the rotor (T ≫ T0 ).
A slow periodic variation of the moment of inertia means that the current natural
frequency of the oscillator is slowly modulated.
We see clearly in Figures 6.13 – 6.14 how oscillations slow down (the natu-
ral period increases) when the weights are moved towards the ends of the rotor,
and vice versa. Hence we can consider the motion of the rotor in conditions of a

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

6.3. CONCLUDING REMARKS 147

2.47 .
ϕ (t )
v ( t)

-2 .47 o o
-97.1 0 97.1
Spectrum ϕ (t )
o
78.2
o 29.5
o 0
17.4 o
7.9 1.8o
0 2 4 6 8 10 o
- 97.1
0 1 2 3T
Period of modulation 1.9112T0, depth of modulation 30%, no friction

Figure 6.13: The phase trajectory and time-dependent graph of the angular veloc-
ity (upper part), the spectrum and the graph of the angular displacement of the
flywheel (lower part) for stationary oscillations at the right boundary of the fourth
interval of parametric instability.

high-order parametric resonance as a frequency modulated oscillation,3 in which

the natural oscillation—the dominating harmonic component—plays the role of a
carrier.
The spectral composition shown in Figures 6.13 and 6.14 gives convincing evi-
dence in favor of this interpretation. The harmonic component with the frequency
nω/2 ≈ ω0 has the greatest amplitude (the carrier). The coefficients Cn−2 and
Cn+2 of lateral spectral components with frequencies (nω/2) ± ω have opposite
signs and (for n ≫ 1) are nearly equal in magnitude. This spectrum is character-
istic of the frequency modulation.

6.3 Concluding Remarks

We have developed in this chapter a theoretical approach to the phenomenon of
parametric resonance in a linear oscillator under a smooth modulation of the pa-
rameter, caused by periodic sinusoidal motion of masses along the rod of the fly-
wheel. The analytical description is complemented by a computerized experimen-
tal investigation. A simple mathematical model of the physical system (based on
a linear differential equation) is used. The model allows a complete quantitative
description of the parametric excitation, which can be verified by the simulations.
3 The amplitude of this oscillation is also slightly modulated with the same period as is the fre-

quency. This means that the modulation is actually combined amplitude-frequency modulation, that is,
not purely the frequency modulation.

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

148 CHAPTER 6. SINUSOIDAL MODULATION OF THE PARAMETER

2.34 .
ϕ (t )
v ( t)

-2.34 o
−101 0 101o ϕ (t )
Spectrum
o
74.3
o
23.8
o 34.8 o 0
o 10.7 2.7o
1.5
1 3 5 7 11 9
o
- 101
0 1 2 3 4T
Period of modulation 2.3872T0 , depth of modulation 30%, no friction

Figure 6.14: The phase trajectory and time-dependent graph of the angular veloc-
ity (upper part), the spectrum and the graph of the angular displacement of the
flywheel (lower part) for stationary oscillations at the left boundary of the fifth
interval of parametric instability.

Visualization of motion simultaneously with plotting the graphs of different vari-

ables and phase trajectories makes the simulation experiments very convincing
and comprehensible. This investigation provides a good background for the study
of more complicated nonlinear parametric systems like a pendulum whose length
is periodically changed (a model of the playground swing), or a pendulum whose
suspension point is driven periodically in the vertical direction. Such systems are
investigated in Part II of the textbook.

6.4 Questions, Problems, Suggestions

6.4.1 Principal Parametric Resonance
6.4.1.1∗ Principal Resonance (n = 1) in the Absence of Friction.
(a) Let the period T of the forced motion of the weights correspond to the
principal resonance n = 1 (T = T0 /2). What expression describes the growth
of the amplitude of parametrically excited oscillations when there is no friction?
How does the rate of the growth in amplitude depend on the depth of modulation
m of the distance of the weights from the axis?
(b) What kind of initial conditions allows the growth of the amplitude from
the very beginning of the modulation if the motion of the weights along the rod is
given by Eq. (6.2)?
(c) For the initial conditions found in part (b) and in the absence of friction,
after what number of periods T of modulation does the amplitude increase by 5

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

6.4. QUESTIONS, PROBLEMS, SUGGESTIONS 149

times its initial value for m = 10%?

(d) What is the difference in your answers if you set the initial deflection to be
on the opposite side of the equilibrium position?
6.4.1.2∗ The Amplitude Growth under the Conditions of Principal Reso-
nance without Friction.
(a) If the frequency of modulation is exactly tuned to the principal resonance
(n = 1), what value of the modulation depth m produces an increase in the am-
plitude of oscillation from 20 to 80 degrees during 12 mean natural periods T0 ?
Verify your answer experimentally.
(b) At what initial conditions does the amplitude of oscillation diminish during
the early stages following the start of the simulation? Produce this case in a simu-
lation experiment. Why does a decrease in the amplitude change after a while into
a growth in the amplitude? Using the plots of the oscillations, explain the physical
reason for the growth or decay in amplitude as they depend on the phase relation-
ship between the initial natural oscillation and the forced variation in the moment
of inertia. In what lapse of time does the initial decrease in amplitude turn into an
unlimited increase?
6.4.1.3∗ The Threshold for Principal Resonance.
(a) Estimate the threshold (minimal) value of the modulation depth m that
produces a growth in the amplitude at parametric resonance (T = T0 /2) for a
given value of the quality factor Q (say, Q = 15). Verify your answer experimen-
tally. Explain the features of the plots and of the phase trajectory in this case of
parametric regeneration.
(b) Letting the frequency of modulation be exactly tuned to that of the princi-
pal resonance n = 1, estimate the threshold (minimal) value of the quality factor
Q that corresponds to stationary oscillations (i.e., to parametric regeneration) for
some moderate value (10 – 15 percent) of the modulation depth m of the distance
of the weights from the axis.
(c) What initial conditions produce steady oscillations from the very begin-
ning?
(d) Make your above-calculated estimation of the threshold Q-value more pre-
cise using a computer experiment. Why are there discrepancies between the theo-
retical predictions and your experimental results?
(e) Describe the character of the plots and of the phase trajectory under con-
ditions of parametric regeneration and explain their features. Is the mode of sta-
tionary oscillations at the threshold stable with respect to small deviations in the
properties of the system? Is it stable with respect to small deviations in the initial
conditions?
(f) For the stationary oscillations at the threshold of principal parametric reso-
nance, what is the ratio of the amplitude of the third harmonic to the amplitude of
the fundamental harmonic? (The fundamental frequency in this case is one half of
the frequency of modulation).
6.4.1.4∗ Amplitude Growth above the Threshold.
(a) For a given value of m (say, for m = 0.15) calculate the minimal value of

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

150 CHAPTER 6. SINUSOIDAL MODULATION OF THE PARAMETER

the quality factor Qmin that corresponds to the threshold of parametric excitation.
Enter the doubled value Q = 2Qmin . How many times is the initial amplitude of
oscillation increased during the first 10 cycles of modulation when the modulation
is exactly tuned to resonance? Does the answer depend on the initial conditions?
Verify your answer by simulating the experiment on the computer.
(b) What is the amplitude of oscillations after the next 10 cycles of modula-
tion? Why does the friction not restrict the growth of the amplitude of parametri-
cally excited oscillations?

6.4.2 The Principal Interval of Parametric Resonance

6.4.2.1∗∗ The Principal Interval of Parametric Resonance in the Absence
of Friction.
(a) Calculate the values of the modulation period T that correspond to the
boundaries of the instability interval at a given modulation depth m and γ = 0
(for the approximation m ≪ 1, say, for m = 10%). How does the width of the
interval depend on the depth of modulation?
(b) Calculate the boundaries of the principal interval of parametric instability
up to terms of the second order in the depth of modulation m. Do the terms of the
second order influence the width of the interval?
(c) What initial conditions should you enter for each boundary of the inter-
val in order to observe steady oscillations of a constant amplitude from the very
beginning of the simulation?
Input some initial deflection. What value of the initial angular velocity ought
you to input at a given deflection in order to observe stationary oscillations?
(d) Verify experimentally your calculated approximate values of T and of the
initial angular velocity for either boundary of the instability interval, and try to
find more precise values.
(e) Explain characteristic peculiarities of the graphs and the phase trajecto-
ries of stationary oscillations corresponding to each boundary of the instability
interval.
(f) For these stationary oscillations, what is the ratio of the amplitude of the
third harmonic to the amplitude of the fundamental harmonic?
6.4.2.2∗∗ Oscillations at a Boundary of the Instability Interval.
(a) Enter the value of the modulation period T that corresponds to one of the
boundaries of the instability interval containing the period of principal parametric
resonance. Remember that at these boundaries stationary oscillations of constant
amplitude are possible. If you then enter arbitrary initial conditions, the amplitude
of oscillations at first either grows or decreases, and the phase trajectory is not a
closed curve, as it should be for a periodic process. Explain why.
(b) Follow how the pattern of oscillations gradually approaches the shape that
you should expect for the chosen boundary of the interval of instability. For a
while the oscillations preserve this shape, but then the amplitude begins to grow
or to decrease again, and the shape of oscillations changes again. Why?

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

6.4. QUESTIONS, PROBLEMS, SUGGESTIONS 151

6.4.2.3∗∗ The Threshold of Excitation within the Instability Interval.

(a) Choose a value T of the period of modulation somewhere within the lim-
its of the interval of instability, e.g., approximately halfway between the resonant
value and one of the boundaries. Evaluate experimentally the growth of the am-
plitude without friction and calculate on this basis the threshold value Qmin of
the quality factor for parametric excitation occurring at the given values T of the
modulation period and m of the modulation depth.
(b) Verify your result experimentally and obtain a more precise Q-value for
the threshold. Compare the observed plots of these stationary oscillations with the
plots of stationary (threshold) oscillations when the system is exactly tuned to res-
onance. What are the differences between the plots (and the phase trajectories) of
stationary oscillations at the threshold within the interval of parametric excitation
with friction and the plots (and the phase trajectories) of stationary oscillations at
the boundaries of the instability interval without friction?
(c) How does friction change the width of the principal (n = 1) interval of
instability? Estimate the values of the modulation period T that correspond to the
boundaries of the instability interval for some given values of m and Q. Verify
your results in a simulation experiment on the computer and try to determine the
boundaries more exactly.
(d) For small values of the modulation depth m ≪ 1, try to calculate the
threshold value of the quality factor for some period of modulation T lying within
the principal interval of parametric instability up to terms of the second order in m.
Compare your theoretical result with the threshold value that you have obtained
experimentally in parts (a) and (b).
6.4.2.4 Oscillations outside the Interval of Parametric Resonance.
Enter a value T of the modulation period lying somewhere outside the limits
of the interval of instability for a given value of the modulation depth m. Con-
vince yourself by simulating the experiment that for any initial conditions the
oscillations eventually die out (even if the friction is very weak), and the oscillator
comes to rest at the equilibrium position in spite of the forced periodic changes of
the moment of inertia.

6.4.3 The Second Parametric Resonance

6.4.3.1∗∗ The Second Resonance (n = 2) in the Absence of Friction.
(a) For relatively small values of the modulation depth m ≪ 1 and for zero
friction, calculate the boundaries of the instability interval for resonance of the
order n = 2 (when the period of modulation T is close to the mean natural period
T0 ). How does the width of the interval depend on the depth of modulation m?
(b) For each boundary of the instability interval, what initial conditions pro-
duce stationary oscillations? Verify your calculated values experimentally.
(c) What is the ratio of the amplitude of the second harmonic to the amplitude
of the fundamental harmonic in stationary oscillations at each of the boundaries
of the instability interval? (We assume that in this case the fundamental frequency
equals the frequency of modulation.)

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

152 CHAPTER 6. SINUSOIDAL MODULATION OF THE PARAMETER

6.4.3.2∗∗∗ The Threshold for the Second Parametric Resonance

(a) Enter the period of modulation corresponding to parametric resonance of
the second order. What initial conditions produce increasing oscillations from the
beginning of the simulation?
Note the characteristic shape of the phase diagram at this resonance. Using
the plots obtained in the simulation experiment, explain qualitatively the physi-
cal reasons for the observed growth of the amplitude. Compare the curves (and
the phase diagram) for sinusoidal modulation with the corresponding curves for
square-wave modulation (studied in the previous simulation program “Parametric
Oscillations of Linear Torsion Pendulum”).
(b) For a given modulation depth m, what is the threshold (minimal) value
Qmin of the quality factor for parametric resonance of the second order? What
value of the period of modulation corresponds to the threshold?
(c) What initial conditions produce stationary oscillations for the threshold
conditions from the beginning of the simulation?
(d) What is the ratio of the amplitude of the second harmonic to the ampli-
tude of the fundamental harmonic in stationary oscillations at the threshold of this
resonance?

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

Part II

Nonlinear Oscillations

153

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

Chapter 7

Free Oscillations of the Rigid

Pendulum

Annotation. Various kinds of free (unforced) motion of the planar rigid pendulum
in the gravitational field (including swinging with arbitrarily large amplitudes and
complete revolutions) are investigated in this chapter both analytically and with
the help of computerized simulations. The simulation experiments reveal many
interesting peculiarities of this famous physical model and complement the ana-
lytical study of the subject in a manner that is mutually reinforcing. Chapter 7 also
includes a set of theoretical and experimental problems to be solved by students,
as well as various assignments that the instructor can offer students for possible
individual work on their own.

7.1 Summary of the Theory

7.1.1 The Physical System
The most familiar example of a nonlinear mechanical oscillator is an ordinary
pendulum in the gravitational field, that is, any rigid body that can swing and
rotate about some fixed horizontal axis (a physical pendulum) or a massive small
bob at the end of a rigid rod of negligible mass (a simple pendulum). We employ a
rigid rod rather than a flexible string in order to examine complete revolutions of
the pendulum as well as its swinging to and fro.
The simple pendulum is a famous physical model frequently encountered in
textbooks and papers primarily due to its important role in the history of physics.
This versatile model is useful and interesting not only in itself as the most familiar
example of a nonlinear mechanical oscillator but more importantly because many
problems in various branches of physics can be reduced to the differential equa-
tion describing the motion of a pendulum. The theory of solitons (solitary wave
disturbances traveling in nonlinear media with dispersion), the problem of super-

155

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

156 CHAPTER 7. OSCILLATIONS OF THE RIGID PENDULUM

Figure 7.1: Schematic diagram of the rigid pendulum on the computer screen.

radiation in quantum optics, and Josephson effects in weak superconductivity are

the most important examples.
A very interesting treatment of the pendulum within a historical and cultural
context can be found in the text of Gregory L. Baker and James A. Blackburn,
“The Pendulum: a Case Study in Physics” [1]. This is a unique book in several
ways. Firstly, it is a comprehensive quantitative study of one physical system,
the pendulum, from the viewpoint of elementary and more advanced classical
physics, modern chaotic dynamics, and quantum mechanics. In addition, pendu-
lum analogs of superconducting devices are also discussed. Secondly, this book
treats the physics of the pendulum in various aspects, showing, for example, that
the pendulum has been intimately connected with studies of the earth’s density,
the earth’s motion, and timekeeping.
A schematic diagram of the pendulum pictured on the screen by the simulation
program is shown in Figure 7.1.
In the state of stable equilibrium the center of mass of the pendulum is located
vertically below the axis of rotation. When the pendulum is deflected from this
position through an angle ϕ, the restoring torque of the gravitational force is pro-
portional to sin ϕ. In the case of small angles ϕ (i.e., for small oscillations of the
pendulum) the values of the sine and of its argument nearly coincide (sin ϕ ≈ ϕ),
and the pendulum behaves like a linear oscillator. In particular, in the absence
of friction it executes simple harmonic motion. However, when the amplitude is
large, the motion is oscillatory but no longer simple harmonic. In this case, a graph
of the angular displacement versus time noticeably departs from a sine curve, and
the period of oscillation noticeably depends on the amplitude.
If the angular velocity imparted to the pendulum at its initial excitation is great
enough, the pendulum at first executes complete revolutions losing energy through
friction, after which it oscillates to and fro.
In this chapter we describe a combined analytical and computerized approach
to the eternal problem of the pendulum motion. Our study is based on the usage of
the relevant simulation program of the software package Physics of Oscillations.

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

7.1. SUMMARY OF THE THEORY 157

The simulations allow us to investigate interesting situations that are inaccessible

in a real laboratory experiment. Special attention is devoted to cases in which the
swing approaches 180◦. Such oscillations are treated on the basis of a physically
justified approach in which the cycle of oscillation is divided into several stages.
The major part of the almost closed circular path of the pendulum is approximated
by the limiting motion (see Section 7.1.6), while the motion in the vicinity of the
inverted position is described on the basis of the linearized equation. The accepted
approach provides additional insight into the dynamics of nonlinear physical sys-
tems. Full revolutions of the rigid planar pendulum are also investigated in detail.

7.1.2 The Differential Equation of Motion for a Pendulum

The equation of rotation of a solid about a fixed axis in the absence of friction in
the case of a physical pendulum in a uniform gravitational field is:
J ϕ̈ = −mga sin ϕ. (7.1)
Here J is the moment of inertia of the pendulum relative the axis of rotation, a is
the distance between this axis and the center of mass, and g is the acceleration of
gravity. The left-hand side of Eq. (7.1) is the time rate of change of the angular
momentum, and the right-hand side is the restoring torque of the force of gravity.
This torque is the product of the force mg (applied at the center of mass) and the
lever arm a sin ϕ of this force. Dividing both sides of Eq. (7.1) by J we have:
ϕ̈ + ω02 sin ϕ = 0, (7.2)
where the notation ω02 = mga/J is introduced.
For a simple pendulum a = l, J = ml2 , and so ω02 = g/l. For a physical
pendulum, the expression for ω02 can be written in the same form as for a simple
pendulum provided we define a quantity l to be given by l = J/(ma). It has the
dimension of length, and is called the reduced or effective length of a physical
pendulum. Since the differential equation of motion of a physical pendulum with
an effective length l is the same as that for a simple pendulum of the same length,
the two systems are dynamically equivalent.
At small angles of deflection from stable equilibrium, we can replace sin ϕ
with ϕ in Eq. (7.2). Then Eq. (7.2) becomes the differential equation of motion
of a linear oscillator (see Eq. (1.2) of Chapter 1). Therefore, the quantity ω0 in
the differential equation of the pendulum, Eq. (7.2), has the physical sense of the
angular frequency of infinitely small oscillations of the pendulum in the absence
of friction.
In the presence of a torque due to viscous friction, we must add a term to the
right-hand side of Eq. (7.2) that is proportional to the angular velocity ϕ̇. Thus,
with friction included, the differential equation of the pendulum assumes the form:
ϕ̈ + 2γ ϕ̇ + ω02 sin ϕ = 0. (7.3)
We see that in our model a pendulum is characterized by two parameters:
The angular frequency ω0 of small free oscillations, and the damping constant

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

158 CHAPTER 7. OSCILLATIONS OF THE RIGID PENDULUM

γ, which has the dimensions of frequency (or of angular velocity). As in the

case of a linear oscillator, it is convenient to use the dimensionless quality fac-
tor Q = ω0 /(2γ) (see Eq. (1.11) of Chapter 1) rather than the damping constant γ
to measure the effect of damping. At small free oscillations of the pendulum, the
value Q/π is the number of complete cycles during which the amplitude decreases
by a factor of e ≈ 2.72.
The principal difference between Eq. (7.3) for the pendulum and the corre-
sponding differential equation of motion for a spring oscillator discussed in Chap-
ter 1 is that Eq. (7.3) is a nonlinear differential equation. The difficulties in obtain-
ing an analytical solution of Eq. (7.3) are caused by its nonlinearity. In the general
case it is impossible to express the solution of Eq. (7.3) in elementary functions,
although in the absence of friction the solution of Eq. (7.2) can be given in terms
of special functions (elliptic integrals).

7.1.3 Dependence of the Period on the Amplitude

The nonlinear character of the pendulum is revealed primarily in dependence of
the period of oscillations on the amplitude. To find an approximate formula for
this dependence, we can expand sin ϕ in Eq. (7.3) into the power series. Keeping
the first two terms, we obtain (for the conservative pendulum with γ = 0):
1
ϕ̈ + ω02 ϕ − ω02 ϕ3 = 0. (7.4)
6
An approximate solution to Eq. (7.4) can be searched as a superposition of the
sinusoidal oscillation ϕ(t) = ϕm cos ωt and its third harmonic ǫϕm cos 3ωt. (We
assume t = 0 to be the moment of maximal deflection). This solution is found in
many textbooks (see, for example, [2]). The fractional contribution ǫ of the third
harmonic equals ϕ2m /192, where ϕm is the amplitude of the principal harmonic
component whose frequency ω differs from the limiting frequency ω0 of small
oscillations by a term proportional to the square of the amplitude:

ω ≈ ω0 (1 − ϕ2m /16), T ≈ T0 (1 + ϕ2m /16). (7.5)

The same approximate formula for the period can be obtained by expanding the
exact solution expressed in terms of elliptic integrals (see, for example, [2], [3], or
[4]) into a power series with respect to the amplitude ϕm .
Equation (7.5) shows that, say, for ϕm = 30◦ (0.52 rad) the fractional in-
crement of the period (compared to the period of infinitely small oscillations)
equals 0.017 (1.7%). The fractional contribution of the third harmonic in this non-
sinusoidal oscillation equals 0.14%, that is, its amplitude equals only 0.043◦.
The relevant simulation program of the package PHYSICS OF OSCILLA-
TIONS allows us to verify this approximate formula for the period. Table 7.1
gives the values of T (for several values of the amplitude) calculated with the help
of Eq. (7.5) and measured in the computational experiment. Comparing the values
in the last two columns, we see that the approximate formula, Eq. (7.5), gives the
value of the period for the amplitude of 45◦ with an error of only 0.13%. However,

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

7.1. SUMMARY OF THE THEORY 159

Table 7.1. Dependence of period on the amplitude

Amplitude T /T0 T /T0

ϕm (calculated) (measured)

30◦ (π/6) 1.0171 1.0175

45◦ (π/4) 1.0386 1.0400
60◦ (π/3) 1.0685 1.0732
90◦ (π/2) 1.1539 1.1803
120◦ (2π/3) 1.2742 1.3730
135◦ (3π/4) 1.3470 1.5279
150◦ (5π/6) 1.4284 1.7622

for 90◦ the error is already 2.24%. The error does not exceed 1% for amplitudes
up to 70◦ .

7.1.4 The Phase Portrait of the Pendulum

The evolution of the mechanical state of the pendulum during its entire motion
can be graphically demonstrated very clearly by a phase diagram, i.e., a graph that
plots the angular velocity ϕ̇ (or the angular momentum I ϕ̇) versus the angular dis-
placement ϕ. If the motion of the physical system is periodic, the system returns
to the same mechanical state after a full cycle, and the representative point, mov-
ing clockwise, generates a closed path in the phase plane. In general, the structure
of a phase diagram tells us a great deal about the possible motions of a nonlinear
physical system.
A general idea about the free motion of the pendulum resulting from various
values of energy imparted to the pendulum is given by its phase portrait, i.e., the
family of phase trajectories.
We can construct a phase portrait for a conservative system (e.g., for the pen-
dulum), without explicitly solving the differential equation of motion of the sys-
tem. The equations for phase trajectories follow directly from the law of the con-
servation of energy.
The potential energy Epot (ϕ) of a pendulum in the gravitational field depends
on the angle of deflection ϕ measured from the equilibrium position:

Epot (ϕ) = mga(1 − cos ϕ). (7.6)

A graph of Epot (ϕ) is shown in the upper panel of Figure 7.2. The potential
energy of the pendulum has a minimal value of zero in the lower stable equilibrium
position (at ϕ = 0), and a maximal value of Emax = 2mga in the inverted position
(at ϕ = ±π) of unstable equilibrium. This maximal value of the potential energy
is assumed to be the unit of energy in Figure 7.2. The dashed line shows the
parabolic potential well for a linear oscillator whose period is independent of the

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

160 CHAPTER 7. OSCILLATIONS OF THE RIGID PENDULUM

E/E max
1 E pot

0.5

- 2p -p . p 2p j
j
2
1

- 2p -p 0 p 2p j
-1
-2

Figure 7.2: The potential well (graph of the potential energy Epot (ϕ)) and the
phase portrait of the conservative planar rigid pendulum. Closed phase trajectories
that enclose the origin of the phase plane correspond to oscillations with different
amplitudes. Trajectories passing over and below separatrix (see text for detail)
correspond to counterclockwise and clockwise revolutions, respectively.

amplitude (and of the energy) and equals the period of infinitely small oscillations
of the pendulum.
In the absence of friction, the total energy E of the pendulum, i.e., the sum
of its kinetic energy, Ekin (ϕ̇) = 12 J ϕ̇2 , and potential energy, Epot (ϕ), remains
constant during the motion:
1 2
J ϕ̇ + mga(1 − cos ϕ) = E. (7.7)
2
This equation gives the relation between ϕ̇ and ϕ, and therefore is the equation
of the phase trajectory that corresponds to a definite value E of total energy.
It is convenient to express Eq. (7.7) in a slightly different form. Recalling that
mga/J = ω02 and defining the quantity E0 = Jω02 /2 (the quantity E0 has the
physical sense of the kinetic energy of a body with the moment of inertia J, rotat-
ing with the angular velocity ω0 ), we rewrite Eq. (7.7):

ϕ̇2 E
2 + 2(1 − cos ϕ) = . (7.8)
ω0 E0
If the total energy E of the pendulum is less than the maximal possible value of
its potential energy (E < 2mga = Emax = 4E0 ), that is, if the total energy is less
than the height of the potential barrier shown in Figure 7.2, the pendulum swings

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

7.1. SUMMARY OF THE THEORY 161

to and fro between the extreme deflections ϕm and −ϕm . These angles correspond
to the extreme points at which the potential energy Epot (ϕ) becomes equal to the
total energy E of the pendulum. If the amplitude is small (ϕm ≪ π/2), the time
dependence of oscillations is nearly sinusoidal, and the corresponding phase tra-
jectory is nearly an ellipse. The elliptical shape of the curve follows from Eq. (7.8)
if we substitute there the approximate expression cos ϕ ≈ 1−ϕ2 /2 valid for small
angles ϕ:
ϕ̇2 ϕ2
+ = 1. (7.9)
Eω02 /E0 E/E0
p (ϕ, ϕ̇). Its horizontal semiaxis
This is the equation of an ellipse in the phase plane
equals the maximal deflection angle ϕm = E/E0 . If the angular velocity ϕ̇
on the ordinate axis is plotted in units of the angular frequency ω0 of small free
oscillations, the ellipse (7.9) becomes a circle.
The shape of the closed phase trajectory gradually changes as the amplitude
and the energy are increased. The width (along ϕ-axis) of the phase trajectory
increases more rapidly than does its height as the total energy E increases to Emax .
The phase trajectory is stretched horizontally because for the same total energy the
amplitude of oscillations in the potential well of the pendulum is greater than it is
in the parabolic potential well of the linear oscillator. The greater the total energy
E (and thus the greater the amplitude ϕm ), the greater the departure of the phase
trajectory from an ellipse and the greater the departure of the motion from simple
harmonic.
With the growth of the angular displacement, the restoring torque for the pen-
dulum does not increase as rapidly as for the linear oscillator: The pendulum is
a system with a “soft” restoring torque. The upper slopes of its potential well are
not as steep as those of the parabola, and at large amplitudes the pendulum spends
more time near the extreme points where its direction of motion is reversed.
The period, while independent of the amplitude for the linear oscillator, grows
with the amplitude for the pendulum. The crests of the graph of ϕ(t) are flattened,
and those of the ϕ̇(t) graph are sharpened. These changes in the shape of time-
dependent graphs of oscillations are clearly visible in Figure 7.3. We also note
the increased period of these oscillations—the time marks on the abscissa axis of
these graphs correspond to the period of infinitely small oscillations.
In the case of a linear oscillator whose potential well is parabolic, time de-
pendencies of both potential and kinetic energies are sinusoidal, and their time
average values are equal to one another.
As the extreme angular displacement approaches 180◦ , the pendulum spends
the greater part of its period near the inverted position, and so the potential energy
of the pendulum is close to its maximal value 2mga most of the time. Only for
the brief time during which the pendulum rotates rapidly through the bottom part
of its circular path is the potential energy converted into kinetic energy. Crests of
the graph of the potential energy Epot (t) become wider than the valleys between
them (see the lower panel of Figure 7.3).
The opposite changes occur with the graph of the kinetic energy Ekin (t). Al-
though maximum values of both potential and kinetic energies are equal to the

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

162 CHAPTER 7. OSCILLATIONS OF THE RIGID PENDULUM

180o
.
ϕ (t) ϕ0 = 175.0o
ϕ (t)
0

o
-180
0 1 2 3 4 5 6T0

E pot E tot

E kin

0 1 2 3 4 5 6T0

Figure 7.3: The graphs of large oscillations in the absence of friction (initial con-
ditions: ϕ(0) = ϕ0 = 175◦ , ϕ̇(0) = 0, amplitude ϕm = 175◦ ).

constant value of total energy Etot , the average value of the potential energy dur-
ing a complete cycle of this motion becomes considerably greater than that of the
kinetic energy, in contrast to the case of small oscillations, for which the time
average values of potential and kinetic energies are equal.
As the total energy imparted to the pendulum is increased so as to approach
the value 2mga from below, the period of oscillations sharply increases and tends
logarithmically to infinity. The shape of the curve of angular velocity versus time
resembles a periodic succession of solitary impulses whose duration is close to
the period T0 of small oscillations. Time intervals between successive impulses
are considerably greater than T0 . These intervals grow longer and longer as the
total energy E is changed so as to approach the maximal allowed potential energy
2mga. Executing such swinging, the pendulum moves very rapidly through the
bottom of its circular path and very slowly at the top, in the vicinity of the extreme
points.
If E > 2mga, the kinetic energy and the angular velocity of the pendulum are
non-zero even at ϕ = ±π. In contrast to the case of swinging, now the angular
velocity does not change its sign. The pendulum executes rotation in a full cir-
cle. This rotation is nonuniform. When the pendulum passes through the lowest
point (through the position of stable equilibrium), its angular velocity is greatest,
and when the pendulum passes through the highest point (through the position of
unstable equilibrium), its angular velocity is smallest.
In the phase plane, rotation of the pendulum is represented by the paths that
continue beyond the vertical lines ϕ = ±π, repeating themselves every full cycle
of revolution, as shown in Figure 7.2.
Upper paths lying above the ϕ-axis, where ϕ̇ is positive and ϕ grows in value,
correspond to counterclockwise rotation, and paths below the axis, along which

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

7.1. SUMMARY OF THE THEORY 163

the representative point moves from the right to the left, correspond to clockwise
rotation of the pendulum.
The phase trajectory that corresponds to the total energy E = 2mga separates
the central region of the phase plane which is occupied by the closed phase trajec-
tories of oscillations from the outer region, occupied by the phase trajectories of
rotations. This boundary is called the separatrix. The separatrix divides the phase
plane of a conservative pendulum into regions that correspond to different types
of motion (see Figure 7.2).
For a conservative system, the equation of a phase trajectory (e.g., Eq. (7.8)
in the case of a pendulum) is always an even function of ϕ̇, because the energy
depends only on ϕ̇2 . Consequently, the phase trajectory of a conservative system
is symmetric about the horizontal ϕ-axis. This symmetry means that the motion
of the system in the clockwise direction is mechanically the same as the motion
in the counterclockwise direction. In other words, the motion of a conservative
system is reversible: If we instantaneously change the sign of its velocity, the
representative point jumps to the symmetric position of the same phase trajectory
on the other side of the horizontal ϕ-axis. In the reverse motion the system passes
through each spatial point ϕ with the same speed as in the direct motion. Since
changing the sign of the velocity (ϕ̇ → −ϕ̇) is the same as changing the sign of
time (t → −t), this property of a conservative system is also referred to as the
symmetry of time reversal.
The additional symmetry of the phase trajectories of the conservative pendu-
lum about the vertical ϕ̇-axis (with respect to the change ϕ → −ϕ) follows from
the symmetry of its potential well: Epot (−ϕ) = Epot (ϕ). (Unlike the symmetry
about the ϕ-axis, this additional symmetry is not a property of all conservative
systems.)
When we include friction in our model, motion of the pendulum becomes irre-
versible, and the above-discussed symmetry of its phase trajectories with respect
to reflections in the coordinate axes of the phase plane vanishes. The influence of
friction on the phase portrait we discuss below (see Section 7.4.1, p. 179).
The angles ϕ and ϕ±2π, ϕ±4π, . . . denote the same position of the pendulum
and thus are equivalent. Thus it is sufficient to consider only a part of the phase
plane, e.g., the part enclosed between the vertical lines ϕ = −π and ϕ = π
(see Figure 7.2). The cyclic motion of the pendulum in the phase plane is then
restricted to the region lying between these vertical lines. We can identify these
lines and assume that when the representative point leaves the region crossing the
right boundary ϕ = π, it enters simultaneously from the opposite side at the left
boundary ϕ = −π (for a counterclockwise rotation of the pendulum).
We can imagine the two-dimensional phase space of a rigid pendulum not
only as a part of the plane (ϕ, ϕ̇) enclosed between the vertical lines ϕ = +π and
ϕ = −π, but also as a continuous surface. We may do so because opposing points
on these vertical lines have the same value of ϕ̇ and describe physically equiva-
lent mechanical states. And so, taking into account the identity of the mechanical
states of the pendulum at these points and the periodicity of the dependence of
the restoring gravitational torque on ϕ, we can cut out this part of the phase plane

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

164 CHAPTER 7. OSCILLATIONS OF THE RIGID PENDULUM

and roll it into a cylinder so that the bounding lines ϕ = +π and ϕ = −π are
joined. We can thus consider the surface of such a cylinder as the phase space
of a rigid pendulum. A phase curve circling around the cylinder corresponds to a
nonuniform rotational motion of the pendulum.

7.1.5 The Phase Portrait in the Simulation Program

Choosing the item of the computer program entitled “Phase Portrait and T (E)”
(under the item “View” of the main menu), you can construct on the screen a
whole family of phase trajectories of the pendulum. This phase portrait represents
possible motions of the pendulum in the absence of friction. Different curves of
this family correspond to various values of the total energy of the pendulum. In
this experiment, simultaneously with drawing a phase trajectory, you can measure
the corresponding period of oscillations or rotations. Doing this for various values
of the energy imparted to the pendulum at the initial excitation, you can obtain
the dependence T (E) of the period on total energy E. Two modes of working
with the computer are available in this section: demonstration and computation
experiment.
To initiate the drawing of a phase curve, type in the desired value of the energy
E/Emax into the text-box labeled “Energy value E/Emax for the next curve,” and
click on the Go command button. Each time you do so, the pendulum is set in
motion from the equilibrium position by a momentary push. By clicking at the
check-box “Demo mode,” you can activate the demonstration mode, in which the
energy imparted to the pendulum is increased by a fixed amount for each subse-
quent curve. You will obtain a phase portrait of the conservative pendulum formed
by the family of phase curves for a set of values of the total energy differing by a
fixed amount (similar to one shown in Figure 7.2).
During the construction of the phase curves, the motion of the representative
point in the phase plane is compared with the graph of the potential energy versus
the angle of deflection. This graph is displayed above the phase plane on the com-
puter screen. At the extreme points of the phase diagram, where the curve crosses
the ϕ-axis, vertical dotted lines are drawn that connect the corresponding points
of the phase diagram and of the potential energy graph. (At these points the total
energy equals the potential energy.)
Simultaneously with the phase trajectory, the simulation program draws
graphs of the time-dependence of the kinetic energy and of the total energy (on the
upper right-hand side of the screen). The graph of the total energy for a conser-
vative system is clearly a horizontal straight line at a distance from the time-axis
corresponding to the given value of the initial energy. For each value of the energy,
the plotting of curves on the screen is terminated automatically when the pendu-
lum completes one full cycle of its motion. Hence the lengths of these horizontal
segments of the graphs of the total energy are equal to the period and give a very
clear indication of the dependence of the period T on the total energy E.
In the lower right-hand corner of the screen, a table is displayed that shows
the values of period T obtained in this computer simulation, for different values

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

7.1. SUMMARY OF THE THEORY 165

T/T0

3.0

Period
2.0

1.0

0 0.5 1.0 1.5 E/E max

Energy

Figure 7.4: The period versus total energy.

of the total energy E. From this table you can form a quantitative judgment about
the dependence of the period on the total energy T (E).
It is also possible to draw a continuous curve through the ends of the hori-
zontal segments of the graphs of the total energy that correspond to one cycle of
oscillation or rotation. To do this, click on the relevant check-box “Plot T (E)” at
the bottom of the panel. This curve gives a graph of dependence of the period T of
oscillations and revolutions on the total energy E imparted to the pendulum. The
axis of energy on this diagram is oriented vertically, the axis of period horizon-
tally. The graph T (E) with traditional orientation of axes—energy E as abscissa,
period T as ordinate—is shown in Figure 7.4.
The initial almost linear growth of the period with E corresponds to the ap-
proximate formula, Eq. (7.5). Indeed, Eq. (7.5) predicts a linear dependence of T
on ϕ2m , and for small amplitudes ϕm the energy is proportional to the square of
the amplitude.
When the energy approaches the value Em , the period grows infinitely. We
note the sharp unlimited growth of the period as the total energy E of the pen-
dulum approaches the maximal possible value Emax = 2mga of the potential
energy. Greater values of the energy correspond to the rotating pendulum. The
period of rotation rapidly decreases with the energy.

7.1.6 The Limiting Motion along the Separatrix

The phase trajectory corresponding to a total energy E that is equal to the max-
imal possible potential energy, namely Epot (π) = 2mga = Emax , is of special
interest. This boundary (the separatrix) divides the phase plane of a conservative
pendulum into regions that correspond to different types of motion. The equation
of the separatrix follows from Eq. (7.7) by setting E = Emax = 2mga, or from

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

166 CHAPTER 7. OSCILLATIONS OF THE RIGID PENDULUM

Eq. (7.8) by setting E = 4E0 = 2Jω02 :

ϕ̇ = ±2ω0 cos(ϕ/2). (7.10)

The limiting motion of a conservative pendulum with total energy E = 2mga

is worthy of a more detailed investigation. In this case the representative point in
the phase plane moves along the separatrix.
When the pendulum with the energy E = 2mga approaches the inverted po-
sition at ϕ = π or ϕ = −π, its velocity tends to zero, becoming zero at ϕ = ±π.
This state is represented in the phase plane by the saddle points ϕ = π, ϕ̇ = 0 and
ϕ = −π, ϕ̇ = 0 where the upper and lower branches of the separatrix, Eq. (7.10),
meet on the ϕ-axis. Both these points represent the same mechanical state of the
system, that in which the pendulum is at rest in the unstable inverted position.
The slightest initial displacement of the pendulum from this point to one side or
the other results in its swinging with an amplitude that almost equals π, and the
slightest initial push causes rotational motion (revolution) of the pendulum in a
full circle. With such swinging, or with such rotation, the pendulum remains in
the vicinity of the inverted position for an extended time.
For the case of motion along the separatrix, i.e., for the motion of the pendu-
lum with total energy E = 2mga = 4E0 , there exists an analytical solution (in
terms of elementary functions) for the angle of deflection ϕ(t) and for the angular
velocity ϕ̇(t). Integration of the differential equation Eq. (7.10) with respect to
time (for the positive sign of the root) at the initial condition ϕ(0) = 0 yields:

−ω0 t = ln tan[(π − ϕ)/4], (7.11)

and we obtain the following expression for ϕ(t):

ϕ(t) = π − 4 arctan(e−ω0 t ). (7.12)

This solution describes a counterclockwise motion beginning at t = −∞ from

ϕ = −π. At t = 0 the pendulum passes through the bottom of its circular path,
and continues its motion until t = +∞, asymptotically approaching ϕ = +π. A
graph of ϕ(t) for this motion is shown in Figure 7.5.
The second solution that corresponds to the clockwise motion of the pendulum
(to the motion along the other branch of the separatrix in the phase plane) can be
obtained from Eq. (7.12) by the transformation of time reversal, i.e., by the change
t → −t. Solutions with different initial conditions can be obtained from Eq. (7.12)
simply by a shift of the time origin (by the substitution of t − t0 for t).
Differentiating ϕ(t) given by Eq. (7.12) with respect to time t, we find the
following time dependence of the angular velocity ϕ̇(t) for the limiting motion of
the pendulum:
2ω0 4ω0
ϕ̇(t) = = ω0 t . (7.13)
cosh(ω0 t) e + e−ω0 t
The graph of this function ϕ̇(t) has the form of an isolated impulse (see Fig-
ure 7.5). In Eq. (7.13) the origin t = 0 is chosen to be the instant at which the pen-
dulum passes through the equilibrium position with the angular velocity ϕ̇ = 2ω0 .

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

7.1. SUMMARY OF THE THEORY 167

p
.
ϕ (t) ϕ (t)
p
2

-1.5T0 -T0 -T0 /2 T0 /2 T0 1.5 T0

0.14 T0
p
2

-p

Figure 7.5: The graphs of ϕ(t) and ϕ̇(t) for the limiting motion (total energy
E = Emax = 2mga = 4E0 ).

This moment corresponds to the peak value of the impulse. The characteristic
width of its profile, i.e., the duration of such a solitary impulse, is of the order of
1/ω0 . For time t = ±T0 /2 on either side of the peak, Eq. (7.13) gives the angular
velocity of only ±0.17 ω0. Therefore the period T0 = 2π/ω0 of small oscillations
gives an estimate for the duration of the impulse on the velocity graph, that is, for
the time needed for the pendulum to execute almost all of its circular path, from
the vicinity of the inverted position through the lowest point to the other side of
the inverted position.
Using the analytical expression for the time dependence of the angular de-
flection given by Eq. (7.12), we can calculate the time interval τ during which
the pendulum moves from one horizontal position to the other, passing through
the lower equilibrium position: τ = 0.28 T0 . During this time the kinetic energy
of the pendulum is greater than its potential energy, so τ is the half-width of the
solitary impulse of the kinetic energy for the motion under consideration.
The wings of the ϕ̇(t) profile for the limiting motion decrease exponentially
as t → ±∞. Actually, for large positive values of t, we may neglect the second
term exp(−ω0 t) in the denominator of Eq. (7.13), and we find that:

ϕ̇(t) ≈ ±4ω0 e−ω0 t . (7.14)

Thus, in the limiting motion of the representative point along the separatrix, when
the total energy E is exactly equal to the height 2mga of the potential barrier, the
speed of the pendulum decreases steadily as it nears the inverted position of unsta-
ble equilibrium. The pendulum approaches the inverted position asymptotically,
requiring an infinite time to reach it. The motion is not periodic.
The mathematical relationships associated with the limiting motion of a pen-
dulum along the separatrix play an important role in the theory of solitons.
In the program that simulates the pendulum motion, there is a section “Spec-
trum of Oscillations.” In this section a Fourier decomposition of periodic oscil-
lations occurring in the absence of friction is performed. Oscillations with large
amplitudes are of special interest because their spectrum is rich of harmonics. You

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

168 CHAPTER 7. OSCILLATIONS OF THE RIGID PENDULUM

ϕ (t)

.
ϕ ( t)

Figure 7.6: Graphs of ϕ(t) (upper panel) and ϕ̇(t) (lower panel) for oscillations
with the amplitude 179.885◦ in the absence of friction, and the graphs of their
harmonics.

can obtain the graphs of the separate harmonic components as well as the graph of
the partial sum of any number of harmonics. You can then compare these graphs
with the graph of the actual motion, which is plotted by the program (in the same
place on the screen) during the numerical simulation. Figure 7.6 shows the graph
of ϕ(t) and its harmonics (upper panel), and the graph of the angular velocity
ϕ̇(t) with its harmonics (lower panel) for oscillations of the pendulum in the ab-
sence of friction with the amplitude 179.885◦. The period T of such oscillations
equals more than five periods of oscillations with an infinitely small amplitude
T =5.276 T0 ).

7.2 Oscillations of the Pendulum with Extremely

Large Amplitudes
The previous analysis of the limiting motion along the separatrix provides a means
for understanding the character of oscillations with amplitudes approaching 180◦,
when total energy of the pendulum is slightly less than the height 2mga of the
potential barrier, the potential energy of the inverted pendulum.
The old problem of large oscillations of a simple planar pendulum continues
to attract attention of the academic community. Dozens of papers on the subject
appeared during the last decade in journals—see, for example, [5]–[9] and ref-
erences therein. In most of the papers various approximation schemes have been
developed to express the large-angle pendulum period by simple formulae in terms
of elementary functions. Each of the authors usually claims that the formula pro-
posed by him is more simple and accurate when compared with other approximate
formulae. A detailed comparison of several approximate expressions that have ap-
peared in recent publications can be found in [10]. The common feature of all

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

7.2. OSCILLATIONS WITH EXTREMELY LARGE AMPLITUDES 169

suggested approximation schemes can be reduced to a search for some empirical

expression for the period T (ϕm ), which gives for large amplitudes ϕm an accept-
able numerical agreement with the values obtained from the exact formula given
by the complete elliptic integral of the first kind, K(q):
Zπ/2
2 ϕm dx
T (ϕm ) = T0 K(sin2 ( )), K(q) = p , (7.15)
π 2 1 − q sin2 x
0
p
where T0 = 2π/ω0 = 2π l/g is the natural period and ω0 is the frequency of os-
cillations with infinitely small amplitude, l is the effective length of the pendulum,
and g is the acceleration due to gravity.
The approximate expressions for the period that can be found in the literature
(see [10] and references therein) give indefinitely increasing errors as the ampli-
tude of the pendulum tends to 180◦ . Moreover, all these exercises with various
approximation schemes give little physical insight into the nonlinear dynamics of
the pendulum behavior at large amplitudes.
In the present section we suggest a radically different approach to the problem
of extremely large amplitudes. Our approach is based on physically clear presen-
tation of large oscillations as consisting of several stages during which the motion
can be described analytically with high precision in terms of elementary functions.
The principal idea of our approach is very simple: The motion of the pendulum in
the close vicinity of the inverted position can be described by a linear differential
equation (if we choose as a variable the angle α = π − ϕ, which the pendulum
makes with the upper vertical line), while the remaining part of the pendulum’s
path (constituting nearly a full circle) is almost indistinguishable from the limiting
motion (motion along the separatrix), for which a simple solution in elementary
functions is available. Precision of the final (very simple) formula for the period,
Eq. (7.22), p. 172, increases as the amplitude approaches 180◦ .

7.2.1 Oscillations with Amplitudes Approaching 180◦

If the pendulum is released with zero initial velocity near the inverted position
(say, at initial angle about 179◦ ), it slowly starts moving toward the down position
with a small initial acceleration, because the torque of gravity, being proportional
to the sine of deviation from the inverted position, is small. After the pendulum
gains some speed, it rapidly makes almost a full circular path through the lower
equilibrium position. When the pendulum occurs on the opposite side of the in-
verted position, its motion gradually slows down as it climbs up along the slope of
the potential barrier to its summit. In the absence of friction the pendulum stops
when its angular distance to the vertical becomes equal to the initial deviation.
From this turning point all the motion repeats in the opposite direction, and after
a period the pendulum occurs at the initial point with zero velocity.
To observe the simulation of oscillations discussed in this section, we should
switch off the viscous friction (using the corresponding check-box on the “Param-
eters” panel), and choose appropriate initial conditions (initial angle about 179◦,

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

170 CHAPTER 7. OSCILLATIONS OF THE RIGID PENDULUM

o
180
ϕ (t ) ϕ0= 179.90o

0
.
ϕ (t )

o
-180
180o
ϕ (t ) ϕ0= 179.99o

0
.
ϕ (t )

o
-180
0 1 2 3 4 5 6T0

Figure 7.7: Graphs of ϕ(t) and ϕ̇(t) for oscillations with the amplitude 179.90◦
and 179.99◦ in the absence of friction, obtained in the simulation experiment.

initial velocity zero). The program allows the user to plot the time dependencies
of ϕ(t) and ϕ̇(t), and to draw the phase trajectory simultaneously with the visu-
alization of oscillations. Graphs of ϕ(t) and ϕ̇(t) for oscillations with amplitudes
179.90◦ and 179.99◦ in the absence of friction are shown in Figure 7.7.
Comparing these graphs, we can see that for most of the angular excursion
from −π to π, these graphs for amplitudes 179.90◦ and 179.99◦ are nearly iden-
tical. We guess that for these stages of motion deflection angle ϕ(t) and angular
velocity ϕ̇(t) are characterized by almost the same time dependence as for the lim-
iting motion along the separatrix, shown in Figure 7.5, p. 167. This dependence
of ϕ(t) on time is described (in elementary functions) by the simple expression
(7.13), p. 166. Hence the duration of this stage of oscillation for all these cases of
large amplitudes approaching 180◦ is about T0 (the period of small oscillations)
and can be calculated with high precision with the help of the same expression
(7.13). The duration of the remaining stage, during which the pendulum lingers
near the inverted position, depends critically on the amplitude ϕm . This is clearly
seen from comparison of the upper and lower panels of Figure 7.7. This duration
increases indefinitely as ϕm → 180◦ . In order to calculate the duration of this
stage for certain large amplitudes, we can make use of the linearized differential
equation, applicable for small deviations from the inverted position. Next we will
do this (see discussion on p. 171).
The closed phase trajectory of oscillatory motion with a large amplitude ϕm
is shown in Figure 7.8. The largest part of the phase trajectory almost coincides
with the separatrix. The representing point goes around the whole closed curve

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

7.2. OSCILLATIONS WITH EXTREMELY LARGE AMPLITUDES 171

.
a ϕ
2ω 0 b .
ϕ

1 αm
ϕc
0 ϕ ϕ
−π ϕm π π
ϕm
−1
αc
−2
t = T/4

Figure 7.8: The phase trajectory of oscillatory motion with a large amplitude ϕm
(a) and its portion (increased) that corresponds to the motion of the pendulum in
the vicinity of the inverted position (b).

during one period T of oscillation. Next we consider one quarter of this curve,
which starts in the phase plane at the initial point of maximal deflection ϕ =
ϕm and initial velocity ϕ̇(t) = 0, and ends at the point ϕ = 0 (marked as t =
T /4 in Figure 7.8a). To calculate this time t = T /4, we choose on this curve
an arbitrary point ϕ = ϕc located not far from the inverted position ϕ = π (see
Figure 7.8b), which divides the curve into two parts. The first part between ϕ =
ϕm and ϕ = ϕc lies in the vicinity of the inverted position, so that duration t1 of
motion along this part can be calculated with the help of a linearized differential
equation of motion (see below). The second part between ϕ = ϕc and ϕ = 0 is
almost indistinguishable from the separatrix, so that duration t2 of motion along
this part can be immediately expressed with the help of Eq. (7.12):
π − ϕc αc 4
ω0 t2 = − ln tan = − ln tan ≈ ln . (7.16)
4 4 αc
Here we introduced the notation αc = π − ϕc for the angle that the pendulum
makes with the upward vertical line at ϕ = ϕc . When ϕc is close to π, the angle
αc is small, so that in Eq. (7.16) we can assume tan(αc /4) ≈ αc /4. Therefore
ω0 t2 ≈ ln(4/αc ).
When considering the motion of the pendulum in the vicinity of the inverted
position, we find it convenient to define the pendulum position (instead of the
angle ϕ) by the angle α of deflection from the position of unstable equilibrium.
This angle equals π − ϕ, so that ϕ = π − α. Substituting angular acceleration
ϕ̈ = −α̈ and sin ϕ = sin α in Eq. (7.2), we find the differential equation for the
pendulum in terms of α. Since near the inverted position α ≪ 1, we can replace
in this equation sin α by α. Thus we get the following linear differential equation

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

172 CHAPTER 7. OSCILLATIONS OF THE RIGID PENDULUM

approximately valid for the pendulum’s motion between ϕ = ϕm and ϕ = ϕc :

α̈ − ω02 α = 0. (7.17)
The general solution to this linear equation can be represented as a superposi-
tion of two exponential functions of time t:
α(t) = C1 eω0 t + C2 e−ω0 t . (7.18)
The initial conditions for the motion from ϕ = ϕm to ϕ = ϕc are α(0) = αm
and α̇(0) = 0. Applying these conditions, we find the constants C1 and C2 in
Eq. (7.18):
1
α(t) = αm (eω0 t + e−ω0 t ) = αm cosh ω0 t. (7.19)
2
To find the duration t1 of motion from ϕ = ϕm to ϕ = ϕc (from α = αm to
α = αc ), we substitute in Eq. (7.19) α(t1 ) = αc :
1 1
αc = αm (eω0 t1 + e−ω0 t1 ) ≈ αm eω0 t1 . (7.20)
2 2
We have omitted the second term in the right-hand side of Eq. (7.20). This is
admissible if the arbitrary angle αc (which divides the phase trajectory into two
parts) is chosen to be large compared to αm . From Eq. (7.20) we get for t1 :
2αc
ω0 t1 = ln . (7.21)
αm
The desired period of oscillations T is four times greater than the duration t1 + t2
of motion from ϕ = ϕm to the lower equilibrium position ϕ = 0. Adding t1 from
Eq. (7.21) and t2 from (7.16), we finally obtain the following expression for the
period of oscillations with large amplitude ϕm approaching 180◦ :
4 2αc 4 2 8
T = 4(t1 + t2 ) = (ln + ln ) = T0 ln . (7.22)
ω0 αm αc π αm
(Here αm = π − ϕm .) We note that both t1 and t2 depend on the value αc of the
angle that we have chosen to divide the trajectory into one part that corresponds
to the motion in the vicinity of the inverted position, and the other that almost
merges with the separatrix. Nevertheless, this dependence on αc disappears when
we add t1 and t2 : The final expression (7.22) for the period is independent of the
arbitrarily chosen value of αc (provided αm ≪ αc ≪ 1).
The approximation given by expression (7.22) is more accurate the closer the
amplitude ϕm to 180◦ . Table 7.2 illustrates the precision of this simple expres-
sion for oscillations of extremely large amplitudes. The values of T in the middle
column are calculated on the basis of exact formula (7.15); the right column cor-
responds to the approximate expression (7.22).
We note that according to this table one cycle of the pendulum oscillation at
large amplitudes covers several periods of small oscillations. As an assignment for
students’ activity, we suggest verifying the values cited in the table by direct mea-
surements of the period in a simulation experiment using the relevant program.

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

7.2. OSCILLATIONS WITH EXTREMELY LARGE AMPLITUDES 173

Table 7.2. Period of large-amplitude oscillations

Amplitude T /T0 T /T0

ϕm (αm ) (exact value) (approximate)

175.000◦ (5.000◦ ) 2.877664 2.876395

177.000◦ (3.000◦ ) 3.202109 3.201597
179.000◦ (1.000◦ ) 3.901065 3.900995
179.900◦ (0.100◦ ) 5.366867 5.366866
179.990◦ (0.010◦ ) 6.832737 6.832737
179.999◦ (0.001◦ ) 8.298608 8.298608

7.2.2 Another Derivation of the Expression for the Period

of Large Oscillations
In the above derivation of expression (7.22) we have arbitrarily chosen some small
angle αc for dividing the motion into stages described by different analytical time
dependencies. Another way is to choose for this conventional boundary of the two
stages, instead of the angular position αc , some arbitrary small angular velocity
ωc ≪ ω0 , which the pendulum gains while moving from the turning point αm at
which its angular velocity is zero. To find the duration t3 of this stage occurring
in the vicinity of the inverted position, we can make use of the above-obtained
solution (7.19) to the linearized Eq. (7.17), according to which
1
α̇(t) = αm ω0 (eω0 t − e−ω0 t ). (7.23)
2
Substituting α̇(t3 ) = ωc in Eq. (7.23) and taking into account that e−ω0 t3 ≪
eω0 t3 , we find
2ωc
ω0 t3 = ln . (7.24)
ω0 αm
The further motion towards the equilibrium position is almost indistinguishable
from the limiting motion. Hence the time dependence of the angular velocity
α̇(t) = −ϕ̇(t) for this stage can be assumed to be the same as for the limiting
motion; see Eq. (7.13). Therefore, for calculating the duration t4 of this stage we
can substitute ϕ̇(t4 ) = ωc in (7.13) and take into account that e−ω0 t4 ≪ eω0 t4 .
This yields
4ω0
ω0 t4 = ln . (7.25)
ωc
Adding t3 from Eq. (7.24) and t4 from (7.25), we finally obtain the same simple
expression (7.22) for the period of oscillations with a very large amplitude ϕm :
 
4 2ωc 4ω0 2 8
T = 4(t3 + t4 ) = ln + ln = T0 ln . (7.26)
ω0 ω0 αm ωc π αm
Again, the arbitrarily chosen angular velocity ωc (ωc ≪ ω0 ), which we have used
to divide the motion on different stages, vanishes from the final expression (7.26).

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

174 CHAPTER 7. OSCILLATIONS OF THE RIGID PENDULUM

7.3 Period of Revolutions and Large Oscillations

7.3.1 The Period of Fast Revolutions
If the total energy of the pendulum is considerably greater than the maximal value
of its potential energy, we can assume all the energy of the pendulum to be the
kinetic energy of its rotation. In other words, we can neglect the influence of the
gravitational field on the rotation and consider this rotation to be uniform and
occurring approximately with the angular velocity Ω received by the pendulum at
the initial excitation. Thus for E ≈ JΩ2 /2 ≫ 2mga the asymptotic dependence
of the period T = 2π/Ω on the initial angular velocity is the inverse proportion:
T ∼ 1/Ω. To find the dependence T (Ω) more precisely, we need to take into
account the variations in the angular velocity caused by gravitation. The angular
velocity of the pendulum oscillates between the maximal value Ω in the lower
position and the minimal value Ωmin in the upper position. The latter can be found
from the conservation of energy:
ω2
q  
Ωmin = Ω2 − 4ω02 ≈ Ω 1 − 2 02 . (7.27)
Ω
For rapid rotation we can assume these oscillations of the angular velocity to
be almost sinusoidal. Then the average angular velocity of rotation is approxi-
mately the half-sum of its maximal and minimal values. Thus,
ω02 ω02
   
2π ω0
Ωav ≈ Ω 1 − 2 , T (Ω) = ≈ T0 1+ 2 . (7.28)
Ω Ωav Ω Ω

7.3.2 Relationship between the Periods of Revolutions

and Large Oscillations
The most interesting peculiarities are revealed if we investigate the dependence
of the period on energy in the vicinity of Em = 2mga. Measuring the period of
oscillations for the amplitudes 179.900◦, 179.990◦, and 179.999◦, we see that du-
ration of the impulses on the graph of the angular velocity very nearly remains the
same, but the intervals between them become longer as the amplitude approaches
180◦ : Experimental values of the period T of such extraordinary oscillations are,
respectively, 5.5 T0 , 6.8 T0 , and 8.3 T0 .
It is also interesting to compare the motions for two values of the total energy
E, which differ slightly from Emax on either side by the same amount, e.g., for
E/Emax = 0.9999 and E/Emax = 1.0001. In the phase plane, these motions oc-
cur very near to the separatrix, the first one inside (oscillations with the amplitude
178.9◦) and the second outside of the separatrix (very slow revolutions). Mea-
suring the periods of these motions, we obtain the values 3.814 T0 and 1.907 T0,
respectively: The period of oscillations is twice the period of rotation almost ex-
actly. The graphs of ϕ(t) and ϕ̇(t) for oscillations and revolutions of the pendulum
whose energy equals Emax ∓ ∆E are shown respectively in the upper and lower
parts of Figure 7.9.

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

7.3. REVOLUTIONS AND LARGE OSCILLATIONS 175

o
180
ϕ (t )

0
.
ϕ (t )
o
-180
o 0 2 4 6 8 10 12
180

ϕ (t )
0
.
ϕ (t )
o
-180
0 2 4 6 8 10 12 T0

Figure 7.9: The graphs of ϕ(t) and ϕ̇(t) for the pendulum excited at ϕ = 0 by
imparting the initial angular velocity of ϕ̇ = ω0 (2 ∓ 1 · 10−6 ).

Next we suggest a theoretical approach, which can be used to calculate the

period of oscillations and revolutions with E ≈ Emax . From the simulation ex-
periments we can conclude that during the semicircular path, from the equilibrium
position up to the extreme deflection or to the inverted position, both of the mo-
tions shown in Figure 7.9 almost coincide with the limiting motion (Figure 7.5,
p. 167). These motions differ appreciably from the limiting motion only in the im-
mediate vicinity of the extreme point or near the inverted position: In the first case
(E < Emax ) the pendulum stops at this extreme point and then begins to move
backwards, while in the limiting motion the pendulum continues moving for an
unlimited time towards the inverted position; in the second case (E > Em ) the
pendulum reaches the inverted position during a finite time.
For the oscillatory motion under consideration, the representative point in the
phase plane generates a closed path during one cycle, passing along both branches
of the separatrix. The pendulum goes twice around almost the whole circle, cov-
ering it in both directions. On the other hand, executing rotation, the pendulum
makes one circle during a cycle of revolutions, and the representative point passes
along one branch of the separatrix (upper or lower, depending on the direction of
rotation). To explain why the period of these oscillations is twice the period of
corresponding revolutions, we must show that the motion of the pendulum with
energy E = Emax − ∆E from ϕ = 0 up to the extreme point requires the same
time as the motion with the energy E = Emax + ∆E from ϕ = 0 up to the
inverted vertical position.
The major part of each of both motions under consideration occurs very nearly
along the same path in the phase plane, namely, along the separatrix from the
initial point ϕ = 0, ϕ̇ ≈ 2ω0 up to some angle ϕ0 whose value is close to π.

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

176 CHAPTER 7. OSCILLATIONS OF THE RIGID PENDULUM

We choose this value ϕ0 arbitrarily. To calculate the time interval required for this
part of the motion, we can assume that the motion (in both cases) occurs exactly
along the separatrix, and take advantage of the corresponding analytical solution,
expressed by Eq. (7.12). Assuming ϕ(t) in Eq. (7.12) to be equal to ϕ0 , we can
find the time t0 during which the pendulum moves from the equilibrium position
ϕ = 0 up to the angle ϕ0 (for both cases):
π − ϕ0 α0
ω0 t0 = − ln tan = − ln tan , (7.29)
4 4
where we introduced the notation α0 = π − ϕ0 for the angle that the pendulum
at ϕ = ϕ0 forms with the upward vertical line. When ϕ0 is close to π, the angle
α0 is small, so that in Eq. (7.29) we can assume tan(α0 /4) ≈ α0 /4. Therefore
ω0 t0 ≈ ln(4/α0 ).
Next we shall consider in detail the subsequent part of the motion that occurs
from this arbitrarily chosen angle ϕ = ϕ0 towards the inverted position, and prove
that the time t1 required for the pendulum with the energy Emax + ∆E (rotational
motion) to reach the inverted position ϕ = π equals the time t2 during which the
pendulum with the energy Emax − ∆E (oscillatory motion) moves from ϕ0 up
to its extreme deflection ϕm , where the angular velocity becomes zero, and the
pendulum begins to move backwards. We emphasize that these time intervals t1
and t2 are equal to one another only if ∆E is the same in both cases.
When considering the motion of the pendulum in the vicinity of the inverted
position, we find it convenient to define its position (instead of the angle ϕ) by the
angle α of deflection from this position of unstable equilibrium. This angle equals
π − ϕ, and the angular velocity α̇ equals −ϕ̇. The potential energy (measured
relative to the lower equilibrium position) depends on α in the following way:

Epot (α) = mga(1 + cos α) ≈ Emax (1 − α2 /4). (7.30)

The latter expression is valid only for small values of α, when the pendulum moves
near the inverted position. Phase trajectories of motion with energies E = Emax ±
∆E near the saddle point (the origin in the new variables α, α̇) can be found from
the conservation of energy with the help of the approximate expression (7.30) for
the potential energy:

1 2 1 α̇2
J α̇ + Emax α2 = ±∆E, or 2 − α2 = ±4ε. (7.31)
2 4 ω0

Here we use the notation ε = ∆E/Emax for the small (ε ≪ 1) dimensionless

quantity characterizing the fractional deviation of energy E from its value Emax
for the separatrix. It follows from Eq. (7.31) that phase trajectories near the saddle
point are hyperbolas whose asymptotas are the two branches of the separatrix that
meet at the saddle point. Part of the phase portrait near the saddle point is shown
in Figure 7.10. The curve 1 for the energy E = Emax + ∆E corresponds to
the rotation of the pendulum. It intersects the ordinate axis when the pendulum
passes through the inverted position. The curve 2 for the energy E = Emax −

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

7.3. REVOLUTIONS AND LARGE OSCILLATIONS 177

.
a0 a

1
2

a0 am -a

Figure 7.10: The phase curves near the saddle point.

∆E describes the oscillatory motion. It intersects the abscissa axis at the distance
αm = π − ϕm from the origin (from the zero point of the axis). This point of
intersection shows the extreme deflection in the oscillations.
The new variable α(t) = π − ϕ(t) near the point α = 0 (for α ≪ 1) satisfies
the linearized differential equation, Eq. (7.17), p. 172, which is valid for the pen-
dulum motion in the vicinity of the inverted position. The general solution to this
linear equation is given by a superposition of two exponential functions of time t,
Eq. (7.18):
α(t) = C1 eω0 t + C2 e−ω0 t . (7.32)
Next we separately consider the two cases of motion of the pendulum with the
energies E = Emax ± ∆E.
1. Rotational motion (E = Emax + ∆E) along the curve 1 from α0 up to
the intersection with the ordinate axis. Let t = 0 be the moment of crossing the
inverted vertical position:
√ α(0) = 0. Hence
√ in Eq. (7.32), C2 = −C1 . Then from
Eq. (7.31) α̇(0) = 2 εω0 , and C1 = ε. To determine duration t1 of the motion,
we assume in Eq. (7.32) α(t1 ) = α0 :
√ √
α0 = ε(eω0 t1 − e−ω0 t1 ) ≈ εeω0 t1 . (7.33)

(Here we can√ choose an arbitrary value of α0 , although a small one, to be large

compared to √ ε, so that the condition e−ω0 t1 ≪ eω0 t1 is fulfilled). Therefore
ω0 t1 = ln(α0 / ε).
2. Oscillatory motion (E = Emax − ∆E) along the curve 2 from α0 up to
the extreme point αm . Let t = 0 be the moment of maximal deflection, when the
phase curve intersects the abscissa axis:√α̇(0) = 0. Hence
√ in Eq. (7.32) C2 = C1 .
Then from Eq. (7.31) α(0) = αm = 2 ε, and C1 = ε. To determine duration
t2 of this motion, we assume in Eq. (7.32) α(t2 ) = α0 . Hence
√ √
α0 = ε(eω0 t2 + e−ω0 t2 ) ≈ εeω0 t2 , (7.34)

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

178 CHAPTER 7. OSCILLATIONS OF THE RIGID PENDULUM

√
and we find ω0 t2 = ln(α0 / ε).
We see that t2 = t1 if ε = ∆E/Emax is the same in both cases. Therefore the
period of oscillations is twice the period of rotation for the values of energy that
differ from the critical value Emax on both sides by the same small amount ∆E.
Indeed, we can assume with great precision that the motion from ϕ = 0 up to
ϕ0 = π − α0 lasts the same time t0 given by Eq. (7.29), since these parts of both
phase trajectories very nearly coincide with the separatrix. In the case of rotation,
the remaining motion from ϕ0 up to the inverted position also lasts the same time
as, in the case of oscillations, does the motion from ϕ0 up to the utmost deflection
ϕm , since t1 = t2 .
The period of rotation Trot is twice the duration t0 + t1 of motion from the
equilibrium position ϕ = 0 up to the ϕ = π. Using the above value for t1 and
Eq. (7.29) for t0 , we find:
2 4 1 4
Trot = 2(t0 + t1 ) = ln √ = T0 ln √ . (7.35)
ω0 ε π ε
√
We note that an arbitrarily chosen angle α0 (however, ε ≪ α0 ≪ 1), which
delimits the two stages of motion (along the separatrix, and near the saddle point
in the phase plane), falls out of the final formula for the period (when we add t0
and t1 ). For ε = 0.0001 (for E = 1.0001Emax) the above formula gives the value
Trot = 1.907 T0, which coincides with the cited above experimental result.
The period of oscillations T is four times greater than the duration t0 + t2 of
motion from ϕ = 0 up to the extreme point ϕm :
4 4 2 8
T = 4(t0 + t2 ) = ln √ = T0 ln . (7.36)
ω0 ε π αm
For αm ≪ 1 (ϕm ≈ π) this formula agrees well with the experimental results:
It yields T = 5.37 T0 for ϕm = 179.900◦, T = 6.83 T0 for ϕm = 179.990◦, and
T = 8.30 T0 for ϕm = 179.999◦. From the obtained expressions we see how both
the period of oscillations T and the period of rotation Trot tend to infinity as the
total energy approaches Emax = 2mga.

7.3.3 Mean Values of the Potential and Kinetic Energies

We can estimate the ratio of the values of the potential and kinetic energies, av-
eraged over a period, if we take into account that most of the time the angular
velocity of the pendulum is nearly zero, and for a brief time of motion the time
dependence of ϕ(t) is very nearly the same as it is for the limiting motion along
the separatrix. Therefore we can assume that during an impulse the kinetic energy
depends on time in the same way it does in the limiting motion. This assumption
allows us to extend the limits of integration to ±∞. Since two sharp impulses
of the angular velocity (and of the kinetic energy) occur during the period T of
oscillations, we can write:
J ∞ 2 J π
Z Z
hEkin i = ϕ̇ (t)dt = ϕ̇(ϕ)dϕ. (7.37)
T −∞ T −π

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

7.4. THE INFLUENCE OF FRICTION 179

.
ϕ

−2π −π 0 π 2π ϕ

Figure 7.11: The phase portrait for revolutions and oscillations of the pendulum
in the presence of friction (Q = 18.12).

The integration with respect to time is replaced here with an integration over the
angle. The mean kinetic energy hEkin i is proportional to the area S of the phase
plane bounded by the separatrix: 2hEkin i = JS/T . We can substitute for ϕ̇(ϕ) its
expression from the equation of the separatrix, Eq. (7.10):
Z π
J ϕ 4 T0
hEkin i = 2ω0 cos dϕ = Jω02 . (7.38)
T −π 2 π T

Taking into account that the total energy E for this motion approximately equals
2mga = 2Iω02 , and Epot = E − Ekin , we find:

hEpot i 2Jω02 π T
= −1= − 1. (7.39)
hEkin i hEkin i 2 T0

For ϕm = 179.99◦ the period T equals 6.83 T0, and so the ratio of mean values of
potential and kinetic energies is 9.7 (compare with the case of small oscillations
for which these mean values are equal).

7.4 The Influence of Friction

7.4.1 The Phase Portrait of the Pendulum in the Presence
of Friction
When we take into account the small amount of friction inevitable in any real
system, the phase portrait of the pendulum changes qualitatively (see Figure 7.11).
The closed phase trajectories corresponding to oscillations of a conservative
system are transformed by friction into shrinking spirals that wind around a focus
located at the origin of the phase plane. This focus represents a state of rest in
the equilibrium position, and is an attractor of the phase trajectories: All phase
trajectories of the damped pendulum spiral in toward the focus, forming an infinite

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

180 CHAPTER 7. OSCILLATIONS OF THE RIGID PENDULUM

2.30 180 o .
ϕ (t ) ϕ (t )

0 0

-2.30 -180 o
-180o 0 180o 0 1 2 3 4 5 6 7 8T0

Figure 7.12: Phase diagram (left) and time dependent graphs of ϕ(t) and ϕ̇(t)
(right) for revolution and subsequent oscillation of the pendulum with friction
(Q = 18.1) excited from the equilibrium position with the initial angular velocity
Ω = 2.3 ω0.

number of loops. The phase curve that formerly passed along the upper branch of
the separatrix does not now reach the saddle point (π, 0). Instead it also begins
to wind around the origin, gradually approaching it. Similarly, the lower branch
crosses the abscissa axis ϕ̇ = 0 to the right of the saddle point (−π, 0), and also
spirals in towards the origin.
The inner part of the spiral, corresponding to small amplitudes, looks much
like the phase trajectory of a linear oscillator experiencing viscous friction. The
size of its gradually shrinking loops diminishes in a geometric progression.

7.4.2 Revolutions Followed by Oscillations

Typical graphs of the pendulum rotation followed by damped oscillations are
shown in Figure 7.12 together with the corresponding phase diagram. The phase
trajectory representing the counterclockwise rotation of the pendulum sinks lower
and lower toward the separatrix with each revolution. When the pendulum has lost
sufficient energy through friction so that it swings rather than rotates, its phase tra-
jectory continues after crossing the separatrix as a gradually shrinking spiral, also
winding around the origin.
When friction is weak, we can make some theoretical predictions for the mo-
tions whose phase trajectories pass close to the separatrix. For example, we can
evaluate the minimal value of the initial velocity that the pendulum must be given
in the lower (or some other) initial position in order to reach the inverted position,
assuming that the motion occurs along the separatrix, and consequently that the
dependence of the angular velocity on the angle of deflection is approximately
given by the equation of the separatrix, Eq. (7.10).
The frictional torque is proportional to the angular velocity: Nfr = −2γJ ϕ̇.
Substituting the angular velocity from Eq. (7.10), we find
ϕ 2mga ϕ
Nfr = ∓4γJω0 cos =∓ cos . (7.40)
2 Q 2

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

7.5. REVIEW OF THE PRINCIPAL FORMULAS 181

2.33 180o
.
ϕ (t )

0 0
ϕ (t )

o
-2.33 o -
o 180
-180 0 180 0 1 2 3 4 5 6 7 8T0

Figure 7.13: Revolution and subsequent oscillation of the pendulum with friction
(Q = 20) excited from the equilibrium position with the initial angular velocity
Ω = 2.3347 ω0.

Hence the work Wfr of the frictional force during the motion from an initial point
ϕ0 to the final inverted position ϕ = ±π is:
±π
mga  ϕ0 
Z
Wfr = Nfr dϕ = −4 1 ∓ sin . (7.41)
ϕ0 Q 2

The necessary value of the initial angular velocity Ω can be found with the
help of the conservation of energy, in which the work Wfr of the frictional force
is taken into account:
 
4  ϕ0 
Ω2 = 2ω02 1 + cos ϕ0 + 1 ∓ sin . (7.42)
Q 2

For ϕ0 6= 0 the sign in Eq. (7.42) depends on the direction of the initial angular
velocity. The exact value of Ω is slightly greater since the motion towards the in-
verted position occurs in the phase plane close to the separatrix but always outside
it, with the angular velocity of a slightly greater magnitude. Consequently, the
work of the frictional force during this motion is a little larger than the calculated
value. For example, with ϕ0 = 0 and the quality Q = 20, the above estimate
yields Ω = ±2.098 ω0, but a more precise value of Ω determined experimentally
by trial and error is ±2.101 ω0.
Figure 7.13 shows the phase trajectory and the graphs of ϕ(t) and ϕ̇(t) for
a similar case in which the initial angular velocity is chosen exactly to let the
pendulum reach the inverted position after a revolution.

7.5 Review of the Principal Formulas

The differential equation of motion for a planar rigid pendulum:

ϕ̈ + 2γ ϕ̇ + ω02 sin ϕ = 0, (7.43)

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

182 CHAPTER 7. OSCILLATIONS OF THE RIGID PENDULUM

where ω0 is the frequency of small free oscillations:

ω02 = mga/J = g/l; l = J/ma. (7.44)

Here m is the mass of the pendulum, a is the distance between the horizontal axis
of rotation (the point of suspension) and the center of mass, J is the moment of
inertia about the same axis, l is the reduced length of the physical pendulum, and
g is the acceleration of gravity.
The equation of a phase trajectory in the absence of friction:

ϕ̇2 E
2 + 2(1 − cos ϕ) = , (7.45)
ω0 E0

where E is the total energy, and

1 2 1 1
E0 = Jω = mga = (Epot )max . (7.46)
2 0 2 4
Here (Epot )max = 2mga is the maximal possible value of the potential energy
of the pendulum, which is its potential energy when it is in the inverted vertical
position.
The equation of the separatrix in the phase plane:

ϕ̇ = ±2ω0 cos(ϕ/2). (7.47)

The angular deflection and angular velocity for the motion of the pendulum that
generates the separatrix in the phase plane are:
2ω0 4ω0
ϕ(t) = π − 4 arctan(e−ω0 t ), ϕ̇(t) = ± = ± ω0 t .
cosh(ω0 t) (e + e−ω0 t )
(7.48)

7.6 Questions, Problems, Suggestions

7.6.1 Small Oscillations of the Pendulum
At small angles of deflection, when sin ϕ ≈ ϕ, the restoring torque of the force
of gravity is approximately proportional to the angle of deflection from the po-
sition of stable equilibrium, and the pendulum behaves like a linear oscillator. In
the absence of friction it executes simple harmonic motion. In the presence of
weak friction, its motion can be considered as a nearly harmonic oscillation with
a slowly decreasing amplitude.
7.6.1.1 The Amplitude, Phase Trajectory, and Energy of Small Oscilla-
tions. Select the case of no friction and use initial conditions which produce os-
cillations of small amplitude. For instance, let the initial deflection be 30◦ and the
initial velocity be zero. In this case the amplitude will be 30◦ .

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

7.6. QUESTIONS, PROBLEMS, SUGGESTIONS 183

(a) What is the maximal value of the angular velocity in these oscillations?
Verify your answer with a computer-simulated experiment.
(b) What initial angular velocity should you give the pendulum initially in the
equilibrium position (ϕ(0) = 0) in order to excite oscillations of the same ampli-
tude (of 30◦ )? Verify your answer with an experiment. Remember that the initial
angular velocity you enter must be expressed in units of the frequency ω0 of small
oscillations. What is the difference between these oscillations and oscillations ex-
cited by an initial deflection?
(c) Convince yourself that at small amplitudes the graphs of the angle of de-
flection versus time and of the angular velocity versus time have shapes which
are close to that of a sine curve. Also convince yourself, that oscillations of the
velocity lead the oscillations of the angular displacement in phase by a quarter
period.
Compare the graphs of time dependence of the deflection angle and of the
angular velocity with the motion of the representative point along the phase tra-
jectory. What is the form of the phase trajectory for small oscillations? With what
scale along the ordinate axis of the phase plane is the phase trajectory approxi-
mately a circle?
(d) What can you say about the time dependence of the kinetic and potential
energies of the pendulum at small amplitudes? Prove that the time average values
of kinetic and potential energy are approximately equal. If the amplitude equals
30 degrees, what is the ratio of total energy E to the maximal possible value of
potential energy E/Emax ?
7.6.1.2∗ Period of Small Oscillations.
For graphs of the time dependencies of the angle of deflection and of the an-
gular velocity, the scale shown on the time axis is in p the appropriate units for a
given pendulum, namely in units of T0 = 2π/ω0 = 2π l/g), which is the period
of small oscillations of the pendulum. That is, the duration between hatch marks
on the time axis is T0 .
(a) Note that at small but finite amplitudes (say about 30◦ ), the period of os-
cillations is a bit longer than T0 . You can make this observation either from the
curves plotted on the screen or from readings of the timer. In the latter case, you
can stop the timer by clicking the “Pause” button or by pressing the Spacebar at
the moment when the pendulum completes a whole number of cycles. As a con-
venience in taking further readings, you may set the timer to zero during a pause
in the simulation by clicking the “Reset Timer” button. Try to measure the period
(in units of T0 ) for several moderate values of the amplitude.
(b) In performing precise measurements of the period in the simulation ex-
periments, which instants are better for starting and stopping the timer: When the
pendulum passes through the equilibrium position or when it reaches the points of
its greatest deflection? Give a convincing explanation of your answer.
(c) Compare the measurement of the period T for a given amplitude ϕ0 ob-
tained from the simulation experiment with the value given by the theoretical ap-
proximation:
T = T0 (1 + ϕ20 /16), (7.49)

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

184 CHAPTER 7. OSCILLATIONS OF THE RIGID PENDULUM

in which the amplitude ϕ0 is expressed in radians. Determine the maximal value

of ϕ0 for which Eq. (7.49) gives the value of T to within one percent. Find the
error the formula yields for an amplitude of 45 degrees.
7.6.1.3 Damping of Small Oscillations.
(a) Prove theoretically that weak viscous friction causes exponential damping
of small free oscillations. At what value of the quality factor Q does the amplitude
halve during four complete cycles? Input the calculated value of Q and verify it in
a simulation experiment on the computer.
(b) Convince yourself that for large viscous friction for which the quality fac-
tor Q is less than the critical value of 0.5, a disturbed pendulum returns to the
equilibrium position without swinging. What is the principal qualitative differ-
ence of the phase trajectories for the cases of weak and strong damping?

7.6.2 Oscillations with Large Amplitudes

7.6.2.1 Comparison of the Pendulum with a Linear Oscillator. For large
angular displacements from the equilibrium position, the nonlinearity of the de-
pendence on the angle ϕ of the restoring gravitational torque is more apparent.
Because sin ϕ < ϕ, the increase in the restoring torque with increasing angular
deflection is not as large for a pendulum as it is for a linear oscillator. Therefore,
a pendulum is referred to as a nonlinear oscillatory system with a “soft” restoring
force.
(a) How do the differences between a pendulum and a linear oscillator reveal
themselves in graphs of the time dependence of the angular deflection and the an-
gular velocity? How do the differences reveal themselves in the phase trajectory?
Give a qualitative physical explanation for the differences.
(b) What are the differences between the pendulum and a linear oscillator
with respect to energy transformations? Compare the phase trajectory with the
graph of potential energy versus deflection angle. The placing of the graphs on the
computer screen (if you choose the item “View,” “Energy Transformations” in the
menu) is especially convenient for such comparison. Pay special attention to the
position of the extreme points on the phase trajectory and on the potential well of
the pendulum. For given initial conditions ϕ(0) = ϕ0 , ϕ̇(0) = Ω, what are the
values of the potential energy and kinetic energy of the pendulum at the extreme
points and at the equilibrium position?
7.6.2.2∗ Oscillations with Large Amplitudes.
(a) Study large oscillations of the pendulum experimentally in the absence of
friction. Note the exact periodicity of these clearly non-sinusoidal oscillations of
the dynamical variables in the conservative system.
When the amplitude exceeds 90◦ , the graph of angular velocity versus time is
nearly a saw-tooth with equilateral triangular teeth. Explain this shape.
The shape of a tooth in the corresponding graph of the angular deflection in
this case is close to a parabola, in contrast to the sinusoidally shaped tooth asso-
ciated with oscillations with small amplitudes. Explain this parabolic shape. Note

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

7.6. QUESTIONS, PROBLEMS, SUGGESTIONS 185

the increase in the period with increasing amplitude. (Hatch marks on the time
axis are separated by T0 , the period of small oscillations.)
(b) Note how the closed phase trajectories of the oscillating pendulum are
stretched horizontally as the energy of the pendulum increases. Explain why these
phase trajectories are different from the elliptical phase trajectories of a linear
oscillator. To do so, use the shapes of the parabolic potential well of a linear os-
cillator and the sinusoidal potential well of the pendulum. Assume the curvature
near the bottom to be the same for both potential wells: The period T0 of small
oscillations of the pendulum should be equal to the period of the linear oscillator.
Remember that the latter period is independent of the energy.
Explain the increase of the period of the pendulum with increasing amplitude,
comparing its potential well with that of a linear oscillator.
(c) At large amplitudes the pendulum passes rapidly through the vicinity of
the equilibrium position (through the sinusoidal bottom of the potential well) and
slowly climbs up the sinusoidal crest of the well, along its nearly horizontal up-
per slopes; then it slowly descends from them. So on the average the pendulum
remains at large deflections longer than does a linear oscillator, whose parabolic
potential well has steadily increasing slopes. Use the shapes of these potential
wells to explain why, during a cycle, the time average values of the potential and
kinetic energies of a pendulum are not equal to one another while those of the
linear oscillator are.
(d)∗ Carefully study the interesting case of oscillations with an amplitude near
180◦ . Set the initial deflection to be 179.999◦, and the initial velocity to be zero.
After remaining near one side of the inverted position for a long time, the pen-
dulum rapidly passes through the bottom of its path, and then remains for a long
time again near the other side of the inverted position.
Compare the time during which the pendulum covers almost all its circular
path (except a small vicinity of the extreme positions) with the period of small free
oscillations of the pendulum. In other words, estimate the duration of a solitary
impulse on the graph of angular velocity versus time. Or, equivalently, estimate
the width of the nearly vertical portion of the nearly rectangular saw-tooth graph
of the angular deflection versus time.
(e)∗ Try to discover what factor determines the width of this nearly rectangular
tooth of the graph ϕ(t), or, equivalently, what factor determines the time interval
between successive impulses in the graph of angular velocity versus time. That
is, try to discover the physical cause that determines the complete period of these
extraordinary oscillations of the pendulum. (Hint: Set the initial deflection of the
pendulum at the successive values 179.999◦, 179.990◦, and 179.900◦, each with
an initial velocity of zero.
(f)∗∗ Try to evaluate theoretically the time interval needed for the pendulum to
reach the extreme deflection of 179.99 degrees at excitation from rest in the lower
stable equilibrium position. Use your results to estimate the period of oscillations
with the amplitude 179.99 degrees. Compare your estimation of the period with
the value of T obtained in the simulation experiment.
(g) Note the character of energy transformations in the motion considered

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

186 CHAPTER 7. OSCILLATIONS OF THE RIGID PENDULUM

above. Total energy E in this motion nearly equals the height 2mga of the poten-
tial barrier. It is the value of the potential energy of the pendulum in the inverted
position (ϕ = ±π). Since the pendulum spends most of its period near the in-
verted position (because the pendulum moves and accelerates very slowly while
in the vicinity of the inverted position), the time averaged value of its potential en-
ergy, taken over a complete oscillation, is much greater than the mean value of its
kinetic energy. In this case potential energy is converted into kinetic energy only
for the short time during which the pendulum makes a rapid turn passing through
the lower equilibrium position of minimal potential energy. Try to evaluate (to an
order of magnitude) the ratio of the values of the potential energy to the kinetic
energy, averaged over a period, during oscillations with an amplitude of 179.99◦.
7.6.2.3∗ Motion along the Separatrix.
(a) When you set the initial deflection to be almost 180 degrees and the initial
velocity to be zero, the phase trajectory of the resulting motion nearly coincides
with the separatrix ϕ̇ = ±2ω0 cos(ϕ/2). The point representing the mechanical
state of the pendulum in the phase plane passes rapidly along the lower branch of
the separatrix, remains for a long time at the left saddle point (−π, 0), and then
returns along the upper branch of the separatrix. What initial conditions should
you choose in order to make the representative point move first along the upper
branch of the separatrix and then along the lower one?
(b) What value of the initial angular velocity Ω (in units of ω0 ) must be ini-
tially given to the pendulum in its lower equilibrium position in order to make
the representative point in the phase plane move along the separatrix? What value
of the initial angular velocity should you input if the pendulum is to be initially
deflected from the equilibrium position by an angle of 60◦ ? 90◦ ? −90◦ ? 120◦ ?
Verify your answers with simulation experiments.
(c) For the limiting motion along the separatrix, calculate the time interval τ
during which kinetic energy of the pendulum is greater than its potential energy.
In other words, for the pendulum making its circular path from one side of the
inverted position to the other, find the lapse of time between the two instants at
which the pendulum passes through the horizontal positions on either side of the
lower equilibrium position. Express this time interval in units of the period T0 of
small oscillations. Verify your calculated value by the experiment on the computer.
7.6.2.4 Large Oscillations with Friction.
(a) Examine the influence of viscous friction on oscillations of large ampli-
tude. Begin with rather weak friction (Q ≈ 20). Note the gradual changes in the
pattern of the graphs as friction slowly decreases the mechanical energy and the
amplitude of the pendulum. In particular, note how the initial triangular saw-tooth
curve of angular velocity, with its sharp nearly rectilinear teeth, as well as the ini-
tial curve of angular deflection with smooth parabolic crests, both evolve into the
sinusoidal curves characteristic of the simple harmonic oscillator.
(b) Under the influence of viscous friction, the topologies of the phase trajec-
tories of a pendulum change. Instead of closed curves corresponding to exactly
periodic oscillations of a conservative pendulum, you see twisting spirals making

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

7.6. QUESTIONS, PROBLEMS, SUGGESTIONS 187

an infinite number of gradually shrinking loops around the focus at the origin of
the phase plane. Note how the form of the loops changes when they recede from
the separatrix. Give a qualitative explanation for the observed changes. (For a lin-
ear oscillator experiencing viscous friction, the shrinking loops of the phase curve
remain similar as the curve approaches the origin.)
(c) Using the program item “Energy Transformations,” note how the rate at
which energy is dissipated depends upon the position of the representative point
in the potential well. At which part of a cycle does the rate of energy dissipation
reach a maximum? Explain your answer.
(d)∗∗ Using the law of energy conservation, calculate the minimal value of the
initial velocity that the pendulum must be given in the lower equilibrium position
in order to reach the inverted position, for the case in which there is no friction and
for the case in which the quality factor Q = 20. What must be the initial velocity
in order to reach the inverted position if the pendulum is initially deflected by the
angle 60 degrees? By 90 degrees?

7.6.3 The Rotating Pendulum

A pendulum makes a full revolution if its total energy exceeds the value 2mga, the
maximal value possible for its potential energy. The influence of the gravitational
force makes this rotation in the vertical plane nonuniform: The angular velocity
is a maximum (in the absence of friction) each time the pendulum passes through
the lower, stable equilibrium position, and a minimum each time the pendulum
passes through the upper, unstable equilibrium position.
7.6.3.1 The Angular Velocity at Revolutions.
(a) Select the case of the absence of friction. Calculate the minimal initial
angular velocity needed to obtain a full revolution of the pendulum when it is ini-
tially at the lower equilibrium position. Note the character of the graph of angular
velocity versus time: As the pendulum revolves, its angular velocity changes peri-
odically (that is, the angular velocity oscillates in time), but the sign of the angular
velocity does not change (that is, the curve does not intersect the time axis).
(b)∗ How does the period of these oscillations change if the initial angular ve-
locity is increased? Calculate the minimal value of the oscillating angular velocity
for a given value of the initial angular velocity. Find the asymptotic dependence
of the period of rotation on the initial angular velocity T (Ω), valid for the values
of total energy E, which are much greater than the potential energy of the inverted
pendulum (E ≫ 2mga).
(c) What initial conditions must be entered in order to obtain a phase trajectory
located above the separatrix in the phase plane? . . . located below the separatrix?
. . . coinciding with the upper or lower branch of the separatrix?
7.6.3.2∗ The Period of Revolutions and Oscillations.
(a) It is especially interesting to compare the period of rotation with the period
of oscillation of the conservative pendulum whose total energy E is close to the
maximal possible value of the potential energy Emax = 2mga. In this case, the

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

188 CHAPTER 7. OSCILLATIONS OF THE RIGID PENDULUM

phase trajectories lie in the vicinity of the separatrix. Using the simulation experi-
ment, perform the measurement of the period for two values of the total energy E,
which are slightly different from Emax by equal amounts on either side of Emax .
For example, first let E/Emax = 0.9999 and then let E/Emax = 1.0001.
It is convenient to use the item “Phase Portrait of the Pendulum” of the com-
puter program to carry out these experiments. For each value of the energy, the
simulation of motion and the plotting of curves on the screen is terminated auto-
matically when the pendulum completes one full cycle of its motion. So the final
reading of the timer gives the value of the period (in units of the natural period T0
of small oscillations) for the simulated motion.
(b) What is the ratio of the periods you have measured in these two cases?
How can you explain this ratio?
(c) When the total energy E of the pendulum is greater than the height Emax =
2mga of the potential barrier, the period of rotation T rapidly decreases as the
energy is increased. The period tends to zero with the growth of the energy. What
is the asymptotic behavior of T (E) when E tends to infinity?
7.6.3.3∗ Rotation of the Pendulum with Friction.
(a) Experimentally examine the rotation of the pendulum in the presence of
weak viscous friction. Note the gradual approach of the phase trajectory to the
separatrix. What is the value of the total energy of the pendulum at the moment
when the phase trajectory crosses the separatrix? Note that before the crossing
(while the pendulum is executing complete revolutions), the kinetic energy and
the angular velocity of the pendulum are never zero.
(b)∗∗ Using the law of energy conservation, evaluate the minimal value of the
initial velocity needed to obtain a complete revolution of the pendulum when it is
initially in the lower position if the quality factor Q = 15. What value of the initial
velocity is needed to obtain two revolutions of the pendulum? Verify your result
in a simulation experiment. Try to improve the approximate theoretical value of
the required initial velocity.

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

Chapter 8

Rigid Planar Pendulum under

Sinusoidal Forcing

Annotation. In Chapter 8 several well-known and recently discovered counterin-

tuitive regular and chaotic modes of the sinusoidally driven rigid planar pendulum
are discussed and illustrated by computer simulations. The software supporting
the investigation offers many interesting predefined examples that demonstrate
various peculiarities of this famous physical model. Plausible physical explana-
tions are suggested for some exotic and unexpected motions. The simulation pro-
gram can also serve as an exploration-oriented tool for discovering new features of
the driven pendulum and gives students an opportunity to perform mini-research
projects on their own.

8.1 Regular Response of a Harmonically Driven

Rigid Pendulum
8.1.1 Introduction
If we ask ourselves what is the most famous instrument in the history of physics,
the first idea may be about the pendulum. We may expect that an ordinary pen-
dulum subjected to periodic forcing will exhibit quite familiar behavior, which
agrees well with our intuition. However, despite the apparent simplicity, this well-
known nonlinear system can display a rich variety of rather complex, as-yet-
unexplored modes of motion, which include various kinds of transient processes,
single- and multiple-period stationary oscillations and complete revolutions, sub-
harmonic and superharmonic resonance responses, bistability and multistability,
intermittency, and transient and stationary chaos. Most of these modes delight the
eye and certainly challenge our physical intuition. By slowly varying the control
parameters of the system (the frequency and amplitude of the drive, and damping
factor), we can observe various kinds of bifurcations manifesting transitions of the

189

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

190 CHAPTER 8. PENDULUM UNDER SINUSOIDAL FORCING

pendulum between dramatically different modes of behavior.

The seemingly simple situation of the forced pendulum is actually quite com-
plex due to the subtle interplay between natural modes of the pendulum (these
modes are described in detail in Chapter 7 and [11]) and the periodic driving
force. The driven pendulum is interesting not only by virtue of its role in the his-
tory of physics, but, maybe more importantly, because it is isomorphic to many
other physical systems, including rf-driven Josephson junctions and phase-locked
voltage-controlled oscillators. The equation of motion of the dissipative, exter-
nally driven pendulum serves as a paradigmatic model of various low-dimensional
nonlinear dynamical systems and plays an important role in explorations of bifur-
cational and chaotic phenomena. Mechanical analogues of such systems allow us
to observe a direct visualization of their motion and thus can be very useful in
gaining an intuitive understanding of complex phenomena.
Numerous nonlinear problems in relation to the forced pendulum are clearly
presented in [12]. The nonlinear phenomena that can be predicted by analytical
methods are described in [13]. Detailed reviews of the experimental and theoret-
ical investigations of various regular and chaotic features of the system are avail-
able in the literature (see, for example, [14]–[15] and references therein).
An obvious way to understand the behavior of a nonlinear mechanical system
is to observe a computer simulation of its motion. Sometimes the simulations can
tell us much more than the equations can and thus contribute greatly in building
our physical intuition. For this purpose we have developed an interactive simu-
lation program, “Rigid pendulum driven by a sinusoidal torque,” included in the
software package “Nonlinear oscillations” that accompanies this textbook. The
program illustrates the motion of the sinusoidally driven pendulum. It simulates
all known modes of the pendulum behavior, and can also serve as a convenient
tool for discovering new features of this seemingly inexhaustible system.

8.1.2 The Physical Model

In this chapter and in the relevant simulation program we consider an ordinary pla-
nar rigid pendulum, say, a weightless rigid rod with a massive bob (point mass) at
one end (a simple or mathematical pendulum), or any other massive body (a phys-
ical pendulum) that can turn about a horizontal axis in a uniform gravitational
field. Being excited, the pendulum can rotate in the vertical plane or swing about
the stable equilibrium position in which its center of mass is below the axis. The
period T0 of infinitely small natural oscillations in the absence of friction is char-
acteristic of the given pendulum and can serve as a convenient unit of time for the
simulation. Natural oscillations gradually dampen due to friction whose braking
torque is assumed in the model to be proportional to the angular velocity of the
pendulum (viscous friction).
The momentary mechanical state of the pendulum is determined by its angular
position ϕ, which is the angle of deflection from the vertical equilibrium position
measured in radians (or degrees), and by the angular velocity ϕ̇ = dϕ/dt mea-
sured in the simulation program in units of the natural angular frequency ω0 of

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

8.1. REGULAR RESPONSE OF A DRIVEN PENDULUM 191

(undamped) infinitely small oscillations of the pendulum (ω0 = 2π/T0 ). We as-

sume that the pendulum is directly driven by an external sinusoidal torque with a
definite frequency ω and some constant amplitude.
The differential equation of motion used here and in the computer program to
simulate the damped driven pendulum is of the form

ϕ̈ + 2γ ϕ̇ + ω02 sin ϕ = ω02 φ0 sin ωt. (8.1)

Here ω is the driving frequency, and γ is the damping factor. To measure the vis-
cous damping, we can use instead of γ a more convenient dimensionless quantity
Q—the quality factor that equals the ratio ω0 /2γ.
The driving torque in the right-hand part of Eq. (8.1) is proportional to
φ(t) = φ0 sin ωt. This means that the dimensionless quantity φ(t) can be used as
a convenient measure of the external torque. Its physical sense can be explained
as follows. Imagine that some small constant (time-independent) external torque
φ is exerted on the pendulum (instead of φ(t) = φ0 sin ωt). This torque φ causes
a static displacement ϕ of the pendulum from the vertical. The sine of this angular
displacement is proportional to the torque. Indeed, for the pendulum in equilib-
rium the time derivatives of ϕ vanish (ϕ̈ = 0 and ϕ̇ = 0), and we conclude from
Eq. (8.1) that under a static torque φ the relation sin ϕ = φ is valid. Hence the
value φ = 1 corresponds to the external torque, which is necessary to hold the
pendulum stationary at horizontal position ϕ = π/2, the position of maximum
restoring torque of gravity.
For a small enough value of a constant torque, the displacement is small
(ϕ ≪ 1), and we can assume sin ϕ ≈ ϕ. That is, ϕ ≈ φ. This means that the
angular displacement of the pendulum under a small static torque just equals this
torque measured in the assumed angular units. In the limit of a very low driving
frequency (when ω → 0), the pendulum adiabatically follows the external torque,
and the low frequency steady-state forced oscillation of the pendulum will occur
just with the amplitude of the driving torque measured in these units (provided
the amplitude is small enough so that the static displacement is proportional to the
torque).

8.1.3 Behavior of the Pendulum under the Slow Varying

Sinusoidal Torque Whose Amplitude is Close to 1
Graphs of potential energy U (ϕ) ∼ (1 − cos ϕ − φϕ) for the pendulum subjected
to a static torque φ are shown in Figure 8.1 for several values of φ. In the absence
of external torque (φ = 0) stable equilibrium positions—minima of U (ϕ)—are
located at ϕ = ±2πn, n = 0, 1, . . . . Natural oscillations of the hanging down
pendulum can occur in any of the equivalent potential wells (say, about the mid-
point ϕ = 0) with the frequency ω0 .
The static torque φ causes a displacement of the equilibrium position to
ϕ = arcsin φ. The pendulum can be in equilibrium under a static torque if φ < 1
(curves 1 and 2 in Figure 8.1); at greater values of φ potential energy, U (ϕ) has no

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

192 CHAPTER 8. PENDULUM UNDER SINUSOIDAL FORCING

3 U(ϕ )
2
1 1
−π 0 π 2π ϕ
2

Figure 8.1: Potential energy U (ϕ) ∼ (1 − cos ϕ − φϕ) of the pendulum subjected
to a static torque φ. Curve 1 – φ = 0, curve 2 – φ = 0.5, curve 3 – φ = 1.

90.0 ο
ϕ (t)

N (t)
0

ο
−90.0
0 1 2 3T
o
(quality 10.0, driving frequency 0.01w0, driving amplitude 57.2958 ,
initial deflection -0.0558 o, initial angular velocity 0.01w0).

Figure 8.2: Steady-state motion of the pendulum at φ0 ≈ 1 under sinusoidal torque

N (t) of a low driving frequency (ω = 0.01 ω0). Harmonic components of ϕ(t)
are also shown.

minima (curve 3 in Figure 8.1), so that equilibrium (as well as an oscillatory mo-
tion) is impossible: The pendulum rotates. When φ → 1, the static displacement
ϕ → π/2 (tends to the horizontal position of the pendulum).
The case of a slow varying sinusoidal torque whose amplitude φ0 ≈ 1 deserves
special investigation. Figure 8.2 shows the time-dependent graph of the steady-
state motion at ω = 0.01 ω0 under the driving torque whose amplitude φ0 slightly
exceeds one radian. Period T = 2π/ω of the driving torque is chosen in this graph
as an appropriate time unit. We note a linear dependence of ϕ(t) on time t when
the external torque N (t) increases with time sinusoidally from zero to its maximal
value φ0 = 1.
Next we try to explain this counterintuitive behavior on the basis of the dif-
ferential equation of the pendulum, Eq. (8.1). For a slow steady-state motion (at
ω ≪ ω0 ), we can ignore the terms with the angular velocity and acceleration in
the differential equation, Eq. (8.1), of the pendulum. In other words, the pendulum
adiabatically follows the slow-varying external torque, remaining all the time in

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

8.1. REGULAR RESPONSE OF A DRIVEN PENDULUM 193

0.03 .
N (t) ϕ (t)

− 0.03
0 1 2 3T

Figure 8.3: Angular velocity at the low-frequency (ω = 0.01 ω0) motion of the
pendulum under the sinusoidal torque N (t) whose amplitude φ0 ≈ 1.

the equilibrium position (in the potential energy minimum), which is displaced
from the vertical by the external torque.
The sine of this angular displacement ϕ(t) is equal to the torque φ(t) =
φ0 sin ωt. This is evident from Eq. (8.1). Therefore for φ0 = 1 we get sin ϕ(t) =
sin ωt, and hence for the time interval (0, T /4) the angle of deflection ϕ(t) = ωt:
When the external torque φ(t) increases sinusoidally, the equilibrium position
ϕ(t) is displaced linearly with time. The angular velocity of the pendulum in this
slow uniform motion equals the driving frequency: ϕ̇(t) = ω.
This means that we can assume a linear function ϕ(t) ≈ ωt for the zero-order
solution to Eq. (8.1) in the time interval (−T /4, T /4). Similarly, for the adjacent
interval (T /4, 3T /4) we can write ϕ(t) ≈ π/2 − ω(t − T /4) = π − ωt. As a
whole, this approximate steady-state periodic solution is characterized by a saw-
tooth pattern with equilateral triangle teeth.
The simulation shows that this rectilinear tooth shape is slightly distorted near
each apex by rapid oscillations occurring after the external torque reaches a max-
imum and the direction of motion of the equilibrium position is reversed.
These rapid oscillations are especially pronounced in the angular velocity plot
(Figure 8.3). The angular velocity ϕ̇ is expressed here in units ω0 of the frequency
of small undamped natural oscillations.
In order to investigate analytically the character of these oscillations, we as-
sume that the angle of deflection for the time interval (T /4, 3T /4) can be ex-
pressed as ϕ(t) ≈ π − ωt + δ(t), where the correction δ(t) to the zero-order
function is small: δ(t) ≪ 1.
Differentiating Eq. (8.1) with respect to time, we obtain the following equation
for the angular velocity ϕ̇(t) = ν:

ν̈ + 2γ ν̇ + ω02 cos ϕ(t) ν = ωω02 cos ωt. (8.2)

In the left-hand part of this equation we can replace ϕ(t) by its zero-order time-
dependence ϕ(t) = π − ωt, and substitute for cos ϕ(t) its approximate expression
cos(π−ωt) = − cos ωt. Thus instead of (8.2) we get an approximate second-order

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

194 CHAPTER 8. PENDULUM UNDER SINUSOIDAL FORCING

.
0.01 ϕ (t)

0
3T/4
T/4
−0.01

Figure 8.4: Approximate behavior of the angular velocity at the low-frequency

steady-state motion of the pendulum.

linear homogeneous equation for the angular velocity ϕ̇(t) = ν:

ν̈ + 2γ ν̇ − ω02 cos ωt (ν + ω) = 0. (8.3)

We can conclude from Eq. (8.3) that, after the oscillations of the angular ve-
locity ν damp out, and ν approaches a constant value, so that its time derivatives
in Eq. (8.3) become negligible, this constant value equals −ω: ϕ̇ = ν → −ω.
During the preceding interval (−T /4, T /4), oscillations of the angular ve-
locity ϕ̇(t) have also damped out, and its constant value at the beginning of the
interval (T /4, 3T /4) approximately equals ω.
In further calculations it is convenient to transfer the time origin to the initial
moment of the interval (T /4, 3T /4), that is, to replace t → (t + T /4), or ωt →
(ωt + π/2) in Eq. (8.3):

ν̈ + 2γ ν̇ − ω02 sin ωt (ν + ω) = 0. (8.4)

Oscillations of the angular velocity at the beginning of the interval

(T /4, 3T /4) are approximately described by a solution to this homogeneous equa-
tion. We note that in Eq. (8.4) sin ωt can be replaced by ωt for the time interval
we are interested in. Even after this simplification Eq. (8.4) cannot be solved an-
alytically. However, we can find numerically its particular solution for the given
time interval with the help of any available mathematical package.
The initial conditions for the starting point of this interval follow from the
known pattern of the velocity graph for the periodic steady-state motion shown in
Figure 8.3: ν = ω, ν̇ = 0.
The graph in Figure 8.4 shows the solution to Eq. (8.4) with these initial
conditions obtained with the help of the “Mathematica” package (ω = 0.01 ω0,
Q = ω0 /2γ = 0.1).
Comparing this graph with those shown in Figure 8.3, we see that approximate
expression, Eq. (8.4), indeed describes qualitatively the general character of oscil-
lations of the angular velocity that take place after the motion of the equilibrium
position is reversed.

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

8.1. REGULAR RESPONSE OF A DRIVEN PENDULUM 195

ο
180
ϕ (t)

0
N (t)

ο
−180
2.9 .
ϕ (t)
N (t)

−2.9
0 1 2T
(quality 3.5, driving frequency 0.03w0, driving amplitude 57.8o,
initial deflection -29.5 o, initial angular velocity -0.343w 0)

Figure 8.5: Rotations and oscillations of the pendulum under a slow-varying si-
nusoidal external torque whose frequency equals 0.03 ω0 and whose amplitude
corresponds to φ0 slightly greater than 1.

If amplitude φ0 of the driving torque and its frequency ω are chosen to be

a little greater than in the case considered above, the pendulum at first follows
again the slowly varying torque, so that deflection angle ϕ(t) also slowly increases
almost linearly with time up to the highest (nearly horizontal) position.
However, instead of reversing its slow motion alongside the equilibrium po-
sition when the latter starts to move back, the pendulum in this case escapes the
shallow potential well over its low right barrier and “slides down” along its bumpy
outer slope (see curves 2 and 3 in Figure 8.1). This means that the pendulum com-
mences a rapid, unidirectional, nonuniform rotation. Figure 8.5 shows the graphs
of angular position ϕ(t) and angular velocity ϕ̇(t) time dependence for this ex-
traordinary counterintuitive motion.
We can evaluate the average angular velocity hϕ̇iav of this rapid rotation by
equating the external torque φ0 at its maximum to the torque of viscous friction:
ω02 φ0 = 2γhϕ̇iav , whence hϕ̇iav ≈ Qω0 . The average period Trot of this rotation
can be estimated as Trot = 2π/hϕ̇iav ≈ T0 /Q, where T0 = 2π/ω0 is the period
of the small natural oscillations.
When t approaches T /2, the external torque N (t) becomes smaller, the pen-
dulum rotation gradually slows down, and finally (when the torque almost van-
ishes) the pendulum becomes trapped in the potential well, within which it ex-
ecutes damped natural oscillations near the equilibrium position, which moves
uniformly backward under the reversed external torque (see Figure 8.5). Then all
the above-described motion repeats in the opposite direction.

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

196 CHAPTER 8. PENDULUM UNDER SINUSOIDAL FORCING

ϕm ϕm o
120
o Q = 10
120
o
0
= 12.5
100 Q=7 100 =8
o
o 0
80 = 12.5 80 Q = 10 o
0 Q=5 =5
60 60 0

40 40
20 20

0.2 0.4 0.6 0.8 1.0 1.2 1.4 0.2 0.4 0.6 0.8 1.0 1.2 1.4 ω/ω 0

Figure 8.6: Approximate theoretical response-frequency curves of the pendulum:

(a) constant driving amplitude φ0 = 12.5◦ ; (b) constant quality factor Q = 10.

8.2 Steady-State Response-Frequency Curves

of the Harmonically Driven Pendulum
8.2.1 Approximate Theoretical Resonance Curve
The steady-state response of a linear oscillator subjected to sinusoidal forcing is
also a sinusoidal motion whose frequency equals the forcing frequency and whose
amplitude depends on the frequency in a resonance manner. For the pendulum, as
long as the driving amplitude is small and the damping is not too weak, the steady-
state oscillation occurs with a small amplitude, so that the amplitude-frequency
resonance curve is rather well approximated by the result for a harmonic oscillator.
For a stronger driving and/or weaker friction, the resonance curve of the pen-
dulum bends toward lower frequencies and even folds, as shown in Figure 8.6. The
nonlinear resonance curve can be approximated considerably well by the follow-
ing heuristic approach [4]: We take the resonance curve of the harmonic oscillator
and replace in it the natural frequency ω0 by ωres (ϕm ). This approach assumes
that the steady-state oscillation is still harmonic, i.e., sinusoidal, but its frequency
depends on the amplitude ϕm .
For this dependence, ωres (ϕm ), we can use the approximate frequency-ampli-
tude relation of the pendulum, valid at moderate values of the amplitude ϕm (see
Chapter 7, Eq. (7.5), p. 158): ωres ≈ ω0 (1 − ϕ2m /16). This approximate depen-
dence of ωres on the amplitude ϕm —the so-called skeleton curve—is shown by
the thin dashed line in Figure 8.6. The resonance peak at its maximum is shifted to
lower frequencies and acquires a shape typical for nonlinear systems with a “soft”
restoring force. Over some critical value of the driving amplitude (for a given
quality factor), the theoretical resonance curve becomes S–shaped with three so-
lutions, only two of which are stable. This folding of the response-frequency curve
(foldover effect) leads to bistability and hysteresis.
Within some interval of driving frequencies the pendulum oscillates either
with a large amplitude or a small amplitude. In-between there is always an unsta-

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

8.2. RESPONSE-FREQUENCY CURVES 197

ϕm

C
150
2
B
100

A 50
1
D
0.4 0.6 0.8 1.0 1.2 ω/ω 0

Figure 8.7: Hysteretic behavior of the amplitude-frequency dependence during the

up and down sweeping of the drive frequency (φ0 = 12.5◦, Q = 10).

ble solution (see the left overhanging slope of resonance peaks shown by dashed
lines in Figure 8.6). Which of the two stable, periodic motions (limit cycles) even-
tually is established depends on the initial conditions.
A convenient traditional way to observe the nonlinear resonance response of
the pendulum is to slowly vary (“sweep”) the driving frequency from one side of
the natural frequency through the resonance peak and to the other side in a process
of continuous steady-state oscillations, while the amplitude of the driving torque
is kept constant.
The pendulum responds differently depending on the direction of the fre-
quency variation—there is an associated hysteresis characterized by abrupt jumps
in the amplitude and phase of the steady-state response. When in the process of
frequency sweeping an abrupt jump occurs from one slope of the folded resonance
peak to the other, not only the amplitude of the steady-state oscillations changes
considerably, but the whole mode undergoes a dramatic change.
Figure 8.7 shows the response-frequency curve (see [58]) obtained with the
help of the simulation program. When we start the sweeping from low driving
frequencies (and at the initial conditions of zero, with the pendulum resting in the
equilibrium position), the observed steady-state response agrees perfectly well
with the theoretical prediction: The forced oscillations occur almost in phase with
the drive, and their amplitude grows gradually while the frequency is increased up
to point A, which is characterized by a vertical tangent to the theoretical curve.
Then an abrupt jump to point B lying on the right slope of the resonance peak oc-
curs. After this jump, the amplitude and phase again agree well with the theoretical
prediction. In the process of further sweeping, the amplitude of the steady-state re-
sponse gradually diminishes, and the pendulum oscillates in almost opposite phase
with respect to the driving torque.

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

198 CHAPTER 8. PENDULUM UNDER SINUSOIDAL FORCING

.
ϕ2( t ) ϕ2( t )
ο
125
ϕ1( t )

0
N (t) .
ϕ1( t )
−125ο
0 1 2 3 4T
(quality 10.0, driving frequency 0.75w0 , driving amplitude 12.5o,
initial deflection -101.1o, initial angular velocity -0.92w 0)

Figure 8.8: Graphs of the angle and angular velocity for small-amplitude (30◦ ) and
large-amplitude (125◦) oscillations at the same parameters of the pendulum and of
the external torque (ω = 0.75 ω0, φ0 = 12.5◦, and Q = 10). Initial conditions are
indicated for the large-amplitude oscillations. These graphs correspond to points
1 and 2 respectively on the response-frequency diagram of Figure 8.7.

8.2.2 Autoresonance, Hysteresis, and Bistability

If we reverse the direction of the frequency sweep, the pendulum’s response on the
way back follows the same curve up to point B. However, the amplitude, instead
of jumping down, continues to increase along the right slope of the theoretical
resonance peak after we have passed through point B.
Figure 8.8 gives an example of this bistability, and of the hysteretic behav-
ior of the pendulum at sweeping the driving frequency. Indeed, during the direct
sweeping from left to right, the steady-state oscillations at point 1 are almost si-
nusoidal in shape (curves ϕ1 (t) and ϕ̇1 (t) in Figure 8.8) and occur nearly in the
same phase with the driving torque N (t), while on the way back, at point 2 (at
the same frequency and amplitude of the drive as at point 1), the oscillations have
a much greater amplitude, and these oscillations lag in phase behind the driving
torque more than a quarter-period.
This phase relationship between the drive and the pendulum is characteristic
of resonance, and is preserved during the slow sweeping of the drive frequency by
virtue of the phase locking. Hence this phenomenon can be called autoresonance.
On this branch of the frequency-response characteristic, the large-amplitude oscil-
lations are no longer harmonic: The graph of ϕ̇2 (t) has a saw-toothed appearance,
while the graph of ϕ2 (t), though resembling a sinusoid, actually consists of nearly
parabolic alternating segments.
Steady-state modes ϕ1 (t) and ϕ2 (t), coexisting at the same frequency and
amplitude of the drive, differ in amplitude and in phase relationships with the
driving torque—they correspond to different slopes of the resonance peak. The
large amplitude mode ϕ2 (t) has a much smaller basin of attraction than ϕ1 (t).

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

8.2. RESPONSE-FREQUENCY CURVES 199

ο
169
ϕ (t)
.
N( t) ϕ (t)
0

.
ϕnat( t )
ο
−169
0 1 2 3T
(quality 30.0, driving frequency 0.442w0 , driving amplitude 4.0o,
initial deflection -147.0o, initial angular velocity -0.555w 0)

Figure 8.9: Graphs of ϕ(t) and ϕ̇(t) for the “bell-ringer” mode of forced oscil-
lations. The graph of ϕ̇nat (t) for natural undamped oscillations (φ0 = 0) of the
same amplitude (about 168◦ ) is also shown for comparison.

8.2.3 Nonlinear Resonance and a “Bell-Ringer Mode”

As we continue to sweep the drive frequency ω from point B towards lower values,
we move to the left in Figure 8.7 almost along the skeleton curve, and can reach
amplitudes that are considerably greater than the estimated theoretical maximum
(see Figure 8.6).
During this sweeping, the phase relationships between the drive and the pen-
dulum that are characteristic for resonance are automatically preserved (autoreso-
nance) due to the phase locking. For weak damping, the steady-state amplitude at
low drive frequencies can be very large (approaching 180◦ ) at moderate and even
quite small drive amplitudes.
Traces in Figure 8.9 give an example of such extraordinary motion of the pen-
dulum, which Peters called a “bell-ringer mode” [18]. In this mode the pendulum
passes rapidly through the lower equilibrium position, but “sticks” near the ex-
treme points of oscillation, spending a very long time moving slowly in the vicin-
ity of unstable equilibrium position (near the saddle point in the phase plane).
These large-amplitude oscillations give an example of nonlinear resonance:
Their period (which equals the drive period) is very close to the period of natu-
ral oscillations of the same amplitude. For large amplitudes, this period can last
several periods of small natural oscillations.
Actually, such forced oscillations in conditions of nonlinear resonance are very
much like natural oscillations of the pendulum that occur at the corresponding
large amplitude. To emphasize this similarity, we also show in Figure 8.9 by a
thin line the graph of ϕ̇nat (t) for natural undamped oscillations (φ0 = 0) of the
same amplitude (about 168◦ ).
We can exploit the similarity between the bell-ringer mode and natural un-
damped oscillations for theoretical calculation of the amplitude at a given resonant
drive frequency. It was shown in Chapter 7, Eq (7.22), p. 172 (and also in [11]) that
for non-sinusoidal natural oscillations with an amplitude ϕm approaching 180◦ the

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

200 CHAPTER 8. PENDULUM UNDER SINUSOIDAL FORCING

period depends on the amplitude as follows:

2 8  πω 
0
T = T0 ln , whence ϕm = π − 8 exp − . (8.5)
π π − ϕm 2 ω
In the simulation presented in Figure 8.9 the frequency of the drive ω = 0.442 ω0.
Substituting this value in Eq. (8.5), we find for the corresponding amplitude of
undamped natural oscillations: ϕm = 167◦ . This theoretical estimate is very close
to the amplitude of the bell-ringer forced oscillations (ϕm ≈ 168◦) observed in
the simulation experiment at the drive frequency ω = 0.442 ω0.
When the drive period is almost equal to the period of large natural oscilla-
tions, full synchronization between the motions (phase-locking) can occur. The
pendulum lags in phase about a quarter period behind the periodic external torque
N (t). Due to the phase-locking, this small torque N (t) at autoresonance is almost
always directed in phase with the angular velocity ϕ̇(t) of the pendulum (see Fig-
ure 8.10) and therefore supplies the pendulum with energy needed to compensate
for frictional losses and to maintain the constant amplitude of large non-sinusoidal
nearly natural oscillations.
An alternative physical explanation of the “bell-ringer mode” is based on con-
sidering the motion of a particle in a time-dependent spatially periodic potential
(see curve 1 in Figure 8.1) whose pattern is “rocking” slightly about the origin
(point 0), so that the right barrier of the well lowers a bit and the left one rises
when the external torque is directed to the right, and vice versa, after a half-period
of the drive. Let us imagine that the particle in the well on its way from left to right
slowly “climbs up” the slope of the right barrier and turns back approximately at
the time when the potential pattern is horizontal (zero external torque), and passes
back through the bottom of the well just after the moment at which the right bar-
rier is at its maximal height. The duration of the particle motion back and forth in
the non-parabolic well depends on the amplitude, and if this duration equals the
period of “rocking” of the potential pattern, phase locking can occur and a steady-
state process can eventually establish. The energy needed to overcome friction is
supplied by the source that “rocks” the potential pattern (that is, by the periodic
external torque).
This “bell-ringer” mode can certainly also be excited by carefully choosing
proper starting conditions. However, to maintain the large-amplitude motion, the
phase relation between the pendulum and the drive torque is critical. This means
that for this mode the basin of attraction in the phase plane of initial conditions
is rather small. Hence it is much easier to reach this mode experimentally by
sweeping down the drive frequency as described above. After each step along this
way, we must wait for transients to settle.
The spectrum of the bell-ringer large-amplitude oscillations, besides the fun-
damental harmonic whose period equals the driving period, also contains several
harmonic components of higher orders. Their frequencies are odd integer mul-
tiplies of the fundamental frequency. Figure 8.10 shows the graphs of angular
velocity ϕ̇(t) and its harmonics for the bell-ringer mode.

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

8.2. RESPONSE-FREQUENCY CURVES 201

2.02 .
ϕ (t)
N( t )

− 2.02
0 1 2 3T
(quality 30.0, driving frequency 0.442w0 , driving amplitude 4.0o,
initial deflection -147.0o, initial angular velocity -0.555w 0)

Figure 8.10: Graphs of angular velocity ϕ̇(t) and its harmonics for the “bell-ringer
mode” of forced oscillations.

8.2.4 Symmetry-Breaking and Period-Doubling Bifurcations,

Chaos and the Crisis
Further decreasing of the drive frequency brings the system to point C of the
amplitude-frequency characteristic (see Figure 8.7, p. 197) at which a symmetry-
breaking bifurcation occurs: The pendulum’s excursion to one side is greater than
to the other side, for example, 173◦ versus 165◦ . This spatial asymmetry of os-
cillations increases as we move further to lower frequencies. In the spectrum of
such asymmetric steady-state oscillations, besides harmonics of odd orders, even-
order harmonics of small amplitudes are present, including a zero-order compo-
nent (constant mid-point displacement). Such asymmetric modes exist in pairs
whose phase orbits (at the same drive frequency) are the mirror images of one
another.
In this way, at a certain frequency, a period-doubling bifurcation occurs: In
each subsequent cycle the maximal deflection to the same side slightly differs from
the preceding one, but after two cycles the motion repeats exactly. This means that
the period of this steady-state oscillation equals two drive periods. Period dou-
bling breaks the original time-translational symmetry of the sinusoidally driven
pendulum: Although the driving torque repeats exactly from cycle to cycle, the
pendulum executes slightly different motions on alternate cycles. For this motion,
the Poincaré section consists of two nearby points in the phase plane visited in
alternation.
What happens after this period-doubling bifurcation in the simulation exper-
iment depends strongly on details of the frequency diminution. If the sweeping
occurs in very small steps, a whole cascade of close-set period-doubling bifurca-
tions can be observed. Each bifurcation in this series doubles the period of motion
and the number of fixed points in its Poincaré map.
This cascade of period-doubling bifurcations converges to a chaotic large-
amplitude oscillation of the pendulum. During these chaotic oscillations of the
bell-ringer type the maximal deflection is close to 180◦ and varies randomly from

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

202 CHAPTER 8. PENDULUM UNDER SINUSOIDAL FORCING

.
ϕ (t) ϕ (t)
180 ο

N (t )
0

ο
−180
0 1 2 3 4 5 6 7 8 9 10T
(quality 10.0, driving frequency 0.52w0 , driving amplitude 12.5o )

Figure 8.11: Transition from large-amplitude nonlinear oscillations (from the ir-
regular “bell-ringer mode”) to small-amplitude ordinary forced oscillations.

cycle to cycle whose duration equals approximately one driving period. Contrary
to a complicated initial transient that leads eventually to a regular motion char-
acterized by a fixed finite set of Poincaré sections, this chaotic regime persists
indefinitely. The Poincaré map consists of two small nearby islands visited in al-
ternation. Within each island the point bounces randomly from cycle to cycle. This
chaotic state is stable in the sense that after a small perturbation the phase trajec-
tory converges to the same region. Attracting regions in the phase space that cor-
respond to chaotic regimes are called strange attractors because they are formed
by fractals—geometric objects of non-integer dimensions. The fractal character
of attractors is essential to the existence of persistent dynamical chaos.
The chaotic oscillatory regime following the bell-ringer mode exists in a very
narrow interval of the driving frequencies, so that a slight perturbation can cause a
crisis leading to an abrupt jump of the amplitude down to point D (see Figure 8.7)
located on the far left outskirt of the resonance peak. If the frequency sweeping is
executed by steps that are not small enough, this jump can occur before the chaotic
regime is established or even before the period-doubling bifurcation occurs.
Actually, this abrupt jump of the amplitude (the crisis) is presented by a long
irregular transient (as in Figure 8.11), during which the motion of the pendulum
undergoes a radical rearrangement.
Details of this transient are very sensitive to the character of perturbation (to
the magnitude and timing of the frequency step). In particular, the initial stage of
the transient may have the character of intermittency: During a long time the pen-
dulum executes an asymmetric oscillation in which its excursion, say, to the left
side is greater than to the right side. Then during several cycles the asymmetry
changes to the opposite, that is, to prolonged oscillations with a greater maxi-
mal deflection to the right side. Such irregular interchanges of the two spatially
asymmetric regimes are characterized by a time scale much longer than the cycle
duration (the drive period), and can occur several times before the crisis.
The crisis leading to the jump down of the amplitude can be initiated, for

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

8.3. SUBHARMONIC AND SUPERHARMONIC RESONANCES 203

example, if irregular amplitude variations lead the pendulum to cross the vertical
(to make a full revolution), after which the pendulum gradually settles down to
the low-frequency and low-amplitude regular (sinusoidal) steady-state oscillation
for which the angular displacement ϕ(t) almost exactly equals the driving torque:
ϕ(t) ≈ φ0 sin ωt. The simulation of motion in Figure 8.11 shows us what can
happen during a transient that accompanies this amplitude jump.
If after this jump down of the amplitude at point D (see Figure 8.7) we con-
tinue to sweep the frequency down, the amplitude and phase of steady-state oscil-
lations again obey the theoretical response-frequency curve, just as they did while
sweeping the frequency from left to right.

8.3 Subharmonic and Superharmonic Resonances

Steady-state forced oscillations of a large amplitude, resembling the bell-ringer
mode, can also occur if the driving frequency is approximately three times greater
than the natural frequency that corresponds to this large amplitude. An example of
such oscillations is shown in Figure 8.12. Their graphs show clearly that the third
harmonic of these steady-state oscillations has the frequency that equals the driv-
ing frequency, while the frequency of the fundamental harmonic equals one third
of the driving frequency. In other words, one cycle of such non-harmonic oscilla-
tions of the pendulum covers three driving periods. Forced period-3 oscillations
of large amplitude occurring under such conditions give an example of nonlin-
ear third-order subharmonic resonance. Subharmonic resonances do not exist in
linear systems.
Similarly to the bell-ringer mode (see Figure 8.10), the pendulum behaves
here very much like during free (unforced) oscillations. The sinusoidal external
torque, being synchronized with the third harmonic of these non-harmonic large-
amplitude natural oscillations, compensates for frictional losses and maintains a
constant angular excursion.
This synchronization (phase-locking) can occur only if at t = 0 (the time mo-
ment when the torque is switched on) large-amplitude natural oscillations already
exist. This means that for given frequency of the external torque (lying within
a definite interval) the third-order subharmonic resonance occurs only for initial
conditions from a certain region (from the basin of attraction of this limit cycle).
Different initial conditions cause the pendulum to eventually settle down into the
low-amplitude period-1 antiphase oscillation that corresponds to the far-off high-
frequency slope of the nonlinear resonance peak.
During slow reduction of the driving frequency under conditions of the third-
order subharmonic resonance, a symmetry-breaking bifurcation takes place, after
which the angular excursion to one side is greater than to the opposite side. Such
spatially asymmetric modes exist in pairs whose phase orbits are the mirror images
of one another. Further reduction of the driving frequency leads to a crisis: After
a long transient the pendulum settles into the ordinary antiphase mode of forced
oscillations that correspond to the right slope of the nonlinear resonance curve.

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

204 CHAPTER 8. PENDULUM UNDER SINUSOIDAL FORCING

ο
135
ϕ (t)
N (t)
0

ο
−135
0 1 2 3 4 5 6T
2.27 .
ϕ (t)
N( t )
0

−2.27
0 1 2 3 4 5 6T
o
(quality 280.0, driving frequency 1.85 w0, driving amplitude 35.0 ,
initial deflection -3.99o, initial angular velocity -2.27w 0)

Figure 8.12: Graphs of the angular deflection ϕ(t), angular velocity ϕ̇(t) and their
harmonics of oscillations at subharmonic resonance of the third order.

At subharmonic resonance of the fifth order (see the graph of ϕ̇(t) and its
harmonics in Figure 8.13), one cycle of the pendulum’s almost natural oscillation
covers five driving periods: The external torque is synchronized with the fifth har-
monic of a period-5 large-amplitude oscillation of the pendulum. On average, this
phase-locking provides a surplus of energy transferred to the pendulum over the
energy returned back to the source of the external torque, thus compensating for
frictional losses.
Gradually reducing the driving frequency under conditions of the fifth-order
subharmonic resonance, we can observe bifurcations of the symmetry-breaking
and period-tripling, after which the period of steady-state forced oscillations
equals 15 driving periods. The set of Poincaré sections consists of 15 fixed points
in 5 groups visited by turn. Each group consists of 3 nearby points.
The subharmonic resonances discussed above occur at rather high drive fre-
quencies, which are equal to an odd integer of the natural frequency. By contrast,
superharmonic resonances can be excited at rather low drive frequencies: Syn-
chronization of the drive with oscillations of the pendulum (phase locking) occurs
if one period of the drive covers an odd integer number of natural periods.
The nature and origin of superharmonic resonances can be explained in the
following way. Let us consider natural nonlinear oscillations of the pendulum in a
potential well that slowly moves back and forth due to sinusoidally varying (with
driving frequency ω) external torque. Under certain conditions an integer number
of natural cycles covers one period of the potential well motion. Figure 8.14 shows
clearly that for the third-order superharmonic resonance just three natural cycles

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

8.3. SUBHARMONIC AND SUPERHARMONIC RESONANCES 205

2.09 .
ϕ (t)
N (t )
0

−2.09
0 1 2 3 4 5 6 7 8 9 10T
(quality 100.0, driving frequency 2.65 w0 , driving amplitude 26.18 o,
initial deflection 142.95 o, initial angular velocity -0.6272w 0)

Figure 8.13: Graphs of angular velocity ϕ̇(t) and its harmonics at subharmonic
resonance of the fifth order.
ο
107 ϕ (t)
N (t)

ο
−107
0 1 2 3T
(quality 25.0, driving frequency 0.26 w0 , driving amplitude 45.0 o,
initial deflection 13.28 o, initial angular velocity 1.253w 0)

Figure 8.14: Plots of angular deflection ϕ(t) and its harmonics at superharmonic
resonance of the third order occurring under the sinusoidal external torque N (t).

are executed during one period of the drive. In this case phase-locking of the po-
tential well motion with natural oscillations can occur. By virtue of this synchro-
nization, the external torque can continuously supply the pendulum with energy
required to compensate for frictional losses and prevent damping of short-period
natural oscillations of the pendulum in the moving potential well. As a result, a
steady-state non-sinusoidal period-1 oscillation (its period equals that of the drive)
is established, whose spectrum is distinguished by the considerable contribution
of the third harmonic.
Depiction of such a motion on the screen with the help of the simulation pro-
gram allows us to develop an intuitive feel for how nonlinear systems generate
high harmonics of the sinusoidal input oscillation. The simulation tells us much
more for understanding this phenomenon than the mathematical equations can do.
Superharmonic resonances are also accompanied by symmetry-breaking bi-
furcations and chaotic regimes. Examples of strange attractors that follow super-
harmonic resonances of the third and fifth order are shown in Figures 8.15a and
8.15b, respectively.

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

206 CHAPTER 8. PENDULUM UNDER SINUSOIDAL FORCING

a b
1.34 1.0

0 0

−1.34 −1.0
o o
−127
o 0 127 −120 o 0 120
a − quality 7.1, driving frequency 0.23w0 , driving amplitude 51.0 o,
initial deflection 7.45o, initial angular velocity 1.22 w 0;
b − quality 22.0, driving frequency 0.1483w0 , driving amplitude 50.0 o,
initial deflection −17.95 o, initial angular velocity −0.50 w 0

Figure 8.15: Phase diagrams (with Poincaré sections) of chaotic oscillations in the
vicinity of superharmonic resonances of the third (a) and fifth (b) orders.

8.4 Other Extraordinary Regular Forced

Oscillations
The original time-translational symmetry in the motion of the sinusoidally driven
pendulum can be broken not only by the above-described period-doubling bifurca-
tions: Under certain conditions regular oscillations of the pendulum have a period
that covers some integer (other than two) number of the drive period.
Figure 8.16 shows the graphs of ϕ(t) and ϕ̇(t) with their harmonics for a
period-3 nonlinear oscillation in which the frequency of the third harmonic co-
incides with the driving frequency. The pendulum makes one oscillation during
each drive period, but the swing differs from one cycle to the next as though the
mid-point were moving with a period that is 3 times the drive period: Maximal
angular excursion of the pendulum equals 171◦ , then 117◦, and then 111◦. After
3 cycles of the external torque all the motion repeats. The set of Poincaré sections
consists of three fixed points visited in turn. If the initial conditions are chosen
somewhere beyond the basin of attraction of this mode, the pendulum eventually
settles into a coexisting simple mode—period-1 spatially symmetric oscillations
with an amplitude of 142◦ that occurs in the opposite phase with respect to the
driving torque.
When the driving frequency is slightly smaller than the natural frequency,
rather counterintuitive steady-state modes can occur in which the motion of the
pendulum resembles beats: The amplitude (and the frequency) of oscillation are
not constant but instead vary slowly with a long period that equals an integer
(odd and rather large) number of driving periods. An example of such steady-
state self-modulated oscillations whose period equals 11 driving cycles is shown
in Figure 8.17. We note the most surprising feature of this mode: The maximal

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

8.4. OTHER EXTRAORDINARY FORCED OSCILLATIONS 207

ο
171
ϕ (t)
N (t)
0

−171 ο
2.11 .
ϕ (t)
N (t)

− 2.11
0 1 2 3 4 5 6T
o
(quality 15.0, driving frequency 0.765 w0 , driving amplitude 30.0 ,
initial deflection 43.48 o, initial angular velocity -2.037w 0)

Figure 8.16: Period-3 nonlinear oscillations. The frequency of the fundamental

harmonic equals one third of the driving frequency.

deflections of the pendulum have no tendency to equalize in the course of time.

Contrary to ordinary transient beats, in these oscillations the variations of ampli-
tude and frequency do not fade: Once established, they continue forever.
We can suggest a simple physical explanation for this exotic mode of self-
modulated oscillations. Let the phase of the sinusoidal driving torque be initially
almost equal to the phase of (natural) oscillations of the angular velocity. That
is, let us assume that the external torque varies with time in such a way that it is
directed along the angular velocity during almost the entire period. In this case
the energy is transferred to the pendulum, and the amplitude of oscillations grad-
ually grows. But with the growing amplitude the natural period of the pendulum
becomes longer. Therefore after a while the oscillations of the angular velocity
accumulate some phase lag with respect to the driving torque. When this phase
lag increases up to 180◦ , that is, the driving torque varies in the opposite phase
with respect to the angular velocity, the energy flow is reversed. This causes the
swing of oscillations to decrease.
Then after a while the phase relations again become favorable for supplying
the energy to the pendulum, and the amplitude grows again. Thus the amplitude
of oscillations is modulated with some (rather long) period.
A small amount of friction can stabilize the period of modulation. If this period
equals an integer number of the driving periods, the phase-locking can occur. By
virtue of this synchronization between the external drive and natural oscillations of
the pendulum, the whole process of self-modulated oscillations becomes exactly
periodic.

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

208 CHAPTER 8. PENDULUM UNDER SINUSOIDAL FORCING

ο .
147 ϕ (t) ϕ (t)
N (t)
0

ο
−147
0 1 2 3 4 5 6 7 8 9 10 11T
(quality 600.0, driving frequency 0.90 w0 , driving amplitude 12.0 o,
initial deflection -5.18o, initial angular velocity -0.677w 0)

Figure 8.17: Period-11 forced steady-state self-modulated oscillations.

147 ο
ϕ (t)
N( t )
0

ο
−147
0 1 2 3 4 5 6 7 8 9 10 11T
(quality 600.0, driving frequency 0.90 w0 , driving amplitude 12.0 o,
initial deflection -5.18o, initial angular velocity -0.677w 0)

Figure 8.18: Graph of period-11 steady-state forced self-modulated oscillations

and graphs of its harmonic components that constitute its spectrum.

The energy dissipation is compensated by a somewhat greater amount of en-

ergy being transferred to the pendulum on average (during a cycle of the modula-
tion) compared to the backward transfer from the pendulum to the source of the
external drive.
Thin lines in Figure 8.18 show harmonic components (the spectrum) of these
period-11 self-modulated oscillations. The fundamental harmonic of this non-
sinusoidal oscillation has the frequency that equals (1/11)th of the driving fre-
quency ω. Its amplitude is an order of magnitude smaller than the amplitude of
the eleventh harmonic component whose frequency equals the driving frequency.
Besides this component, harmonics with frequencies 7/11, 9/11 and 13/11 of the
driving frequency contribute considerably to the resulting oscillation. The ampli-
tudes of odd harmonics are listed in Table 8.1.
Another example of a periodic steady-state forced motion of a pendulum is
presented in Figure 8.19. During one period of the external torque the pendulum
makes 6 fast revolutions to one side, then its rotation slows down, and it makes 6

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

8.4. OTHER EXTRAORDINARY FORCED OSCILLATIONS 209

Table 8.1: Amplitudes of odd harmonics for period-11 oscillations

No Amplitude (radians) Amplitude (degrees) Velocity (units of ω0 )

1 0.091 5.214 0.007

3 0.137 7.850 0.039
5 0.269 15.41 0.121
7 0.585 33.52 0.374
9 1.164 66.69 0.951
11 0.883 50.59 0.885
13 0.428 24.52 0.507
15 0.013 0.745 0.018

revolutions to the opposite side.

From the angular velocity graph of this extraordinary motion (see the lower
panel of Figure 8.19) we suppose that the time dependence of ϕ̇(t) can be rep-
resented as a superposition of a slow periodic component (varying almost sinu-
soidally with the drive period T ) and a small fast component distorting this slow
variation. We suppose that the slow variation of ϕ̇(t) is caused by the slow vary-
ing external torque, while the additional fast oscillations of ϕ̇(t) appear by virtue
of the gravitational force that influences the pendulum rotation. Hence to a first
approximation this extraordinary behavior of the pendulum can be explained by
neglecting the force of gravity. Omitting the last term (whose origin is related to
gravity) in the left-hand part of Eq. (8.1), we get the following linear first-order
equation for the angular velocity ν(t) = ϕ̇(t):

ν̇ + 2γν = ω02 φ0 sin ωt. (8.6)

The steady-state periodic solution to this equation can be represented as follows:

ω 2 φ0 2γ
ν(t) = −νm cos(ωt + δ), νm = p 0 , δ = arctan . (8.7)
ω 2 + 4γ 2 ω

Hence the angular velocity ν(t) varies sinusoidally with drive frequency ω and
amplitude νm given by Eq. (8.7). Actually ν(t) corresponds to the slow compo-
nent of ϕ̇(t) averaged over the period of fast rotation: ν(t) = hϕ̇(t)iav . According
to Eq. (8.7), its amplitude νm ≈ ω02 φ0 /ω equals 3.6 ω0 for the values φ0 = 0.8 and
ω = 0.22 ω0 that were used in the simulation experiment shown in Figure 8.19,
while the phase lag δ = arctan(2γ/ω) = arctan(ω0 /Qω) ≈ 0.3. These val-
ues agree rather well with the experiment. To evaluate the minimal period ∆t of
fast rotation, we can divide the full angle 2π by the average angular velocity νm ,
whence ∆t/T = ω/νm = 0.06, which also agrees well with the experimental
graph in Figure 8.19.

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

210 CHAPTER 8. PENDULUM UNDER SINUSOIDAL FORCING

ο
180
ϕ (t)
N(t)
0

ο
−180
0 . 1 2T
4.5
ϕ (t)
N( t )
0

− 4.5
0 1 2T
(quality 15.0, driving frequency 0.22 w0 , driving amplitude 45.8 o,
o
initial deflection -63.36 , initial angular velocity -4.183w 0)

Figure 8.19: Stationary period-1 fast bidirectional revolutions of the pendulum.

We can evaluate the amplitude of fast oscillations of the angular velocity ϕ̇(t)
on the basis of the energy conservation. Let ϕ̇max and ϕ̇min be the maximum and
minimum values of ϕ̇(t) during the stage of fastest rotation. Kinetic energy of the
rotating pendulum at the lowest point (which is proportional to ϕ̇2max ) is greater
than at the inverted position approximately by the difference in the potential en-
ergy at these points. From these considerations we find:

ω2 ω 2 φ0 ω 2 φ0
 
ϕ̇max, min = νm 1 ± 20 , νm = p 0 ≈ 0 . (8.8)
νm ω 2 + 4γ 2 ω

According to this estimate, the fractional difference (ϕ̇max − ϕ̇min )/νm at φ0 =

0.8 and ω = 0.22 ω0 equals 0.15, again in a good agreement with the experiment.
An alternative physical explanation of this counterintuitive mode may be for-
mulated by considering the motion of a particle in a time-dependent periodic po-
tential (see curve 1 in Figure 8.1) whose lateral barriers are slowly rising and
falling with time. Contrary to the similar approach in the explanation of the bell-
ringer mode (see Section 8.2.3), now the potential pattern is “rocking” about the
origin (point 0) with a large amplitude. After escaping the potential well by cross-
ing its falling barrier, the particle starts to slide down along the bumpy slope cross-
ing the barriers until the next barrier rises high enough to slow down the particle
and to force its backward non-uniform motion.
To maintain this exotic steady-state periodic motion (to provide the phase
locking), the phase relation between the pendulum and the periodic variation of
the potential pattern (that is, the drive torque N (t) time dependence) is critical.

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

8.5. CONCLUDING REMARKS 211

This means that the mode can be excited only by choosing the initial conditions
carefully. In other words, this limit cycle is characterized by a small basin of at-
traction.

8.5 Concluding Remarks

The dynamic behavior of the planar forced pendulum discussed in this chapter
is richer in various modes than we might expect for such a simple physical sys-
tem relying on our intuition. Its nonlinear large-amplitude motions can hardly be
called simple. Variations of the parameters result in different regular and chaotic
types of dynamical behavior. The simulation program “Rigid pendulum driven by
a sinusoidal torque” offers many interesting predefined examples (besides those
discussed above) that illustrate various peculiarities of this famous physical model
in vivid computer simulations. Visualization of the motion simultaneously with
plotting the graphs of different variables and phase trajectories makes the simula-
tion experiments very convincing and comprehensible.
We have touched on only a small portion of the steady-state modes and regu-
lar motions of the sinusoidally driven rigid pendulum. The pendulum’s dynamics
exhibits a great variety of other counterintuitive rotational, oscillatory, and com-
bined (both rotational and oscillatory) multiple-periodic stationary states (attrac-
tors), whose basins of attraction are sometimes characterized by a surprisingly
complex (fractal) structure. Computer simulations also reveal intricate sequences
of bifurcations, leading to numerous intriguing chaotic regimes. Most of these fea-
tures remain beyond the scope of this chapter. With good reason we can say that
this familiar and apparently simple physical system seems almost inexhaustible.

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

Chapter 9

Pendulum with a
Square-Wave Modulated
Length

Annotation. The phenomenon of parametric resonance in a rigid planar pendulum

caused by a square-wave modulation of its length is investigated both analytically
and with the help of a computer simulation. Characteristics of parametric reso-
nance are found and discussed in detail. The role of nonlinear properties of the
pendulum in restricting the resonant swinging is emphasized. The boundaries of
parametric excitation as functions of the modulation depth and the quality factor
are determined. Stationary oscillations at these boundaries and at the threshold
conditions are investigated.

9.1 The Investigated Physical System

Periodic excitation of a physical system is called parametric forcing if it is real-
ized by variation of some parameter that characterizes the system. In particular,
a pendulum can be excited parametrically by a given vertical motion of its sus-
pension point. This apparently simple physical system exhibits a surprisingly vast
variety of possible regular and chaotic motions. Hundreds of texts and papers are
devoted to investigation of the pendulum with vertically oscillating pivot: See, for
example, [16], [17] and references therein. In the frame of reference associated
with the pivot, such forcing of the pendulum is equivalent to periodic modulation
of the gravitational field. A widely known curiosity in the behavior of an ordinary
rigid planar pendulum whose pivot is forced to oscillate along the vertical line
is the dynamic stabilization of its inverted position, occurring when the driving
amplitude and frequency lie in certain intervals (see [17]–[21] and Chapter 10).
Another familiar method of parametric excitation that we explore in this chap-

213

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

214 CHAPTER 9. PENDULUM WITH MODULATED LENGTH

ter consists of a periodic variation of the length of the pendulum. In many text-
books and papers (see, for example, [22]–[26]) such a system is considered as a
simple model of a playground swing. Indeed, the swing can be treated as a phys-
ical pendulum whose effective length changes periodically as the child squats at
the extreme points, and straightens each time the swing passes through the equi-
librium position. It is easy to illustrate this phenomenon of the swing pumping in
a classroom by the following simple experiment. Let a thread with a bob hanging
on its end pass through a little ring fixed in a support. You can pull by some small
distance the other end of the thread that you are holding in your hand each time
the swinging bob passes through the middle position, and release the thread to
its previous length each time the bob reaches its extreme positions. These peri-
odic variations of the pendulum’s length with the frequency twice the frequency
of natural oscillation cause the amplitude to increase progressively.
A remarkable description of an exotic example illustrating this mode of para-
metric excitation can be found in [1], p. 27. In Spain, in the cathedral of a northern
town Santiago de Compostela, there is a famous O Botafumeiro, a very large in-
cense burner suspended by a long rope, which can swing through a huge arc.
The censer is pumped by periodically shortening and lengthening the rope as it is
wound up and then down around the rollers supported high above the floor of the
nave. The pumping action is carried out by a squad of priests, called tiraboleiros,
or ball swingers, each holding a rope that is a strand of the main rope that goes
from the pendulum to the rollers and back down to near the floor. The tiraboleiros
periodically pull on their respective ropes in response to orders from the chief
verger of the cathedral. One of the more terrifying aspects of the pendulum’s mo-
tion is the fact that the amplitude of its swing is very large, and it passes through
the bottom of its arc with a high velocity, spewing smoke and flames.
In this chapter we consider a pendulum with modulated length that can swing
in the vertical plane in the uniform gravitational field (Figure 9.1). To allow ar-
bitrarily large swinging and even full revolutions, we assume that the pendulum
consists of a rigid massless rod (rather than a flexible string) with a massive small
bob on its end. The effective length of the pendulum can be changed by shifting
the bob up and down along this rod. Periodic modulation of the effective length by
such mass redistribution can cause, under certain conditions, a growth of initially
small natural oscillations. This phenomenon is called parametric resonance.

9.1.1 The Square-Wave Modulation of the Pendulum Length

In this chapter we are concerned with a periodic square-wave (piecewise constant)
modulation of the pendulum length. The square-wave modulation provides an al-
ternative (compared to the sinusoidal modulation) and may be a more straightfor-
ward way to understand and describe quantitatively the phenomenon of parametric
resonance. A relevant computer program [27] developed by the author simulates
such a physical system and aids greatly in investigating the phenomenon.
In the case of the square-wave modulation, abrupt, almost instantaneous incre-
ments and decrements in the length of the pendulum occur sequentially, separated

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

9.1. THE INVESTIGATED PHYSICAL SYSTEM 215

Figure 9.1: Modulation of the moment of inertia by periodic displacements of the

massive bob up and down along the rod of the pendulum.

by equal time intervals. We denote these intervals by T /2, so that T equals the
period of the length variation (the period of modulation). It is easy to understand
how the square-wave modulation can produce considerable oscillation of the pen-
dulum if the period and phase of modulation are chosen properly.
For example, suppose that the bob is shifted up (toward the axis) at an instant
at which the pendulum passes through the lower equilibrium position, when its an-
gular velocity reaches a maximum value. While the weight is moved radially, the
angular momentum of the pendulum with respect to the pivot remains constant.
Thus the resulting reduction in the moment of inertia is accompanied by an incre-
ment in the angular velocity, and the pendulum gets additional energy. The greater
the angular velocity, the greater the increment in energy. This additional energy is
supplied by the source that moves the bob along the rod of the pendulum.
On the other hand, if the bob is instantly moved down along the rod of the
swinging pendulum, the angular velocity and the energy of the pendulum dimin-
ish. The decrease in energy is transferred back to the source. In order that incre-
ments in energy occur regularly and exceed the amounts of energy returned, i.e., in
order that, as a whole, the modulation of the length regularly feeds the pendulum
with energy, the period and phase of modulation must satisfy certain conditions.
In particular, the greatest growth of the amplitude occurs if effective length of
the pendulum is reduced each time the pendulum crosses the equilibrium position,
and is increased back at greatest elongations, when the angular velocity is almost
zero. Therefore this radial displacement of the bob into its former position causes
nearly no decrement in the kinetic energy. The resonant growth of the amplitude
occurs if two cycles of modulation are executed during one period of natural os-
cillations. This is the principal parametric resonance. The time history of such
oscillations for the case of a very weak friction (Q = 1500) is shown in Figure 9.2
together with the square-wave variation of the pendulum length.
In a real system the growth of the amplitude at parametric resonance is re-
stricted by nonlinear effects. In a nonlinear system like the pendulum, the natural

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

216 CHAPTER 9. PENDULUM WITH MODULATED LENGTH

o
103
j (t)
-D l
0

-103o
0 5 10 15 20 25 30 35 40T
Period of modulation T0 = T /2, depth of modulation ml = 10%, quality factor Q = 1500

Figure 9.2: Initial exponential growth of the amplitude of oscillations at parametric

resonance of the first order (n = 1) under the square-wave modulation, followed
by beats.

period depends on the amplitude of oscillations. As the amplitude grows, the nat-
ural period of the pendulum becomes longer. However, in the accepted model the
drive period (period of modulation) remains constant. If conditions for paramet-
ric excitation are fulfilled at small oscillations and the amplitude is growing, the
conditions of resonance become violated at large amplitudes—the drive slips out
of resonance. The drive will then drift out of phase with the pendulum. The phase
relationships between the modulation and oscillations of the pendulum change
gradually to those favorable for the backward transfer of energy from the pendu-
lum to the source of modulation. This causes gradual reduction of the amplitude.
The natural period becomes shorter, and conditions for the growth of the ampli-
tude restore. Oscillations of the pendulum acquire the character of beats, as shown
in Figure 9.2. Due to friction these transient beats gradually fade, and the ampli-
tude tends to a finite constant value.
Details of the process of resonant growth followed by a nonlinear restriction
of the amplitude for a parametrically excited pendulum (T = T0 /2) with con-
siderable values of the modulation depth and friction (ml =15%, Q = 5.0) are
shown in Figure 9.3. The vertical segments of the phase trajectory and of the
ϕ̇(t) graph correspond to instantaneous increments and decrements of the angular
velocity ϕ̇ at the instants at which the bob is shifted up and down, respectively.
The curved portions of the phase trajectory that spiral in toward the origin cor-
respond to damped natural motions of the pendulum between the jumps of the
bob. The initially fast growth of the amplitude (described by the expanding part of
the phase trajectory) gradually slows down, because the natural period becomes
longer. After reaching the maximum value of 78.3◦, the amplitude alternatively
decreases and increases within a small range, slowly approaching its final value of
about 74◦ . The initially unwinding spiral of the phase trajectory simultaneously
approaches the closed limit cycle, whose characteristic shape can be seen in the
left panel of Figure 9.3.
It is evident that the energy of the pendulum is increased not only when two

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

9.1. THE INVESTIGATED PHYSICAL SYSTEM 217

1.68 .
ϕ ( t)

−1.68 o
−78.3 0 78.3
ϕ (t)
-D l
0

o
−78.3
0 2 4 6 8 10 12 14 16 18 20 T
Period of modulation T = T0 /2, depth of modulation m = 15%, quality Q = 5.0

Figure 9.3: The phase diagram (ϕ − ϕ̇ plane) and time-dependent graphs of an-
gular velocity ϕ̇(t) and angle ϕ(t) for the process of resonant growth followed by
nonlinear restriction of the amplitude.

full cycles of variation in the parameter occur during one natural period of oscil-
lation, but also when two cycles occur during three, five, or any odd number of
natural periods (resonances of odd orders). We shall see later that the delivery of
energy, though less efficient, is also possible if two cycles of modulation occur
during an even number of natural periods (resonances of even orders).

9.1.2 Conditions and Peculiarities of Parametric Resonance

There are several important differences that distinguish parametric resonance from
the ordinary resonance caused by an external force exerted directly on the system.
Variations of the length cannot take a resting pendulum out of equilibrium: In
contrast to the direct forcing, parametric excitation can occur only if (even small)
natural oscillations already exist. Parametric resonance is possible when one of
the following conditions for the frequency ω (or for the period T ) of a parameter
modulation is fulfilled:
2ω0 nT0
ω = ωn = , T = Tn = , n = 1, 2, . . . (9.1)
n 2
Parametric resonance is possible not only at the frequencies ωn given in Eq. (9.1),
but also in ranges of frequencies ω lying on either side of the values ωn (in the
ranges of instability). These intervals become wider as the depth of modulation is
increased.
An important difference between parametric excitation and forced oscillations
is related to the dependence of the growth of energy on the energy already stored

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

218 CHAPTER 9. PENDULUM WITH MODULATED LENGTH

in the system. While for a direct forced excitation the increment in energy during
one period is proportional to the amplitude of oscillations, i.e., to the square root
of the energy, at parametric resonance the increment in energy is proportional to
the energy itself, stored in the system.
Energy losses caused by friction are also proportional to the energy already
stored. In the case of direct forced excitation, energy losses restrict the growth of
the amplitude because these losses grow with the energy faster than does the in-
vestment in energy arising from the work done by the external force. In the case of
parametric resonance, both the investment in energy caused by the modulation of
a parameter and the frictional losses are proportional to the energy stored, and so
their ratio does not depend on the amplitude. Therefore, parametric resonance is
possible only when a threshold is exceeded, that is, when the increment in energy
during a period (caused by the parameter variation) is larger than the amount of
energy dissipated during the same time. The critical (threshold) value of the mod-
ulation depth depends on friction. However, if the threshold is exceeded, the fric-
tional losses of energy cannot restrict the growth of the amplitude. With friction,
stationary oscillations of a finite amplitude eventually establish due to nonlinear
properties of the pendulum.

9.2 The Threshold of Parametric Excitation

9.2.1 The Energy Supplied by the Square-Wave Modulation
We can use arguments employing the conservation laws to evaluate the modulation
depth that corresponds to the threshold of the principal parametric resonance. Let
the changes in the length l of the pendulum occur between l1 = l0 (1 + ml )
and l2 = l0 (1 − ml ), where ml is the dimensionless depth of modulation (or
modulation index). To calculate the change in total energy of the pendulum during
a period, we should not worry about the potential energy. Indeed, after a period
the pendulum occurs again in the vertical position with the bob at the same height,
hence after a period its potential energy is the same. Thus we should calculate only
the change in kinetic energy.
Next we calculate the fractional increment in energy ∆E/E during one cycle
of modulation, namely, between two consecutive passages through the equilibrium
position in opposite directions. At the first passage, the energy E1 equals v12 /2 per
unit mass of the bob, where v1 is the bob’s velocity. At this time moment the bob
is shifted up, so the length of the pendulum changes from l0 (1+ml ) to l0 (1−ml ).
During abrupt radial displacements of the bob along the pendulum rod, the angular
momentum L = J ϕ̇ = M l2 ϕ̇ is conserved (M is mass of the bob, J = M l2 is the
moment of inertia about the pivot). Therefore the angular velocity changes at this
moment from ϕ̇1 to (1 + ml )2 /(1 − ml )2 ϕ̇1 . This means that the linear velocity
v of the bob changes from v1 = l0 (1 + ml )ϕ̇1 to l0 (1 + ml )/(1 − ml ) v1 . Then
the pendulum moves from the vertical ϕ = 0 up to the maximum deflection ϕm ,

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

9.2. THE THRESHOLD OF PARAMETRIC EXCITATION 219

whose value can be calculated using the energy conservation:

 2
1 2 1 + ml
v = g l0 (1 − ml )(1 − cos ϕm ). (9.2)
2 1 1 − ml

When the frequency and phase of the modulation have those values that are favor-
able for the most effective delivery of energy to the pendulum, the abrupt back-
ward displacement of the bob toward the end of the rod occurs at the instant when
the pendulum attains its greatest deflection (more precisely, when the pendulum is
very near it). At this instant the angular velocity of the pendulum is almost zero.
Hence this action produces no change in the kinetic energy. At this time moment
the bob is shifted down, and the length of the pendulum becomes l0 (1 + ml ). The
pendulum starts its backward motion with zero velocity. Velocity v2 in the equi-
librium position, which is gained during this motion, again can be calculated, like
in Eq. (9.2), on the basis of energy conservation:
1 2
v = g l0 (1 + ml )(1 − cos ϕm ). (9.3)
2 2
From Eqs.(9.2)–(9.3) we find:
 3  3
1 + ml 1 + ml
v22 = v12 , E2 = E1 , (9.4)
1 − ml 1 − ml

where E2 = v22 /2 is the kinetic energy (per unit mass) after a period T of modu-
lation. Hence
 3
∆E E2 1 + ml
= −1= − 1 ≈ 6 ml . (9.5)
E E1 1 − ml

The last approximate expression in Eq. (9.5) is valid for small values of the mod-
ulation depth ml ≪ 1. That is, the fractional increment of total energy ∆E/E
during one period T of modulation approximately equals 6ml . The sequence of
energy values En at consecutive passages through the equilibrium position forms
a geometric progression. A process in which the increment in energy ∆E during a
period is proportional to the energy E stored (dE/dt ≈ 6ml E/T ) is characterized
on average by the exponential growth of the energy with time:
6ml
E(t) = E0 exp( t) = E0 exp(αt). (9.6)
T
In this case of tuning to the principal resonance, the index of growth α is propor-
tional to the depth of modulation ml of the pendulum length: α = 6ml /T .

9.2.2 Regime of Parametric Regeneration

When the modulation is exactly tuned to the principal resonance (T = T0 /2), the
decrease of energy is caused almost only by friction. Dissipation of energy due to

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

220 CHAPTER 9. PENDULUM WITH MODULATED LENGTH

.
ϕ( t )
-D l

ϕ( t )
T = 0.5 T0
ml = 10%

Q = 5.2
0 1 2 3 4T

Figure 9.4: The phase trajectory (left) and the time-dependent graphs of stationary
oscillations (right) at the threshold condition ml ≈ π/(6Q) for T = T0 /2. The
square-wave modulation of the pendulum length is also shown.

viscous friction during an integer number of natural half-cycles (for t = nT =

nT0 /2) is described by the following expression:
E(t) = E0 exp(−2γt). (9.7)
Comparing equations (9.6) and (9.7), we obtain the following estimate for the
threshold (minimal) value (ml )min of the depth of modulation corresponding to
the excitation of the principal parametric resonance:
1 1 π
(ml )min = γT = γT0 = . (9.8)
3 6 6Q
Here we introduced the dimensionless quality factor Q = ω0 /(2γ) to characterize
viscous friction in the system.
The phase trajectory and the plots of time dependence of the angle and an-
gular velocity of parametric oscillations of a small amplitude occurring at the
threshold conditions, Eq. (9.8), are shown in Figure 9.4. We can see on the graphs
and the phase trajectory only abrupt increments in the magnitude of the angular
velocity occurring twice during the period of oscillation (when the bob is shifted
upward). The downward shifts of the bob occur at instants when the angular ve-
locity is almost zero. Therefore the corresponding decrements in velocity are too
small to be visible on the graphs. This mode of steady oscillations (which have
a constant amplitude in spite of the dissipation of energy) is called parametric
regeneration. Computer simulations show that regime of parametric regeneration
is stable with respect to small variations in initial conditions: At different initial
conditions the phase trajectory and graphs acquire after a while the same charac-
teristic shape. However, this regime is unstable with respect to variations of the
pendulum parameters. If the friction is slightly greater or the depth of modulation
slightly smaller than Eq. (9.8) requires, oscillations gradually damp in spite of the
modulation. Otherwise, the amplitude grows.

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

9.2. THE THRESHOLD OF PARAMETRIC EXCITATION 221

o
64.0
j (t)

o
-64.0
0 10 20 30 40 50 60 70 80 T
Period of modulation T = T0 /2, depth of modulation ml = 5%, quality factor Q = 50

Figure 9.5: Gradually fading beats of the amplitude of oscillations at parametric

resonance of the first order (n = 1).

For the third resonance (T = 3T0 /2) the threshold value of the depth of
modulation is three times greater than its value for the principal resonance:
(ml )min = π/(2Q). In this instance two cycles of the parametric variation oc-
cur during three full periods of natural oscillations. Radial displacements of the
pendulum bob again happen at the time moments most favorable for pumping the
pendulum—up at the equilibrium position, and down at the extreme positions. The
same investment in energy occurs during an interval that is three times longer than
the interval for the principal resonance.

9.2.3 Transients over the Threshold

When the depth of modulation exceeds the threshold value, the energy of ini-
tially small oscillations during the first stage increases exponentially with time.
For the principal parametric resonance this initial growth is shown in Figure 9.5.
The growth of the energy again is described by Eq. (9.6). However, now the index
of growth α is determined by the amount by which the energy delivered through
parametric modulation exceeds the simultaneous losses of energy caused by fric-
tion: α = 6ml /T − 2γ. If the swing is small enough, the energy is proportional
to the square of the amplitude. Hence the amplitude of parametrically excited os-
cillations initially also increases exponentially with time: a(t) = a0 exp(βt). The
index β in the growth of amplitude is one half the index of the growth in energy.
For the principal resonance, when the investment in energy occurs twice during
one natural period of oscillation, we have β = 3ml /T − γ = 6ml /T0 − γ =
3ml ω0 /π − γ.
If the threshold is exceeded, the amplitude grows, conditions of resonance are
violated, and this causes a gradual reduction of the amplitude. Then the natural
period becomes shorter, and oscillations of the pendulum acquire the character of
beats. Due to friction, these transient beats gradually fade, and eventually steady-
state oscillations of a finite amplitude establish (Figure 9.5). We note again that
the growth of amplitude is restricted by nonlinear properties of the pendulum,
namely, by the dependence of the natural period T0 on the amplitude. For small

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

222 CHAPTER 9. PENDULUM WITH MODULATED LENGTH

o
180

j (t) -D l

o
-180
0 5 10 15 20 25 30 35 40T
Period of modulation T = T0 /2 and feedback, depth of modulation ml = 5%, quality factor Q = 50

Figure 9.6: Parametric pumping of the pendulum with the usage of a feedback
loop that provides the most effective delivery of energy to the pendulum.

and moderate values of the amplitude ϕm this dependence is approximately given

by T0 (ϕm ) ≈ Tsmall (1 + ϕ2m /16), where Tsmall is the period of infinitely small
natural oscillations. In contrast with the ordinary resonance caused by direct pe-
riodic forcing in a linear isochronous system, friction alone cannot restrict the
growth of the amplitude at parametric resonance. In an idealized linear system
the amplitude of parametric oscillations over the threshold grows indefinitely (see
[28]–[29]).

9.2.4 Parametric Swinging with the Feedback

In the above analysis we assumed that the period T of modulation remains the
same as the amplitude increases. At exact tuning to the principal resonance this
period equals T0 /2, where T0 is the period of small natural oscillations. When
we apply the model of the pendulum with modulated length for explaining the
pumping of a playground swing, we should take into account that the child on the
swing may notice the lengthening of the natural period as the amplitude increases,
and can react correspondingly, increasing the period of pumping to stay in phase
with the swing.
This intuitive reaction may be considered as a kind of feedback loop: The
child determines the time instants to squat and to stand depending on the actual
position of the swing. We can include this feedback loop in our model by requiring
that instantaneous upward shifts of the bob of the pendulum occur exactly at the
time moments, at which the pendulum crosses the equilibrium position, and that
backward shifts of the bob to the end of the rod occur exactly at extreme positions
of the pendulum. Such manipulations provide the optimal control for the most
effective and rapid pumping.
Figure 9.6 shows the graph of progressively growing oscillations occurring un-
der this optimal control with a feedback. Initially the period of modulation T sat-
isfies conditions of the principal parametric resonance at small swing (T = T0 /2).
We note how the period T of the square-wave modulation increases with the am-

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

9.3. AUTORESONANCE, BIFURCATIONS, MULTISTABILITY 223

plitude due to the feedback. After the amplitude reaches 180◦, the pendulum exe-
cutes full revolutions.
Certainly, the priests that pump O Botafumeiro also use the feedback. They
gradually increase the period of modulation as the amplitude grows, and then
probably reduce the depth of modulation to the level sufficient to compensate for
frictional losses and to maintain the desirable swing.

9.3 Parametric Autoresonance, Bifurcations,

Multistability
9.3.1 Autoresonance
Is it possible to excite large oscillations of the pendulum at a small excess of
the drive over the threshold without the feedback, that is, without appropriately
adjusting the period and phase of modulation as the amplitude grows? It occurs
that under certain conditions a spontaneous phase locking between the drive and
the pendulum motion becomes possible: The pendulum can automatically adjust
its amplitude to stay matched with the drive. By sweeping the drive period ap-
propriately, we can control the amplitude of the pendulum. This phenomenon is
called autoresonance. Autoresonance allows us to both excite and control a large
resonant response in nonlinear systems by a small forcing.
We can start pumping the pendulum by modulating its length with period
T = T0 /2, which corresponds to resonant condition at an infinitely small swing.
Then, in the process of oscillations, we slowly increase the period of modulation.
This can be done in small steps. After each increment of the period we wait for
a while so that transients almost fade away. During this time the amplitude in-
creases just to the amount that provides matching of an increased natural period
of the pendulum with the new period of modulation. Thus in each step of this
sweeping the pendulum remains locked in phase with the drive.

9.3.2 Bifurcations of Symmetry Braking and Period Doubling

To illustrate the phenomenon of parametric autoresonance in a computer simula-
tion (Figure 9.7), we choose the following values for the pendulum parameters:
Depth of modulation 5%, quality factor Q = 20. When the period of modulation is
gradually increased from T = 0.5T0 up to T = 0.9T0 , the pendulum swings with
an amplitude of 153◦ . At T = 0.90T0 a bifurcation of symmetry breaking occurs:
The pendulum swings to one side through an angle of 161◦ , while its excursion to
the other side is only 146◦.
This asymmetry in the swing increases up to T = 0.913T0, when a bifurca-
tion of period doubling occurs: During two cycles of modulation the pendulum
executes one asymmetric oscillation between the values 161.5◦ and 146.9◦, while
during the next two cycles the pendulum swings between 161.5◦ and 144.2◦. We

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

224 CHAPTER 9. PENDULUM WITH MODULATED LENGTH

2.13 .
ϕ ( t)

−2.13 ο ο
−161 0 161
ϕ (t)

0
-D l
ο
− 161
0 1 2 3 4T
Period of modulation T = 0.913 T0, depth of modulation m = 5%, quality Q = 20,
ο
initial deflection 142 , initial angular velocity 0.308ω0

Figure 9.7: Bifurcation of period doubling in parametric autoresonance. Phase

diagram (left) and time-dependent graphs of the angular velocity ϕ̇(t) and angle
of deflection ϕ(t) from the equilibrium position (right).

note that the latter elongation is slightly smaller than in the preceding cycle. Then
the process repeats.
Thus one period of the pendulum motion now covers four periods of exci-
tation. These oscillations are illustrated in Figure 9.7. We note that the closed
phase trajectory is formed by two nearby almost merging loops. Such asymmetric
regimes exist (for the same values of ml and Q) in pairs, whose phase orbits are
mirror images of one another.
Further increasing of the drive period by tiny steps causes a whole condensing
cascade of nearby period doubling bifurcations, which ends at T = 0.9148T0
by a crisis: Oscillations of the pendulum become unstable, finally it turns over
the upper equilibrium, and then, after long irregular transient oscillations with
gradually diminishing amplitude, the pendulum eventually comes to rest in the
downward vertical position.
Stationary parametric oscillations of the pendulum with large amplitude that
are locked in phase with the drive and occur at a rather small or moderate modula-
tion (like those described above and shown in Figure 9.7), can be excited not only
by slowly sweeping the drive period, but also by appropriate initial conditions.
The system eventually comes to a certain periodic regime (limit cycle, or attrac-
tor), if initial conditions are chosen within the basin of attraction of this regime.
In nonlinear systems different periodic regimes may coexist at the same values of
parameters. This property is called multistability.
An example of multistability is shown in Figure 9.8. Curve 1 (upper side of
Figure 9.8) describes stationary periodic oscillations of the pendulum with a fi-
nite amplitude corresponding to the principal parametric resonance. One period
of these oscillations covers two cycles of excitation. Curves 2 and 3 (lower side

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

9.4. THEORY OF PARAMETRIC EXCITATION 225

2.70 .
ϕ ( t)
1
0
1

−2.70
2.70 3 3
.
ϕ (t)
0

-D l
−2.70 ο
2 ο
2
−180 0 180 0 1 2 3 4T
Period of modulation T = 0.56T0, depth of modulation m = 15%, quality Q = 20

Figure 9.8: Stationary periodic oscillations and rotations, occurring at the same
values of the system parameters.

of Figure 9.8) correspond to period-1 unidirectional rotations of the pendulum

in clockwise and counterclockwise directions, respectively. The pendulum makes
one revolution during each period of modulation. One more attractor is repre-
sented by a single point at the origin of the phase plane, which describes the state
of rest of the pendulum in the downward vertical position. Each of these different
stationary modes, coexisting at the same values of all parameters of the pendulum
and the drive, is characterized by a certain basin of attraction in the phase plane of
initial states.

9.4 Quantitative Theory of Parametric Excitation

9.4.1 Differential Equation for Parametric Oscillations
Next we consider a more rigorous mathematical treatment of parametric reso-
nance under square-wave modulation of the parameter. During the time intervals
(0, T /2) and (T /2, T ), the length of the pendulum is constant, and its motion
can be considered as a free oscillation described by a corresponding differential
equation. However, the coefficients in this equation are different for the adjacent
time intervals (0, T /2) and (T /2, T ):

ϕ̈ + 2γ ϕ̇ + ω12 sin ϕ = 0, ω1 = √ ω0
1+m
for 0 < t < T /2, (9.9)
ϕ̈ + 2γ ϕ̇ + ω22 sin ϕ = 0, ω2 = √ ω0
1−m
for − T /2 < t < 0. (9.10)
p
Here ω0 = g/l0 is the natural frequency of small oscillations for the pendulum
with mean length l0 , and γ is the damping constant characterizing the strength
of viscous friction. For a slow pendulum traveling in air, the linear dependence

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

226 CHAPTER 9. PENDULUM WITH MODULATED LENGTH

of drag on velocity is a reasonable approximation. When damping is caused by

the drag force exerted on the pendulum bob, and this force is proportional to the
linear velocity of the bob, the frictional torque about the pivot is proportional to
l2 . Since the moment of inertia is also proportional to l2 , the damping constant γ
in this model remains the same when the length of the pendulum changes, that is,
its values in Eqs. (9.9) and (9.10) are equal.
At each instant tn = nT /2 (n = 1, 2, . . . ) of an abrupt change in the
length of the pendulum, we must make a transition from one of these equations
(9.9)–(9.10) to the other. During each half-period T /2 the motion of the pendulum
is a segment of some natural oscillation. An analytical investigation of parametric
excitation can be carried out by fitting to one another known solutions to equations
(9.9)–(9.10) for consecutive adjacent time intervals.
The initial conditions for each subsequent time interval are chosen accord-
ing to the physical model in the following way. Each initial value of the angular
displacement ϕ equals the value ϕ(t) reached by the oscillator at the end of the
preceding time interval. The initial value of the angular velocity ϕ̇ is related to the
angular velocity at the end of the preceding time interval by the law of conserva-
tion of the angular momentum:

(1 + ml )2 ϕ̇1 = (1 − ml )2 ϕ̇2 . (9.11)

In Eq. (9.11), ϕ̇1 is the angular velocity at the end of the preceding time interval,
when the moment of inertia of the pendulum has the value J1 = J0 (1 + ml )2 , and
ϕ̇2 is the initial value for the following time interval, during which the moment of
inertia equals J2 = J0 (1 − ml )2 . The change in the angular velocity at an abrupt
variation of the inertia moment from the value J2 to J1 can be found in the same
way.
We may use here the conservation of angular momentum, as expressed in
Eq. (9.11), because at sufficiently rapid displacement of the bob along the rod
of the pendulum, the influence of the torque produced by the force of gravity is
negligible. In other words, we can assume the pendulum to be freely rotating about
its axis. This assumption is valid provided the duration of the displacement of the
bob constitutes a small portion of the natural period.
Considering conditions for which equations (9.9)–(9.10) yield solutions with
increasing amplitudes, we can determine the ranges of frequency ω near the val-
ues ωn = 2ω0 /n, within which the state of rest is unstable for a given modulation
depth ml . In these ranges of parametric instability an arbitrarily small deflection
from equilibrium is sufficient for the progressive growth of small initial oscilla-
tions.

9.4.2 The Mean Natural Period at Large Depth of Modulation

The threshold for the parametric excitation of the pendulum is determined above
for the resonant situations in which two cycles of the parametric modulation oc-
cur during one natural period or during three natural periods of oscillation. The

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

9.5. RANGES FOR PARAMETRIC RESONANCE 227

estimate obtained, Eq. (9.8), is valid for small values of the modulation depth ml
of the pendulum length.
For large values of the modulation depth ml , the notionp of a natural period
needs a more precise definition. Let T0 = 2π/ω0 = 2π l0 /g be the period of
oscillation of the pendulum when its massive bob is fixed in the middle position,
for which the effective length equals l0 . The period
√ is somewhat longer when the
weight is moved further from the axis: T1 = T0 1 + ml ≈ T0 (1 + √ ml /2). The
period is shorter when the weight is moved closer to the axis: T2 = T0 1 − ml ≈
T0 (1 − ml /2).
It is convenient to define the natural average period Tav not as the arithmetic
mean 12 (T1 + T2 ), but rather as the period that corresponds to the arithmetic mean
frequency ωav = 12 (ω1 + ω2 ), where ω1 = 2π/T1 and ω2 = 2π/T2 . So we define
Tav by the relation:
2π 2T1 T2
Tav = = . (9.12)
ωav (T1 + T2 )
Indeed, the period T of the parametric modulation that is exactly tuned to any of
the parametric resonances is determined not only by the order n of the resonance,
but also by the depth of modulation ml . In order to satisfy the resonant conditions,
the increment in the phase of natural oscillations during one cycle of modulation
must be equal to π, 2π, 3π, . . . , nπ, . . . . During the first half-cycle the phase
of oscillation increases by ω1 T /2, and during the second half-cycle by ω2 T /2.
Consequently, instead of the approximate condition expressed by Eq. (9.1), we
obtain:
ω1 + ω2 π Tav
T = nπ, or T = Tn = n =n . (9.13)
2 ωav 2
Thus, for a parametric resonance of some definite order n, the condition for exact
tuning can be expressed in terms of the two natural periods, T1 and T2 . This con-
dition is T = nTav /2, where Tav is defined by Eq. (9.12). For small and moderate
values of ml it is possible to use approximate expressions for the average natural
frequency and period:
 
ω0 1 1 3 3
ωav = √ +√ ≈ ω0 (1 + m2l ), Tav ≈ T0 (1 − m2l ).
2 1 + ml 1 − ml 8 8
(9.14)
The difference between Tav and T0 reveals itself in terms proportional to the
square of the depth of modulation ml .

9.5 Frequency Ranges for Parametric Resonance

9.5.1 Resonances of Odd Orders
To find the boundaries of the frequency ranges of parametric instability surround-
ing the resonant values T = Tav /2, T = Tav , T = 3Tav /2, . . . , we can consider
stationary oscillations of indefinitely small amplitude that occur when the period

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

228 CHAPTER 9. PENDULUM WITH MODULATED LENGTH

.
ϕ (t )

ϕ (t )
ml = 10% -D l

T = 0.4507 T0

-1 0 1 2 3T

Figure 9.9: Phase trajectory and time-dependent graphs of stationary parametric

oscillations at the lower boundary of the principal interval of instability (the period
of modulation T near Tav /2).

of modulation T corresponds to one of the boundaries. These periodic stationary

oscillations can be represented as an alternation of natural oscillations with the
periods T1 and T2 .

9.5.2 Main Interval of Parametric Instability

We examine first the vicinity of the principal resonance occurring at T = Tav /2.
Suppose that the period T of the parametric square-wave modulation is a little
shorter than the resonant value T = Tav /2, so that T corresponds to the left
boundary of the interval of instability. In this case a little less than a quarter of
the mean natural period Tav elapses between consecutive abrupt increases and de-
creases of the pendulum length. Stationary regime with a constant swing in the
absence of friction can be realized only if the abrupt increments and decrements
of the angular velocity are equal in magnitude. The graphs of the angle ϕ(t) and
angular velocity ϕ̇(t) for this periodic stationary process have the characteristic
symmetric patterns shown in Figure 9.9. The segments of the graphs of free oscil-
lations (which occur within time intervals during which the length of the pendu-
lum is constant) are alternating parts of sine or cosine curves with the periods T1
and T2 . These segments are symmetrically truncated on both sides.
To find conditions at which such stationary oscillations take place, we can
write the expressions for ϕ(t) and ϕ̇(t) during the adjacent intervals in which
the oscillator executes natural oscillations, and then fit these expressions to one
another at the boundaries. Such fitting must provide a periodic stationary process.
We let the origin of time, t = 0, be the instant when the bob is shifted down-
ward. The angular velocity is abruptly decreased in magnitude at this instant (see
Figure 9.9). Then during the interval (0, T /2) the graph describes a natural oscil-

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

9.5. RANGES FOR PARAMETRIC RESONANCE 229

√
lation with the frequency ω1 = ω0 / 1 + m. Since the graph is symmetric with
respect to time moment T /4, we can write the corresponding time dependencies
of ϕ(t) and ϕ̇(t) in the following form:

ϕ1 (t) = −A1 cos ω1 (t−T /4), ϕ̇1 (t) = A1 ω1 sin ω1 (t−T /4), 0 < t < T /2.
(9.15)
Similarly, during the interval (−T /2, 0) the graph
√ in Figure 9.9 is a segment of
natural oscillation with the frequency ω2 = ω0 / 1 − m:

ϕ2 (t) = −A2 sin ω2 (t+T /4), ϕ̇2 (t) = −A2 ω2 cos ω2 (t+T /4), −T /2 < t < 0.
(9.16)
To determine the values of constants A1 and A2 , we use the conditions that
must be satisfied when the segments of the graph are joined together, and take into
account the periodicity of the stationary process. At t = 0 the angle of deflection
is the same for both ϕ1 and ϕ2 , that is, ϕ1 (0) = ϕ2 (0). The angular velocity
at t = 0 undergoes a sudden change, which follows from the conservation of
angular momentum: (1 + ml )2 ϕ̇1 (0) = (1 − ml )2 ϕ̇2 (0), see Eq. (9.11). From
these conditions of fitting the graphs we find the following equations for A1 and
A2 :
A1 cos(ω1 T /4) = A2 sin(ω2 T /4). (9.17)
A1 (1 + ml )2 ω1 sin(ω1 T /4) = A2 (1 − ml )2 ω2 cos(ω2 T /4). (9.18)
These homogeneous equations (9.17)–(9.18) for A1 and A2 are compatible only
if the following condition is fulfilled:

(1 + ml )2 ω1 sin(ω1 T /4) sin(ω2 T /4) = (1 − ml )2 ω2 cos(ω1 T /4) cos(ω2 T /4).

(9.19)
This is the equation that determines period T of modulation (for a given value
ml of the depth of modulation) which corresponds to the left boundary of the
interval of parametric instability. Next we rearrange Eq. (9.19) to the following
form, which is more convenient for obtaining a numeric solution for the unknown
variable T :
(q + 1) cos(ωav T /2) = (q − 1) cos(∆ωT /4), (9.20)
where ωav = (ω1 + ω2 )/2, and ∆ω = ω2 − ω1 . In Eq. (9.20) we have introduced
a dimensionless quantity q which depends on the depth of modulation ml :
 3/2
1 + ml
q= . (9.21)
1 − ml
To find the left boundary T− of the instability interval that contains the princi-
pal parametric resonance, we search for a solution T to Eq. (9.20) in the vicinity
of T = T0 /2. We replace T in the argument of the cosine on the left-hand side
of Eq. (9.20) by Tav /2 + ∆T . Since ωav Tav = 2π, we can write the cosine as
− sin(ωav ∆T /2). Then Eq. (9.20) becomes:
q−1 ∆ω(Tav /2 + ∆T )
sin(ωav ∆T /2) = − cos . (9.22)
q+1 4

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

230 CHAPTER 9. PENDULUM WITH MODULATED LENGTH

m (%)

80

60

40

20

0.5 1.0 1.5 2.0 2.5 3.0

T/T0

Figure 9.10: Intervals of parametric instability at square-wave modulation of the

pendulum length in the absence of friction.

This equation for the unknown quantity ∆T can be solved numerically by itera-
tion. We start with ∆T = 0 as an approximation of the zeroth order, substituting
it into the right-hand side of Eq. (9.22). Then the left-hand side of Eq. (5.22) gives
us the value of ∆T to the first order. We substitute this first-order value into the
right-hand side of Eq. (9.22), and on the left-hand side we obtain ∆T to the sec-
ond order. This procedure is iterated until a self-consistent value of ∆T for the
left boundary is obtained. Performing such calculations for various values of the
modulation depth ml , we obtain the whole left boundary T− (ml ) for the first in-
terval of parametric instability. Below we explain how the right boundary of this
interval can be calculated, as well as the boundaries of other intervals.
The intervals of instability in the plane T — ml for the first six parametric
resonances, calculated numerically with the help of the above-described proce-
dure, are shown in Figure 9.10. This is an analog of the Incze-Strutt diagram of
parametric instability for a system that is described by the Mathieu equation, say,
for a pendulum with vertical oscillations of the suspension point.
To observe stationary oscillations that correspond to the left boundary of the
instability interval (see Figure 9.9) in the simulation, it is insufficient to choose for
period T of modulation a self-consistent solution to Eq. (9.22) for a given value
of modulation depth ml . After period T is calculated, the initial conditions should
also be chosen properly. This can be done on the basis of Eq. (9.15), according to
which for an arbitrary initial displacement ϕ(0) the initial angular velocity should
have the value ϕ̇1 (0) = ω1 tan(ω1 T /4)ϕ1 (0).
For the right boundary of the main interval of instability, the period T of
the parametric square-wave modulation is a little longer than the resonant value
T = Tav /2. In this case a little more than a quarter of the mean natural period
Tav elapses between consecutive abrupt increases and decreases of the pendulum
length. The graphs of the angle ϕ(t) and angular velocity ϕ̇(t) for this periodic
stationary process are shown in Figure 9.11.

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

9.5. RANGES FOR PARAMETRIC RESONANCE 231

.
ϕ (t )

ϕ (t )
ml = 10% -D l
T = 0.5455 T0

-1 0 1 2 3T

Figure 9.11: Stationary parametric oscillations at the upper boundary of the prin-
cipal interval of instability (near T = Tav /2).

We can write the corresponding time dependencies of ϕ(t) and ϕ̇(t) for the
time interval (0, T /2) in the following form:

ϕ1 (t) = B1 sin ω1 (t − T /4), ϕ̇1 (t) = B1 ω1 cos ω1 (t − T /4),

0 < t < T /2.
(9.23)
During the interval (−T /2, 0) the graph√ in Figure 9.11 is a segment of natural
oscillation with the frequency ω2 = ω0 / 1 − m:

ϕ2 (t) = −B2 cos ω2 (t + T /4), ϕ̇2 (t) = B2 ω2 sin ω2 (t + T /4), −T /2 < t < 0.
(9.24)
Further calculations are similar to those for the left boundary already described
after Eqs. (9.15)–(9.16). It occurs that ∆T for the right boundary is determined
as a solution to the equation that differs from Eq. (9.22) by the opposite sign
on its right-hand side. Solving it numerically by iterations for various values of
ml , we obtain the right boundary of the principal interval (n = 1) of parametric
instability, Figure 9.10.
To obtain approximate analytical solutions to Eq. (9.22) that are valid for small
values of the modulation depth ml , we can simplify the expression on its right-
hand side by assuming that q ≈ 1 + 3ml , q − 1 ≈ 3ml . We may also assume the
value of the cosine to be approximately 1. On the left-hand side of Eq. (9.22), the
sine can be replaced by its small argument, in which ωav = 2π/Tav . This yields
the following approximate expressions for both boundaries of the main interval
that are valid up to terms to the second order in ml :

3m2l
   
1 3ml 1 3ml
T∓ = 1∓ Tav = 1∓ − T0 . (9.25)
2 π 2 π 8

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

232 CHAPTER 9. PENDULUM WITH MODULATED LENGTH

.
8 5 ϕ (t) 5 8
7 6 6 7

2 3 2 3
1 4 1 4 1

ϕ ( t)
Left boundary: -D l
0
ml = 20%
T = 1.3928T0
−1 0 1 2 3T
. 1 8 1
1 8 ϕ(t)
2 7 2 7

3 6 3 6
4 5 4 5
ϕ ( t)
Right boundary:
-D l
ml = 20%

T = 1.5592T0
−1 0 1 2 3T

Figure 9.12: The phase trajectory and the graphs of the angular velocity and the
deflection angle of stationary parametric oscillations at the left and right bound-
aries of the interval of instability near T = 3Tav /2.

9.5.3 Third-Order Interval of Parametric Instability

The boundaries of the instability intervals that contain higher order parametric
resonances can be determined in a similar way. At the third order resonance (n =
3) two cycles of variation of the pendulum length occur during approximately
three natural periods of oscillation (T ≈ 3Tav /2). The phase trajectories and the
time-dependent graphs of stationary oscillations at the left and right boundaries
of the third interval are shown in Figure 9.12. The phase orbit of the periodic
oscillation closes after two cycles of modulation. This orbit is formed by two
concentric ellipses that correspond to small natural oscillations of the pendulum
with frequencies ω1 and ω2 . The representative point moves clockwise along this
orbit, jumping from one ellipse to the other each time the bob is shifted along

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

9.5. RANGES FOR PARAMETRIC RESONANCE 233

the pendulum rod. The numbers in Figure 9.12 make it easier to follow how the
representative point describes this orbit: Equivalent points of the phase orbit and
the graph of angular velocity are marked by equal numbers.
Considering conditions at which the graphs of natural oscillations with fre-
quencies ω1 and ω2 on the left boundary fit one another for adjacent time intervals
and produce the periodic process shown in Figure 9.12, we get the same equa-
tions (9.17)–(9.18) for A1 and A2 , as well as Eq. (9.22) for the period of modula-
tion. Actually, this is true for all intervals of parametric instability of odd orders.
Similarly, for the right boundary we get the same equations for B1 and B2 as
in case n = 1, and also Eq. (9.22) with the opposite sign for determination of
the corresponding period of modulation T . However, if we are interested in the
third interval, we should search for a solution to these equations in the vicinity of
T = 3Tav /2, as well as for any other interval of odd order n in the vicinity of
T = nTav /2. The boundaries of intervals of the third and fifth orders, obtained by
a numerical solution, are also shown in Figure 9.10.
For small values of the depth of modulation ml , we can find approximate
analytical expressions for the lower and the upper boundaries of the third interval
that are valid up to quadratic terms in ml :
3m2l
 
3 ml  3 ml
T∓ = 1∓ Tav = 1∓ − T0 , ml ≪ 1. (9.26)
2 π 2 π 8
In this approximation, the third interval has the same width (3ml /π)T0 as does
the interval of instability in the vicinity of the principal resonance. However, this
interval is distinguished by greater asymmetry: Its central point is displaced to the
left of the value T = 32 T0 by 169
m2l T0 .

9.5.4 Parametric Resonances of Even Orders

For small and moderate square-wave modulation of the pendulum length, para-
metric resonance of the order n = 2 (one cycle of the modulation during one
natural period of oscillation) is relatively weak compared to the above-considered
resonances n = 1 and n = 3. In the case in which n = 2 the abrupt shifts of the
bob induce both an increase and a decrease of the energy only once during each
natural period. The growth of oscillations occurs only if the increase in energy at
the instant when the bob is shifted up is greater than the decrease in energy when
the bob is shifted down. This is possible only if the bob is shifted up when the
angular velocity of the pendulum is greater in magnitude than it is when the bob is
shifted down. For T ≈ Tav , these conditions can be fulfilled only because there is
a (small)√difference between the natural periods T1 and T2 of the pendulum,
√ where
T1 = T0 1 + ml is the period with the bob shifted down and T2 = T0 1 − ml
is the period with the bob shifted up. This difference in the natural periods is
proportional to ml .
The growth of oscillations at parametric resonance of the second order is
shown in Figure 9.13. We note the asymmetric character of oscillations at n = 2
resonance: The angular excursion of the pendulum to one side is greater than to

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

234 CHAPTER 9. PENDULUM WITH MODULATED LENGTH

1.03 .
ϕ (t )

−1.03 o o
− 54.3 0 54.3
-D l ϕ (t )
T = 0.985 T0
ml = 20% 0
Q = 50
o
− 54.3
0 2 4 6 8 10 12 T

Figure 9.13: The phase trajectory and the graphs of angular velocity ϕ̇(t) and
angle ϕ(t) of oscillations corresponding to parametric resonance of the second
order n = 2 (T = Tav ).

the other. In this case, the investment in energy during a period is proportional to
the square of the depth of modulation ml , while in the cases of resonances with
n = 1 and n = 3 the investment in energy is proportional to the first power of ml .
Therefore, for the same value of the damping constant γ (the same quality fac-
tor Q), a considerably greater depth of modulation is required here to exceed the
threshold of parametric excitation. The growth of the amplitude again is restricted
by the nonlinear properties of the pendulum.
The interval of instability in the vicinity of n = 2 resonance (for small values
of ml ) is considerably narrower compared to the corresponding intervals of n = 1
and n = 3 resonances. Its width is also proportional only to the square of ml .
To determine the boundaries of this interval of instability, we can consider, as
is done above for other resonances, stationary oscillations for T ≈ T0 formed by
alternating segments of free oscillations with the periods T1 and T2 . The phase
trajectory and the graphs of the angular velocity ϕ̇(t) and the angle ϕ(t) of such
stationary periodic oscillations for one of the boundaries are shown in Figure 9.14.
During oscillations occurring at the boundary of the instability interval, the abrupt
increment and decrement in the angular velocity exactly compensate each other.
To describe these stationary oscillations with small amplitude, we can use
the following expressions for ϕ(t) and ϕ̇(t) in the interval (0, −T /2) (see Fig-
ure 9.14):

ϕ1 (t) = −A1 cos ω1 (t−T /4), ϕ̇1 (t) = A1 ω1 sin ω1 (t−T /4), 0 < t < T /2,
(9.27)

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

9.5. RANGES FOR PARAMETRIC RESONANCE 235

.
ϕ ( t)

-D l ϕ(t )
T = 0.9336T0
ml = 25% 0

−1 0 1 2 3T

Figure 9.14: Stationary parametric oscillations at the left boundary of the interval
of instability of the second order n = 2 (near T = Tav ≈ T0 ).

and during the interval (−T /2, 0):

ϕ2 (t) = A2 cos ω2 (t + T /4), ϕ̇2 (t) = −A2 ω2 sin ω2 (t + T /4), −T /2 < t < 0.
(9.28)
The conditions for joining the graphs at t = 0 are the same as for other resonances,
namely, at t = 0 we require ϕ1 (0) = ϕ2 (0), and the angular velocity undergoes a
sudden change, which follows from the conservation of angular momentum (see
Eq. (9.11)). These conditions yield the following equations for A1 and A2 :

A1 cos(ω1 T /4) = −A2 cos(ω2 T /4), (9.29)

A1 (1 + ml )2 ω1 sin(ω1 T /4) = A2 (1 − ml )2 ω2 cos(ω2 T /4). (9.30)

These homogeneous equations (9.29)–(9.30) for A1 and A2 are compatible only
if the following condition is fulfilled:

(1 + ml )2 ω1 sin(ω1 T /4) cos(ω2 T /4) = −(1 − ml )2 ω2 sin(ω2 T /4) cos(ω1 T /4).

(9.31)
This is the equation that determines period T of modulation (for a given value ml
of the depth of modulation) which corresponds to the left boundary of the second
interval of parametric instability. We transform Eq. (9.31) to the following form,
which is convenient for a numeric solution by iteration:

(q + 1) sin(ωav T /2) = (q − 1) sin(∆ωT /4), (9.32)

where q depends on the depth of modulation ml according to Eq. (9.21). Next

we replace T in the argument of the sine on the left-hand side of Eq. (9.32) by
Tav + ∆T . Since ωav Tav = 2π, we can write this sine as − sin(ωav ∆T /2). Then

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

236 CHAPTER 9. PENDULUM WITH MODULATED LENGTH

Eq. (9.32) becomes:

q−1 ∆ω(Tav + ∆T )
sin(ωav ∆T /2) = − sin . (9.33)
q+1 4
This equation for ∆T can be solved numerically by iteration with the help of the
above-described procedure. Its self-consistent solutions for various values of the
modulation depth ml give the left boundary of the n = 2 instability interval. After
period T for this boundary is calculated, the initial conditions that provide station-
ary oscillations can be chosen on the basis of Eq. (9.27), according to which, for
an arbitrary initial displacement ϕ(0), the initial angular velocity should have the
value ϕ̇1 (0) = ω1 tan(ω1 T /4)ϕ1 (0).
The right boundary of the second interval is given by an equation, which dif-
fers from Eq. (9.33) by the opposite sign on its right-hand side. Both boundaries
are shown on the diagram in Figure 9.10 together with intervals of higher even
orders, which are obtained with the help of similar numeric calculations.
We note how the intervals of even resonances (n =2, 4, 6) are narrow at small
values of the modulation depth ml in contrast to the intervals of odd orders. With
the growth of ml the even intervals expand and become comparable with the in-
tervals of odd orders.
For small and moderate values of the depth of modulation ml ≪ 1, an approx-
imate analytical expression for both boundaries of the second interval of instability
can be found as a solution to Eq. (9.33) (and to an equation with the opposite sign
for the other boundary):
   
3 2 3 3
T∓ = 1 ∓ ml Tav = T0 + ∓ − m2l T0 , (9.34)
4 4 8
i.e., T− = T0 (1 − 98 m2 ), T+ = T0 (1 + 83 m2l ). As mentioned above, the width of
this interval of instability T+ − T− = 32 m2l T0 is proportional to the square of the
modulation depth.

9.5.5 Intersections of the Boundaries at Large Modulation

Figure 9.10 shows that at certain values of ml both boundaries of intervals with
n > 2 coincide (we may consider that they intersect). This means that at these
values of ml the corresponding intervals of parametric instability disappear. Such
values of ml correspond to the natural periods of oscillation T1 and T2 , whose
ratio is 2 : 1, 3 : 1, and 3 : 2.
For the first intersection (ratio 2 : 1) exactly one half of the natural oscillation
with period T1 is completed during the first half of the modulation cycle (see
Figure 9.15). On the phase diagram, the representing point traces a half of the
smaller ellipse (1 — 2), and then abruptly jumps down to the larger ellipse (2 —
3). During the second half of the modulation cycle the oscillator executes exactly
a whole natural oscillation with period T2 = T1 /2, so that the representing point
passes in the phase plane along the whole larger ellipse (3 — 4), and then jumps
up to the smaller ellipse along the same vertical segment (4 — 5).

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

9.6. INTERVALS IN THE PRESENCE OF FRICTION 237

.
8 ϕ ( t) 7 8
7
5 6 1 2 1
2 1 5 6

4
3 3 4
ϕ (t)

T = 1.26491 T0
ml= 60%

-D l
0 1 2 3 4T

Figure 9.15: The phase trajectory and time-dependent graphs of angular velocity
ϕ̇(t) and angle ϕ(t) for stationary oscillations at the intersection of both bound-
aries of the third interval.

During the next modulation cycle the representing point first generates the
other half of the smaller ellipse (5 — 6), and then again the whole larger ellipse
(7 — 8). Therefore during any two adjacent cycles of modulation the represent-
ing point passes once along the closed smaller ellipse and twice along the larger
one, returning finally to the initial point of the phase plane. We see that such an
oscillation is periodic for arbitrary initial conditions. This means that for the cor-
responding values of the modulation depth ml and the period of modulation T the
growth of amplitude is impossible even in the absence of friction (the instability
interval vanishes).
Similar explanations can be suggested for other cases in Figure 9.10 in which
the boundaries of the instability intervals intersect.

9.6 Intervals of Parametric Excitation in the

Presence of Friction
When there is friction in the system, the intervals of the period of modulation
that correspond to parametric instability become narrower, and for strong enough
friction (below the threshold) the intervals disappear. Above the threshold, ap-
proximate values for the boundaries of the firstpinterval are given by Eq. (9.25)
provided we substitute for ml the expression m2l − (ml )2min with the thresh-
old value (ml )min = π/(6Q) defined by Eq. (9.8). The proof can be found in
the following
p Section 9.6.1. For the third interval, we can use Eq. (9.26), substi-
tuting m2l − (ml )2min for ml , with (ml )min = π/(2Q). When ml is equal to
the corresponding threshold value (ml )min , the interval of parametric resonance

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

238 CHAPTER 9. PENDULUM WITH MODULATED LENGTH

ml (%)

80

60

40

20

T/T0
0.5 1.0 1.5

Figure 9.16: Intervals of parametric excitation at square-wave modulation of the

pendulum length without friction, for Q = 7, Q = 5, and for Q = 3.

disappears.
The boundaries of the second interval of parametric resonance in the presence
2
of friction are approximately
p given by Eq. (9.34) provided we substitutep for ml
the expression m4l − (ml )4min with the threshold value (ml )min = 2/(3Q),
which corresponds to the second parametric resonance (see Section 9.6.1).
The diagram in Figure 9.16 shows the boundaries of the first three intervals of
parametric resonance for Q = 3, Q = 5, and Q = 7 (and also in the absence of
friction). We note the “island” of parametric resonance of the third order (n = 3)
at Q = 7. This resonance disappears when the depth of modulation exceeds 48%
and reappears when ml exceeds approximately 66%.
In the presence of friction, for any given value ml of the depth of modulation,
only several first intervals of parametric resonance (where ml exceeds the thresh-
old) can exist. We note that in case the equilibrium of the system is unstable due to
modulation of the parameter, parametric resonance can occur only if at least small
oscillations are already excited. Indeed, when the initial values of ϕ and ϕ̇ are ex-
actly zero, they remain zero over the course of time. This behavior is in contrast to
that of resonance arising from direct forcing, when the amplitude increases with
time even if initially the system is at rest in the equilibrium position (if the initial
conditions are zero).

9.6.1 Boundaries of Instability for Resonances of Odd Orders

Stationary oscillations occurring at the left boundary of the instability interval in
the vicinity of the principal parametric resonance in the presence of friction are
shown in Figure 9.17 (compare with Figure 9.9). Twice during the full cycle of

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

9.6. INTERVALS IN THE PRESENCE OF FRICTION 239

.
ϕ ( t)

ϕ ( t)
ml = 10% -D l
T = 0.456 T0
Q = 10.0
-1 0 1 2 3T

Figure 9.17: Stationary oscillations in the presence of friction at the left boundary
of the principal instability interval.

modulation the angular velocity abruptly increases, and twice it decreases. The
increments are greater than the decrements, so that as a whole the energy received
by the pendulum exceeds the energy given away. This surplus compensates for
the dissipation of the energy that occurs at natural oscillation during the intervals
between the abrupt displacements of the bob along the rod of the pendulum.
To find conditions at which such stationary oscillations take place, we can
write the expressions for ϕ(t) and ϕ̇(t) during the adjacent intervals when the
pendulum executes damped natural oscillations, and then fit these expressions to
one another at the boundaries. Contrary to the frictionless pendulum (see Fig-
ure 9.9, p. 228), now the phase trajectory is not symmetric with respect to the
ordinate axis (Figure 9.17). We choose as the time origin t = 0 the instant when
the bob is shifted down, and the angular velocity decreases in magnitude. Then
during the interval (0, T /2)
√ the graph describes a damped natural oscillation with
the frequency ω1 = ω0 / 1 + ml . We can represent this motion as a superposi-
tion of damped natural oscillations of sine and cosine type with some constants
A1 and B1 :
ϕ1 (t) =(A1 sin ω1 t + B1 cos ω1 t) e−γt ,
(9.35)
ϕ̇1 (t) ≈(A1 ω1 cos ω1 t − B1 ω1 sin ω1 t) e−γt .
The latter expression for ϕ̇(t) is valid for relatively weak friction (γ ≪ ω0 ). To
obtain it, we differentiate ϕ(t) with respect to the time, considering the exponen-
tial factor e−γt to be approximately constant. Indeed, at weak damping the main
contribution to the time derivative originates from the oscillating factors sin ω1 t
and cos ω1 t in the expression for ϕ(t).
Similarly, during the interval (−T /2, 0) the graph in Figure 9.17 is a segment

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

240 CHAPTER 9. PENDULUM WITH MODULATED LENGTH

of damped natural oscillation with the frequency ω2 :

ϕ2 (t) =(A2 sin ω2 t + B2 cos ω2 t) e−γt ,

(9.36)
ϕ̇2 (t) ≈(A2 ω2 cos ω2 t − B2 ω2 sin ω2 t) e−γt .

To determine the values of constants A1 , A2 , and B1 , B2 , we use the condi-

tions that must be satisfied when the segments of the graph are joined together,
and take into account the periodicity of the stationary process. At t = 0 the an-
gle of deflection is the same for both ϕ1 and ϕ2 , that is, ϕ1 (0) = ϕ2 (0). From
this condition we get B2 = B1 . We later denote these equal constants by B. The
angular velocity at t = 0 undergoes a sudden change, which follows from the con-
servation of angular momentum: (1+ml )2 ϕ̇1 = (1−ml )2 ϕ̇2 , see Eq. (9.11). This
condition gives us the following relation between A1 and A2 : A2 = qA1 = qA
(further on we denote A1 as A), where the factor q depends on modulation depth
ml according to Eq. (9.21).
For stationary periodic oscillations, corresponding to the principal resonance,
as well as to all resonances of odd orders n = 1, 3, . . . in Eq. (9.13), the condi-
tions of periodicity are:

ϕ1 (T /2) = −ϕ2 (−T /2), (1 + m)2 ϕ̇1 (T /2) = −(1 − m)2 ϕ̇2 (−T /2). (9.37)

Substituting ϕ and ϕ̇ in Eq. (9.37), we obtain the system of homogeneous equa-

tions for the unknown quantities A and B:

(pS1 − qS2 )A + (p C1 + C2 )B = 0,
(9.38)
q(p C1 + C2 )A − (p qS1 − S2 )B = 0,

where p = exp(−γT ). In Eq. (9.38) the following notations are used:

C1 = cos(ω1 T /2), C2 = cos(ω2 T /2),

S1 = sin(ω1 T /2), S2 = sin(ω2 T /2).

The homogeneous system of Eqs. (9.38) for A and B has a non-trivial (non-zero)
solution only if its determinant is zero:

2qC1 C2 − (1 + q 2 )S1 S2 + q(p + 1/p) = 0. (9.39)

This condition for the existence of a non-zero solution to Eqs. (9.38) gives us an
equation for the unknown variable T , which enters Eq. (9.39) as the arguments
of sine and cosine functions in S1 , S2 and C1 , C2 , and also as the argument of
the exponent in p = e−γT . The desired boundaries of the interval of instability
T− and T+ are given by the roots of Eq. (9.39). To find approximate solutions T
to this transcendental equation, we transform it into a more convenient form. We
first represent in Eq. (9.39) the products C1 C2 and S1 S2 as follows:
1 ∆ωT 1 ∆ωT
C1 C2 = (cos + cos ωav T ), S1 S2 = (cos − cos ωav T ).
2 2 2 2
(9.40)

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

9.6. INTERVALS IN THE PRESENCE OF FRICTION 241

Then, using the identity cos α = 2 cos2 (α/2) − 1, we reduce Eq. (9.39) to the
following form:
p
(q + 1) cos(ωav T /2) = ± (q − 1)2 cos2 (∆ωT /4) − q(p + 1/p − 2). (9.41)

To find the boundaries of the interval that contains the principal resonance,
we should search for a solution T of Eq. (9.41) in the vicinity of T = T0 /2 ≈
Tav /2. If for a given value of the quality factor Q (Q enters p = e−γT ) the depth
of modulation ml exceeds the threshold value, Eq. (9.41) has two solutions that
correspond to the desirable boundaries T− and T+ of the instability interval. These
solutions exist if the expression under the radical sign in Eq. (9.41) is positive. Its
zero value corresponds to the threshold conditions:

(q − 1)2 1
cos2 (∆ωT /4) = p + − 2. (9.42)
q p
To evaluate the threshold value of Q for small values of the modulation depth
ml ≪ 1, we may assume here q ≈ 1 + 3ml (see Eq. (9.21)), and cos(∆ωT /4) ≈
1. On the right-hand side of Eq. (9.42), in p = e−γT , we can consider γT ≈
γT0 /2 = π/(2Q) ≪ 1, so that p + 1/p − 2 ≈ (γT )2 = (π/2Q)2 . Thus, for the
threshold of the principal parametric resonance we obtain
π π
Qmin ≈ (ml )min ≈ . (9.43)
6ml 6Q
At the threshold the expression under the radical sign in Eq. (9.41) is zero.
Both its roots (the boundaries of the instability interval) merge. This occurs when
the cosine on the left-hand side of Eq. (9.41) is zero, that is, when its argument
equals π/2:
T π π 1
ωav = , or T = = Tav ,
2 2 ωav 2
so that the threshold conditions (9.43) correspond to exact tuning to resonance,
when T = Tav /2.
To find the boundaries T− and T+ of the instability interval, we represent
T in the argument of the cosine function on the left-hand side of Eq. (9.41) as
Tav /2 + ∆T . Since ωav Tav = 2π, we can write this cosine as − sin(ωav ∆T /2).
Then Eq. (9.41) becomes:
s
1 ∆ω( 12 Tav + ∆T ) (p − 1)2
sin(ωav ∆T /2) = ∓ (q − 1)2 cos2 −q .
q+1 4 p
(9.44)
For zero friction p = 1, and Eq. (9.44) coincides with Eq. (9.21). The diagram
in Figure 9.16 is obtained by numerically solving this equation for ∆T by itera-
tion. Boundaries of the instability for intervals of higher odd orders n = 3, 5, . . .
are calculated similarly by representing T in Eq.(9.41) as nTav /2 + ∆T . They
are also shown in Figure 9.16 for several values of the quality factor Q. For large

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

242 CHAPTER 9. PENDULUM WITH MODULATED LENGTH

values of the modulation depth ml these boundaries almost merge with the corre-
sponding boundaries in the absence of friction.
To find an approximate solution of Eq. (9.44) that is valid for small values
of the modulation depth ml ≪ 1 up to terms to the second order in ml , we can
simplify the expression under the radical sign on the right-hand side of Eq. (9.38),
assuming q ≈ 1 + 3ml , (q − 1)2 ≈ 9m2l , and the value of the cosine function
to be 1. The last term of the radicand can be represented as (π/6Q)2 ≈ (ml )2min .
On the left-hand side the sine can be replaced with its small argument, where
ωav = 2π/Tav . Thus we obtain:
 
∆T 1 Tav 1
q q
≈∓ m2l − (ml )2min , or T∓ = 1∓ m2l − (ml )2min .
Tav 2π 2 π
(9.45)
For the case of zero friction (ml )min = 0, and these approximate expressions
for the boundaries of the instability interval reduce to Eq. (9.25). For the threshold
conditions ml = (ml )min , and both boundaries of the interval merge, that is, the
interval disappears.
After the substitution of one of the roots T− or T+ of Eq. (9.25) into (9.38),
both equations for A and B become equivalent and allow us to find only the ratio
A/B. Nevertheless, these oscillations have a definite shape, which is determined
by the ratio of the amplitudes A and B of the sine and cosine functions whose
segments form the typical pattern of the stationary parametric oscillation (see Fig-
ures 9.9 and 9.11).

9.6.2 Resonances of Even Orders

To describe stationary oscillations occurring on the boundaries of instability inter-
vals of even orders, we can use the same expressions for ϕ(t) and ϕ̇(t), Eqs. (9.35)
and (9.36). The conditions of joining the graphs at t = 0 are also the same. How-
ever, the conditions of periodicity at the instants −T /2 and T /2 for resonances
of even orders differ from Eqs. (9.37) by the opposite sign. This yields, instead of
Eq. (9.41), the following equation for the boundaries of instability intervals:
q
(q + 1) sin(ωav T /2) = ± (q − 1)2 sin2 (∆ωT /4) − q(p + 1/p − 2). (9.46)

For the interval of the second order, we should search for its solution T in the
vicinity of T0 ≈ Tav . If for a given value of the quality factor Q (Q enters
p = e−γT ) the depth of modulation ml exceeds the threshold value, Eq. (9.46) has
two solutions that correspond to the boundaries T− and T+ of the instability inter-
val. These solutions exist if the expression under the radical sign in Eq. (9.46) is
positive. Its zero value corresponds to the threshold conditions, that is, to (ml )min
for a given Q or Qmin for a given ml :

(q − 1)2 (p − 1)2
sin2 (∆ωTav /4) = . (9.47)
q p

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

9.7. CONCLUDING REMARKS 243

The threshold conditions fulfil at exact tuning to second resonance, when T =

Tav . To estimate the threshold value of Q for small values of the modulation depth
ml , we may assume here q ≈ 1 + 3ml , sin(∆ωT /4) ≈ ∆ωT0 /4, and ∆ω ≈
ml ω0 . On the right-hand side of Eq. (9.47), in p = e−γT , we can consider γT ≈
γT0 = π/Q ≪ 1, so that p + 1/p − 2 = (p − 1)2 /p ≈ (γT )2 = (π/Q)2 . Thus,
for the threshold of the second parametric resonance we obtain:
r
2 2
Qmin ≈ 2 , (ml )min ≈ . (9.48)
3ml 3Q
To find the boundaries T− and T+ of the second instability interval, we repre-
sent T in the argument of the sine function on the left-hand side of Eq. (9.46) as
Tav + ∆T . Since ωav Tav = 2π, we can write this sine as − sin(ωav ∆T /2). Then
Eq. (9.46) becomes:
s
ωav ∆T 1 ∆ω(Tav + ∆T ) (p − 1)2
sin =∓ (q − 1)2 sin2 −q . (9.49)
2 q+1 4 p

This form of the equation is convenient for numerical solution by iteration. For
the zero friction p = 1, and Eq. (9.49) coincides with Eq. (9.33). To obtain an
approximate solution to Eq. (9.49), valid for small values of the modulation depth
ml up to the terms of the second order of ml , we can simplify the expression
under the radical sign on the right-hand side of Eq. (9.49), assuming q ≈ 1 + 3ml ,
(q − 1)2 ≈ (3ml )2 , and sin[∆ω(Tav + ∆T )/4] ≈ ∆ωTav /4 = πml /2. The last
term of the radicand can be represented as (2/3Q)2 ≈ (ml )4min . On the left-hand
side the sine can be replaced by its small argument, where ωav = 2π/Tav . Thus
for the boundaries of the second instability interval we get:
 
∆T 3 3
q q
≈∓ 4 4
ml − (ml )min , or T∓ = 1 ∓ 4 4
ml − (ml )min Tav .
Tav 4 4
(9.50)

9.7 Concluding Remarks

We have shown in this chapter that a pendulum whose length is subject to square-
wave modulation by mass reconfiguration gives a very convenient example in
which the phenomenon of parametric resonance in a nonlinear system can be
clearly explained physically with all its peculiarities. The threshold of paramet-
ric excitation is easily determined on the basis of energy considerations.
In a linear system, if the threshold of parametric excitation is exceeded, the
amplitude of oscillations increases exponentially with time. In contrast to forced
oscillations, linear viscous friction is unable to restrict the growth of the ampli-
tude at parametric resonance. In real systems like the pendulum, the growth of
the amplitude is restricted by nonlinear effects that cause the natural period to
depend on the amplitude. During parametric excitation the growth of the ampli-
tude causes an increment in the natural period of the pendulum. The system slips

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

244 CHAPTER 9. PENDULUM WITH MODULATED LENGTH

out of resonance, the swing becomes smaller, and conditions of resonance restore.
These transient beats fade out due to friction, and oscillations of finite amplitude
eventually establish.
Computer simulations aid substantially in understanding the restriction of the
amplitude growth over the threshold caused by nonlinear properties of the pen-
dulum. The simulations illustrate the phenomenon of parametric autoresonance,
stationary periodic oscillatory and rotational regimes that are possible due to the
phase locking between the drive and the pendulum. The simulations also reveal
bifurcations of symmetry breaking and intriguing sequences of period doubling.
The boundaries of parametric instability for a pendulum with the square-wave
modulated length are investigated quantitatively by rather modest mathematical
means.

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

Chapter 10

Rigid Pendulum with

Oscillating Pivot

10.1 Introductory Notes

Familiar and apparently simple dynamical systems for which our intuition may
seem to be well developed can behave in very complicated and even irregular
ways, in spite of the quite simple and exact nature of governing physical laws and
the deterministic character of relevant differential equations.
Various kinds of motion of the pendulum whose axis is driven periodically
in the vertical direction are of special interest. Depending on the frequency and
amplitude of this constrained oscillation of the suspension point, this seemingly
simple system exhibits a rich variety of nonlinear phenomena characterized by
amazingly different types of motion. Some modes of such a parametrically forced
pendulum are quite simple indeed and agree well with our intuition, while others
are very complicated and counterintuitive.
When the external frequency is approximately twice the natural frequency
of the pendulum, the lower state of equilibrium becomes unstable, and the sys-
tem leaves it executing oscillations whose amplitude increases progressively, pro-
vided the driving amplitude exceeds some threshold value. This well-known phe-
nomenon is called parametric resonance. In contrast to the case of ordinary reso-
nance caused by a direct influence of some periodic external force, over the thresh-
old friction is unable to restrict the growth of parametrically excited oscillations.
The growth of the amplitude is restricted because the period of natural oscillations
increases with the amplitude due to nonlinear properties of the pendulum. Para-
metric resonance is possible when two driving cycles occur during approximately
one, two, three and any other integer number of natural periods. With increasing
friction, parametric resonances of higher orders become weaker and disappear.
Besides the principal parametric resonance, excited when two driving cy-
cles occur during approximately one natural oscillation, parametric resonances

245

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

246 CHAPTER 10. PENDULUM WITH OSCILLATING PIVOT

of higher orders are possible when approximately two driving cycles occur during
two, three and any other integer number of natural periods. At small (and moder-
ate) driving amplitudes, parametrically excited oscillations in all these cases are
very much like the natural ones — their frequency is close to the natural frequency
of the pendulum. The forced oscillation of the pivot at resonant conditions sup-
plies the pendulum with energy needed to compensate for frictional losses, thus
preventing these almost natural oscillations from damping. With increasing fric-
tion, parametric resonances of higher orders become weaker and disappear.
Another possible kind of regular motion is a synchronized non-uniform uni-
directional rotation in a full circle with a period that equals the period of the con-
strained motion of the axis or an integer multiple of this period. More complicated
regular modes of the parametrically forced pendulum are formed by combined ro-
tational and oscillating motions synchronized with the pivot. Different competing
modes can coexist at the same values of the driving amplitude and frequency.
Which mode is activated depends on the starting conditions.
Behavior of the pendulum whose axis is forced to oscillate with a frequency
from a certain interval (and with large enough amplitude) can be irregular, chaotic.
The pendulum makes several revolutions in one direction, then swings for a while
with permanently changing amplitude, then rotates again in the former or in the
opposite direction, and so forth. At first sight such essentially unpredictable, ran-
dom behavior contradicts the well-known uniqueness of solution to a differential
equation of motion with given initial conditions. Within the scope of classical me-
chanics, which naturally includes the concept of mechanical determinism, chaotic
behavior of simple dynamical systems considered admissible only as a result of
external random perturbations of the system, i.e., as something introduced from
the outside, from the environment. Discovery of random behavior and intrinsic
irregular, chaotic oscillations in deterministic dynamical systems of different na-
ture (physical, chemical, biological) is one of the most prominent recent scientific
sensations. It is remarkable that such a simple mechanical system as a pendu-
lum whose pivot is forced to oscillate regularly can exhibit at some conditions
a chaotic behavior, illustrated by a strange attractor in the phase plane. Chaotic
modes of the parametrically driven pendulum have been intensively investigated
over past decades (see, for example, [30]–[35] and references therein).

10.2 Kapitza’s Pendulum — Dynamic Stabilization

Another well-known interesting feature in the behavior of a rigid pendulum whose
suspension point is constrained to vibrate with a high frequency along the verti-
cal line is the dynamic stabilization of its inverted position. When the frequency
and/or the amplitude of these vibrations are large enough (the necessary condi-
tions are determined below), the inverted pendulum shows no tendency to turn
down. Moreover, at small and moderate deviations from the vertical inverted po-
sition the pendulum tends to return to it. Being deviated, the pendulum executes
relatively slow oscillations about the vertical line on the background of rapid os-

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

10.2. KAPITZA’S PENDULUM — DYNAMIC STABILIZATION 247

ο
30.0
p −ϕ (t)
z(t)
0

ο
−30.0
0 2 4 6 8 10 12 14 16 T
(quality 25.0, driving frequency 15.0w0, amplitude of the axis 20.0%,
initial deflection 180o, initial angular velocity -1.0w0)

Figure 10.1: Graphs of the angular deflection from the inverted (upside-down)
position for the pendulum whose pivot oscillates at a high frequency along the
vertical line.

cillations of the suspension point. An example of the graphs of such oscillations of

the inverted pendulum obtained in the computer simulation of the motion is shown
in Figure 10.1. We note how the rapid vibrations with the frequency of the pivot
superimpose on the slow oscillation of the pendulum and distort its smooth shape.
In the presence of friction these slow oscillations gradually damp away, and the
pendulum eventually comes to the vertical inverted position.
This type of dynamic stability was first pointed out by Stephenson almost a
century ago [36]. In 1951 such extraordinary behavior of the pendulum was ex-
plained and investigated experimentally in detail by Pjotr Kapitza [37], and the
corresponding physical device is now widely known as “Kapitza’s pendulum.”
Simple hand-made devices are often used in lectures to show this fascinating phe-
nomenon of classical mechanics. An old electric shaver’s mechanism (or a jig saw)
can serve perfectly well to force the pivot of a light rigid pendulum vibrating with
a high enough frequency and sufficient amplitude to make the inverted position
stable (Figure 10.2). The hand holds the shaver in the position that provides the
vertical direction of the pivot oscillations. If the rod is turned into the inverted ver-
tical position, it remains there until the axis is vibrating. When the rod is slightly
deflected to one side and released, it oscillates slowly about the inverted position.
Such a demonstration inevitably evokes vivid response, astonishing those students
who see it for the first time.
Below is a citation from the paper of Kapitza [38] published in the Russian
journal Uspekhi:

Demonstration of oscillations of the inverted pendulum is very im-

pressive. Our eyes cannot follow the fast small movements caused by
vibrations of the pivot, so that behavior of the pendulum in the in-
verted position seems perplexing and even astonishing . . . When we
carefully touch the rod of the pendulum trying to deviate it from the
vertical, the finger feels the resistance produced by the vibrational

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

248 CHAPTER 10. PENDULUM WITH OSCILLATING PIVOT

Figure 10.2: Demonstration of the inverted pendulum’s dynamic stabilization by

vertical vibration of the pivot (left), and small slow oscillations about the inverted
position (right).

torque. After acquaintance with the experiment on dynamic stabi-

lization of the inverted pendulum we reasonably conclude that this
phenomenon is as much instructive as the dynamic stabilization of a
gyroscope, and should be necessarily included in lecture demonstra-
tions on classical mechanics.

After Kapitza, this simple but very curious and intriguing system attracted the
attention of many researchers, and the theory of the phenomenon may seem to be
well elaborated (see, for example, [39]). Nevertheless, more and more new fea-
tures in behavior of this inexhaustible system are reported regularly. Many related
papers were published in the American Journal of Physics ([40]–[49]).
However, in the abundant literature on the subject (a vast bibliography can be
found in [50]) it is almost impossible to find a simple and clear interpretation of
this interesting phenomenon. Understanding the dynamic stabilization of an in-
verted pendulum is certainly a challenge to our intuition. The principal aim of
this chapter is to present a quite simple qualitative physical explanation for the
phenomenon. We also focus on an approximate quantitative theory (leading to the
well-known concept of the effective potential for the slow motion of the pendu-
lum) that can be developed on the basis of the suggested approach to the problem.
Finally, we show that the loss of dynamic stability at large amplitudes of the pivot
is closely related to the commonly known conditions of parametric instability of
the non-inverted pendulum.

10.3 The Physical Model of the Investigated System

For simplicity we consider a light rigid rod of length l with a heavy small bob
of mass m on its end and assume that the rod has zero mass. Let the axis of the
pendulum be forced to execute a given harmonic oscillation along the vertical line
with a frequency ω and an amplitude a, i.e., let the motion of the axis be described

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

10.4. PARAMETRIC RESONANCE 249

by the following equation:

z(t) = a sin ωt or z(t) = a cos ωt. (10.1)

Depending on the problem under consideration, either sine or cosine time depen-
dence may be more convenient for calculations.
The force of inertia Fin (t) exerted on the bob in the non-inertial frame of
reference also has a sinusoidal dependence on time:

Fin (t) = −mz̈(t) = maω 2 sin ωt or Fin (t) = maω 2 cos ωt. (10.2)

This force of inertia is directed downward during the time intervals for which
z(t) < 0, i.e., when the axis is below the middle point of its oscillations. We
see this directly from the equation for Fin (t), Eq. (10.2), whose right-hand side
depends on time exactly as the z-coordinate of the axis. Therefore during the
corresponding half-period of the oscillation of the pivot this additional force is
equivalent to some strengthening of the force of gravity. During the other half-
period the axis is above its middle position (z(t) > 0), and the action of this
additional force is equivalent to some weakening of the gravitational force. When
the frequency and/or amplitude of the pivot are large enough (when aω 2 > g), for
some part of the period the apparent gravity is even directed upward.
On the basis of this approach, taking into account the periodic variations of the
apparent gravity, we can easily explain, say, the physical reason for the ordinary
parametric swinging of the pendulum, when its pivot is driven vertically with a
frequency approximately twice the frequency of natural oscillations.

10.4 Parametric Resonance

Taking into account the modulation of the apparent gravitational force in the non-
inertial reference frame associated with the pivot, we can easily understand the
reason for the growth of oscillations in conditions of parametric resonance. The
principal parametric resonance takes place if approximately two cycles of modu-
lation occur during one period of natural oscillations. Indeed, let us consider the
time interval during which the pendulum moves from the utmost deflection toward
the lower equilibrium position (a quarter of the natural period). During this time
interval let the pivot in its constrained oscillation be below the midpoint, z(t) < 0
(a half-period of the pivot oscillation). Due to the additional apparent gravity dur-
ing this interval the pendulum gains a greater speed than it would have gained in
the absence of the pivot’s motion.
During the further motion of the pendulum away from the equilibrium posi-
tion, the pivot is above its midpoint (z(t) > 0, Figure 10.3), so that the force of
inertia reduces the apparent gravity. Thus the pendulum reaches a greater angular
displacement than it would have reached otherwise. During the second half-period
of the pendulum’s motion the swing increases again, and so on, until the stationary
motion is established due to violation of the resonance conditions at large swing.

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

250 CHAPTER 10. PENDULUM WITH OSCILLATING PIVOT

o
78.0

z (t) j (t)
0

o
78.0
0 2 4 6 8 10 12 14 16T
.
j
1.27 z (t)

t = 10
0
j t =1 t=0

-1.27 o o
-78.0 0 78.0
(quality 10.0, driving frequency 1.95w0, amplitude of the axis 10.0%,
initial deflection 10o, initial angular velocity -0.2w0)

Figure 10.3: The time history ϕ(t), phase trajectory (with Poincarè sections), and
the spatial trajectory of the pendulum bob in conditions of parametric resonance.
The period T of the pivot oscillations is used as a unit of time.

The most effective growth of the amplitude occurs when the frequency of the
pivot oscillation is twice the natural frequency of the pendulum, and if the pivot,
moving upward, crosses its middle position at the moment at which the pendulum
crosses its equilibrium position.
The growth of the energy is provided by the work done by the source that
makes the pivot oscillate. The swing of the pendulum increases if the gain of
energy during the period exceeds frictional losses, that is, when the amplitude of
the pivot is greater than some threshold value.
In contrast to the ordinary resonance (which is excited by an external force
whose frequency equals the natural frequency of the pendulum), friction cannot
restrict the growth of amplitude at parametric resonance if the threshold is ex-
ceeded.
The growth of the pendulum swing in conditions of parametric resonance is
restricted due to nonlinear properties of this system. Indeed, the natural period of
the pendulum increases when the angular excursion becomes greater. This growth
of the natural period at the fixed period of the pivot oscillations leads to violation
of resonance conditions as the swing grows. The amplitude starts to diminish, and
the resonance conditions restore, so that the amplitude grows again, and so on.
Due to friction these transient beats gradually fade away, and periodic oscillations
of a finite swing finally establish.

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

10.5. PHYSICAL REASONS FOR STABILITY 251

The lower left panel of Figure 10.3 shows the phase trajectory of the pendulum
at resonance conditions. The Poincarè sections correspond to the time instants at
which the oscillating pivot reaches its lowest position. During the transient process
the expanding phase trajectory approaches the closed curve, that is, approaches
the limit cycle that corresponds to steady-state periodic oscillations. During this
transient the Poincarè sections condense to the two fixed points in the phase plane.
The lower right panel of Figure 10.3 shows the corresponding trajectory of the
pendulum’s bob in space. The simulation is based on a numerical integration of
the exact differential equation, Eq. (10.11), p. 257, for the momentary angular
deflection ϕ(t). This equation includes the torque of the force of gravity and the
instantaneous value of the torque exerted on the pendulum by the force of inertia
Fin (t) that depends explicitly on time t.
Excitation of the principal parametric resonance at finite amplitudes of the
pivot oscillation is possible not only when two driving cycles occur during exactly
one natural oscillation of the pendulum (or, generally, during an integer number n
of natural periods), but rather in intervals of the pivot frequencies ω in the vicinity
of ω = 2ω0 and, generally, near ω = 2ω0 /n. The intervals of parametric instabil-
ity are characterized by “tongues” in the parameter’s plane. These “tongues” are
discussed in more detail in Section 10.10.2, p. 281.

10.5 Physical Reasons for Stability of the Inverted

Pendulum
In the case of oscillations of the axis with high enough frequency, the mean value
of the force of inertia, averaged over the short period of these oscillations, is zero,
but the value of its torque about the axis, averaged over the period, is not zero.
Next we show why. This non-zero mean torque of the force of inertia explains the
pendulum stabilization in the inverted position.
Let us begin with the case in which the rod of the pendulum is oriented hori-
zontally, i.e., at the right angle ψ = π/2 to the direction of oscillations of the axis
(Figure 10.4a).
To better understand the influence of the force of inertia upon the system, we
first forget for a while about the force of gravity. If the bob has zero initial velocity,
in the inertial reference frame in the absence of gravity, it stays practically at the
same level while the axis A oscillates between the extreme points 1 and 2 and
the rod turns down and up through a (small) angle, as shown in the upper panel
of Figure 10.4a. (For the sake of clarity the amplitude of the pivot and hence this
angle are exaggerated in this figure.)
In the non-inertial frame of reference associated with the oscillating axis, the
same motion of the rod is shown in the lower panel of (Figure 10.4a): The bob of
the pendulum moves up and down along an arc of a circle and occurs in positions
1 and 2 (lower panel of Figure 10.4a) at the instants at which the oscillating axis
reaches its extreme positions 1 and 2 respectively (upper panel of Figure 10.4a).
In position 1 the force of inertia F1 exerted on the bob is directed downward, and

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

252 CHAPTER 10. PENDULUM WITH OSCILLATING PIVOT

a b
2
2
2
y = 90
y

1 1
1 1
F1
y = 90 F1 F2

A 2
F2

2 A y

Figure 10.4: The forces of inertia exerted on the pendulum in the non-inertial
reference frame at the extreme positions 1 and 2 of the oscillating axis A.

in the position 2 the force F2 of the same magnitude is directed upward. The arm
of the force in the positions 1 and 2 is the same. Therefore the torques of the force
of inertia in positions 1 and 2 are equal and opposite. The same is true for all pairs
of symmetric intermediate positions of the pivot.
Hence the torque of this force, averaged over the period of oscillations, is
zero. In the absence of gravity this orientation of the pendulum (perpendicularly
to the direction of oscillations) corresponds to a dynamic equilibrium position (an
unstable one, as we shall see later).
Now let us consider the case in which on average the rod is deflected through
an arbitrary angle ψ from the direction of oscillations, and the axis oscillates be-
tween extreme points 1 and 2, as shown in the upper panel of Figure 10.4b. In the
non-inertial frame of reference associated with the oscillating axis, the bob moves
at these oscillations between the points 1 and 2 in the lower panel of Figure 10.4b
along an arc of a circle whose center coincides with the axis A of the pendulum.
We note that the rod has the same simultaneous orientations in both reference
frames at the instant 1 as well as at the instant 2.
When the axis is displaced downward (to the position 1) from its mid-point,
the force of inertia F1 exerted on the bob is also directed downward. In the other
extreme position 2 the force of inertia F2 has an equal magnitude and is directed
upward. However, now the torque of the force of inertia in the position 2 is greater
than in the position 1 because the arm of the force in this position is greater. There-
fore on average over a period of the pivot’s vibration the force of inertia creates

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

10.6. APPROXIMATE QUANTITATIVE THEORY 253

a torque about the axis that tends to turn the pendulum upward, into the vertical
inverted position, in which the rod is parallel to the direction of oscillations.
Certainly, if the pendulum makes an acute angle with respect to the downward
vertical position, the mean torque of the force of inertia tends to turn the pendulum
downward.
Thus, the torque of the force of inertia, averaged over a period of rapid oscil-
lations of the pivot, tends to align the pendulum along the direction of constrained
oscillations of the axis. The right-hand panel (b) of Figure 10.4 presents an utterly
simple and clear explanation of the origin of this torque. Kapitza [37] called this
torque vibrational, but we can also call it inertial, because its origin is related to
the pseudo force of inertia that arises due to the constrained rapid vibrations of the
axis. For given values of the driving frequency and amplitude, this torque depends
only on the angle of the pendulum’s deflection from the direction of the pivot’s
vibration. This mean inertial torque does not depend on the time explicitly, and
its influence on the pendulum can be considered exactly in the same way as the
influence of other ordinary external torques, such as the torque of the gravitational
force.
This mean torque of the alternating force of inertia gives the desired explana-
tion for the physical reason of existence of the two stable equilibrium positions
that correspond to the two preferable orientations of the pendulum’s rod along the
direction of the pivot’s vibration. The principal idea is utterly simple: Although
the mean value of the force of inertia Fin (t), averaged over the short period of
these oscillations, is zero, the value of its torque about the axis, averaged over this
period, is not zero. This is clearly seen from the right-hand panel of Figure 10.4.
With gravity, the inverted pendulum is stable with respect to small deviations
from this position provided the mean torque of the force of inertia is greater in
magnitude than the torque of the force of gravity that tends to tip the pendulum
down.
Next we show that this stabilization √ occurs when the following condition is
fulfilled: a2 ω 2 > 2gl, or (a/l)(ω/ω0 ) > 2 (see, e.g., [49]). However, this is only
an approximate criterion for dynamic stability of the inverted pendulum, which
is valid at small amplitudes of forced vibrations of the pivot (a ≪ l). In Sec-
tion 10.11 we consider a more rigorous mathematical theory of the phenomenon
that allows us to establish a more precise criterion, which is also valid for low
frequencies and large amplitudes of constrained oscillations of the pivot.

10.6 An Approximate Quantitative Theory of the

Inverted Pendulum
Now we can determine the approximate quantitative conditions (valid at a ≪ l
and ω ≫ ω0 ) that provide the dynamic stabilization of the inverted equilibrium
position in the presence of the force of gravity. Rapid vertical oscillations of the
axis make the inverted position stable, if at small deflections from this position
the torque of the force of inertia, averaged over the period of the pivot oscillations

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

254 CHAPTER 10. PENDULUM WITH OSCILLATING PIVOT

(this torque tends to return the pendulum to the inverted position), is greater in
magnitude than the torque of the gravitational force that tends to tip the pendulum
down.
We can consider the motion of the pendulum whose axis is vibrating with
a high frequency as a superposition of two components: A ‘slow’ component,
whose variation during a period of constrained vibrations is small, and a ‘fast’ (or
vibrational) component. Let us imagine an observer who does not notice (or does
not want to notice) the vibrational component of this compound motion. If this
observer uses, for example, a stroboscopic illumination with the interval between
the flashes that equals the period of constrained vibrations of the pendulum’s axis,
he can see only the slow component of the motion. Our principal interest is to
determine this slow component.
When the rod of the pendulum is deflected from the downward vertical posi-
tion on the average through an angle ψ, the instantaneous value ϕ(t) of the deflec-
tion angle is subjected to additional rapid sinusoidal oscillation with the frequency
ω about this average value ψ = hϕ(t)i because of the constrained oscillation of
the axis. This can be clearly seen from the plots of the angular deflection and
velocity (see Figure 10.1). Therefore we can try to search for the instantaneous
angle of deflection ϕ(t) as the sum of a slowly varying function ψ(t) and an addi-
tional fast term δ(t) whose mean value is zero. This fast angle δ(t) oscillates with
the high frequency ω of the pivot vibration. The amplitude of this oscillation is
proportional to the sine of the momentary value of the average slow angle ψ(t):
z(t) a
ϕ(t) = ψ(t) + δ(t) = ψ(t) − sin ψ = ψ(t) − sin ψ sin ωt. (10.3)
l l
Here a is the amplitude of forced vibrations of the axis, l is the length of the pen-
dulum. (When the axis is above its middle position, z is positive and the additional
angle δ = −(z/l) sin ψ is negative.) Later on we shall find the differential equa-
tion for this unknown slow varying function ψ(t) that describes the motion of the
pendulum, averaged over the period of rapid oscillations.
The torque of the force of inertia depends on its momentary value
maω 2 sin ωt, Eq. (6.3), and on the sine of the angle ϕ. The oscillations of the
axis cause only small deviations of the momentary deflection angle ϕ from its av-
erage value ψ (i.e., δ(t) ≪ 1 for all t), and so for the sine of the deflection angle
we can write the following approximate expression:
sin ϕ = sin(ψ + δ) ≈ sin ψ + δ cos ψ. (10.4)
With the help of this equation, we can find the approximate value of the gravi-
tational torque about the point of suspension (about the axis of the pendulum),
averaged over the period of rapid oscillations of the axis:
h−mgl sin ϕi = −mglhsin(ψ + δ)i = −mgl sin ψ, (10.5)
because the average value of δ(t) is zero: hδ(t)i = 0. We see that the mean torque
of the gravitational force is the same as in the case of a pendulum with the im-
movable suspension point: The oscillating second term in the expansion for the

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

10.6. APPROXIMATE QUANTITATIVE THEORY 255

momentary angle, Eq. (10.4), being multiplied to a constant gravitational force,

gives no contribution to the mean torque. However, when we take the time aver-
age for the torque of the oscillating force of inertia, the first term in the expansion,
Eq. (10.4), vanishes, but the oscillating second term gives a nonzero contribution.
This occurs by virtue of the sinusoidal dependence on time both of δ(t) and the
force of inertia Fin (t) in Eq. (6.3):

a
hFin (t)l sin(ψ + δ)i = −maω 2 l cos ψ sin ψhsin2 ωti =
l
1
− ma2 ω 2 cos ψ sin ψ. (10.6)
2
The nonzero contribution of this term to the torque of the force of inertia arises
because the average value of the sine squared is equal to 1/2: hsin2 ωti = 1/2.
An explanation of this nonzero average torque on the physical level is given by
the right-hand panel of Figure 10.4. For ψ > π/2 the average value of the torque
of the force of inertia is positive: If the pendulum makes an acute angle with the
upward vertical direction, this torque tends to turn the pendulum up.
Comparing the right-hand sides of Eqs. (10.5) and (10.6), we see that the
torque of the force of inertia can exceed in magnitude the torque of the gravi-
tational force tending to tip the pendulum down, when the following condition is
fulfilled:
a2 ω 2 > 2gl. (10.7)
Thus, the inverted position of the pendulum is stable
√ if the maximal velocity
ωa of the vibrating axis is greater than the velocity 2gl attained by a body dur-
ing a free fall from the height that equals the pendulum length l. We can write this
criterion of stability in another form, using the expression ω02 = g/l for the fre-
quency ω0 of small natural oscillations of the pendulum in the absence of forced
vibrations of the axis. Substituting g = lω02 in Eq. (10.7) we get:
a ω √
· > 2. (10.8)
l ω0
According to this approximate criterion of stability, Eq. (10.8), the product of
the dimensionless fractional amplitude of forced oscillations of the axis a/l and
the dimensionless fractional frequency of these oscillations ω/ω0 must exceed the
square root of 2. For instance, for the pendulum whose length l = 20 cm and the
frequency of forced oscillations of the axis f = ω/2π = 100 Hz, the amplitude a
must be greater than 3.2 mm. To provide the dynamic stabilization of the inverted
pendulum within some finite interval of the angles of deflection from the vertical
position, the product of the dimensionless amplitude of forced
√ oscillations of the
axis and the dimensionless frequency must be greater than 2 by a finite value.
We emphasize that expressions (10.7) and (10.8) give only an approximate
criterion of inverted pendulum stability, which is valid for fast enough vertical
oscillations of the pivot, whose frequency is much greater than the frequency of
natural oscillations of this pendulum in the gravitational field. Later on we will

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

256 CHAPTER 10. PENDULUM WITH OSCILLATING PIVOT

find an improved and extended criterion of dynamic stability that also holds for
low frequencies of the pivot.
For a physical pendulum, the condition of dynamic stability in the inverted
position is expressed by the same Eq. (10.7) or (10.8) provided we imply by the
quantity l the reduced length of the pendulum I/md, where I is the moment of
inertia with respect to the axis of rotation, m is the mass, and d is the distance
between the axis and the center of mass. We note that the criterion (10.7) or (10.8)
is independent of friction.
The critical minimum value of the product of the driving amplitude and fre-
quency aω found above, Eq. (10.8), agrees with the lower boundary of stability of
the inverted pendulum obtained by approximating the exact nonlinear equation of
motion by the Mathieu equation, the solutions of which are widely documented
in the extensive literature concerning the problem (see, for example, [40], [41]).
However, the investigation based on the Mathieu equation and infinite Hill’s de-
terminants gives little physical insight into the problem and, more importantly,
is restricted to the motion within small angles from the vertical. Conversely, the
above explanation shows clearly the physical reason for the dynamic stabilization
of the inverted pendulum and is free from the restriction of small angles.
In particular, on the basis of the above developed approach, for given values
of the frequency ω and the amplitude a of forced oscillations of the axis, we can
find the maximal admissible mean angle of deflection from the inverted vertical
position θmax = π − ψ0 for which the pendulum will return to this position. To do
this, we should equate the right-hand sides of Eqs. (10.5) and (10.6) that determine
the average values of the torque of the gravitational force, which tends to tip the
pendulum down, and of the torque of the force of inertia, which tends to return the
pendulum to the inverted position:
 2
2gl ω0 l
cos θmax = − cos ψ0 = =2 . (10.9)
a2 ω 2 ω a

The greater the product ωa of the frequency ω and the amplitude a of con-
strained vibrations of the axis, the closer the angle θmax to π/2. Being deflected
from the vertical position by an angle smaller than θmax , the pendulum will ex-
ecute relatively slow oscillations about this inverted position. This slow motion
occurs under the mean torque of both the force of inertia and the force of gravity.
Rapid oscillations with the frequency of forced vibrations of the axis superimpose
on this slow motion of the pendulum. With friction, the slow motion gradually
damps, and the pendulum wobbles up, settling eventually in the inverted position.
Similar behavior of the pendulum can be observed when it is deflected from
the lower vertical position. But in this case the frequency of slow oscillations
is greater than in the case of the inverted pendulum. Actually, this frequency is
greater than the frequency of natural oscillations in the absence of forced vibra-
tions of the axis, because in this case the averaged torque of the force of inertia
tends to return the pendulum to the lower vertical position and is added to the

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

10.7. EXACT DIFFERENTIAL EQUATION 257

torque of the gravitational force. This means that an ordinary clock with a pendu-
lum will be always ahead of time if it is subjected to a fast vertical vibration.
When we take into account the mean torque of the force of inertia, the fre-
quencies ωup and ωdown of small slow oscillations about the inverted position and
the lower vertical position are given by the following expressions:

2 a2 ω 2 a2 ω 2
ωup = − ω02 , 2
ωdown = + ω02 . (10.10)
2l2 2l2
Substituting ω0 = 0 into these formulas, we get the expression for the fre-
quency of small slow oscillations of the pendulum with vibrating axis in the ab-
sence of the gravitational force. These oscillations can occur about either of the
two symmetrical stable equilibrium positions located opposite one another along
the direction of forced oscillations of the axis. For vertical oscillations of the axis
in the field of gravity, the force of gravity increases the average restoring torque of
the force of inertia (and consequently the frequency of slow oscillations) about the
lower equilibrium position, and the force of gravity decreases the average restor-
ing torque (and the frequency of slow oscillations) about the upper equilibrium
position.
It is worth mentioning that the results regarding the behavior of the pendulum
with a vertically vibrating axis are obtained here without the differential equa-
tion for the system under consideration. (This equation is discussed in the next
section). These results are valid if the amplitude of this constrained vibration of
the axis is small compared with the pendulum length, and its frequency is much
greater than the frequency of small natural oscillations of the pendulum.
As we mentioned above, at certain intervals of the system parameters (in the
intervals of parametric instability) the lower position of the pendulum becomes
unstable. However, parametric resonance, as well as the modes of chaotic behav-
ior, occur at such frequencies of constrained oscillations of the pivot that do not
satisfy the condition ω ≫ ω0 of applicability of the approach used in this chapter.
(For the principal parametric resonance, ω ≈ 2ω0 .) Therefore the existence of
parametric resonance does not disprove our conclusion about the stability of the
lower equilibrium position.

10.7 Exact Differential Equation for the Pendulum

with Oscillating Pivot
The exact differential equation for the pendulum with oscillating pivot includes,
besides the torque of the force of gravity, the instantaneous (not averaged over the
fast period) value of the torque exerted on the pendulum by the force of inertia
that depends explicitly on time t:
a
ϕ̈ + 2γ ϕ̇ + (ω02 − ω 2 sin ωt) sin ϕ = 0. (10.11)
l

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

258 CHAPTER 10. PENDULUM WITH OSCILLATING PIVOT

The second term of Eq. (10.11) takes into account the braking frictional torque,
assumed to be proportional to the momentary angular velocity ϕ̇ in the mathemat-
ical model of the simulated system. The damping constant γ is inversely propor-
tional to the quality factor Q commonly used to characterize the viscous friction:
Q = ω0 /2γ. In the absence of gravity the parametrically driven pendulum is
described by Eq. (10.11) with ω0 = 0. Since in this case the notion of natu-
ral frequency loses its sense, it is impossible to use the quality factor defined as
Q = ω0 /2γ to characterize friction, but instead we can use another dimensionless
quantity ω/2γ, where ω is the driving frequency.
We note that oscillations about the inverted position can be formally described
by the same differential equation, Eq. (10.11), with negative values of ω02 = g/l.
This is clearly seen if by ϕ in Eq. (10.11) we mean the deviation of the pendulum
from the upward vertical. In other words, we can consider ω02 as a control param-
eter whose variation is physically equivalent to changing the gravitational force
exerted on the pendulum. When this control parameter is diminished through zero
to negative values, the constant (gravitational) torque in Eq. (10.11) first turns to
zero and then changes its sign to the opposite. Such a “gravity” tends to bring
the pendulum into the inverted position ϕ = π, destabilizing the position ϕ = 0
of the unforced pendulum: The inverted position with ω02 < 0 in Eq. (10.11) is
equivalent to the hanging-down position with the positive value of ω02 of the same
magnitude.
Experimental verification of approximate expressions (10.10) for frequencies
ωup and ωdown of small slow oscillations about the inverted position and the lower
vertical position is given by the graphs in Figure 10.5, obtained in the simula-
tion. The simulation is based on a numerical integration of the exact differen-
tial equation, Eq. (10.11), for the momentary angular deflection ϕ(t). To make
the verification easier, the pivot frequency ω was chosen to be 16 ω0 , so that
(a2 /2l2 )ω 2 = 3.0 ω02 . Thus Eq. (10.10) provides the value ωdown = 2ω0 for
the frequency of slow oscillations about the downward position, which is exactly
twice the natural frequency. Then the period of slow oscillations Tdown must equal
one half of the period T0 of natural oscillations in the absence of pivot vibrations
(Tdown = T0 /2). Figure 10.5 shows that the pendulum executes exactly two cy-
cles of slow oscillations during one period T0 , which in this case (at ω = 16 ω0)
equals 16 periods T = 2π/ω of pivot vibrations. (The units T are used for the
time scale.) For
√ the frequency of slow oscillations about the upward √ vertical posi-
tion ωup = 2ω0 , so that their period should equal (Tup = T0 / 2). This value
of the period is also in good agreement with the lower graph in Figure 10.5.

10.8 Effective Potential Function for a Pendulum

with the Pivot Vibrating at High Frequency
The graphs in Figure 10.5 show clearly that the smooth motion is distorted by
the high frequency oscillations most of all near the utmost deflections of the pen-
dulum, and these distortions are relatively small while the pendulum crosses the

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

10.8. EFFECTIVE POTENTIAL FUNCTION 259

ο
20.0

z (t) ϕ (t)
0

ο
−20.0
0 2 4 6 8 10 12 14 16 T
(no friction, driving frequency 16.0w0, amplitude of the axis 15.3%,
o
ο initial deflection 0.00, initial angular velocity 0.50w0)
21.0
π − ϕ (t)
z (t)
0

ο
−21.0
0 2 4 6 8 10 12 14 16 T
(no friction, driving frequency 16.0w0, amplitude of the axis 15.3%,
o
initial deflection 180.0, initial angular velocity 0.50w0)

Figure 10.5: The graphs of the momentary angular deflection ϕ(t) for oscillations
of the rigid planar pendulum with vibrating axis about the dynamically stabilized
lower and upper equilibrium positions respectively, obtained by a numerical in-
tegration of the exact differential equation, Eq. (10.11). The sinusoidal graphs of
the axis motion z(t) = −a cos ωt are also shown.

equilibrium position. This proves that the momentary deflection angle ϕ(t) can
be represented approximately as a superposition of the slow varying mean angle
ψ(t) and the high frequency term whose angular amplitude is proportional to sine
of ψ(t) (see Eq. (10.3)). Indeed, the angular amplitude of the rapid (second) term
in Eq. (10.3) is the greatest at the extreme deflections of the pendulum, and this
amplitude vanishes when the pendulum in its smooth motion crosses each of the
vertical positions.
An observer that doesn’t notice the rapid oscillating motion of the pendu-
lum can consider simply that the system moves in an effective potential field
U = U (ψ). Such a potential function that governs the smooth motion of the
pendulum averaged over the rapid oscillations was first introduced by Landau
[4], and derived by several different methods afterwards (see, for example, [47]
– [49]). Certainly, some subtle details in the motion of the pendulum revealed by
the simulations are lost in the approximate analysis, which refers only to the slow
component of the investigated motion. Nevertheless, this analysis allows us to
clearly interpret the principal features of the physical system under consideration,
and even to evaluate such typically nonlinear properties as the dependence of the
period on the amplitude of slow oscillations.
The approximate differential equation for the slow motion of the pendulum
can be written under the assumption that the angular acceleration ψ̈(t) in this

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

260 CHAPTER 10. PENDULUM WITH OSCILLATING PIVOT

θ max U (ψ )
θ max

Utot
Uin
Ugr

−π −π /2 0 π /2 ψ0 π ψ

Figure 10.6: Graphs of the gravitational potential energy Ugr , mean potential en-
ergy Uin in the field of the force of inertia, and of the total potential energy Utot (ψ)
for the pendulum with vertically oscillating pivot.

slow motion is determined by the mean torque N (ψ) exerted on the pendulum in
the non-inertial frame of reference associated with its axis:
1 a2 2
ψ̈ = −ω02 sin ψ − ω cos ψ sin ψ. (10.12)
2 l2
The mean torque in the right-hand side of Eq. (10.12) is calculated approx-
imately under the assumption that the slowly varying angular coordinate ψ(t) is
‘frozen.’ To facilitate interpretation of the slow motion described by this nonlinear
differential equation, we can introduce a potential function U (ψ) that corresponds
to the mean torque N (ψ) exerted on the pendulum. The torque is determined
by the derivative of this potential function: N (ψ) = −dU (ψ)/dψ. The above-
mentioned observer who doesn’t notice the rapid oscillating motion of the pen-
dulum can consider simply that the system moves in an effective potential field
U = U (ψ). From the right-hand panel of Eq. (10.12) we conclude that the effec-
tive potential consists of two parts Ugr (ψ) and Uin (ψ) that describe the influence
of the force of gravity and the force of inertia, respectively:
1
U (ψ) = Ugr (ψ) + Uin (ψ) = mgl(1 − cos ψ) + ma2 ω 2 (1 − cos 2ψ). (10.13)
4
The graphs of Ugr (ψ) and Uin (ψ) are shown in Figure 10.6. They both have a
sinusoidal shape, but the period of Uin (ψ) is just one half of the period of Ugr (ψ).
Their minima at ψ = 0 coincide, thus generating the principal minimum of the
total potential function U (ψ) = Utot (ψ). This minimum corresponds to the stable
lower equilibrium position of the pendulum. But the next minimum of Uin (ψ) is
located at ψ = π, where Ugr (ψ) has its maximum corresponding to the inverted
position of the pendulum.

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

10.8. EFFECTIVE POTENTIAL FUNCTION 261

If the criterion (10.7) or (10.8) is fulfilled, the amplitude of Uin (ψ) is greater
than that of Ugr (ψ). Then the potential function U (ψ) has (in addition to the ab-
solute minimum at ψ = 0, which corresponds to the lower equilibrium position)
relative minima at ψ = ±π. Both additional minima correspond to the same in-
verted position of the pendulum. Oscillations of a particle trapped in an additional
minimum describe the behavior of the inverted pendulum. Slow small oscillations
occurring near the bottom of a potential well are almost harmonic.
Frequencies of these oscillations can be found from the differential equa-
tion (10.12) for the slow motion, assuming sin ψ ≈ ψ, cos ψ ≈ 1 in the vicinity
of ψ = 0 and sin ψ = sin(π − θ) ≈ θ, cos ψ ≈ −1 near ψ = ±π:

1 a2 2 1 a2 2
ψ̈ = −(ω02 + ω )ψ, θ̈ = −(−ω02 + ω )θ. (10.14)
2 l2 2 l2
It follows from (10.14) that frequencies ωdown and ωup of small slow oscil-
lations about the lower (ψ = 0) and upper (ψ = ±π) equilibrium positions are
given by the same expressions: Eqs. (10.10), p. 257, obtained earlier on the basis
of a simple physical approach.
The slopes of the shallow additional potential wells are not as steep as the
slopes of the principal well at ψ = 0. Therefore the frequency ωup of slow small
oscillations about the inverted position is smaller than the frequency ωdown of
small oscillations within the principal well (about the lower vertical position), in
accordance with the above expressions (10.10) and with the simulations repre-
sented by graphs in Figure 10.5. Certainly, some subtle details in the motion of
the pendulum revealed by the simulations are lost in our approximate analysis,
which refers only to the slow component of the investigated motion. Nevertheless,
this analysis allows us to clearly interpret principal features of the physical system
under consideration.
The maxima of the total potential energy U (ψ) are determined by Eq. (10.9).
The tops of the potential barrier between the two wells occur at deflections ±ψ0
(ψ0 > π/2) from the lower vertical position and ±θmax (θmax < π/2) from
the upper equilibrium position (Figure 10.6). At these positions of the pendulum
the mean torque of gravity is balanced by the mean torque of the force of iner-
tia. However, these equilibrium positions are unstable: The slightest disturbance
makes the pendulum to slowly slip down into one of the wells and oscillate there,
moving from one slope to the other and back. The pattern of such slow oscillations
(averaged over the fast period of constrained vibrations) is far from a sine curve.
The pendulum stays for a prolonged time near the summit of the potential barrier
at the utmost deflection, and then moves rather fast towards the other utmost de-
flection to linger there again before the backward fast motion. The simulation of
such a motion is shown in Figure 10.7.
The results discussed above are obtained by a decomposition of motion on
slow oscillations and rapid vibrations with the driving frequency. Hence these
results are approximate and valid when the amplitude of constrained vibration of
the axis is small compared to the pendulums length (a ≪ l) and their frequency
is high enough (ω ≫ ω0 ). It follows from the graph U = U (ψ) (see Figure 10.6)

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

262 CHAPTER 10. PENDULUM WITH OSCILLATING PIVOT

ο
96.0
π − ϕ (t)
z (t)
0

ο
−96.0
0 2 4 6 8 10 12 14 16T
(quality 600.0, driving frequency 15.0w0, amplitude of the axis 20.0%,
o
initial deflection 103 , initial angular velocity 0.428 w0)

Figure 10.7: The graphs of oscillations of the pendulum about the inverted position
with maximal possible angular excursion.

that the lower equilibrium position is always stable, and the upper one is stable
if additional minima exist on the curve of potential energy U = U (ψ). These
minima appear when condition (10.7) or (10.8) is fulfilled.
As we already mentioned earlier, for some intervals of the pivot frequency
(intervals of parametric instability) the lower equilibrium position becomes un-
stable — the phenomenon of parametric resonance occurs at which small initial
oscillations increase progressively. This conclusion does not follow from the in-
vestigation based on a decomposition of motion on slow and rapid components.
This is by no means surprising because parametric resonance occurs at such driv-
ing frequencies (for the principal parametric resonance ω ≈ 2ω0 ) for which this
decomposition is not applicable. In the next chapter we will show that the inverted
(dynamically stabilized) position can also become unstable: At large enough am-
plitude of the pivot oscillations the pendulum is involved in oscillations about the
inverted position with the period covering two cycles of excitation. This so-called
“flutter” mode of oscillations is closely related to ordinary parametric parametric
resonance of the hanging-down pendulum. We emphasize that parametric reso-
nance, “flutter” mode, and other complicated regimes occur at such frequencies
and amplitudes of the pivot, for which the decomposition of motion on the slow
and rapid components is not applicable.
A more accurate and enhanced criterion of dynamic stabilization of the in-
verted pendulum, valid in a wider region of system parameters, will be described
in the following chapter.

10.9 Subharmonic Resonances of High Orders

In this section we investigate in detail recently discovered kinds of motion of the
pendulum with oscillating pivot — namely the so called subharmonic resonances.
The boundaries of the region in the parameter space are determined in which these
resonances can exist, and their relationship with the dynamic stabilization of the
inverted pendulum is discussed. An enhanced and more exact criterion of the in-

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

10.9. SUBHARMONIC RESONANCES OF HIGH ORDERS 263

verted pendulum stability is determined, which is valid in particular for relatively

low frequencies and large amplitudes of the excitation.
The natural slow oscillatory motion in the effective potential well is almost
periodic (exactly periodic in the absence of friction). When the driving ampli-
tude and frequency lie within certain ranges, the pendulum, instead of gradually
approaching the equilibrium position (either dynamically stabilized inverted posi-
tion or ordinary downward position) by the process of damped slow oscillations,
can be trapped in an n-periodic limit cycle locked in phase to the rapid forced
vibration of the axis. In such oscillations the phase trajectory repeats itself after n
driving periods T . Since the motion has period nT , and the frequency of its funda-
mental harmonic equals ω/n (where ω is the driving frequency), this phenomenon
can be called a subharmonic resonance of n-th order.
For the inverted pendulum with a vibrating pivot, periodic oscillations of this
type were first described by Acheson [51], who called them “multiple-nodding”
oscillations. Computer simulations show that the pendulum motion in this regime
reminds us of some kind of original dance. We note that these modes of regular
periodic oscillations are not specific for the inverted pendulum with a vibrating
pivot. Similar ‘dancing’ oscillations can also be executed (at appropriate values
of the driving parameters) about the ordinary (downward-hanging) equilibrium
position.
Actually, the origin of these modes is independent of gravity, because such
“multiple-nodding” oscillations, synchronized with the pivot, can also occur in the
absence of gravity about any of the two equivalent dynamically stabilized equilib-
rium positions of the pendulum with a vibrating axis (see [52]). Synchronization
of these modes with the pivot oscillations creates conditions for supplying the en-
ergy to the pendulum needed to compensate for dissipation, and the whole process
becomes exactly periodic.

10.9.1 Multiple-Nodding Oscillations of the Parametrically

Driven Pendulum
An example of multiple-nodding stationary oscillations whose period equals eight
periods of the axis is shown in Figure 10.8. The left-hand upper panel of the
figure shows the spatial trajectory of the pendulum’s bob at these multiple-nodding
oscillations of the inverted pendulum.
The left-hand lower panel shows the closed looping trajectory in the phase
plane (ϕ, ϕ̇). The right-hand panel of Figure 10.8, alongside the graphs of ϕ(t)
and ϕ̇(t), also shows their harmonic components and the sinusoidal graphs of the
pivot oscillations.
The spectrum of these period-8 oscillations is rich in harmonics. The funda-
mental harmonic with frequency ω/8 whose period equals eight driving periods
dominates the spectrum. We may treat this low-frequency component of the spec-
trum as a subharmonic (as an ‘undertone’) of the driving oscillation. This har-
monic describes the above-discussed smooth component ψ(t) of the compound
period-8 oscillation.

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

264 CHAPTER 10. PENDULUM WITH OSCILLATING PIVOT

ο
79.0
π − ϕ (t)
z (t)
0

.
-ϕ -79.0
ο
.
-ϕ ( t ) .
z (t)

0 0
π −ϕ

ο ο 0 1 2 3 4 5 6 7 8T
-79.0 79.0
(quality 400.0, no gravity, amplitude of the axis 26.5%,
o
initial deflection 176.73 , initial angular velocity -0.2297w0)

Figure 10.8: The spatial path, phase orbit with Poincaré sections, and graphs of
large-amplitude stationary period-8 oscillations. The graphs are obtained by a nu-
merical integration of the exact differential equation, Eq. (10.11), with ω0 = 0, for
the momentary angular deflection ϕ(t). Thin lines show separate harmonics. The
fundamental harmonic with the frequency ω/8 dominates the spectrum. The sev-
enth and ninth harmonics have nearly equal amplitudes. Graphs of the axis motion
−z(t) and −ż(t) are also shown.

We note that at large swing the third harmonic (frequency 3ω/8) is noticeable
(see Figure 10.8). This spectral component reflects the non-harmonic character of
slow oscillations in the non-parabolic well of the effective potential.
The seventh and ninth harmonics with nearly equal amplitudes give consid-
erable contribution into the spectrum of these period-8 oscillations. Strange as it
may seem from the first sight, the 8th harmonic with the driving frequency has
zero amplitude, that is, this harmonic is absent in the spectrum. However, this pe-
culiarity also can be easily explained on the basis of the approach developed in
this paper.
Indeed, in Eq. (10.3), p. 254, which represents the momentary angular position
of the pendulum ϕ(t) as a superposition of slow and fast motions, the rapid com-
ponent with the driving frequency enters the expression for ϕ(t) being multiplied
by the sine of the slow varying coordinate ψ(t). Therefore the rapid component
has a varying amplitude, which even changes its sign each time the pendulum
crosses the equilibrium position.
Actually, the rapidly oscillating second term in Eq. (10.3) is not a harmonic
component in the spectrum of the resulting periodic oscillation, because harmon-
ics of a periodic function are characterized by constant amplitudes.

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

10.9. SUBHARMONIC RESONANCES OF HIGH ORDERS 265

Next we show that the approximate approach based on the effective potential
for the slow motion provides a simple qualitative physical explanation for such an
extraordinary (and even counterintuitive at first sight) behavior of the pendulum.
Moreover, for subharmonic resonances with n ≫ 1 this approach yields rather
good quantitative results.
The approximate theory developed above allows us to predict conditions at
which these n-periodic oscillations can occur. For small amplitudes of the slow
oscillations, the corresponding minimum of the effective potential can be approx-
imated by a parabolic well in which the smooth motion is almost harmonic.
The natural slow oscillatory motion in the effective potential well is almost
periodic (exactly periodic in the absence of friction). A subharmonic resonance of
order n can occur if one cycle of this slow motion covers approximately n driving
periods, that is, when the driving frequency ω is close to an integer multiple n of
the natural frequency of slow oscillations near either the inverted or the ordinary
equilibrium position: ω = nωup or ω = nωdown . In this case the phase locking can
occur, in which one cycle of the slow motion is completed exactly during n driving
periods. Synchronization of these modes with the oscillations of the pivot creates
conditions for systematically supplying the pendulum with the energy needed to
compensate for dissipation, and the whole process becomes exactly periodic.
To estimate the frequency of the slow motion (the fundamental frequency), we
can use Eq. (10.10), p. 257. As an example, next we consider the pendulum in the
absence of gravity, or, which is essentially the same, in the limiting case of very
high driving frequencies ω ≫ ω0 (ω/ω0 → ∞). In this limit both equilibrium
positions (ordinary and inverted) are equivalent, and the dimensionless driving
amplitude a/l is the only parameter to be predicted as a required condition for the
subharmonic resonance of order n (of n-periodic oscillations of the pendulum,
synchronized with the pivot).
√ for ω0 = 0 the frequency of slow oscillations
According to Eq. (10.10),
is given by ωslow = a/(l 2)ω. For the “quadruple-nodding” mode the slow
motion√period equals eight periods of the axis, so that ωslow = ω/8, whence
a/l = 2/8 = 0.177. This value agrees rather well with the predictions of a more
sophisticated quantitative theory of these modes based on the linearized differen-
tial equation of the system (see Section 10.9.2 below, Eq. (10.22) with n = 8),
which gives for such period-8 small oscillations in the absence √ of gravity the fol-
lowing expression for the driving amplitude: amin = 63/(32 130) l = 0.173 l.
The latter value agrees perfectly with the simulation experiment in the limit of
extremely small angular excursions.
Estimating conditions for n-periodic oscillations with the help of Eq. (10.10),
we assume the slow motion of the pendulum in the effective potential well to be
simple harmonic, which is true only if this motion is limited to a small vicin-
ity of the bottom of this well. Therefore we get the lower limit for the driving
amplitude at which n-periodic oscillations of only infinitely small amplitude can
occur. Smooth non-harmonic oscillations of a finite angular excursion that extends
over the slanting slopes of the non-parabolic effective potential well are character-
ized by a greater period than the small-amplitude harmonic oscillations occurring

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

266 CHAPTER 10. PENDULUM WITH OSCILLATING PIVOT

ο
55.0
π − ϕ (t)
z (t)
0

. ο
-ϕ - 55.0
.
-ϕ ( t ) .
z (t)

0
0 π −ϕ

ο
- 55.0 ο 0 55.0 0 1 2 3 4 5 6T
(quality 400.0, no gravity,
o
amplitude of the axis 26.5%,
initial deflection 175.77 , initial angular velocity -0.1708w0)

Figure 10.9: The spatial path, phase orbit with Poincaré sections, and graphs of
stationary period-6 oscillations. The graphs are obtained by a numerical integra-
tion of the exact differential equation, Eq. (10.11) with ω0 = 0, for the momentary
angular deflection ϕ(t). Thin lines show separate harmonics. The fundamental
harmonic with the frequency ω/6 dominates the spectrum. The fifth and seventh
harmonics have nearly equal amplitudes. Graphs of the axis motion −z(t) and
−ż(t) are also shown.

within the parabolic bottom of this well. Therefore large-amplitude period-8 os-
cillations shown in Figure 10.8 (their swing equals ∼ 80◦ ) occur at a considerably
greater value of the driving amplitude (a = 0.265 l).
For small angular excursions of the pendulum occurring at driving amplitudes
slightly greater than amin = 0.173 l, the spectrum of period-8 oscillations is
formed by the principal harmonic and also by the seventh and ninth harmonics
whose frequencies are close to the driving frequency. Their amplitudes equal, re-
spectively, 11.2% and 6.7% of the principal harmonic amplitude. These values
observed in the corresponding simulation experiment agree perfectly well with
the theoretical values; see Section 10.9.2 below, Eq. (10.24) on p. 271. For the
oscillations of a large swing shown in Figure 10.8, the amplitudes of these har-
monics slightly differ from the above values, and the contributions of the third,
fifth, and eleventh harmonics are also noticeable.
As noted above, in the case of period-8 oscillations of a small swing, the ap-
√ on the effective potential predicts for the driving amplitude a/l a
proach based
value of 2/8 = 0.177, which is rather close to the exact low-amplitude theoreti-
cal limit (a/l = 0.173). To obtain the slow oscillations of a smaller period (say, of
six driving periods; see Figure 10.9), we should increase the driving amplitude. In-

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

10.9. SUBHARMONIC RESONANCES OF HIGH ORDERS 267

√
deed, when ωslow = ω/6, Eq. (10.10) yields a greater value a/l = 2/6 = 0.236.
However, for such period-6 oscillations this predicted value agrees somewhat
worse with the theory based on the linearized differential equation of the system.
This theory (see Section 10.9.2 below, Eq. (10.22) with n = 6) gives for small
period-6 oscillations in the √
absence of gravity a value of the minimal driving am-
plitude of amin = 35/(18 74) l = 0.226 l, which perfectly agrees with the
corresponding simulation experiment. Not surprisingly, for the n-periodic oscil-
lation with a small n we cannot expect good quantitative predictions from the
effective potential approach because in such cases the period of a “smooth” mo-
tion contains only a few driving periods. The “rapid” component of the motion
here is not rapid enough for good averaging.
With gravity, these complex n-periodic multiple-nodding modes exist both for
the inverted and non-inverted pendulum. Assuming ωdown, up = ω/n (n driving
cycles during one cycle of the slow oscillation), we find for the minimal normal-
ized driving amplitudes (for the boundaries of the subharmonic resonances in the
presence of gravity) the values
p
mmin = 2(1/n2 ∓ k), (10.15)

where we have introduced a notation k = (ω0 /ω)2 . This dimensionless parameter

k (inverse normalized drive frequency squared), being physically less meaning-
ful than ω/ω0 , is nevertheless more convenient for further investigation, because
the improved criterion acquires a simpler form in terms of k. As we already indi-
cated above (see Section 10.7, p. 258), negative values of this parameter k corre-
spond to negative g values (negative ω02 values) in the exact differential equation,
Eq. (10.11), and can be treated as referring to the inverted pendulum. Then the
boundaries of subharmonic resonances can be expressed both p for the hanging-
down and inverted pendulums by the same formula: mmin = 2(1/n2 − k). The
limit of this expression at n → ∞ gives the above-mentioned approximate con-
dition of stability of the inverted pendulum, Eq. (10.8), p. 255, now expressed in
terms of parameter k: √
mmin = −2k. (10.16)
(In this expression k < 0, because it is applicable to the inverted pendulum.)
Indeed, the lower limit of stability of the inverted pendulum, Eq. (10.8), p. 255,
can be regarded as the condition of subharmonic resonance of an infinite order in
the inverted position — on the edge of stability the frequency of slow oscillation
about the inverted position tends to zero.

10.9.2 Spectrum of Small-Amplitude n-Periodic Oscillations

As we have seen from the above described simulations, the spectrum of station-
ary n-periodic oscillations of a small angular excursion consists primarily of the
fundamental harmonic A sin(ωt/n) with the frequency ω/n, and two high har-
monics of the orders n − 1 and n + 1. This spectrum composition is consistent

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

268 CHAPTER 10. PENDULUM WITH OSCILLATING PIVOT

ο
58.0
π − ϕ (t)
z (t)
0

.
-ϕ -58.0
ο
.
-ϕ ( t )

0
0 π −ϕ

ο
-58.0ο 0 58.0 0 1 2 3 4 5T
(no friction, no gravity, amplitude of the axis 31.9%,
o
initial deflection 180.0, initial angular velocity -0.1004w0)

Figure 10.10: The spatial path, phase orbit with Poincaré sections, and time-
dependent graphs of the subharmonic resonance of the fifth order. The graphs are
obtained by a numerical integration of the exact differential equation, Eq. (10.11),
for the momentary angular deflection of the pendulum ϕ(t). Separate harmon-
ics are shown by thin lines. The fundamental harmonic (frequency ω/5) domi-
nates the spectrum. Next the fourth and sixth harmonics (frequencies 4ω/5 and
6ω/5) contribute to a considerable extent. At large swing the second harmonic
(frequency 2ω/5) is also noticeable.

with the above-considered representation of the pendulum motion as a superposi-

tion of slow and rapid motions given by Eq. (10.3), p. 254. Indeed, according to
this equation with sin ψ ≈ ψ, in this approximation
ϕ(t) =ψ(t) − m sin ψ cos ωt ≈ ψ(t) − mψ cos ωt =
ω ω
=A sin( t) − mA sin( t) cos ωt = (10.17)
n n
ω mA n−1 mA n+1
=A sin( t) + sin( ωt) − sin( ωt).
n 2 n 2 n
This kind of spectral composition is clearly seen from the plots in Figures 10.8
and 10.9, as well as from Figure 10.10 for the resonance of the fifth order. While
the pendulum crosses the equilibrium position, both high harmonics add in the
opposite phases and thus almost don’t distort the smooth motion (described by
the principal harmonic). Near the utmost deflections the phases of high harmonics
coincide, and thus here their sum causes the most serious distortions of the smooth
motion.
According to Eq. (10.17), both high harmonics must have equal amplitudes
(m/2)A. However, we see from the above-mentioned plots that these amplitudes

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

10.9. SUBHARMONIC RESONANCES OF HIGH ORDERS 269

are slightly different. Therefore we can try to improve the approximate solution
for ϕ(t), Eq. (10.17), as well as the theoretical values for the lower boundaries
of subharmonic resonances, Eq. (10.15), by assuming for the possible solution a
similar spectrum but with unequal amplitudes, An−1 and An+1 , of the two high
harmonics (for n > 2, the case of n = 2 will be considered separately):

ϕ(t) = A1 sin(ωt/n)+An−1 sin[(n−1)ωt/n]+An+1 sin[(n+1)ωt/n]. (10.18)

Since oscillations at the boundaries have infinitely small amplitudes, we can

use instead of Eq. (10.11) the following linearized (Mathieu) equation:

ϕ̈ + 2γ ϕ̇ + (ω02 − mω 2 sin ωt)ϕ = 0. (10.19)

Substituting ϕ(t), Eq. (10.18), into this equation (with γ = 0) and expand-
ing the products of trigonometric functions, we obtain a system of approximate
equations for the coefficients A1 , An−1 and An+1 :

2(kn2 − 1)A1 + mn2 An−1 − mn2 An+1 = 0,

mn2 A1 + 2[n2 (k − 1) + 2n − 1]An−1 = 0, (10.20)
−mn2 A1 + 2[n2 (k − 1) − 2n − 1]An+1 = 0.

The homogeneous system has a nontrivial solution if its determinant equals

zero. This condition yields an equation for the corresponding critical (minimal)
driving amplitude mmin at which n-period mode ϕ(t), Eq. (10.18), can exist.
Solving the equation, we find:

2 [n6 k(k − 1)2 − n4 (3k 2 + 1) + n2 (3k + 2) − 1]

m2min = . (10.21)
n4 [n2 (1 − k) + 1]

Then, for this critical driving amplitude mmin , the fractional amplitudes An−1 /A1
and An+1 /A1 of high harmonics for a given order n of the subharmonic resonance
can easily be found as the solutions to the homogeneous system of equations,
Eqs. (10.20).

10.9.3 Lower Boundary of the Dynamic Stabilization

As we already mentioned, it is possible to identify the lower boundary of the
dynamic stabilization of the inverted pendulum with the condition of subharmonic
resonance of an infinite order. Therefore the limit of mmin , Eq. (10.21), at n → ∞
gives an improved formula for the lower boundary of the dynamic stabilization
of the inverted
√ pendulum instead of the commonly known approximate criterion
mmin = −2k, which is valid for k ≪ 1 (see Eq. (10.16), p. 267):
p
mmin = amin /l = −2k(1 − k) (k < 0). (10.22)

The minimal amplitude mmin that provides the dynamic stabilization is shown
as a function of parameter k = (ω0 /ω)2 (inverse normalized driving frequency

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

270 CHAPTER 10. PENDULUM WITH OSCILLATING PIVOT

m
0.8
n=2
0.7
n=3
0.6

0.5
n=6
0.4
n=
8

0.3

0.2

0.1

0 k
-0.2 -0.1 0 0.1 0.2 0.3 0.4

Figure 10.11: The normalized driving amplitude m = a/l versus k = (ω0 /ω)2
(inverse normalized driving frequency squared) at the boundaries of the dynamic
stabilization of the inverted pendulum (the left curve marked as n → ∞), and at
subharmonic resonances of several orders n (see text for details).

squared) by the left curve (n → ∞) in Figure 10.11. The other curves to the right
from this boundary show the dependence on k of minimal driving amplitudes for
which the subharmonic resonances of several orders can exist (the first curve for
n = 6 and the others for n values diminishing down to n = 2 from left to right).
At positive values of k these curves correspond to the subharmonic resonances of
the parametrically excited hanging-down pendulum.
Subharmonic oscillations of a given order n (for n > 2, case n = 2 will be
considered separately) are possible to the left of k = 1/n2 , that is, for the driving
frequency ω > nω0 .
The curves in Figure 10.11 show that when the driving frequency ω is in-
creased beyond the value nω0 (i.e., as parameter k is decreased from the critical
value 1/n2 toward zero), the threshold driving amplitude (over which n-order
subharmonic oscillations are possible) rapidly increases. The limit of very high
driving frequency (ω/ω0 → ∞), in which the gravitational force is insignificant
compared with the force of inertia (or, which is essentially the same, the limit of
zero gravity ω0 /ω → 0), corresponds to k = 0, that is, to the points of intersection
of the curves in Figure 10.11 with the m-axis.
The continuations of these curves further to negative k values describe the
transition through zero gravity to the “gravity” directed upward, which is equiv-

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

10.9. SUBHARMONIC RESONANCES OF HIGH ORDERS 271

alent to the case of an inverted pendulum in the ordinary (directed downward)

gravitational field. Therefore these curves at negative k values give the threshold
driving amplitudes for subharmonic resonances of the inverted pendulum.1
The curve n = 2 in Figure 10.11 corresponds to the upper boundary of
dynamic stabilization for the inverted pendulum: As we show below, when this
boundary is exceeded, the inverted pendulum destabilizes and is trapped into the
limit cycle of stationary oscillations with period 2T (so-called “flutter mode”).
Curves that lie between the lower (n → ∞) and upper (n = 2) boundaries of
dynamic stabilization correspond to subharmonic oscillations of different orders.
These oscillations do not disprove the criterion of the inverted pendulum’s sta-
bility because the pendulum is trapped in an n-periodic limit cycle only if the
initial state belongs to the basin of attraction that corresponds to this limit cycle.
Otherwise the pendulum eventually comes to rest in the inverted position.
A complete investigation of the parametrically excited pendulum is compli-
cated by the extensive set of parameters that characterize the system (ω0 , ω, a,
γ). A considerable simplification is achieved p by eliminating one of the parame-
ters, namely, the natural frequency ω0 = g/l, when we turn to studying the
pendulum in the absence of gravity. This simplified model is also useful for qual-
itative understanding of the pendulum’s behavior in the presence of gravity in
cases of high driving frequency and/or large driving amplitude, when the gravi-
tational force plays the role of a small addition to the force of inertia. Many of
the above-mentioned complicated counterintuitive modes are not related to the
force of gravity, and can be studied in their purest form when they are observed
in the simple device with the oscillating pivot in the absence of gravity, which is
described by Eq. (10.11) with ω0 = 0.
The points of intersection of the curves in Figure 10.11 with the m-axis, cor-
responding to the threshold conditions at zero gravity (k = 0), give, according to
Eq. (10.21), the following values of the normalized driving amplitudes:
√ 2
2(n − 1)
mmin = √ . (10.23)
n n2 + 1
2

The fractional amplitudes An−1 /A1 and An+1 /A1 of the most important high
harmonics of ϕ(t) [expressed approximately by Eq. (10.18)] for the case of zero
gravity (k = 0) are given by the following formulas:
An−1 n+1 An+1 n−1
= √ √ , = √ √ . (10.24)
A1 2
2 n + 1(n − 1) A1 2
2 n + 1(n + 1)

For n = 8 (quadruple-nodding
√ oscillations), Eq. (10.21) yields mmin =
amin /l = 63/(32 130) = 0.173. This critical value of the driving amplitude was
already mentioned in Section 10.9.1, p. 263, and it agrees exactly with the sim-
ulation experiment for period-8 small oscillations. The above Eqs. (10.24) also
1 Actually the curves in Figure 10.11 are plotted not according to Eq. (10.21), but rather with the

help of a somewhat more complicated formula (not cited here), which is obtained by holding one more
high order harmonic component in the trial function ϕ(t).

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

272 CHAPTER 10. PENDULUM WITH OSCILLATING PIVOT

ο
59.0
z (t) π − ϕ (t)

-59.0ο
0.39 .
-ϕ ( t )

0
0

ο -0.39
-59.0 0 59.0ο 0 1 2 3 4 5 6 7 8T
(quality 100.0, no gravity, amplitude of the axis 38.5%,
o
initial deflection 178.43 , initial angular velocity -0.3104w0)

Figure 10.12: The spatial path, phase orbit, and graphs of period-4 oscillations.
This example shows ‘double-nodding’ oscillations about one of the dynamically
stabilized equilibrium positions in the absence of gravity.

√ seventh and ninth harmonics: A7 /A1 =

yield√the fractional contributions of the
9/(7 130) = 0.113, A9 /A1 = 7/(9 130) = 0.068 — the values that also agree
perfectly well with the simulations based on numerical integration of the exact
differential equation. √
For n = 6 Eq. (10.23) yields mmin = amin /l = 35/(18 74) = 0.226, and
for fractional contributions of√the fifth and seventh harmonics Eq.√(10.24) gives,
respectively, A5 /A1 = 7/(5 74) = 0.163, and A7 /A1 = 5/(7 74) = 0.083.
These theoretical values agree quite well with the simulations (see Section 10.9.1).
For n = 4 (double-nodding √ oscillations) Eq. (10.23) and Eq.√ (10.24) yield
mmin = amin /l √ = 15/(8 34) = 0.321, A 3 /A1 = 5/(3 34) = 0.286,
A5 /A1 = 3/(5 34) = 0.103. If in the approximate solution we also take into
account the seventh harmonic, for zero gravity and zero friction we find more ac-
curate values of the critical driving amplitude: mmin = amin /l = 0.320, and of
the fractional contributions of high harmonics: A3 /A1 = 0.288, A5 /A1 = 0.102,
A7 /A1 = 0.015. We can compare these values with results of the simulation ex-
periment: amin /l = 0.320, A3 /A1 = 0.287, A5 /A1 = 0.101, A7 /A1 = 0.016.
Critical values for the driving amplitudes that provide small steady-state para-
metric oscillations with odd n values (n = 3, 5, . . . ) calculated on the basis of
a linearized theory also show good agreement with the simulations described in
Section 10.9.1 (p. 263). Thus, the boundaries mmin and amplitudes of high har-
monics An−1 /A1 and An+1 /A1 for subharmonic resonances of different orders n
calculated above with the help of an approximate theory agree perfectly well with
the simulation experiments based on numerical integration of the exact differen-

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

10.9. SUBHARMONIC RESONANCES OF HIGH ORDERS 273

ο
51.0
z (t) π − ϕ (t)

ο
-51.0
0.36 .
-ϕ ( t )

0 0

-0.36
ο
-51.0 0 51.0 ο 0 1 2 3 4 5 6T
(no friction, no gravity, amplitude of the axis 45.0%,
initial deflection 180.0,o initial angular velocity -0.2820w0)

Figure 10.13: The spatial path, phase orbit, and graphs of period-3 steady-state
oscillations without gravity in the absence of friction.

tial equation, Eq. (10.11). Figures 10.8, 10.9, and 10.10 show the graphs of these
subharmonic oscillations for n = 8, n = 6, and n = 5, respectively. Figures 10.12
and 10.13 show the graphs for n = 4 and n = 3, obtained in the simulations.
Friction introduces a phase shift between forced oscillations of the pivot and
harmonics of the steady-state n-periodic motion of the pendulum. By virtue of
this phase shift the pendulum is supplied with energy needed to compensate for
frictional losses. With friction, the direct and backward spatial paths of the pendu-
lum do not coincide, and the symmetry of the phase trajectory with respect to the
ordinate axis is destroyed. This is clearly seen from a comparison of Figures 10.8,
10.9 and 10.12 for subharmonic resonances in the presence of weak friction with
Figures 10.10 and 10.13, which refer to an idealized case in which friction is ab-
sent.

10.9.4 Subharmonic Resonances of Fractional Orders

In this section we discuss several new exotic modes of regular behavior of the
parametrically driven pendulum, kindred to the above-described subharmonic res-
onances. These modes were discovered recently in the simulation experiments.
Figure 10.14 shows a regular period-8 motion of the pendulum, which can
be characterized as a subharmonic resonance of a fractional order, specifically, of
the order 8/3 in this example. Here the amplitude of the fundamental harmonic
(whose frequency equals ω/8) is much smaller than the amplitude of the third
harmonic (frequency 3ω/8). This third harmonic dominates the spectrum, and can
be regarded as the principal one, while the fundamental harmonic can be regarded

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

274 CHAPTER 10. PENDULUM WITH OSCILLATING PIVOT

ο
59.0
z (t) ϕ (t)

ο
-59.0
1.97 . .
ϕ (t) z (t)

-1.97 ο ο
-59.0 0 59.0 0 1 2 3 4 5 6 7 8T
(quality 30.0, driving frequency 5.0w0, amplitude of the axis 40.0%,
o
initial deflection 3.42 , initial angular velocity 1.5876w0)

Figure 10.14: The spatial path, phase orbit, and graphs of stationary oscillations
that can be treated as a subharmonic resonance of a fractional order 8/3. Graphs of
separate harmonics are shown by thin lines. The third harmonic (frequency 3ω/8)
dominates the spectrum.

as its third subharmonic (that is, as an ‘undertone’). Considerable contributions to

the spectrum are given also by the fifth and eleventh harmonics of the fundamental
frequency.
Approximate boundary conditions for small-amplitude stationary oscillations
of this type (n/3-order subresonance) can be found analytically from the lin-
earized differential equation by a method similar to that used above for n-order
subresonance: We can try as ϕ(t) a solution consisting of spectral components
with frequencies 3ω/n, (n − 3)ω/n, and (n + 3)ω/n:
ϕ(t) = A3 sin(3ωt/n) + An−3 sin[(n − 3)ωt/n] + An+3 sin[(n + 3)ωt/n].
(10.25)
Substituting this trial function ϕ(t) into Eq. (10.19) (with γ = 0) and expand-
ing the products of trigonometric functions, we obtain a system of equations for
the coefficients A3 , An−3 and An+3 . The condition of existence of a non-trivial
solution to the system yields the following expression for the minimal driving
amplitude at which the corresponding subresonance of n/3-order can occur:
√
3 2(n2 − 32 )
mmin = √ . (10.26)
n2 n2 + 3 2
(Compare Eq. (10.26) with a similar expression, Eq. (10.23), for the critical driv-
ing amplitude of the integer-order subharmonic resonances.)
For the critical amplitude amin of the (8/3) fractional order subharmonic res-
onance, Eq. (10.26) yields (for the case of the absence of gravity) the value

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

10.9. SUBHARMONIC RESONANCES OF HIGH ORDERS 275

√
amin /l = 165/(32 146) = 0.427. (In the presence of gravity this resonance
is illustrated by Figure 10.14.) For the fractional amplitudes An−3 and An+3 of
harmonics in the trial function, Eq. (10.25) with √ n = 8, this approximate lin-
earized√theory gives the values A5 /A3 = 33/(5 146) = 0.546, A11 /A3 =
15/(11 146) = 0.113. More precise values (which agree well with the simu-
lations) are obtained by also including the thirteenth harmonic in the trial func-
tion, Eq. (10.25): amin /l = 0.419, A5 /A3 = 0.560, A11 /A3 = 0.111,
A13 /A3 = 0.044.
The analytical results of calculations for n ≥ 8 agree well with the simula-
tions, especially if one more high harmonic is included in the trial function ϕ(t).
If the driving amplitude exceeds the critical value, the angular excursion of the
pendulum at these modes increases, and additional harmonics appear in the spec-
tra of such oscillations.
Similar (though more complicated) calculations of the critical driving ampli-
tudes and spectrum on the basis of a linearized differential equation are also pos-
sible for various modes of the parametrically driven pendulum in the presence of
gravity and friction.

10.9.5 Coexistence of Subharmonic Resonances of Different

Orders n
As we have shown above, for subharmonic resonances of high orders in the√case
of zero gravity (n ≫ 1), Eq. (10.23) yields the approximate value mmin ≈ 2/n
obtained with the help of the simple approach which treats the condition of n-order
subharmonic resonance as the coincidence of n driving periods with one period
of the slow motion of the pendulum near the bottom of the effective potential
well. The fractional amplitudes of both high harmonics An−1 /A1 and An+1 /A1 ,
√ by Eq. (10.24), are almost equal at n ≫ 1 and approach the common value
given
1/( 2n) = mmin /2, in accordance with Eq. (10.17), which describes the n-
period subharmonic oscillations as a superposition of the slow and rapid motions.
The approach based on the effective potential provides us not only with a qual-
itative understanding of these complex periodic modes, but also, being applicable
for large-amplitude motions, explains the coexistence of several n-periodic modes
with different n values at identical values of the system parameters.
For the oscillations of a large swing shown in Figure 10.8, the contribution
of the third harmonic to the spectrum is also noticeable. In our approximate ap-
proach, the appearance of this spectral component is explained by deviations in
the shape of the effective potential well (in which the slow oscillation is executed)
from a parabolic well, that is, by the non-harmonic character of the slow oscilla-
tion with a large angular excursion.
For a large angular excursion, the smooth motion of the pendulum occurs in
the non-parabolic effective potential well with a ‘soft’ restoring force, in which the
period becomes longer if we increase the amplitude. By virtue of this dependence
of the periods of non-harmonic smooth motions on the swing, different modes
(modes with different values of n) can coexist at the same amplitude of the pivot.

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

276 CHAPTER 10. PENDULUM WITH OSCILLATING PIVOT

Figure 10.15: The principal harmonic amplitudes for n = 8 and n = 6 modes

versus the driving amplitude m = a/l given by an approximate theory (see text
for details) and by the simulation experiment.

Indeed, the period of a large-amplitude slow oscillation can be equal to, say, six
driving periods, while the period of oscillation with a somewhat greater amplitude
in the same non-parabolic potential well can be equal to eight driving periods.
Figures 10.8 and 10.9 show the simulations of such coexisting period-8 and
period-6 modes, respectively, obtained at the identical parameters of the system
(zero gravity, a/l = 0.265). That is, both smooth motions occur in the same po-
tential well. In which of these competing modes the pendulum eventually becomes
trapped in a certain simulation depends on the starting conditions. The set of ini-
tial conditions that leads, after an interval in which transients decay, to a given
dynamic equilibrium (to the same steady-state periodic motion, or attractor) in the
limit of large time, constitutes the basin of attraction of this attractor. The coexist-
ing periodic motions in Figures 10.8 and 10.9 represent competing attractors and
are characterized by different domains of attraction.
Figure 10.15 shows the dependence on the driving amplitude m = a/l of the
fundamental harmonic amplitudes A1 for both n = 8 and n = 6 modes.
To estimate how the swing of oscillations executed at the subharmonic reso-
nance of a given order n depends on the excess (a−amin ) of the driving amplitude
a over the critical (threshold) value amin , and how the fractional amplitude of the
third harmonic depends on the swing, we can expand sin ψ and sin 2ψ in the dif-
ferential equation that describes the slow motion, Eq. (10.12), p. 260, in a power
series, preserving the two first terms:
1 1 2
ψ̈ + ω02 (ψ − ψ 3 ) + m2 ω 2 (ψ − ψ 3 ) = 0. (10.27)
6 2 3
Here we again use the notation m = a/l for the normalized driving amplitude.
We can try to search for the solution of Eq. (10.27) in the form of a superposition
of the fundamental and third harmonics:
ψ = A1 sin ω1 t + A3 sin 3ω1 t. (10.28)

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

10.10. UPPER BOUNDARY OF DYNAMIC STABILITY 277

Substituting ψ from Eq. (10.28) into Eq. (10.27), and equating to zero the coef-
ficient of sin ω1 t, we find how the frequencies of slow oscillations depend on the
amplitude A1 :

2 1 2 2 1 1
ωdown, up = m ω (1 − A21 ) ± ω02 (1 − A21 ). (10.29)
2 2 8
This expression reduces to Eq. (10.10) if A1 → 0. Equating the frequencies
ωdown, up to the fundamental harmonic frequency ω/n, we obtain an approximate
dependence of the fundamental harmonic amplitude A1 on the excess of the nor-
malized driving amplitude over its critical value (m − mmin). For the case ω0 = 0
(absence of gravity) we find:
√ q p
A1 = 2 1 − m2min /m2 ≈ 2 1 − mmin /m. (10.30)

The latter approximate expression is valid if the driving amplitude only moder-
ately exceeds the critical value (if (m − mmin ) ≪ mmin ). For n = 8 and n = 6
the dependencies of A1 on m are plotted by solid curves in Figure 10.15 together
with experimental values of A1 obtained by numerical simulations. If the driving
amplitude m is greater than mmin = 0.226 for n = 6, each of the subharmonic
oscillations with n = 8 and n = 6 can exist at the same values of the driving
amplitude.
The amplitude A3 of the third harmonic in Eq. (10.28) can be estimated sim-
ilarly by equating to zero the coefficient of cos 3ω1 t, when ψ from Eq. (10.28)
is substituted into Eq. (10.27). It is convenient to express A3 as a function of the
amplitude A1 of the slow motion:
A31
A3 = . (10.31)
3(16 − 7A21 )
The corresponding graph is shown by a solid line in Figure 10.16. The points
refer to the simulation of the subharmonic oscillations with n = 8.

10.10 The Upper Boundary of Dynamic Stabiliza-

tion and the Principal Parametric Resonance
When the amplitude a of the pivot vibrations is increased beyond a certain critical
value amax , the dynamically stabilized inverted position of the pendulum loses
its stability. After a disturbance the pendulum does not come to rest in the up
position, no matter how small the release angle, but instead eventually settles into
a limit cycle, executing finite amplitude steady-state oscillation (about the inverted
vertical position). The period of such oscillation is twice the driving period, and
its swing grows as the excess of the drive amplitude over the threshold amax is
increased.
This loss of stability of the inverted pendulum was first described in 1992 by
Blackburn et al. [44] and demonstrated experimentally in [45]. The authors [44]

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

278 CHAPTER 10. PENDULUM WITH OSCILLATING PIVOT

Figure 10.16: The third harmonic amplitude for n = 8 mode versus the amplitude
of the principal harmonic given by the approximate theory (see text) and by the
simulation experiment.

called these limit-cycle oscillations the “flutter” mode. The graphs and the double-
lobed phase trajectory of such oscillations are shown in Figure 10.17.

10.10.1 The “Flutter” Mode and Ordinary Parametric

Resonance
Obviously, the steady-state “flutter” oscillations can be regarded as a special case
of subharmonic resonances, specifically, the case with n = 2. As we already men-
tioned, for small values of n it is impossible to correctly represent the pendulum
motion as consisting of the slow and rapid components. The driving amplitude
amax is not small compared with the length l of the pendulum. Consequently, this
case occurs beyond the limits of applicability of the approach based on the ef-
fective potential. This approach cannot explain the destabilization of the inverted
pendulum, as well as the loss of stability of the hanging-down pendulum at con-
ditions of ordinary parametric resonance. (In the latter case the driving amplitude
can be small, but the necessary driving frequency is not high enough for the sepa-
ration of rapid and slow motions.)
However, the simulation shows (see Figure 10.17) a very simple spectral com-
position of period-2 steady oscillations occurring over the upper boundary of
dynamic stability. Namely, the spectrum consists of the fundamental harmonic
whose frequency equals ω/2 (half the driving frequency ω) with a small addi-
tion of the third harmonic with the frequency 3ω/2. We note that large-amplitude
oscillations of the non-inverted pendulum in conditions of the principal paramet-
ric resonance are characterized by a similar spectrum (see Figure 10.18). This
similarity of the spectra is by no means occasional: Both the ordinary parametric
resonance and the period-2 “flutter” mode that destroys the dynamic stability of
the inverted state belong essentially to the same branch of possible steady-state
period-2 oscillations of the parametrically excited pendulum. We can treat this
branch as a subharmonic resonance of order n = 2.

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

10.10. UPPER BOUNDARY OF DYNAMIC STABILITY 279

ο
63.0
z (t) π − ϕ (t)

.
-ϕ - 63.0
ο
. .
-ϕ ( t) z (t)

0
0 π −ϕ

-63.0ο 0 63.0
ο 0 1 2 3 4 5 6
(quality 10.0, no gravity, amplitude of the axis 56.0%,
o
initial deflection 117.3 , initial angular velocity 0.07636 w0)

Figure 10.17: Stationary period-2 oscillations occurring over the upper boundary
of dynamic stability (the “flutter” mode). The spectrum consists of the fundamen-
tal harmonic (frequency ω/2) and the third harmonic (frequency 3ω/2).

Therefore the upper boundary of dynamic stability for the inverted pendu-
lum can be found directly from the linearized differential equation of the system,
Eq. (10.11), by the same method that is commonly used for determination of con-
ditions that lead to the loss of stability of the non-inverted pendulum through exci-
tation of ordinary parametric resonance (the ranges of parametric instability; see,
for example, [4]). We can apply the linearized Eq. (10.19) to this problem because
at the boundary of dynamic stability the amplitude of oscillations is infinitely
small. The periodic solution to Eq. (10.19), which corresponds to the boundary
of instability, can be represented as a superposition of the fundamental harmonic
whose frequency ω/2 equals half the driving frequency, and the third harmonic
with the frequency 3ω/2:

ϕ(t) = A1 sin(ωt/2) + A3 sin(3ωt/2). (10.32)

The phases of harmonics in (10.32) correspond to pivot oscillations in the form

z(t) = a cos ωt. Substituting ϕ(t) from Eq. (10.32) into the linearized differen-
tial equation, Eq. (10.19), with γ = 0, and expanding the products of trigono-
metric functions, we obtain an expression in which we should equate to zero
the coefficients of sin(ωt/2) and sin(3ωt/2). Thus we get a system of homo-
geneous equations for the coefficients A1 and A3 , which has a nontrivial solution
when its determinant equals zero. This requirement yields a quadratic equation
for the desired normalized critical driving amplitude amax /l = mmax . The rele-
vant root of this equation (in the case ω0 = 0, which corresponds to the absence
of gravity or to the high frequency limit of the pivot oscillations with gravity) is

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

280 CHAPTER 10. PENDULUM WITH OSCILLATING PIVOT

ο
65.0
ϕ (t)
z (t)
0

. − 65.0
ο
ϕ
1.57 1.57 0 1 . 2 3 4 5 6
ϕ (t)

0 0
ϕ

−1.57 ο −1.57
ο
− 65.0 0 65.0 0 1 2 3 4 5 6
(quality 8.0, driving frequency 3.35w0, amplitude of the axis 40.0%,
o
initial deflection -2.79 , initial angular velocity -1.357w0)

Figure 10.18: The spatial path, phase orbit, and graphs of stationary oscillations
with period 2T (ordinary parametric resonance). The spectrum consists of the
fundamental harmonic (frequency ω/2) and its third harmonic (frequency 3ω/2),
whose graphs are shown by thin lines.

√
mmax = 3( 13 − 3)/4 = 0.454. The corresponding ratio √ of amplitudes of the
third harmonic to the fundamental one equals A3 /A1 = ( 13 − 3)/6 = 0.101.
A somewhat more complicated calculation in which the higher harmonics
in ϕ(t) (up to the seventh) are taken into account yields for mmax and A3 /A1
the values that coincide (within the assumed accuracy) with those cited above.
These values agree well with the simulation experiment in conditions of the ab-
sence of gravity (ω0 = 0) and very small angular excursion of the pendulum.
When the normalized amplitude of the pivot m = a/l exceeds the critical value
mmax = 0.454, the swing of the period-2 “flutter” oscillation (amplitude A1 of
the fundamental
√ harmonic) increases in proportion to the square root of this ex-
cess: A1 ∼ a − amax . This dependence follows from the nonlinear differential
equation of the pendulum, Eq. (10.11), if sin ϕ in it is approximated as ϕ − ϕ3 /6,
and also agrees well with the simulation experiment for amplitudes up to 45◦ (see
Figure 10.19).
As the amplitude a of the pivot is increased over the value 0.555 l, bifurcation
of the symmetry-breaking occurs: The angular excursions of the pendulum to one
side and to the other become different, destroying the spatial symmetry of the
oscillation and hence the symmetry of the phase orbit.
As the pivot amplitude is increased further, after a = 0.565 l the system un-
dergoes a sequence of period-doubling bifurcations, and finally, at a = 0.56622 l
(for Q = ω/2γ = 20), the oscillatory motion of the pendulum becomes replaced,

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

10.10. UPPER BOUNDARY OF DYNAMIC STABILITY 281

Figure 10.19: Amplitude A1 of the fundamental harmonic in the “flutter” mode

over the upper boundary amax of dynamic√stabilization. The solid curve corre-
sponds to the theoretical dependence A1 ∼ a − amax .

at the end of a very long chaotic transient, by a regular unidirectional period-1

rotation.

10.10.2 Boundaries of the Second-Order Subharmonic

Resonance
Similar (though more complicated) theoretical investigation of the boundary con-
ditions for period-2 stationary oscillations in the presence of gravity allows us to
obtain the dependence of the critical (destabilizing) amplitude m of the pivot on
the driving frequency ω. In terms of parameter k = g/(lω 2 ) this dependence has
the following form:
1 p g
mmax = ( 117 − 232k + 80k 2 − 9 + 4k), k= . (10.33)
4 lω 2
The graph of this boundary as function of k is shown in Figure 10.11 (p. 270)
by the curve marked as n = 2. The critical driving amplitude tends to zero at
k → 1/4 (at ω → 2ω0 ). This condition corresponds to ordinary parametric reso-
nance of the hanging-down pendulum: At small driving amplitudes this resonance
is excited if the driving frequency equals the doubled natural frequency. If the
driving frequency exceeds 2ω0 (that is, if k < 0.25), a finite driving amplitude is
required for infinitely small steady parametric oscillations even in the absence of
friction.

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

282 CHAPTER 10. PENDULUM WITH OSCILLATING PIVOT

√
The curve n = 2 intersects the ordinate axis at m = 3( 13 − 3)/4 = 0.454.
This case (k = 0) corresponds to the above-mentioned limit of a very high driv-
ing frequency (ω/ω0 → ∞) or zero gravity (ω0 = 0), so that m = 0.454 gives
the upper limit of stability for each of the two dynamically stabilized equivalent
equilibrium positions: If m > 0.454 at g = 0, the “flutter” mode is excited. The
continuation of this curve to the region of negative values of k = g/(lω 2) corre-
sponds to the transition from ordinary downward gravity through zero to “nega-
tive,” or upward “gravity,” or, equivalently, to the case of the inverted pendulum
in an ordinary (directed down) gravitational field.
Thus, the same formula, Eq. (10.33), gives the driving amplitude (as a function
of the driving frequency) at which both the equilibrium position of the hanging-
down pendulum is destabilized due to excitation of ordinary parametric oscilla-
tions, and the dynamically stabilized inverted equilibrium position is destabilized
due to excitation of period-2 “flutter” oscillations. We can treat this as an indica-
tion that both phenomena belong to the same branch of the pendulum’s regular
behavior: They are closely related and have a common physical nature. All the
curves that correspond to subharmonic resonances of higher orders (n > 2) lie
between this curve and the lower boundary of dynamic stabilization of the in-
verted pendulum (curve n → ∞ in Figure 10.11).
Actually, Eq. (10.33) in the vicinity of k = 1/4 (ω = 2ω0 ) gives both bound-
aries of the instability interval that surrounds the principal parametric resonance,
which occurs at k = 1/4, that is, at ω = 2ω0 . For k > 1/4 (ω < 2ω0 ) Eq. (10.33)
yields negative m whose absolute value |m| corresponds to stationary oscillations
at the other boundary (to the right of k = 0.25, see Figure 10.11). Such oscil-
lations are also represented by two harmonic components with frequencies ω/2
and 3ω/2, but their phases differ from those in Eq. (10.32) — these harmonics are
of cosine type (for m > 0). For negative m, which physically means simply the
opposite phase of the pivot oscillations, harmonics of oscillations at the second
boundary will be of a sine type, just as in (10.32).
In Figure 10.20 the boundaries of the principal interval of parametric instabil-
ity form the “tongue” shown by curves 1 and 2 as functions of normalized driving
frequency ω/ω0 (instead of the more convenient but physically less meaningful
quantity k = (ω0 /ω)2 used in Figure 10.11).
For the hanging-down pendulum, in the absence of friction the critical ampli-
tude tends to zero as the frequency of the pivot approaches 2ω0 from either side.
Curve 3 shows in the parameters plane (ω/ω0 , a/l) the tongue-shaped region of
principal parametric resonance in the presence of friction (for Q = ω0 /2γ = 5.0).
The non-inverted vertical position of the pendulum whose pivot is vibrating at
frequency 2ω0 loses stability when the normalized amplitude of this vibration ex-
ceeds the threshold value of 1/2Q. This curve almost merges with curves 1 and 2
as the frequency ω deviates from the resonant value 2ω0 . (For a detailed discus-
sion of the role of friction see Section 10.10.3.) In the high-frequency limit, for
which the role of gravity is negligible, the normalized critical pivot amplitude a/l
tends to the value 0.454 that corresponds to destabilization of the two symmetric
equilibrium positions in the absence of gravity.

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

10.10. UPPER BOUNDARY OF DYNAMIC STABILITY 283

m
1.2
m = a/l
0.3
1
0.25
0.8
0.2
4
0.6
0.15
0.454 Q=5
0.4
0.1
Q = 10
5
3
0.2 0.05
1 Q=

8
2
w w
0
0 2 4 6 8 10 w 0 1.6 2. 0 2.4 2.8 w0

Figure 10.20: The boundaries of parametric instability (driving amplitude versus

normalized driving frequency). Curves 1 and 2 correspond to frequency bound-
aries of the principal interval of parametric instability (ω ≈ 2ω0 ) for the non-
inverted pendulum in the absence of friction, curve 3 — the same with friction
(Q = 5.0). Curves 4 and 5 correspond to the upper and lower boundaries of dy-
namic stability for the inverted pendulum. Curve 3 is also shown in an increased
scale (right).

Curve 4 of this diagram corresponds to destabilization of the inverted pendu-

lum by excitation of the “flutter” oscillations. The smaller the frequency of the
pivot, the greater the critical amplitude at which the inverted position becomes
unstable. Actually curve 4 for the boundary of the “flutter” mode is the con-
tinuation (through infinite values of the driving frequency) of curve 2 (or curve
3 in the presence of friction). The latter is the boundary of ordinary parametric
resonance of the non-inverted pendulum. This relationship between the two phe-
nomena becomes especially obvious if we compare curve 4 with its equivalent in
Figure 10.11, which is the curve marked as n = 2 at negative k values.
Curve 5 in Figure 10.20 shows in the parameter plane the lower boundary
of dynamic stabilization of the inverted pendulum, which is defined by crite-
rion (10.22). Thus, the region of stability of the inverted pendulum occupies the
shaded panel of the parameter plane between curves 5 and 4.

10.10.3 The Influence of Friction

For small (and moderate) driving amplitudes, the principal parametric resonance
occurs at a driving frequency (frequency of the pivot) whose value is approxi-
mately twice the natural frequency of the pendulum: ω ≈ 2ω0 . We can calculate
the threshold of parametric excitation of the hanging-down pendulum by vertical

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

284 CHAPTER 10. PENDULUM WITH OSCILLATING PIVOT

oscillations of the pivot in the condition of the principal resonance by equating the
work done by the force of inertia during a cycle of a steady-state motion of the
pendulum to the energy dissipated due to friction.
For the calculation of the threshold, it is convenient to consider that the pivot’s
motion is described by a sine function z(t) = −a sin ωt. This specific phase of
the pivot oscillation can be provided by an appropriate choice of the time origin.
In this case the small steady-state oscillations at the threshold are approximately
described by a cosine function: ϕ(t) = C1 cos(ωt/2).
The torque of the force of inertia is Fin l sin ϕ, and the elementary work dW
done by this torque during an infinitesimal time interval dt is

Fin l sin ϕdϕ = Fin l sin ϕ ϕ̇dt =

a 2 a (10.34)
−I ω sin ωt sin ϕ ϕ̇dt ≈ −I ω 2 sin ωt ϕϕ̇dt.
l l
Here I is the pendulum’s moment of inertia. Integrating this expression over the
period of the pendulum motion T = 2π/ω0 = 4π/ω, we find the total work
done during T , that is, the increment ∆E in the total energy E during two driving
periods due to the parameter variation:

∆E = Iω 2 C12 (a/l)ω/2π. (10.35)

The work of the frictional force determines the dissipation of mechanical en-
ergy. The elementary (negative) work dW done by the torque of this force during
dt is −2Iγ(ϕ̇)2 dt. Integrating this work over the period of oscillation, we find
−Iγc21 ω 2 π. We note that both the frictional losses and the energy supplied by
oscillations of the pivot are proportional to the square of the amplitude C1 . This
means that over the threshold, friction cannot restrict the growth of the ampli-
tude. Equating the absolute values of the work done by the force of inertia and by
the frictional force yields ω(a/l) = 2γ. Since at resonance ω ≈ 2ω0 , we obtain
the following approximate expression for the threshold value of the normalized
amplitude of the pivot:
athres γ 1
mthres = = = . (10.36)
l ω0 2Q
If this threshold value is exceeded, parametric resonance occurs in some interval of
driving frequencies extending on both sides of the resonant frequency ωres = 2ω0 .
For a given value of the driving amplitude, the wider the interval, the smaller the
friction. To find the boundaries of parametric instability in the presence of friction,
we should include the damping term 2γ ϕ̇ into the linearized differential equation
of the pendulum, Eq. (10.19). With friction, the solution to this equation includes,
in contrast to Eq. (10.32), both sine and cosine terms:

ϕ(t) = A1 sin(ωt/2)+A3 sin(3ωt/2)+B1 cos(ωt/2)+B3 cos(3ωt/2). (10.37)

Substituting ϕ(t) given by Eq. (10.37) into Eq. (10.19), we obtain the ho-
mogeneous system of approximate equations for A1 , A3 , and B1 , B3 . Desired

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

10.10. UPPER BOUNDARY OF DYNAMIC STABILITY 285

167 ο
ϕ (t)
z (t)
0

. −167 ο
ϕ 0 1 2 3 4T
1.97 1.97 . .
ϕ (t) ϕnat( t )

0
ϕ

−1.97 ο −1.97
ο
−167 0 167 0 1 2 3 4T
(quality 40.0, driving frequency 0.88w0, amplitude of the axis 10.0%,
initial deflection 150.67 o, initial angular velocity -0.4651w0)

Figure 10.21: The spatial path, phase orbit, and graphs of stationary oscillations
with large swing at ω < 2ω0 . The spectrum consists of the fundamental har-
monic (frequency ω/2) and its third harmonic (frequency 3ω/2), whose graphs
are shown by thin lines.

boundaries of parametric instability are found from the condition of existence of

a non-trivial solution to this system. Expressions for the boundaries that follow
from this calculation are rather complicated and not cited here. The correspond-
ing graph (for Q = 5) is shown by curve 3 in Figure 10.20.
It occurs that friction influences noticeably the boundaries of parametric in-
stability only in the vicinity of ω = 2ω0 , that is, at small amplitudes of the pivot
amplitude m = a/l. These boundaries are shown in detail by three thick curves
in the right panel of Figure 10.20 for Q = 5, Q = 10, and for the absence of fric-
tion (Q → ∞). Thin curves are plotted according to the following approximate
expression (10.38), which is valid for small driving amplitudes (m = a/l ≪ 1) in
the vicinity of ω = 2ω0 :
p
ω1,2 = (2 ± 2 m2 − 1/(2Q)2 + 7m2 /2)ω0 . (10.38)

If the driving parameters lie in the region inside these “tongues,” the hanging
equilibrium position is unstable, and the pendulum leaves it after the slightest
perturbation. The growth of the amplitude is restricted by nonlinear effects (by
dependence of the natural frequency on the amplitude). The growth of the natural
period with the amplitude at large swing causes a violation of conditions that
are favorable for parametric resonance (two excitation cycles during one natural
period). As a result, an oscillation of a finite swing is established after fading of
the transient beats.

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

286 CHAPTER 10. PENDULUM WITH OSCILLATING PIVOT

However, by virtue of the same nonlinear properties of the pendulum, station-

ary parametric oscillations with an amplitude approaching 180◦ are possible in the
system with small friction even at small values of the driving amplitude. An ex-
ample of such parametric oscillations is shown in Figure 10.21. Due to the growth
of the natural period of the pendulum at large swing, synchronization of its motion
with the pivot oscillations (phase locking) can occur at ω < 2ω0 . If the pendulum
swing equals 167◦ , the driving frequency that provides the phase locking should
be only 0.88 ω0 (compare with condition ω = 2ω0 of parametric resonance at
small swing).
Parametric oscillations shown in Figure 10.21 are very close in their properties
to the nonlinear natural non-damping oscillations of the same swing. To illustrate
this similarity, the thin line in the lower panel of Figure 10.21 shows for com-
parison the graph of angular velocity ϕ̇nat (t) for natural oscillations of the same
swing in the absence of friction. As we can clearly see from the graphs, the period
of these natural oscillations is the same as the period of corresponding parametric
oscillations in the system with friction, and the shape of graphs ϕ̇nat (t) and ϕ̇(t)
is exactly coinciding.
The role of the pivot oscillations in this case reduces to supplying the pen-
dulum with energy needed to compensate for frictional losses. As a whole, such
parametrically excited oscillations have much in common with the so-called ‘bell-
ringer mode,’ which can be observed at direct forced excitation of a rigid pendu-
lum by a sinusoidal external torque (see Chapter 9).

10.11 An Enhanced Criterion for Kapitza’s

Pendulum Stability
In this section an enhanced and more precise criterion for dynamic stabilization of
the parametrically driven inverted pendulum is obtained: The boundaries of sta-
bility are determined with greater precision and are valid in a wider region of the
system parameters than the above-discussed approximate results. As we already
mentioned, the lower boundary of stability is associated with the phenomenon
of subharmonic resonances in this system, namely, with the subharmonic reso-
nance of an infinitely large order. The upper boundary of dynamic stabilization of
the inverted pendulum is related to the ordinary parametric resonance (i.e., to the
parametric destabilization of the lower equilibrium position). These relationships
allow us to determine the boundaries of dynamic stabilization in a wider region
of the system parameters (including relatively low frequencies and large ampli-
tudes of excitation), and with a greater precision compared to previous results.
Computer simulation of the physical system aids the analytical investigation and
proves the theoretical results.
Among new discoveries regarding the inverted pendulum, the most impor-
tant for finding the stability boundaries are the destabilization of the (dynam-
ically stabilized) inverted position at large driving amplitudes through excita-
tion of period-2 (“flutter”) oscillations [44]–[45], and the existence of n-periodic

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

10.11. ENHANCED CRITERION OF STABILITY 287

. .
z(t) ϕ (t)

0 2 4 6 8 10 12 14 16 T
time
amplitude

frequency
0 1 2 3ω

Figure 10.22: Angular velocity ϕ̇(t) time dependence (with the graphs of separate
harmonics) and its spectrum for subharmonic resonance of the eighth order.

multiple-nodding regular oscillations [51]. The relationship of “flutter” oscilla-

tions in the inverted pendulum with ordinary parametric resonance of the hang-
ing pendulum is discussed in [49]. A physical interpretation of periodic multiple-
nodding oscillations as subharmonic parametric resonances is given above in Sec-
tion 10.9.1 (see also [52]).

10.11.1 Subharmonic Resonances at Arbitrary Frequencies

and Amplitudes of the Pivot Vibration
In the above discussion in Section 10.9.2 of conditions at which subharmonic res-
onances can occur, we assumed that steady-state oscillations whose period equals
n periods of the pivot vibration consist of the principal harmonic with the fre-
quency ω/n, and of two harmonic components with frequencies (1 − 1/n)ω and
(1 + 1/n)ω; see Eq. (10.18), p. 269, for the trial function ϕ(t). This assumption
is valid if the pivot is forced to oscillate at sufficiently small amplitudes and high
enough frequencies. The assumption is also supported by an example shown in
Figure 10.22 of such stationary subharmonic oscillations whose period equals 8
cycles of the pivot’s motion. The graphs show time dependence of the angular
velocity ϕ̇(t) together with its harmonics, and the spectrum of the velocity.
The fundamental harmonic component whose period equals 8 driving peri-
ods dominates the spectrum. We may treat it as a subharmonic (as an ‘under-
tone’) of the driving oscillation. This principal harmonic of the frequency ω/n
describes the smooth component ψ(t) of the compound period-8 oscillation:
ψ(t) = A sin(ωt/n).
However, Figure 10.22, as well as Figure 10.8 (see p. 264), obtained by nu-
meric integration of the exact differential equation for the investigated system,
show the presence of small contributions in ϕ(t) of harmonics with frequencies

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

288 CHAPTER 10. PENDULUM WITH OSCILLATING PIVOT

(2n − 1)ω/n and (2n + 1)ω/n (the graphs of these harmonics are small and hence
not shown in the above-mentioned figures).
We may also include such harmonic components in the trial function in our
attempt to improve the desired criterion and to make this criterion applicable in a
wider region of the system parameters:
ω n−1 n+1
ϕ(t) = A1 sin( t) + An−1 sin( ωt) + An+1 sin( ωt) +
n n n
2n − 1 2n + 1
+A2n−1 sin( ωt) + A2n+1 sin( ωt). (10.39)
n n
Since oscillations at the threshold of the subharmonic resonance have in-
finitely small amplitudes, we can use instead of the exact differential equation,
Eq. (10.11), the linearized (Mathieu) equation, Eq. (10.19), p. 269. Substituting
ϕ(t) from Eq. (10.39) into this equation (with γ = 0) and expanding the prod-
ucts of trigonometric functions, we obtain the following system of approximate
equations for the coefficients A1 , An−1 and An+1 , A2n−1 and A2n+1 :

2(kn2 − 1)A1 + mn2 An−1 − mn2 An+1 = 0,

mn2 A1 + 2[n2 (k − 1) + 2n − 1]An−1 − mn2 A2n−1 = 0,
−mn2 A1 + 2[n2 (k − 1) − 2n − 1]An+1 + mn2 A2n+1 = 0, (10.40)
mn2 An−1 + 2[n2 (k − 4) + 4n − 1]A2n−1 = 0,
mn2 An+1 + 2[n2 (k − 4) − 4n − 1]A2n+1 = 0.

Here k is the parameter that enters into Eq. (10.19): k = g/(lω 2). The ho-
mogeneous system (10.40) has a nontrivial solution if its determinant equals zero.
This condition yields an equation (not cited here) for the corresponding thresh-
old (minimal) normalized driving amplitude mmin = amin /l at which n-periodic
mode ϕ(t) given by expression (10.39) can exist.
The equation for the threshold driving amplitude can be solved numerically
with the help of, say, the Mathematica package by Wolfram Research, Inc.). Then,
after substituting this critical driving amplitude mmin into the system (10.40),
fractional amplitudes An−1 /A1 , An+1 /A1 , A2n−1 /A1 and A2n+1 /A1 of high
harmonics for a given order n can be found as the solutions to the homogeneous
system of Eqs. (10.40).
If we ignore the contribution of harmonics with frequencies (2n − 1)ω/n and
(2n + 1)ω/n in ϕ(t), that is, assume A2n−1 and A2n+1 to be zero, system (10.40)
simplifies considerably. The corresponding approximate solution can be found
above in Section 10.9.2, Eq. (10.21).
For the full system (10.40) the final expressions for mmin and for the ampli-
tudes of harmonics are too bulky to be cited here. We have used them in Fig-
ure 10.23 for plotting the curves of mmin as functions of parameter k = (ω0 /ω)2
(inverse normalized driving frequency squared). The curves in Figure 10.23 cor-
respond to subharmonic oscillations of different orders n (thin curves).
To verify our analytical results for subharmonic oscillations in a computer
simulation, we choose a value k = −0.3, corresponding to the drive frequency

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

10.11. ENHANCED CRITERION OF STABILITY 289

m
1.5

1.25

0.75

0.5 n=4
n=5 n=3
n=6 n=2
n
8

0.25

k
-0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6

Figure 10.23: The normalized driving amplitude m = a/l versus k = (ω0 /ω)2
(inverse normalized driving frequency squared) at the lower boundary of the dy-
namic stabilization of the inverted pendulum (the left curve marked as n → ∞),
and at subharmonic resonances of several orders n (see text for details).

ω = 1.826 ω0, for which the approach based on separation of slow and rapid
motions of the pendulum is obviously inapplicable.
The above-described calculation applied to the subharmonic oscillation of or-
der n = 8 predicts for the threshold normalized drive amplitude mmin = amin /l
a value 87.73% of the pendulum’s length.
Results of the simulation (based on numerical integration of the exact dif-
ferential equation of the system) are presented in Figure 10.24. This simulation
perfectly confirms the theoretical prediction.
The set of Poincaré sections in the phase plane consists of eight fixed points,
and the phase orbit becomes closed after eight cycles of the pivot oscillations. The
fractional amplitudes of harmonics obtained in the simulation agree perfectly well
with the theoretical prediction.

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

290 CHAPTER 10. PENDULUM WITH OSCILLATING PIVOT

.
-ϕ
0.02 . .
z (t) −ϕ ( t )

0
π −ϕ

−0.02 ο ο
−2.0 0 2.0 ο0 1 2 3 4 5 6 7 8T
2.0
z (t) π −ϕ ( t)

ο
−2.0
0 1 2 3 4 5 6 7 8T
(quality 5000, driving frequency 1.82574w0, amplitude of the axis 87.7265%,
initial deflection 180 o, initial angular velocity 0.005w0)

Figure 10.24: Phase trajectory with Poincaré sections, angular velocity −ϕ̇(t) and
angle π − ϕ(t) time dependencies (with the graphs of separate harmonics) for
subharmonic resonance of the eighth order. For convenience of presentation, the
angle is measured from the inverted position.

10.11.2 Improved Lower Boundary of Dynamic Stabilization

As we already noted, the criterion of stability for the inverted pendulum can be
related to the condition of the subharmonic resonance of an infinitely large order
n, which occurs in the vicinity of the inverted position ϕ = ±π. Hence the limit
of mmin at n → ∞ gives an improved formula for the lower boundary of dynamic
stabilization of the inverted pendulum.
If we use the approximate solution of (10.40) in which the contribution of
higher harmonics with frequencies (2n − 1)ω/n and (2n + 1)ω/n p is ignored, the
limit n → ∞ gives for the lower boundary the value mmin = −2k(1 − k) in the
region k < 0 (see Eq. (10.22), p. 269). For the solution of the full system (10.40)
with higher harmonics A2n−1 and A2n+1 included, the limit of mmin at n → ∞
yields the following expression:
r
k(k − 1)(k − 4)
mmin = 2 , (10.41)
3k − 8
which should
√ be used instead of the commonly known approximate criterion
mmin = −2k, given by Eq. (10.7), Eq. (10.8), or Eq. (10.16). The minimal
amplitude mmin = amin /l that corresponds to the improved criterion (10.41) of
dynamic stabilization is shown as a function of k = (ω0 /ω)2 by the thick left

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

10.11. ENHANCED CRITERION OF STABILITY 291

curve marked as n → ∞ in Figure 10.23. This curve is localized wholly in the

region of negative k values.
To compare the improved criterion (10.41) with the commonly known approx-
imate criterion of the inverted pendulum stability given by Eq. (10.16), the latter
is also shown in Figure 10.23 by the dashed thin left curve (red in the electronic
version). We note how these two curves diverge dramatically at low frequencies
and large amplitudes of the pivot oscillations.
The other curves to the right from this boundary show the dependence on k
of minimal driving amplitudes for which the subharmonic resonances of several
orders can exist (the first curve for n = 6, and the others for n values dimin-
ishing down to n = 2 from left to right). At negative k values these curves give
the threshold drive amplitudes for subharmonic oscillations about the inverted po-
sition. Case k = 0 corresponds to zero gravity (or infinitely high frequency of
the pivot vibration). Points of intersection of the curves with the ordinate axis on
this diagram give minimal drive amplitudes for which in the absence of gravity
subharmonic oscillations of certain order n can exist about any of the two dynam-
ically stabilized positions (Figure 10.8, p. 264 shows an example of such period-8
oscillations) in the absence of gravity.
Continuations of the curves to positive k values correspond to subharmonic
parametric resonances (multiple-nodding oscillations) about the downward equi-
librium position. The curve for n = 2 corresponds to ordinary parametric res-
onance, in which two cycles of excitation take place during one full oscillation
of the pendulum. In the absence of friction the threshold drive amplitude for this
resonance tends to zero at ω → ω0 /2, that is, at k → 1/4. From Figure 10.23 we
see clearly that the curve, corresponding to n = 2 subharmonic oscillations of the
inverted pendulum (the “flutter” mode), and the principal parametric resonance of
ordinary (hanging) pendulum belong to essentially the same branch of period-2
regular behavior. Indeed, in the k > 0 region this branch gives the boundaries of
the ordinary parametric resonance of the hanging-down pendulum, while in the
k < 0 region this branch gives the upper boundary of dynamic stabilization for
the inverted pendulum (see Section 10.10 and Section 10.10.2).
We note that the existence of subharmonic oscillations in the same region of
the k—m plane does not disprove criterion (10.41) of the inverted pendulum sta-
bility. Indeed, the pendulum is trapped into an n-periodic subharmonic limit cycle
(with n > 2) only if the initial state belongs to a certain small basin of attraction
that corresponds to this limit cycle. Otherwise the pendulum eventually comes
to rest in the inverted position (or to unidirectional rotation, if the pendulum is
released beyond a certain critical initial deviation).
For experimental verification of the improved criterion (10.22) we again
choose relatively low drive frequency ω = 1.826 ω0 (k = −0.3), for which dis-
tinctions between the conventional and improved criteria are especially noticeable.
At this frequency Eq. (10.41) gives for the lower boundary of stability the drive
amplitude amin = 0.868 l. The upper boundary (see Section 10.10) at k = −0.3
equals amax = 0.929 l. Computer simulations in Figure 10.25 show how within
this narrow region of stability, at a = 0.875 l (just over the lower boundary, Fig-

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

292 CHAPTER 10. PENDULUM WITH OSCILLATING PIVOT

ο
0.14 11.0
z (t) ϕ (t)
(a)

0 0

ο
−0.14 ο −ο 11.0
−11.0 0 11.0 0 2 4 6 8 10 12 14 16 T
(quality 50.0, axis frequency 1.8257, axis ampl. 87.5, initial defl. 170.0 degr., ang. velocity 0.0)
ο
0.14 10.0
z (t) ϕ (t)
( b)

0 0

ο
−0.14 ο −10.0
ο
−10.0 0 10.0 0 2 4 6 8 10 12 14 16 T
(quality 50.0, axis frequency 1.8257, axis ampl. 92.5, initial defl. 170.0 degr., ang. velocity 0.0)

Figure 10.25: Phase trajectories and time histories of gradually damping oscilla-
tions about the inverted position just over the lower (a) and just below the upper
(b) boundaries of dynamic stabilization.

ure 10.25a) and at a = 0.925 l (just below the upper boundary, Figure 10.25b), the
pendulum, being initially deviated through 10◦ and released with zero velocity, in
both cases shown in Figure 10.25 returns gradually to the inverted position.
We note the peculiarities of the transients shown in Figure 10.25 that lead to
the state of rest in the inverted position. In both cases Poincaré sections, corre-
sponding to time instants tq = qT (q = 0, 1, 2 . . ., T = 2π/ω – period of the
pivot oscillations), are located near the ϕ-axis and gradually condense, approach-
ing the origin of the phase plane. This origin corresponds to the state of rest in the
inverted position.
Just over the lower boundary of stability the graph of time history in Fig-
ure 10.25a resembles, during a limited time interval, the corresponding graph of
a subharmonic resonance (of a multiple-nodding oscillation) of some high order
(compare with Figure 10.24): The pendulum executes many ‘nods’ on one side of
the inverted position, then on the other side with a somewhat smaller amplitude,
and so on, gradually approaching the upper vertical. These intermittent damping
‘nods’ are described by one-sided shrinking loops of the phase orbit that pass from
one side of the phase plane to the other each time the pendulum crosses the verti-
cal line. The greater the number of such ‘nods,’ the closer to the upper boundary:
We remember that this boundary corresponds to the subharmonic resonance of
infinite number in which the pendulum makes an infinite number of ‘nods’ before
crossing the upper vertical.

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

10.11. ENHANCED CRITERION OF STABILITY 293

3.97 .
. ϕ (t)
z (t)

0 0

-3.97 ο ο
-180 0 180
ϕ (t)

z (t)
ο
-180
0 2 4 6 8 10 12 14 16 T
(quality 40.0, driving frequency 1.82574w0, amplitude of the axis 80.0%,
initial deflection 179 o, initial angular velocity 0.0w0)

Figure 10.26: Chaotic oscillations below the boundary of stability.

Conversely, just below the upper boundary (Figure 10.25b), the character of
gradually damping oscillations about the inverted position reminds us of the “flut-
ter” mode (see Section 10.10) with its double-lobed phase curve (Figure 10.17):
One cycle of the pendulum oscillations covers approximately two drive periods
and is represented by a double-lobed phase orbit. This orbit shrinks gradually
around the origin of the phase plane. We note that near the upper boundary the
graph of time history and the phase trajectory (Figure 10.25b) resemble those of
a “flutter” oscillation. Over this upper boundary the pendulum eventually occurs
trapped into the period-2 steady-state oscillation instead of returning to the vertical
position.
If the initial deflection exceeds some critical value, at first the pendulum goes
slowly further from the vertical, then executes random revolutions to one and
the other side, and eventually (after a long ‘tumbling’ chaotic transient) becomes
trapped into period-1 unidirectional rotation. The smaller this critical deviation,
the closer the drive parameters are to the boundary of stability. For k = −0.3,
m = 0.875 and Q = 50 the initial deviation from the inverted position should
not exceed 14◦ . Friction reduces the basin of attraction of the equilibrium in the
inverted state: At Q = 20 the initial deviation should not exceed 10◦ .
At smaller than 0.868 l values of the drive amplitude the inverted pendulum is
unstable. Figure 10.26 shows how at k = −0.3 and a = 0.800 l the pendulum,
being released at only 1◦ deviation from the inverted position, occurs eventually in
a chaotic regime (‘tumbling’ chaos). The graphs in Figure 10.26 show the initial
stage of the time history. The set of Poincaré sections in the phase plane gives an
impression of the further random behavior, characterized by a strange attractor. We

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

294 CHAPTER 10. PENDULUM WITH OSCILLATING PIVOT

note that the inverted pendulum at these drive parameters should be stable accord-
ing to conventional criterion, Eq. (10.22), which at k = −0.3 gives a considerably
smaller value amin = 0.775 l (compare with improved value amin = 0.868 l) for
the lower boundary of stability.

10.11.3 Improved Upper Boundary of Dynamic Stabilization

The curve n = 2 in Figure 10.23 (its part in k ≤ 0 region) corresponds to the
upper boundary of dynamic stabilization for the inverted pendulum: After a dis-
turbance the pendulum does not come to rest in the up position, no matter how
small the release angle, but instead eventually settles into a limit cycle, executing
finite amplitude steady-state oscillation (about the inverted vertical position). The
period of such an oscillation is twice the driving period, and its swing grows as
the excess of the drive amplitude over the threshold amax is increased.
As we already mentioned, this loss of stability of the inverted pendulum has
been first described in 1992 by Blackburn et al. [44] and demonstrated experi-
mentally in [45]. The authors [44] called these limit-cycle oscillations the “flutter”
mode. Above we have shown (see Section 10.10.1, p. 278) that the “flutter” mode
and the principal parametric resonance belong to the same branch of the period-
2 stationary regime (this unambiguously follows from Figure 10.23). Hence the
same analytical method can be used to calculate conditions of their excitation.
Simulations show a very simple spectral composition for both, namely a superpo-
sition of the fundamental harmonic whose frequency ω/2 equals half the driving
frequency, the third harmonic with the frequency 3ω/2, and maybe a tiny admix-
ture of the fifth harmonic:

ϕ(t) = A1 cos(ωt/2) + A3 cos(3ωt/2) + A5 cos(5ωt/2). (10.42)

The phases of harmonics in (10.42) correspond to pivot oscillations in the

form z(t) = a cos ωt. Substituting ϕ(t) into the linearized differential equation,
Eq. (10.19), p. 269, with γ = 0 and expanding the products of trigonometric
functions, we obtain an expression in which we should equate the coefficients of
cos(ωt/2), cos(3ωt/2), and cos(5ωt/2) to zero. Thus we get a system of homo-
geneous equations for the coefficients A1 , A3 , and A5 of harmonics in the trial
function (10.42):

(4k − 2m − 1)A1 − 2mA3 = 0,

−2A1 + (4k − 9)A3 − 2mA5 = 0, (10.43)
−2mA3 + (4k − 25)A5 = 0.

This system has a nontrivial solution when its determinant equals zero. If we ne-
glect the contribution of the fifth harmonic in ϕ(t), Eq. (10.42), that is, if we let
A5 = 0, we get the following approximate expression for the upper boundary of
stability (see also Eq. (10.33), p. 281):
1 hp i
mmax = (4k − 9)(20k − 13) + 4k − 9 . (10.44)
4

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

10.11. ENHANCED CRITERION OF STABILITY 295

If the fifth harmonic is included, the requirement for a non-trivial solution

to the system (10.43) yields a cubic equation for the desired normalized critical
driving amplitude amax /l = mmax . The relevant root of this equation (too cum-
bersome to be shown here) is used for plotting the curve n = 2 in Figure 10.23.
However, for the interval of k values under consideration (−0.8 — 0.6) the ap-
proximate expression (10.44) gives a curve that is visually indistinguishable from
the curve n = 2 in Figure 10.23. √
The curve n = 2 intersects the ordinate axis at m ≈ 3( 13 − 3)/4 = 0.454.
This case (k = 0) corresponds to the above-mentioned limit of a very high driving
frequency (ω/ω0 → ∞) or zero gravity (ω0 = 0), so that m = 0.454 gives
the upper limit of stability for each of the two dynamically stabilized equivalent
equilibrium positions: If m > 0.454 at g = 0, the “flutter” mode is excited.
The lower and upper boundaries of the dynamic stability gradually converge
while the drive frequency is reduced: Figure 10.23 shows that the interval between
mmin and mmax shrinks to the left, when |k| is increased. Both boundaries merge
at k ≈ −1.41 (ω ≈ 0.8423ω0) and m ≈ 2.451. The diminishing island of dy-
namic stability of the pendulum in the inverted state vanishes in the surrounding
sea of chaotic motions.
The improved theoretical values for the lower and upper boundaries of stabil-
ity are obtained here for the frictionless system (γ = 0). Computer simulations
(based on numerical integration of exact differential equation, Eq. (7.15), show
that relatively weak friction (Q ≥ 15 – 30) does not noticeably influence these
boundaries. This can be easily explained physically if we take into account that in
conditions of dynamic stabilization (sufficiently large frequency or amplitude of
parametric excitation) the role of inertial forces is much more important.
Conversely, the basin of attraction for equilibrium of the pendulum in the in-
verted position is sensitive to friction: The interval of initial deviations within
which the pendulum returns eventually to the inverted position becomes smaller
as friction is increased.

.
-ϕ o π −ϕ
0.06 6.0
π −ϕ ( t )
z (t)

0
π −ϕ

o
−0.06 o −6.0
−6.0 0 6.0 0 2 4 6 8 10 12 14 16 T
(quality 50.0, driving frequency 1.0w0, amplitude of the axis 191.0%,
o
initial deflection 178.0 , initial angular velocity 0.0w0)

Figure 10.27: Dynamic stability of the inverted pendulum at low frequency and
large amplitude of the pivot oscillations.

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

296 CHAPTER 10. PENDULUM WITH OSCILLATING PIVOT

For further experimental verification of the improved values for both upper
and lower boundaries of the inverted pendulum’s stability, and for comparison of
the improved criterion, Eq. (10.41), with the conventional one, Eq. (10.22), we
now choose k = −0.5 (drive frequency ω = 1.4142 ω0). At this k value the
improved criterion, Eq. (10.41), yields for the low boundary of stability the value
mmin = 1.1920 (the amplitude of the pivot oscillation amin equals 119.20% of the
pendulum length l), while the conventional one, Eq. (10.22), gives only mmin =
1.00 (100% of the pendulum length l: amin = l). This value of the pivot amplitude
is far below the real boundary of dynamic stability.
The improved theoretical value for the upper boundary at k = −0.5 is mmax
= 1.2226 (122.26%), while the approximate theory (in which a harmonic with
the frequency 5ω/2 is not taken into account, see Section 10.10) gives a slightly
greater value: According to Eq. (10.44) mmax = 1.2265 (122.65%). Simulations
show that below m = 115.68% the motion is chaotic (‘tumbling’ chaos); in the
interval m = 115.69% – 119.19% the pendulum, after a long chaotic transient, is
trapped in period-1 non-uniform unidirectional rotation (in contradiction with the
conventional criterion, Eq. (10.22), which predicts stability of the inverted position
in this interval), and only in the interval m = 119.20% – 122.27% the pendulum,
being released at a small deviation from the inverted position, eventually comes
to rest, in perfect accordance with the improved criterion, Eq. (10.41).
Then, over the upper boundary of stability, within the interval m = 122.28%
– 123.02%, the pendulum occurs in a “flutter” mode; at m = 123.03% – 147.01%

−ϕ
2.04 2.04
.
−ϕ ( t )

0 0
0 π −ϕ
.
z (t )
−2.04 −2.04
−255 ο 255ο255 ο
π −ϕ (t)
z (t )
0

ο
−255
0 1 2 3 4 5 6T
(quality 30.0, driving frequency 1.0w0, amplitude of the axis 191.0%,
o
initial deflection 179.2 , initial angular velocity 1.286w0)

Figure 10.28: Period-2 oscillation of a large amplitude about the inverted position:
The phase orbit, spatial trajectory, and graphs of angular velocity ϕ̇(t) and angle
ϕ(t) time dependencies (with the graphs of separate harmonics).

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

10.11. ENHANCED CRITERION OF STABILITY 297

executes unidirectional rotation; at m = 147.02% – 150.6% the pendulum, after

a long chaotic transient, comes to period-2 oscillation about the inverted position
with an amplitude of approximately 260◦ (similar to the oscillation shown in Fig-
ure 10.28); at m ≥ 150.7% the pendulum eventually settles into the unidirectional
rotation.
Further simulations shown in Figures 10.27–10.29 refer to the case of espe-
cially low frequency of the pivot oscillations: ω = ω0 (k = −1). The theoretical
values for the lower and upper boundaries of stability, according to the improved
criteria, at this frequency are mmin = 1.9069 (190.69% of the pendulum length)
and mmax = 1.9138 (191.38%), respectively. Figure 10.27 shows how in this
narrow interval the pendulum, being released at 178◦ , first goes further from the
vertical for about 6◦ maximum angular excursion and then steadily approaches the
inverted position by the process of gradually damping “flutter”-like oscillations.
The basin of attraction for equilibrium in the inverted position is rather small:
At slightly different initial conditions the pendulum, after a long transient, occurs
in a steady-state large-amplitude (approximately 255◦ ) period-2 oscillation about
the inverted position. In this extraordinary motion, the angular excursion of the
pendulum from one extreme position to the other takes one period of excitation
and is greater than a full circle (about 510◦). Therefore, otherwise we can treat
this regime as alternating clockwise and counterclockwise revolutions over the

.
-ϕ
0.03 . .
-ϕ (t ) z (t )

0
0 π −ϕ

−0.03 o
−3.0o 0 3.0
π −ϕ (t ) z(t )

−3.0o
0 1 2 3 4 5 6T
(quality 50.0, driving frequency
o
1.0w0, amplitude of the axis 191.4%,
initial deflection 182.7 , initial angular velocity -0.0005w0)

Figure 10.29: The phase orbit and graphs of ϕ(t) and ϕ̇(t) for “flutter” oscillations
of the inverted pendulum just over the upper boundary of stability.

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

298 CHAPTER 10. PENDULUM WITH OSCILLATING PIVOT

top. The phase orbit (with two fixed points of Poincaré sections) and the spatial
trajectory of the pendulum bob in Figure 10.28 give an impression of such an ex-
traordinary pendulum motion, which coexists with the state of rest in the inverted
position. Just over the upper boundary of stability (191.38%), the pendulum even-
tually settles into the “flutter” mode (Figure 10.29).

10.12 Complicated Regular Motions of the

Parametrically Driven Pendulum
Behavior of the planar rigid pendulum excited parametrically by vertical oscilla-
tions of its pivot is much richer in various modes than we can expect for such
a simple physical system relying on our intuition. Its nonlinear large-amplitude
motions can hardly be called ‘simple.’ The simulations show that variations of
the parameter set (dimensionless driving amplitude a/l, normalized driving fre-
quency ω/ω0 , and quality factor Q) result in different regular and chaotic types
of dynamical behavior. Variations in the initial conditions can result in different
regimes that coexist at the same values of the system parameters. This property of
nonlinear systems is usually referred to as multistability.
One more type of a counterintuitive steady-state regular behavior is shown in
Figure 10.30. This mode can be characterized as resulting from a multiplication of
the period of a subharmonic resonance, specifically, as tripling of the sixth-order
subharmonic resonance described in Section 10.9.1. Comparing Figure 10.30 with

ο
75.0
z (t) π − ϕ (t)

ο
-75.0
3.09 . .
-ϕ ( t ) z (t)

0
0

ο -3.09
-75.0 0 75.0 ο 0 1 2 3 4 5 6T
(quality 10.0, driving frequency 9.1w0, amplitude of the axis 34.7%,
initial deflection -173.1o, initial angular velocity 2.121w0)

Figure 10.30: The spatial path, phase orbit, and time-dependent graphs of ϕ(t)
and ϕ̇(t) for stationary period-18 regular oscillations. The graphs show three con-
secutive cycles of six driving periods each.

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

10.12. COMPLICATED REGULAR MOTIONS 299

73.0ο
z (t) ϕ (t)

ο
- 73.0
1.70 . .
ϕ (t) z (t)

0
0

-1.70
-73.0 ο 0 73.0 ο 0 2 4 6 8 10T
(quality 60.0, driving frequency 3.60w0, amplitude of the axis 38.80%,
o
initial deflection -32.3 , initial angular velocity -0,051w0)

Figure 10.31: The spatial path of the pendulum’s bob, phase orbit, and time-
dependent graphs of ϕ(t) and ϕ̇(t) with graphs of their harmonic components
for stationary period-10 oscillations.

Figure 10.9, p. 266, we see that in both cases the motion is quite similar during
any cycle consisting of six consecutive driving periods, but in Figure 10.30 the
motion during each subsequent cycle of six periods is slightly different from the
preceding cycle.
After three cycles of six driving periods the phase orbit becomes closed and
then repeats itself, so that the period of this stationary motion equals 18 driving
periods. However, the harmonic component whose period equals six driving pe-
riods dominates the spectrum (just like in the spectrum of period-6 oscillations
in Figure 10.9, p. 266), while the fundamental harmonic (frequency ω/18) of a
small amplitude is responsible only for tiny divergences between the adjoining
cycles consisting of six driving periods each. Such multiplications of the period
are characteristic of large amplitude oscillations at subharmonic resonances both
for the inverted and hanging-down pendulum.
Figure 10.31 shows an example of stationary oscillation about the lower equi-
librium position with a period that equals ten driving periods. This large ampli-
tude motion can be treated as originating from a period-2 oscillation (that is, from
ordinary principal parametric resonance) by a five-fold multiplication of the pe-
riod. The harmonic component with half the driving frequency (ω/2) dominates
the spectrum. But in contrast to the preceding example, the divergences between
adjoining cycles consisting of two driving periods each are generated by the con-
tribution of a harmonic component with the frequency 3ω/10 rather than of the
fundamental harmonic (frequency ω/10) whose amplitude is much smaller.
Figure 10.32 shows one more example of complicated steady-state oscilla-

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

300 CHAPTER 10. PENDULUM WITH OSCILLATING PIVOT

118ο
ϕ (t)
z (t)
0

ο
-118
1.54 . .
ϕ (t) z (t)

0
0

ο
-1.54
-118 0 118ο 0 2 4 6 8 10T
(quality 30.0, driving frequency
o
2.0 w 0, amplitude of the axis 20.0%,
initial deflection 117.9 , initial angular velocity 0.012w0)

Figure 10.32: The spatial path, phase orbit, and graphs of stationary period-30
oscillations. The graphs show three consecutive cycles of ten driving periods each.
One full period consists of three almost identical cycles, each of which covers 10
drive periods.

tion of the parametrically driven pendulum. This period-30 motion can be treated
as generated from the period-2 principal parametric resonance first by five-fold
multiplication of the period (resulting in period-10 oscillation), and next by multi-
plication (tripling) of the period. Such large-period stationary regimes are charac-
terized by small domains of attraction consisting of several disjoint islands in the
phase plane of initial conditions.
Other numerous modes of regular, periodic behavior are formed by unidirec-
tional period-2 or period-4 (or even period-8) rotation of the pendulum or by os-
cillations alternating with revolutions to one or to both sides in turn. Such modes
have periods constituting several driving periods.

10.13 Chaotic Motions of the Parametrically Driven

Pendulum
At large enough driving amplitudes the pendulum whose pivot is forced to oscil-
late in the vertical direction exhibits different chaotic regimes. Chaotic behavior
of various nonlinear systems has been a subject of intense interest during recent
decades, and the parametrically forced pendulum serves as an excellent physical
model for studying general laws of the dynamical chaos.
Next we describe several different kinds of chaotic regimes, which for the time
being have not been extensively investigated in the literature. Poincaré mapping,

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

10.13. CHAOTIC MOTIONS OF THE PENDULUM 301

ο
129
ϕ (t) z (t)

ο
-129
1.79 . .
ϕ (t) z (t)

0
0

-1.79
-129 ο 0 129ο 0 2 4 6 8 10 12T
(quality 3.10, driving frequency 1.80w 0, amplitude of the axis 40.0%,
initial deflection 118.8o, initial angular velocity -0.534w0)

Figure 10.33: Chaotic attractor whose Poincaré sections form a two-band set of
points.

that is, a stroboscopic picture of the phase plane for the pendulum taken once
during each driving cycle after initial transients have died away, gives an obvi-
ous and convenient means to distinguish between regular periodic behavior and
persisting chaos. A steady-state subharmonic motion of order n would be seen in
the Poincaré map as a systematic jumping of the representative point among n
fixed mapping points. When the pendulum exhibits a chaotic motion, the points of
Poincaré sections wander randomly, never repeating some pattern exactly. Their
behavior in the phase plane gives an impression of the strange attractor for the
motion in question.
Figure 10.33 shows an example of a purely oscillatory two-band chaotic at-
tractor for which the set of Poincaré sections consists of two disjoint islands. This
attractor is characterized by a fairly large domain of attraction in the phase plane.
The two islands of the Poincaré map are visited regularly (strictly in turn) by the
representing point, but within each island the point wanders irregularly from cy-
cle to cycle. This means that for this kind of motion the flow in the phase plane
is chaotic, but the distance between any two initially close phase points within
this attractor remains limited in the progress of time: The greatest distance in the
phase plane is determined by the size of these islands of the Poincaré map.
Figure 10.34 shows the chaotic attractor that corresponds to a slightly reduced
friction compared to the case shown in Figure 10.33, while all other parameters are
unchanged. Gradual reduction of friction causes the islands of Poincaré sections
to grow and coalesce, and finally to form a strip-shaped set occupying consider-
able region of the phase plane. As in the preceding example, each cycle of these
oscillations (consisting of two driving periods) slightly but randomly varies from

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

302 CHAPTER 10. PENDULUM WITH OSCILLATING PIVOT

167ο
ϕ (t) z (t)

ο
-167
2.09 . .
ϕ (t) z (t)

0 0

ο -2.09
ο
-167 0 167 0 2 4 6 8 10 12T
(quality 3.20, driving frequency 1.80w 0, amplitude of the axis 40.0%,
initial deflection 76.0o, initial angular velocity 1.287w0)

Figure 10.34: Chaotic attractor with a strip-like set of Poincaré sections.

the preceding one.

However, in this case the large and almost constant amplitude of oscillations
occasionally (after a large but unpredictable number of cycles) considerably re-
duces or, vice versa, increases (sometimes so that the pendulum makes a full
revolution over the top). These decrements and increments occasionally result in
switching the phase of oscillations: Motion of the pendulum, say, to the right side
that occurred during even driving cycles is replaced by the motion in the opposite
direction. During long intervals between these rare events the motion of the pen-
dulum is purely oscillatory with only slightly (and randomly) varying amplitude
of the oscillation.
This kind of intermittent irregular behavior differs from the well-known so-
called ‘tumbling’ chaotic attractor (see Figure 10.26, p. 293) that exists over a
relatively broad range of parameter space. The tumbling attractor is characterized
by random oscillations (whose amplitude varies strongly from cycle to cycle),
often alternated with full revolutions to one or the other side.
Figure 10.35 illustrates one more kind of a strange attractor. In this example
the motion is always purely oscillatory, and nearly repeats itself after each six
driving periods. The six bands of Poincaré sections make two groups of three
isolated islands each. The representing point visits these groups in alternation.
Moreover, the representing point visits the islands of each group in a quite definite
order, but within each island the points continue to bounce randomly from one
place to another without any apparent order.
The six-band attractor has a rather extended (and very complicated in shape)
domain of attraction. Nevertheless, at these values of the control parameters the
system exhibits multiple asymptotic states: The chaotic attractor coexists with sev-

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

10.14. CONCLUDING REMARKS 303

ο
142
ϕ (t) z (t)

ο
-142
1.85 . .
ϕ (t) z (t)

0
0

-1.85
ο ο
-142 0 142 0 2 4 6 8 10 12T
(quality 3.055, driving frequency
o
1.57w 0 , amplitude of the axis 40.0%,
initial deflection -83.3 , initial angular velocity -1.104w0)

Figure 10.35: An oscillatory six-band chaotic attractor.

eral periodic regimes. One of these periodic regimes is shown in Figure 10.36. The
period of regular asymmetric oscillations in this example equals four driving peri-
ods. Such asymmetric regimes exist in pairs, whose phase orbits are mirror images
of one another.
Chaotic regimes exist also for purely rotational motions. Poincaré sections for
such rotational chaotic attractors can make several isolated islands in the phase
plane. A possible scenario of transition to such chaotic modes from unidirectional
regular rotation lies through an infinite sequence of period-doubling bifurcations
occurring when a control parameter (the driving amplitude or frequency or the
braking frictional torque) is slowly varied without interrupting the motion of the
pendulum. However, there is no unique route to chaos for the more complicated
chaotic regimes described above.

10.14 Concluding Remarks

In this chapter we have shown that at sufficiently high frequency and small am-
plitude of the pivot oscillations, many remarkable peculiarities in behavior of the
pendulum can be clearly explained on the basis of the method of averaging in
which rapid and slow motions of the pendulum are separated. In particular, this
approach and the related concept of the effective potential for slow motion are
very useful for understanding the dynamic stabilization of the inverted pendulum,
as well as for explanation of the origin of recently discovered subharmonic reso-
nances of high orders. Coexistence of resonances of different orders at identical
parameters of the system also can be easily explained on this basis. Corresponding
approximate quantitative theory of parametric excitation allows us to calculate the

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

304 CHAPTER 10. PENDULUM WITH OSCILLATING PIVOT

ο
145
ϕ (t) z (t)

- 145ο
1.88 . .
ϕ (t) z (t)

0
0

ο
-1.88
ο
-145 0 145 0 1 2 3 4 5 6 7 8T
(quality 3.055, driving frequency 1.57w0, amplitude of the axis 40.0%,
initial deflection 98.8o, initial angular velocity -0.083w0)

Figure 10.36: Regular period-4 asymmetric oscillations that coexist with the
chaotic oscillations shown in Figure 10.35.

spectrum of subharmonic resonances and to find their boundaries in the parameter

space of the system.
The relationship between subharmonic resonances and the phenomenon of
dynamic stabilization of the inverted pendulum reveals a way of obtaining an im-
proved criterion of dynamic stabilization, Eq. (10.22), which is valid in a wider
region of frequencies and amplitudes of the pivot oscillations, including values of
these parameters for which the method of averaging (of separation of rapid and
slow motions) is inapplicable.
Relying on the common nature of the second-order subharmonic resonance
and ordinary parametric resonance, we have presented a physical explanation for
the recently discovered phenomenon of destabilization of the (dynamically stabi-
lized) inverted position: At sufficiently large amplitudes of the pivot oscillations
the inverted pendulum becomes trapped into the limit cycle of “flutter” oscillations
with period 2T .
This phenomenon of “flutter” oscillations is completely analogous to destabi-
lization of the hanging-down pendulum when conditions of ordinary parametric
resonance are fulfilled. Ordinary parametric resonance and the “flutter” mode be-
long to the same branch of 2-periodic stationary oscillations, so that the criterion
of destabilization for the inverted position (10.33) can be obtained by the same
method as for destabilization of the downward position at ordinary parametric
resonance. Results of corresponding analytical calculations for the boundaries of
the stability region agree perfectly well with computer simulations based on a nu-
meric integration of the exact differential equation that describes the rigid planar
pendulum with vertically oscillating pivot.

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

10.14. CONCLUDING REMARKS 305

In this chapter we have touched only a small part of existing stationary states,
regular and chaotic motions of the parametrically driven pendulum. The pendu-
lum’s dynamics exhibits a great variety of other asymptotic rotational, oscillatory,
and combined (both rotational and oscillatory) multiple-periodic stationary states
(attractors), whose basins of attraction are characterized by a surprisingly com-
plex (fractal) structure. Computer simulations also reveal intricate sequences of
bifurcations, leading to numerous intriguing chaotic regimes. Most of them re-
main beyond the scope of this chapter, and those mentioned here are still await-
ing a plausible physical explanation. With good reason, we can suppose that this
seemingly simple physical system is really inexhaustible.

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

Chapter 11

Torsion Pendulum with Dry

and Viscous Damping

Annotation. Free and forced oscillations of a torsion spring pendulum damped

by viscous and dry friction are investigated analytically and with the help of nu-
merical simulations. A simplified mathematical model is assumed (Coulomb law),
which nevertheless can explain many peculiarities in behavior of various oscilla-
tory systems with dry friction. The amplitude of free oscillations diminishes under
dry friction linearly, and the motion stops after a final number of cycles. The am-
plitude of a sinusoidally driven pendulum with dry friction grows at resonance
without limit if the threshold is exceeded. At strong enough non-resonant sinu-
soidal forcing, dry friction causes transients that typically lead to definite limit
cycles — periodic steady-state regimes of symmetric non-sticking forced oscilla-
tions that are independent of initial conditions. However, at the subharmonic sinu-
soidal forcing, interesting peculiarities of the steady-state response are revealed,
such as multiple coexisting regimes of asymmetric oscillations that depend on
initial conditions. Under certain conditions, simple dry friction pendulum shows
complicated stick-slip motions and chaos.

11.1 Basics of the Theory

11.1.1 Introduction
Various mechanical vibration systems with combined viscous and dry (Coulomb)
friction are of considerable importance in numerous applications of dynamics in
engineering.
When friction is viscous, the spring oscillatory systems are described by linear
differential equations. This case allows an exhaustive explicit analytical solution
that is usually studied in undergraduate courses at universities and can be found
in most textbooks on general physics. However, the influence of dry friction on

307

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

308 CHAPTER 11. PENDULUM WITH DRY AND VISCOUS DAMPING

oscillatory systems remains as a rule beyond the scope of the academic literature
and traditional physics courses.
Dry friction results in a nonlinearity. With dry friction, the system acquires a
non-smooth, discontinuous nonlinear character. If the coefficient of dry friction
is sufficiently small, the oscillating body slides under harmonic forcing and its
velocity is zero only for the instants at which the direction of motion reverses.
This kind of motion of a dry friction oscillator with no stick phase is usually
referred to as a pure slip motion, or non-sticking motion. At strong enough dry
friction, sticking may occur: The body remains at rest for a finite time during the
driving cycle after its velocity reaches zero. A detailed historical review of dry
friction and stick-slip phenomena can be found in [53].
Dry friction as a nonlinearity is the current focus of research activities. Even
the simplest dry friction model, the Coulomb friction, can explain the principal
peculiarities in the motion of a dry friction oscillator. Damping of free oscilla-
tions under dry friction is very clearly described in the textbook of Pippard [3]
(see also [54]). Different approaches to the problem are discussed in [55], [56].
Den Hartog [57] was the first to solve, in 1930, the periodic sliding response
of a harmonically forced oscillator with both viscous and dry-friction damping.
Later on, the analytical solutions of non-sticking responses were widely discussed
in the contemporary scientific literature (see [59]–[66] and references therein).
The problem was treated by using a number of various analytical and numerical
techniques. In recent years, there has been an increasing interest in periodic and
chaotic motions of discontinuous dynamical systems because of their important
role in engineering (see, for example, [67]).
In the literature the analytical solution to the problem of oscillations in a sys-
tem with dry friction is usually obtained by a simple method of stage-by-stage in-
tegration of the differential equations that describe the system. These equations are
linear for the time intervals occurring between consecutive turning points, if the
simplest (Coulomb) model is assumed for dry friction. The intervals are bounded
by the instants at which the velocity is zero. The complete solution is obtained by
fitting these half-cycle solutions to one another for adjoining time intervals. By
virtue of the piecewise linear nature of the relevant differential equations, explicit
solutions can be found for the time intervals between the successive turnarounds.
In our approach to the problem we try to rely primarily on the physics un-
derlying the investigated phenomena. In this chapter we are concerned with free
oscillations of a torsion spring pendulum, and with forced oscillations of the pen-
dulum kinematically driven by an external sinusoidal force, including cases of
damping caused by dry (Coulomb) friction, and both by viscous and dry friction.
Mathematically, the pendulum driven by an external force is equivalent to the
spring-mass system with the body residing on the horizontally oscillating base.
The simple formulae of analytical solutions are confirmed by graphs obtained in
computer simulations. New results cover quantitative description of the resonant
growth of oscillations under sinusoidal forcing, and closed-form analytical solu-
tions at sub-resonant frequencies. These solutions correspond to multiple asym-
metric steady-state regimes coexisting at the same values of the system parame-

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

11.1. BASICS OF THE THEORY 309

ters. Characteristics of such regimes depend on the initial conditions. Our analyt-
ical and numerical solutions are illustrated by a simplified version of the relevant
simulation program (Java applet) available on the web [58].

11.1.2 The Physical System

The rotating component of the torsion spring oscillator investigated in the paper is
a balanced flywheel whose center of mass lies on the axis of rotation (Figure 11.1),
similar to devices used in mechanical watches.

-d +d
- q0 +q 0

q (t)
j (t)

Figure 11.1: Schematic diagram of the driven torsion oscillator with dry friction.

A spiral spring with one end attached to the flywheel flexes when the flywheel
is turned. The other end of the spring is attached to the exciter — a driving rod,
which can be turned by an external force about the axis common with the flywheel
axis. The spring provides a restoring torque whose magnitude is proportional to
the angular displacement of the flywheel relative to the driving rod. In other words,
we assume that the flywheel is in equilibrium (the spring is unstrained) when the
rod of the flywheel is parallel to the driving rod.
In the case of unforced (free, or natural) oscillations in an isolated system, the
motion is initiated by an external influence which occurs before a particular in-
stant. This influence determines the initial mechanical state of the system, that is,
the displacement and the velocity of the oscillator at the initial instant. These ini-
tial conditions determine the amplitude and phase of subsequent free oscillations.
The frequency and damping rate of free oscillations are determined solely by the
physical properties of the system, and do not depend on the initial conditions.
Oscillations are called forced if an oscillator is subjected to an external peri-
odic influence whose effect on the system can be expressed by a separate term, a
periodic function of the time, in the differential equation of motion. We are inter-
ested in the response of the system to the periodic external force. The behavior of
oscillatory systems under periodic external forces is one of the most important is-
sues in the theory of oscillations. A noteworthy distinctive characteristic of forced

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

310 CHAPTER 11. PENDULUM WITH DRY AND VISCOUS DAMPING

oscillations is the phenomenon of resonance, in which a small periodic disturbing

force can produce an extraordinarily large response in the oscillator. Resonance
is found everywhere in physics and so a basic understanding of this fundamental
problem has wide and various applications.
The phenomenon of resonance depends upon the whole functional form of the
driving force and occurs over an extended interval of time rather than at some
particular instant. In this paper we draw attention to peculiarities of resonance in
an oscillator with dry friction.
In our model of an oscillatory system, free oscillations of the flywheel occur
when the driving rod is immovable (θ = 0). Forced oscillations are excited when
the driving rod rotates back and forth sinusoidally about its middle position θ = 0
between the angles −θ0 and θ0 (see Figure 11.1): θ(t) = θ0 sin ωt. This mode
differs from the dynamical mode usually considered in textbooks, according to
which oscillations are excited by a given external force exerted on the system. Our
mode can be called kinematical, because in this mode oscillations are excited by
forcing one part of the system (the driving rod) to execute a given motion (in our
case a simple harmonic motion). This kinematical mode is especially convenient
for observation, because the motion of the exciter can be seen simultaneously with
oscillations of the flywheel.
In the Coulomb model of dry friction, as long as the system is moving, the
magnitude of dry friction is assumed to be constant, and its direction is opposite
that of the velocity, that is, its direction changes each time the direction of the ve-
locity changes. When the system is at rest, the force of static dry friction takes on
any value from some interval −Fmax to Fmax . The actual value of static frictional
force can be found from the requirement of balancing the other forces exerted on
the system. In other words, the force of static friction adjusts itself to make equi-
librium with other external forces acting on the body. The magnitude of the force
of kinetic dry friction is assumed in this model to be equal to the limiting force
Fmax of static friction.
In real physical systems dry friction is characterized by more complicated
dependencies on the relative velocity (see, for example, [68], [69]). The limiting
force of static friction is usually greater than the force of kinetic friction. When the
speed of a system increases from zero, kinetic friction at first decreases, reaches
a minimum at some speed, and then gradually increases with a further increase
in speed. These peculiarities are ignored in the idealized z-characteristic of dry
friction. Nevertheless, this idealization allows us to understand many important
features of oscillations in real physical systems.
In our model of a torsion oscillator some amount of dry friction can exist in the
bearings of the flywheel axis. Because the magnitude of static frictional torque can
assume any value up to Nmax , there is a range of values of angular displacement
called the stagnation interval or dead zone in which static friction can balance
the restoring elastic torque of the strained spring. At any point within this interval
the system can be at rest in a state of neutral equilibrium, in contrast to a single
position of stable equilibrium provided by the spring in the case of an oscillator
with viscous friction.

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

11.1. BASICS OF THE THEORY 311

The stagnation interval extends equally to either side of the point at which the
spring is unstrained. The stronger the dry friction in the system, the more extended
the stagnation interval. The boundaries of the interval ±d are determined by the
limiting torque Nmax of static friction. In Figure 11.1 these boundaries −b and +b
are shown for the case in which the driving rod is in its middle position θ = 0.

11.1.3 The Differential Equation of the Oscillator

The rotating flywheel of the torsion oscillator is simultaneously subjected to the
restoring torque −D(ϕ − θ) produced by the spring, the torque −B ϕ̇ of viscous
friction that is proportional to the angular velocity, and the torque Nfr of kinetic
dry friction. When the exciter is forced to move periodically according to θ(t) =
θ0 sin ωt, the differential equation describing the rotational motion of the flywheel
with the moment of inertia J is thus

J ϕ̈ = −D(ϕ − θ0 sin ωt) − B ϕ̇ + Nfr . (11.1)

The torque Nfr is directed oppositely to angular velocity ϕ̇, and is constant in
magnitude while the flywheel is moving, but may have any value in the interval
from −Nmax up to Nmax while the flywheel is at rest:

−Nmax for ϕ̇ > 0,
Nfr (ϕ̇) = −Nmax sign ϕ̇ = (11.2)
Nmax for ϕ̇ < 0.

Here Nmax is the limiting value of the static frictional torque. It is convenient to
express the value Nmax in terms of the maximal possible deflection angle d of the
flywheel at rest, when the driving rod (see Figure 11.1) is immovable at its middle
position θ = 0: Nmax = Dd. The angle d corresponds to the boundary of the
stagnation interval. Dividing all terms of Eq. (11.1) by J, we get

ϕ̈ + 2γ ϕ̇ + ω02 d sign ϕ̇ + ω02 ϕ = ω02 θ0 sin ωt. (11.3)

The damping constant γ is a measure of the intensity of viscous p friction. It is

introduced here by the relation 2γ = B/J. The frequency ω0 = D/J char-
acterizes undamped natural oscillations. The sign ϕ̇ function is meant to take the
undetermined values between 1 and −1 at zero argument, which corresponds to
stick phase. The actual value of the static dry friction torque is such that the sys-
tem is in equilibrium. The differential equation for an oscillator with dry friction,
Eq. (11.1), as well as Eq. (11.3), is nonlinear because the torque Nfr (ϕ̇) abruptly
changes when the sign of ϕ̇ changes at the extreme points of oscillation. This is
the so-called Filippov system [70]. In the idealized case of the z-characteristic
this is a piecewise smooth system, and we may consider the following two linear
equations instead of Eq. (11.3):

ϕ̈ + 2γ ϕ̇ + ω02 (ϕ + d) = ω02 θ0 sin ωt for ϕ̇ > 0, (11.4)

ϕ̈ + 2γ ϕ̇ + ω02 (ϕ − d) = ω02 θ0 sin ωt for ϕ̇ < 0. (11.5)

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

312 CHAPTER 11. PENDULUM WITH DRY AND VISCOUS DAMPING

Whenever the sign of the angular velocity ϕ̇ changes, the pertinent equation of
motion also changes. The nonlinear character of the problem reveals itself in al-
ternate transitions from one of the linear Eqs. (11.4)–(11.5) to the other.

11.1.4 Damping of Free Oscillations under Dry Friction

For the case of free (unforced) oscillations the right-hand side of Eqs. (11.4)–
(11.5) is zero. The case in which the dry friction is absent (the dead zone vanishes:
d = 0) and damping of free oscillations occurs solely due to viscous friction, is
discussed in Chapter 1. For this idealized case the differential equation of motion
becomes linear. It has a well known analytical solution, according to which the
amplitude of free oscillations under viscous friction decreases exponentially with
time. That is, the consecutive maximal deflections of the oscillator from its equi-
librium position form a diminishing geometric progression because their ratio is
constant.
In an idealized linear system such oscillations continue indefinitely, their am-
plitude asymptotically approaching zero. The duration of exponential damping
can be characterized by a conventional decay time τ = 1/γ. The exponential
character of damping caused by viscous friction follows from the proportional-
ity of friction to velocity. Some other relationship between friction and velocity
produces damping with different characteristics.
The role of dry friction in the damping of free oscillations is considered in
detail in Chapter 2. The solution to Eqs. (11.4)–(11.5) for non-zero dry friction
(d 6= 0) can be found by using the method of the stage-by-stage integration of
each of the linear equations for the half-cycle during which the direction of motion
is unchanged. These solutions are then joined together at the instants of transition
from one equation to the other in such a way that the displacement at the end
point of one half-cycle becomes the initial displacement at the beginning of the
next half-cycle. This array of solutions continues until the end point of a half-cycle
lies within the dead zone.
An important feature of free oscillations damped by dry friction is that the
motion completely ceases after a finite number of cycles. As the system oscillates,
each subsequent change of its velocity occurs at a smaller displacement from the
mid-point of the stagnation interval. Eventually the turning point of the motion
occurs within the stagnation interval, where static friction can balance the restor-
ing torque of the spring, and so the motion abruptly stops. At which point of the
interval this event occurs depends on the initial conditions, which may vary from
one situation to the next.
In systems with both dry and viscous friction the damping of oscillations can
also be investigated by the stage-by-stage solving of the equations of motion and
by using the mechanical state at the end of the previous half-cycle as the initial
conditions for the next in turn half-cycle. The phase trajectory consists in this case
of the shrinking alternating halves of spiral loops that are characteristic of a linear
damped oscillator. The focal points of these spirals alternate between the bound-
aries of the stagnation interval. The loops of the phase trajectory are no longer

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

11.2. DRIVEN OSCILLATOR WITH DRY FRICTION 313

o
2.13 126
.
ϕ(t)
ϕ(t )
0 0

θ ( t)
o
-2.13 - 126
o o
-126 0 126 0 1 2 3 4 5 6 T0
Drive amplitude 25 degrees, dead zone 15 degrees, initial displacement -15 degrees

Figure 11.2: Phase trajectory with Poincaré sections (left) and graphs of ϕ(t) and
ϕ̇(t) (right) for oscillations at resonance with dry friction.

equidistant. Nevertheless their shrinking does not last indefinitely: The phase tra-
jectory in this case also terminates after some finite number of turns around the
origin when it reaches the stagnation interval on the ϕ-axis.

11.2 Sinusoidally Driven Oscillator with Dry

Friction
11.2.1 Resonance in the Oscillator with Dry Friction under
Sinusoidal Excitation
In this section we analyze forced oscillations of the torsion spring pendulum in
conditions of resonance, that is, when the frequency of excitation ω equals natural
frequency ω0 of the oscillator (T = T0 = 2π/ω0 ). Generally at large enough dry
friction sticking may occur: The flywheel remains at rest for a finite time after the
velocity reaches zero. However, if the amplitude of excitation θ0 in Eqs. (11.4)–
(11.5) exceeds some threshold value, the motion of the flywheel is purely sliding
(non-sticking), and in the absence of viscous friction the amplitude of oscilla-
tions grows indefinitely. An example of such resonant oscillations is shown in
Figure 11.2. The phase trajectory and the graphs of ϕ(t) and ϕ̇(t) are obtained by
computer simulation that is based on numeric integration of Eqs. (11.4)–(11.5).
We note the linear growth of the amplitude: The succession of maximal deflec-
tions of the flywheel forms an arithmetic progression. Next we find analytically
the threshold for excitation of such growing oscillations, and calculate the incre-
ment of the amplitude after each driving cycle.
We choose for simplicity the initial deflection ϕ(0) of the flywheel coinciding
with the left boundary of the dead zone, that is, ϕ(0) = −d, and initial angular
velocity zero: ϕ̇(0) = 0. Such initial conditions provide the sliding (non-sticking)
motion from the very beginning with two turnarounds during each cycle of exci-
tation. During the first half of the excitation period (0 < t < T0 /2) the angular

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

314 CHAPTER 11. PENDULUM WITH DRY AND VISCOUS DAMPING

velocity is positive (ϕ̇(t) > 0), and we should use Eq. (11.4). The solution to this
equation (with γ = 0), satisfying the above indicated initial conditions, can be
written as follows:
1 1
ϕ(t) = − θ0 (ω0 t cos ω0 t−sin ω0 t)−d, ϕ̇(t) = θ0 ω02 t sin ω0 t, 0 < t < T0 /2.
2 2
(11.6)
According to (11.6), the next maximal elongation to the right side occurs at
t = T0 /2 and equals 21 πθ0 − d. This elongation is greater in magnitude than
the preceding (initial) elongation d to the left side by 12 πθ0 − 2d.
To find the increment in the amplitude during the second half-cycle of excita-
tion, when the flywheel rotates in the opposite direction, we should use Eq. (11.5).
An analytical solution to this equation is given below in Section 11.2.2. It occurs
that the increment in amplitude during the second half-cycle is the same as during
the first half-cycle. Therefore during the whole cycle the increment in amplitude
equals πθ0 − 4d. Specifically, for θ0 = 25◦ and d = 15◦ (the values correspond-
ing to the simulation shown in Figure 11.2) the amplitude should increase during
each cycle by 18.54◦. The simulation in Figure 11.2 shows that during the first
six cycles the amplitude increased by 126◦ − 15◦ = 111◦ , which gives for in-
crement during one cycle the value 18.5◦ , in good agreement with the theoretical
prediction.

11.2.2 Analytical Solution for the Second Half-Cycle of the

Resonant Excitation
For the first half-cycle of excitation at resonance (ω = ω0 ) the motion of the
flywheel is given by Eq. (11.6), if initially the flywheel is at rest (ϕ̇(0) = 0) exactly
at the left side of the stagnation zone: ϕ(0) = −d. If we take some arbitrary initial
deflection ϕ(0) = ϕ0 to the left side from the equilibrium (ϕ0 < 0), which lies
beyond the dead zone (|ϕ0 | > d) and initial velocity ϕ̇(0) = 0, the motion of the
flywheel will also be non-sticking from the very beginning, and during the time
interval 0 < t < T0 /2 will be described by the following expression:
1
ϕ(t) = (ϕ0 + d) cos ω0 t − θ0 (ω0 t cos ω0 t − sin ω0 t) − d, 0 < t < T0 /2.
2
(11.7)
At the end of the first half-cycle (at t = T0 /2) the angular velocity of the flywheel
becomes zero, while its deflection to the right side reaches ϕ1 = −ϕ0 + 21 θ0 π−2d.
These values of ϕ and ϕ̇ should be used as the initial conditions at t = T0 /2
for the differential Eq. (2.5) that describes (with γ = 0) the second half-period
T0 /2 < t < T0 of the forced motion, during which ϕ̇ < 0. To solve this equation,
it is convenient to move the time origin t = 0 to T0 /2. In these new notations
Eq. (2.5) takes the following form:

ϕ̈ + ω02 (ϕ + d) = −ω02 θ0 sin ω0 t. (11.8)

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

11.2. DRIVEN OSCILLATOR WITH DRY FRICTION 315

The solution to Eq. (11.8), satisfying initial conditions ϕ(0) = ϕ1 and ϕ̇(0) = 0,
can be written as follows:
1
ϕ(t) = (ϕ1 − d) cos ω0 t + θ0 (ω0 t cos ω0 t − sin ω0 t) + d. (11.9)
2
To find the angular position ϕ(T0 ) and the angular velocity ϕ̇(T0 ) of the flywheel
at the end of the first cycle of excitation, we should substitute t = T0 /2 into
Eq. (11.9):
1
ϕ(T0 ) = −ϕ1 + d − θ0 π + d = ϕ0 − θ0 π + 4d. (11.10)
2
Hence the magnitude of angular elongation to the left increased during the first
cycle of excitation by the value |ϕ(T0 ) − ϕ0 | = πθ0 − 4d. This increment is in-
dependent of the initial deflection ϕ0 . The succession of maximal deflections at
resonance in the oscillator with dry friction forms an increasing arithmetic pro-
gression.
In case d = 0 (zero width of the dead zone, that is, absence of dry friction) the
solution given by Eq. (11.7) takes the following form:
1
ϕ(t) = ϕ0 cos ω0 t − θ0 (ω0 t cos ω0 t − sin ω0 t). (11.11)
2
Obviously, for initial conditions ϕ(0) = ϕ0 , ϕ̇(0) = 0 this solution is valid for
any t value, not only for the first half-cycle of excitation 0 < t < T0 /2. According
to Eq. (11.11), in the absence of any friction (dry and viscous), the amplitude of
resonant forced oscillations changes in magnitude during one cycle of excitation
by the same amount, πθ0 . If the oscillator is excited from the state of rest in the
equilibrium position, its amplitude grows linearly from the very beginning. This
growth continues indefinitely. From the energy considerations, this can be easily
explained by certain phase relations between rotary oscillations of the flywheel
and the sinusoidally varying torque exerted on the flywheel by the spring: This
torque always acts in the direction of rotation, thus increasing the energy of the
flywheel.
However, if the initial displacement of the flywheel is positive (ϕ(0) > 0),
the external torque at the initial stage is directed against the angular velocity, and
the amplitude of oscillations diminishes, in spite of the exact tuning to resonance,
through value πθ0 during each cycle. The energy is transferred from the oscilla-
tor to the exciter. This situation is illustrated in Figure 11.3. After the amplitude
reduces to zero, the phase relations between the exciting rod and the flywheel be-
come favorable for the transfer of energy to the oscillator, and the amplitude starts
to grow indefinitely. In the absence of dry friction, the initial linear reduction and
further growth of the amplitude occur equally fast, in contrast to the case with dry
friction (see Figure 11.6, p. 319), in which friction speeds up the reduction and
slows down the growth of the amplitude.

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

316 CHAPTER 11. PENDULUM WITH DRY AND VISCOUS DAMPING

o
100 .
ϕ (t ) ϕ (t )
θ ( t)
0

o
-100
0 1 2 3 4 5 6 7 8 9 10 T0

Figure 11.3: Oscillations at resonance without friction (with initial displacement).

11.2.3 The Threshold of the Resonant Growth

The growth of oscillations amplitude occurs if the value of increment πθ0 − 4d
during a cycle of excitation is positive. Hence the threshold of resonance (θ0 )min
for the oscillator with dry friction is given by the following condition:
4 4
θ0 > d, (θ0 )min = d. (11.12)
π π
For given width d of the dead zone (for given dry friction), Eq. (11.12) defines
the critical (minimal) value (θ0 )min of the drive amplitude, which provides non-
sticking forced oscillations of the flywheel after a rather short transient.
However, during the transient, depending on the initial conditions, sticking
is possible. For θ0 > (θ0 )min , after the transient is over, at the initial moment
tn = nT = nT0 of each cycle of excitation in-turn, angular velocity ϕ̇(tn ) of the
flywheel is zero: ϕ̇(nT ) = 0. This means that Poincaré sections in the phase plane
(corresponding to time moments tn = nT ) approach the abscissa axis during the
transient and remain on its negative side further on. Since the increment in the
elongation is the same for each cycle, the points of Poincaré sections on the axis
are equidistant (see Figure 11.2).
Stationary periodic oscillations at the threshold conditions are shown in Fig-
ure 11.4. At arbitrary initial values of ϕ and ϕ̇ the phase trajectory eventually
approaches a limit cycle similar to the cycle shown in the left-hand side of Fig-
ure 11.4. The amplitude of steady-state forced oscillations at the threshold de-
pends on initial conditions. If initial velocity is zero (ϕ̇(0) = 0), steady-state os-
cillations occur from the very beginning, without any transient, in the case where
initial displacement ϕ(0) is negative and lies beyond the dead zone, that is, if
ϕ(0) < 0, |ϕ(0)| ≥ d. The amplitude of these oscillations equals |ϕ(0)|. This
mode of oscillations is unstable with respect to variations in parameters θ0 and
d: A slight increment of the drive amplitude or decrement in the dead zone width
causes an indefinite growth of the amplitude.
If the amplitude of the exciter θ0 is smaller than the critical value (θ0 )min
given by Eq. (11.12), but greater than the dead zone width d, a steady-state regime

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

11.2. DRIVEN OSCILLATOR WITH DRY FRICTION 317

o
0.46 25.0 .
ϕ(t ) ϕ(t )

0 0

-0 .46 o -25.0
o θ t ( )
o
-25.0 0 25. 0 0 1 2 3T0
Drive amplitude 19.1 degrees, dead zone 15 degrees, initial displacement -25 degrees

Figure 11.4: Oscillations at the threshold conditions at resonance with dry friction.

o
1.52 86.3
ϕ(t )

0 0

θ ( t)
o
-1.52 - 86.3
- 86.3 o 0 86.3o 0 2 4 6 8 10 12T0
Drive amplitude 15 degrees, dead zone 5 degrees, quality 10, initial deflection -15 degrees

Figure 11.5: Oscillations at resonance with dry and viscous friction.

with two sliding phases and two sticking phases establishes after the transient is
over. For θ0 smaller than the dead zone width d, the flywheel, depending on the
initial conditions, either remains immovable from the very beginning, or makes
several movements with sticking and then finally stops at some point of the dead
zone.

11.2.4 Resonance in the Presence of Dry and Viscous Friction

The resonant growth of amplitude over the threshold is restricted if some amount
of viscous friction is present in the system. In a dual-damped system steady-state
oscillations with a constant amplitude eventually establish for arbitrary initial con-
ditions. An example of resonant oscillations in the system with both dry and vis-
cous friction is shown in Figure 11.5. Eqs. (11.4)–(11.5) allow us to calculate the
amplitude a of such resonant symmetric steady-state oscillations.
We choose the time origin t = 0 at the beginning of the next-in-turn drive
cycle. At this moment the flywheel occurs at the extreme displacement to the left
side (ϕ(0) = −a) and has the angular velocity zero (ϕ̇(0) = 0). During the first
half-cycle of the drive (0 < t < T0 /2) it moves to the right, so that ϕ̇ is positive
during this interval. Therefore we should use Eq. (11.4) with ω = ω0 in its right-

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

318 CHAPTER 11. PENDULUM WITH DRY AND VISCOUS DAMPING

hand part. It is convenient to introduce instead of ϕ(t) a new unknown function

ψ(t) = ϕ(t) + d, which, according to (11.4), satisfies the following equation:

ψ̈ + 2γ ψ̇ + ω02 ψ = ω02 θ0 sin ω0 t. (11.13)

We can search for its periodic partial solution in the form ψ(t) = A cos ω0 t. This
function satisfies Eq. (11.13), if A = −(ω0 /2γ)θ0 = −Qθ0 . Next we add to this
partial solution the general solution of the corresponding homogeneous equation:

ψ(t) = −Qθ0 cos ω0 t + (C cos ω0 t + S sin ω0 t) exp(−γt). (11.14)

It follows from the initial condition ψ̇(0) = 0 that in (11.14) S = (γ/ω0 )C. To
find C, we require that in the steady-state symmetric regime elongations to both
sides should be equal: ϕ(0) = −ϕ(T0 /2). From this condition we get
2d 2d
C= = . (11.15)
1 − exp(−γT0 /2) 1 − exp(−π/2Q)
Substituting these C and S values in Eq. (11.14), we obtain the time dependence of
the angular displacement ϕ(t) = ψ(t)−d for the first half-cycle of excitation. The
desired amplitude a of this steady-state resonant oscillation is given by −ϕ(0):
   
2 4d
a = Qθ0 − d − 1 ≈ Q θ0 − . (11.16)
1 − exp(−π/2Q) π
The latter approximate expression is valid in the case of rather weak viscous fric-
tion, when Q ≫ 1. In the absence of dry friction (at d = 0) the growth of ampli-
tude at resonance is restricted due to viscous friction by the value Qθ0 , which is Q
times greater than the amplitude of the driving rod θ0 , in accordance with the first
term in Eq. (11.16). With dry friction, the steady-state amplitude is approximately
Q times greater than the excess of the drive amplitude θ0 over the threshold 4d/π.
We emphasize that dry friction alone is unable to restrict the growth of ampli-
tude over the threshold at ω = ω0 . Nevertheless, Eq. (11.16) shows that when dry
friction is added to the system with viscous friction, the steady-state amplitude at
resonance is smaller than Qθ0 . From the numerical simulation (Figure 11.5) we
see that with θ0 = 15◦ and Q = 10 the resonant amplitude equals only 86.3◦,
if the dead zone d equals 5◦ (compare with Qθ0 = 150◦ at d = 0). This ex-
perimental value 86.3◦ is in good agreement with the theoretical result expressed
by (11.16), according to which the steady-state amplitude should be 86.2◦.
In conditions of exact tuning to resonance (at ω = ω0 ) the energy is trans-
ferred to the oscillator from the external source (from the exciter) with maximal
efficiency, if at the beginning of each excitation cycle the flywheel occurs at an ex-
treme elongation to the left-hand side. Indeed, in this case the sinusoidally varying
external torque exerted on the flywheel by the exciter acts during the whole cycle
in the direction of the flywheel rotation, and over the threshold (at πθ0 > 4d) over-
comes the torque of dry friction: The amplitude grows linearly (see Figure 11.2)
increasing during a cycle by πθ0 − 4d. Conversely, if at the beginning of the ex-
citation cycle the flywheel occurs at an extreme elongation to the right-hand side,

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

11.2. DRIVEN OSCILLATOR WITH DRY FRICTION 319

o
1.53 90
ϕ (t )

θ (t )
0 0

o
-1.53 o -90
-90 0 90o 0 1 2 3 4 5 6 7 8 9 10 11 12 T
Drive frequency 1.0 w0 , drive amplitude 6.366 degrees, dead zone 2.5 degrees,
initial angle 90 degrees, initial angular velocity 0.0 w0

Figure 11.6: Phase diagram with Poincaré sections and graph of ϕ(t) of oscilla-
tions with dry friction at resonance with initial deflection ϕ(0) = +90◦.

the external torque of the spring during the whole cycle is directed against the
flywheel’s angular velocity together with the frictional torque. In this case the am-
plitude reduces during each cycle by the amount πθ0 + 4d. After the amplitude
reduces to zero, the phase relations between the flywheel and exciter change to
the opposite and become favorable for the transfer of energy to the oscillator: The
amplitude begins to grow.
An example of such behavior is shown in Figure 11.6. At the drive amplitude
θ0 = 6.366◦ and the dead zone 2.5◦ , the amplitude linearly reduces during each
cycle of the initial stage of the process by πθ0 + 4d = 30◦ . After 3 full driving cy-
cles the amplitude diminishes from initial 90◦ to zero. During the further resonant
growth the amplitude linearly increases during each cycle by πθ0 − 4d = 10◦ , and
after next 9 cycles becomes 90◦ .

11.2.5 Non-Resonant Forced Oscillations

In the case of exact tuning to resonance, in contrast to the oscillator with viscous
damping, dry friction alone is unable to restrict the growth of the amplitude of
forced oscillations over the threshold. In non-resonant cases (ω 6= ω0 ) of harmonic
excitation, after a transient of a finite duration, steady-state oscillations of constant
amplitude can establish due to dry friction even in the absence of viscous friction.
Non-resonant forced oscillations in the oscillator with dry friction received sig-
nificant attention in the literature. Since the pioneer’s work of Den Hartog [57]
in 1930, several researchers [59]–[63] have investigated the system analytically
and numerically, and obtained exact solutions, describing the steady-state non-
sticking motion with two turnarounds per cycle for a harmonically excited dry
friction oscillator.
An example of such non-resonant oscillations in the system with a consider-
able amount of dry friction is shown in Figure 11.7. The periodic motion consists
of two non-sticking phases of equal duration T /2. The angular velocity is negative
in one phase and positive in the other. Unfortunately, it is impossible to express

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

320 CHAPTER 11. PENDULUM WITH DRY AND VISCOUS DAMPING

o
1.04 80.6
ϕ (t )

0 0
θ ( t)
.
ϕ (t )
o
-1.04 o -80.6
- 80.6 0 80.6o 0 1 2 3T
Drive amplitude 45 degrees, drive frequency 0.7w0, dead zone 20 degrees

Figure 11.7: Phase diagram and graphs of ϕ(t) and ϕ̇(t) of non-sticking steady-
state oscillations with dry friction at ω = 0.7ω0 .

ϕ(t) in a closed analytical form, because the turnaround points dividing the two
phases are determined by a transcendental equation. To find the amplitude a(ω)
of this symmetric oscillation, it is sufficient to consider only one phase between
successive turnarounds, which is described by differential Eq. (2.5). The calcu-
lations are similar to those described above for the resonant case (though more
complicated). Using periodicity and symmetry of the desired solution, we find the
following dependence of the steady-state amplitude on the driving frequency ω
and amplitude θ0 = 45◦ of the excitation:
s
1 d2 (ω0 /ω)2 sin2 π(ω0 /ω)
a(ω) = θ0 − . (11.17)
(1 − ω 2 /ω02 ) θ02 (cos π(ω0 /ω) + 1)2

For the frequency of excitation ω = 0.7ω0 , drive amplitude θ0 = 45◦ , and dead
zone d = 20◦ we get from (11.17) for the steady-state amplitude the value 80.63◦,
which is in perfect agreement with the numerical simulation illustrated by Fig-
ure 11.7.
Expression (11.17) for the steady-state amplitude of non-sticking oscillations
coincides (in somewhat different notations) with results published earlier in the
literature [59], [63]. Frequency-response resonant curves (amplitude-frequency
characteristics) given by (11.17) for the oscillator with dry friction are shown
in Figure 11.8 for several values of relative width d/θ0 of the dead zone (in the
frequency region ω > 0.5ω0 ).
We emphasize that expression (11.17) is valid only for sliding (non-sticking)
symmetric motions of the oscillator. Such motions are possible if the following
simple implicit condition (see [59]) on the parameters is fulfilled:
d  ω0 2
a(ω, θ0 , d) ≥ . (11.18)
θ0 ω
Solving Eq. (11.18) numerically for the unknown d at ω = 0.7ω0 and θ0 = 45◦
(these values were used for the simulation shown in Figure 11.7), we find that

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

11.2. DRIVEN OSCILLATOR WITH DRY FRICTION 321

a/q0
10

8
1 1
6
2 2
4
d=0 3 3
2
4 4
0
0.6 0.8 1.0 1.2 1.4 w/w0
1- d = 0, 2 - d/q0 = 0.35, 3- d/q0 = 0.5, 4- d/q0 = 0.6

Figure 11.8: Amplitude–frequency characteristics of sinusoidally driven spring

oscillator with dry friction damping.

the maximal width dmax of the dead zone for which steady-state non-sticking
symmetric motions are possible equals 32.5◦ . A closed-form formula for the do-
main of steady-state symmetric non-sticking oscillatory responses was obtained
in [60]. It provides the minimum driving torque amplitude (θ0 )min required to
prevent sticking for given width d of the dead zone and given drive frequency ω:
s
2 
(ω/ω0 )2 sin2 (πω0 /ω)
 2 
ω0
(θ0 )min = d −1 1+ . (11.19)
ω2 (1 + cos(πω0 /ω)2

Certainly, this equation can also be used to find the maximal width dmax of the
dead zone for which steady-state non-sticking symmetric motions are possible at
given frequency ω and amplitude θ0 of the driving torque. Substituting ω = 0.7ω0
and θ0 = 45◦ in (11.19), we get dmax = 32.5◦, in accordance with the above
estimate (11.18).
The upper part of Figure 11.9 illustrates oscillations occurring on this edge
of such a non-sticking regime (dead zone 32.5◦ ). For initial conditions ϕ(0) =
0, ϕ̇(0) = 0, sticking occurs several times during a short transient, which ends
with non-sticking symmetric steady-state oscillations. According to (11.17), their
amplitude must equal 66.3◦, in good agreement with the simulation.
For comparison, the lower part of Figure 11.9 shows the steady-state oscilla-
tions at the same values of the frequency and amplitude of the exciter (ω = 0.7ω0
and θ0 = 45◦ ), but for a somewhat greater amount of dry friction (dead zone 37◦ ).
In this case sticking occurs twice during each cycle of excitation.
Not surprisingly, the amplitude of steady-state symmetric forced oscillations
with sticking observed in the simulation is smaller than the theoretical value that
Eq. (11.17) predicts (55◦ as opposed to 62◦ ).

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

322 CHAPTER 11. PENDULUM WITH DRY AND VISCOUS DAMPING

o
0.96 66.3
ϕ (t )

0 0
.
ϕ (t )

o θ ( t)
-0.96 o - 66.3
o
-66.3 0 66.3 0 1 2 3 4 5T
Drive frequency 0.7w0, drive amplitude 45 degrees, dead zone 32.5 degrees
o
0.85 55.0
ϕ (t )

0 0
θ ( t)
.
o
ϕ (t )
-0.85 o -55.0
o
-55.0 0 55.0 0 1 2 3T
Drive frequency 0.7w0 , drive amplitude 45 degrees, dead zone 37 degrees

Figure 11.9: Phase diagram and graphs of ϕ(t) and ϕ̇(t) of the transient that leads
to non-sticking steady-state oscillations at ω = 0.7ω0 and critical width d =
32.5◦ of the dead zone (upper part), and steady-state symmetric oscillations with
sticking at d = 37◦ (lower part).

11.3 Harmonic Excitation at Sub-Resonant

Frequencies
Generally characteristics of steady-state behavior of the periodically forced os-
cillator with dry friction, as well as of the oscillator with viscous friction, are
uniquely defined by the system parameters (natural frequency and the quality fac-
tor), and by the frequency and amplitude of the excitation. Certain exceptions are
revealed if the frequency ω of sinusoidal excitation coincides with one of subhar-
monics of the natural frequency: ω = ω0 /n, where n is an integer number.
Analytical steady-state solutions at sub-harmonic excitation were considered
for the first time in [64]. Here we suggest a simpler and physically more transpar-
ent approach to the problem, and discuss peculiarities of such oscillations in more
detail. Simple closed-form solutions are illustrated by time-dependent graphs and
phase orbits obtained with the help of computer simulations.
According to Eq. (11.17), at frequencies of excitation ω = ω0 /2, ω = ω0 /4,

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

11.3. EXCITATION AT SUB-RESONANT FREQUENCIES 323

1.4 a/q0
1.2 d=0

1.0

0.8 d/q0 = 0.1

0.6

0.4

0.2

0.0 0.1 0.2 0.3 0.4 0.5

w/w0
1.4 a/q0
1.2 d=0

1.0
d/q0 = 0.03
0.8

0.6

0.4

0.2

0.0 0.1 0.2 0.3 0.4 0.5 w/w0

1.4
a/q0
1.2 d=0

1.0

0.8 d/q0 = 0.01

0.6

0.4

0.2

0.0 0.1 0.2 0.3 0.4 0.5 w/w 0

Figure 11.10: Frequency–response curves for the oscillator with dry friction given
by Eq. (11.17) at excitation frequencies ω < ω0 /2 for several values of the dead
zone width (d/θ0 = 0.1, d/θ0 = 0.03, d/θ0 = 0.01).

. . . the amplitude of steady-state non-sticking symmetric oscillation, indepen-

dently of the dead zone width d, should be equal to the amplitude of forced
steady-state oscillation in the absence of friction (that is, at d = 0): a(ω) =
θ0 /(1 − ω 2 /ω02 ). (Certainly, this arbitrary value of d should satisfy the condi-
tion (11.19) for non-sticking motions, which at ω = ω0 /n gives d ≤ θ0 /3.)
Figure 11.10 shows that frequency-response curves for different d values at
ω = ω0 /2 graze the curve for d = 0. Computer simulations testify that in these
cases steady-state oscillations are generally asymmetric: The angular excursion to
one side is greater than to the other. This means that at ω = ω0 /n occurrence of
special analytical solutions can be expected. Below we show that in contrast with

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

324 CHAPTER 11. PENDULUM WITH DRY AND VISCOUS DAMPING

0.91 .
ϕ (t )

θ ( t)
0

-0.91 o
-95.0 0 95.0o
ϕ (t )
Drive frequency 0.5w 0 ,
o
drive amplitude 60,
o
dead zone 5.0 , 0
initial angle 65.0 o,
θ ( t)
initial ang. velocity 0.0w 0 ,
o
max. elongation 65.0 , o
o
min. elongation -95.0 -95.0
0 1 2T

Figure 11.11: Phase diagram and time-dependent graphs of ϕ̇(t) and ϕ(t) with the
graphs of their harmonics for non-sticking asymmetric steady-state oscillations at
ω = ω0 /2 and small width of the dead zone (d = 5.0◦ ).

the general case of forced oscillations, for which the steady-state regime is de-
scribed by the unique solution that is independent of the initial conditions, a con-
tinuum of asymmetric non-sticking solutions exists at ω = ω0 /2n. Each solution
gives an asymmetric limit cycle (attractor) that corresponds to initial conditions
from a certain basin of attraction.
To study the excitation of the oscillator at ω = ω0 /n, it is more convenient
to choose further on for the time origin t = 0 the moment at which the exciter
reaches its maximal deflection θ0 , that is, to assume for θ(t) the following time de-
pendence: θ(t) = θ0 cos ωt. With this choice, as we will see later, the turnarounds
in the steady-state motion of the oscillator occur approximately at t = 0 and
t = T /2. This simplifies the form of analytical solutions.
For definiteness we restrict further discussion to the case n = 2. If ω = ω0 /2,
the spectrum of steady-state asymmetric oscillations at sufficiently small dry fric-
tion (narrow dead zone) consists primarily of the principal harmonic with the fre-
quency of excitation ω and its second harmonic, whose frequency 2ω equals the
natural frequency ω0 of the oscillator (see Figure 11.11). A small admixture of the
third harmonic is also noticeable.
Mathematically, the principal harmonic corresponds to the forced periodic par-
tial solution of the nonhomogeneous differential equation of motion (11.3) with
γ = 0 and with the sinusoidal forcing term ω02 θ0 cos ωt whose frequency ω equals
ω0 /2. This partial solution is 43 θ0 cos ωt. The torque of dry friction moves the
mid-point of this oscillation to −d (left boundary of the dead zone) if ϕ̇ > 0
and to d if ϕ̇ < 0. This periodic (square-wave) displacement of the mid-point
caused by dry friction explains the appearance of the third harmonic in the solu-
tion shown in Figure 11.11. The second harmonic with the frequency 2ω = ω0

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

11.3. EXCITATION AT SUB-RESONANT FREQUENCIES 325

is the general solution of the homogeneous equation that corresponds to (11.3).

This general solution describes natural oscillations with the frequency ω0 = 2ω,
and can be represented (at γ = 0) as A cos 2ωt + B sin 2ωt, where A and B are
arbitrary constants. In contrast to a system with viscous friction, now this general
solution — oscillation with the natural frequency — does not damp out in the
course of time during the transient. The simulation shows (in accordance with the
requirement ϕ̇(0) = ϕ̇(T /2) = 0) that in the steady-state regime the phase of this
second harmonic is such that B = 0 (see Figure 11.11). Hence the asymmetric
steady-state motion at ω = ω0 /2 can be approximately described by the following
equations:
4
ϕ+ (t) = 3 θ0 cos ωt + A+ cos 2ωt − d, ϕ̇ > 0, (11.20)
4
ϕ− (t) = 3 θ0 cos ωt + A− cos 2ωt + d, ϕ̇ < 0, (11.21)
and
ϕ̇+ (t) = − 43 ωθ0 sin ωt − 2ωA+ sin 2ωt, ϕ̇ > 0, (11.22)
ϕ̇− (t) = − 34 ωθ0 sin ωt − 2ωA− sin 2ωt, ϕ̇ < 0. (11.23)
One condition on constants A+ and A− follows from the requirement of continu-
ity of ϕ(t) at the turnaround points, when the sign of velocity reverses. These are
the moments t = 0 and t = T /2 (see Figure 11.11). From ϕ+ (0) = ϕ− (0) we
get A+ − A− = 2d, or A+ = A− + 2d (condition ϕ− (T /2) = ϕ+ (T /2) yields
the same relation between A+ and A− ). Therefore only one of these constants
remains arbitrary.
The steady-state regime described by Eqs. (11.20)–(11.23) occurs from the
very beginning (that is, without any transient) if the initial conditions are chosen
properly. At t = 0 we get from (11.22) or (11.23) that the required initial angular
velocity equals zero: ϕ̇(0) = 0. This value is independent of the dead zone width
d and the drive amplitude θ0 . Since during the first half-cycle ϕ̇(0) is negative, for
the required initial displacement we should use Eq. (11.21), which yields ϕ0 =
ϕ(0) = 34 θ0 + A− + d. We see that the arbitrary constants A+ and A− , which
determine the contribution of the second harmonic into the steady-state motion,
depend on an arbitrary initial displacement ϕ0 :
4 4
A+ = ϕ0 − θ0 + d, A− = ϕ0 − θ0 − d. (11.24)
3 3
This means that in the system with dry friction, in contrast to the oscillator with
viscous friction, different initial displacements generally lead to different regimes
of steady-state oscillations (to different limit cycles). Substituting these values
of A+ and A− in (11.20)–(11.23), we get the closed-form analytical solutions
for asymmetric steady-state subresonant regimes of the dry-friction oscillator at
ω = ω0 /2. Below we show that such steady-state regimes occur from the very
beginning if the value of ϕ0 belongs to a certain interval.
Now we can derive some interesting properties of the discussed steady-state
solutions. Extreme elongations correspond to the turnaround points and hence oc-
cur at t ≈ 0 and at t ≈ T /2. Maximum displacement to the right-hand side occurs

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

326 CHAPTER 11. PENDULUM WITH DRY AND VISCOUS DAMPING

0.72 .
ϕ (t )

θ ( t)
0

-0.72 o o
-85.0 0 85.0
ϕ (t )
Drive frequency 0.5w 0 ,
drive amplitude 60 o,
dead zone 5.0 o, 0
o
initial angle 80.0 , θ ( t)
initial ang. velocity 0.0w 0 ,
o
max elongation 80.0, o
min elongation -80.0o -85.00 1 2T

Figure 11.12: Phase diagram and time-dependent graphs of ϕ̇(t) and ϕ(t) with the
graphs of their harmonics for non-sticking symmetric steady-state oscillations at
ω = ω0 /2 and small width of the dead zone (d = 5.0◦ ).

at t ≈ 0 and, according to Eq. (11.20) or (11.21), equals ϕmax = 34 θ0 + A− + d

(or, equivalently, ϕmax = 34 θ0 + A+ − d). Extreme elongation to the left-hand
side occurs t ≈ T /2 and equals |ϕmin | = 34 θ0 − A− − d = 43 θ0 − A+ + d.
We get that the total angular excursion ϕmax + |ϕmin | between the extreme
points for all possible solutions (11.20)–(11.21) equals 38 θ0 :

8
ϕmax + |ϕmin | = ϕmax − ϕmin = θ0 . (11.25)
3
It depends solely on the drive amplitude θ0 , and does not depend on the intensity
of dry friction (on the width d of the dead zone).
The difference between the extreme elongations characterizes the asymmetry
of this steady-state regime:
4
ϕmax − |ϕmin | = 2(A+ − d) = 2(A− + d) = 2(ϕ0 − θ0 ). (11.26)
3
The extreme elongations to both sides are equal to one another if ϕ0 − 34 θ0 = 0.
In this case A+ = d and A− = −d, and the second harmonic in the oscillation
described by Eqs. (11.20)–(11.21) vanishes. Such symmetric steady-state oscilla-
tion with the amplitude 43 θ0 occurs only if the initial displacement ϕ0 equals 43 θ0
and the initial velocity zero.
The phase trajectory and time-dependent graphs of ϕ̇(t) and ϕ(t) (together
with the graphs of their harmonics) for such symmetric oscillation are shown in
Figure 11.12. (These graphs at d ≪ θ0 almost merge with the graphs of their
principal harmonics.)

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

11.3. EXCITATION AT SUB-RESONANT FREQUENCIES 327

The initial conditions that lead to the greatest asymmetry of the limit cycle
can be found as follows. We are interested in the unsticking regime with two
turnarounds per one excitation cycle. These turnarounds occur near t = 0 and
t = T /2, when the angular velocity ϕ̇ changes sign. We can rely on physical
considerations in finding the condition for a turnaround occurring without sticking
for a finite time interval. Indeed, to avoid sticking at this point, the restoring torque
of the spring exerted on the flywheel should be greater or at least equal to the
greatest possible torque of dry friction. The desired condition corresponds to the
equality of these two torques. When the torque exerted by the spring equals the
torque of static friction in magnitude, the angular acceleration ϕ̈ of the flywheel
equals zero. Next we consider this condition for each of the turnarounds, occurring
at t = 0 and t = T /2.

• For the first turnaround occurring at t = 0 we should require ϕ̈(0) = 0 using

Eq. (11.21) that corresponds to negative angular velocity ϕ̇(t) < 0 (see Fig-
ure 11.11). From this requirement we find immediately A− = − 31 θ0 . Sub-
stituting this value to Eq. (11.24), we get the first (lower) boundary ϕ0(lower)
of admissible initial deflections:

ϕ0(lower) = θ0 + d. (11.27)

With this initial displacement one of the two possible most asymmetric
steady-state oscillations occurs, in which the extreme deflections are:
5
ϕmax = θ0 + d, ϕmin = − θ0 + d. (11.28)
3

• To find the other boundary of the admissible initial deflections ϕ0(upper) ,

we should require ϕ̈(T /2) = 0 using Eq. (11.20). This yields A+ = 31 θ0 ,
and for the second (upper) boundary of admissible initial deflections we get:
5
ϕ0(upper) = θ0 − d. (11.29)
3
With this initial displacement the other of the two possible most asymmetric
steady-state oscillations occurs, in which the extreme deflections are:
5
ϕmax = θ0 − d, ϕmin = −θ0 − d. (11.30)
3

To verify these theoretical predictions with the help of numerical simulations,

we choose the drive amplitude θ0 = 60◦ , which means that in the steady-state
regime the total angular excursion between the extreme points ϕmax + |ϕmin |
should equal 38 θ0 = 160◦ independently of the intensity of dry friction. Width d
of the dead zone in this simulation equals 5◦ . One of the two possible steady-state
oscillations with the greatest asymmetry (see Figure 11.11) occurs from the very
beginning (without any transient), according to (11.27), at the initial conditions

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

328 CHAPTER 11. PENDULUM WITH DRY AND VISCOUS DAMPING

0.91 .
ϕ( t )
θ ( t)

-0.91 o o
-95.0 0 95.0
Drive frequency 0.5w 0 , ϕ (t )
o
drive amplitude 60,
o
dead zone 5.0, 0
o
initial angle 95.0, θ ( t)
initial ang. velocity 0.0w 0 ,
o
max. elongation 95.0,
o o
min. elongation -65.0 - 95.0
0 1 2T

Figure 11.13: Phase diagram and time-dependent graphs of ϕ̇(t) and ϕ(t) with
their harmonics for non-sticking asymmetric steady-state oscillations at ω = ω0 /2
and small width of the dead zone (d = 5.0◦ ).

ϕ(0) = θ0 + d = 65◦ , ϕ̇(0) = 0. The extreme elongations should be ϕmax =

θ0 + d = 65◦ and ϕmin = − 53 θ0 + d = −95◦ . These theoretical predictions agree
perfectly well with the computer simulation (see Figure 11.11).
Steady-state oscillation with equal elongations to both sides (ϕmax =
|ϕmin | = 80◦ ) at the same values of the system parameters (θ0 = 60◦ , d = 5◦ )
occurs if the initial displacement ϕ0 = 43 θ0 = 80◦ . This case is illustrated by the
simulation shown in Figure 11.12.
The second case of the greatest asymmetry (ϕ̇(T /2) = 0) occurs, according to
(11.29), at the initial conditions 53 θ0 − d = 95◦ , ϕ̇(0) = 0. Extreme elongations
in this case, according to (11.30), are ϕmax = 53 θ0 − d = 95◦ and ϕmin =
−θ0 − d = −65◦. These asymmetric oscillations are illustrated by the simulation
shown in Figure 11.13.
Three different limit cycles of non-sticking oscillations that are shown in
Figures 11.11–11.13 correspond to the same values of the system parameters
ω = ω0 /2, θ0 = 60.0◦ , and d = 5.0◦ . Actually, at ω = ω0 /2 there exists a
continuum of different steady-state non-sticking motions with the same total an-
gular excursion 83 θ0 , proportional to the drive amplitude θ0 . This is a manifesta-
tion of multistability — a typical feature of nonlinear systems. If the initial angular
velocity ϕ̇(0) equals zero, and the initial angular displacement ϕ0 lies in the in-
terval from ϕ0(lower) = θ0 + d to ϕ0(upper) = 53 θ0 − d, the steady-state motion
starts without a transient. The character of these steady-state oscillations varies
in this interval of initial displacements from one of the most asymmetric cases
at ϕ0(lower) = θ0 + d (see Figure 11.11) through the symmetric case occurring at
ϕ0 = 43 θ0 (Figure 11.12) to the other most asymmetric case at ϕ0(upper) = 53 θ0 −d
(Figure 11.13). If the initial displacement lies beyond this interval, or the initial

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

11.3. EXCITATION AT SUB-RESONANT FREQUENCIES 329

o
1.66 142 .
ϕ ( t) ϕ (t ) θ ( t)

0 0

o
-1.66 o -142
- 142 0 142o 0 1o 2 3 4 5 6T
o o
Drive amplitude 60.0, dead zone 5.0, initial displacement 0.0, initial velocity 0.0 w0

Figure 11.14: Phase diagram and time-dependent graphs of ϕ(t) and ϕ̇(t) for a
transient at ω = ω0 /2, θ0 = 60.0◦ , d = 5.0◦ , and at the initial conditions
ϕ(0) = 0, ϕ̇(0) = 0.

velocity is not equal to zero, one of the limit cycles from the same continuum
is eventually established after a transient process, during which oscillations with
sticking for finite time intervals take place. An example of such a transient occur-
ring at θ0 = 60.0◦, d = 5.0◦ , and initial conditions ϕ(0) = 0, ϕ̇(0) = 0 is shown
in Figure 11.14.
Not surprisingly, if even a small amount of viscous friction is present in the
system, the above-considered asymmetric regimes can be observed only at the ini-
tial stage: After a long transient oscillations become symmetric, like those shown
in Figure 11.12. Indeed, the asymmetry is caused by the contribution of the second
harmonic, which corresponds to natural oscillations with the frequency ω0 = 2ω.
Mathematically, this second harmonic is the general solution of the homogeneous
differential equation. In the presence of viscous friction, these natural oscillations
damp out during the transient.
Steady-state non-sticking sub-resonant forced oscillations of the above-con-
sidered type exist if dry friction is not strong enough for the given drive ampli-
tude. To find this restriction, it is sufficient to equate the expressions for the lower
and upper boundaries of the interval of admissible initial deflections given by ex-
pressions (11.27) and (11.29): ϕ0(lower) = ϕ0(upper) (or for extreme elongations
ϕmax = θ0 + d and |ϕmin | = 35 θ0 − d at any of the most asymmetric cases).
This yields dmax = θ0 /3. At d = θ0 /3 there exists only one (symmetric) limit
cycle with the amplitude ϕmax = 43 θ0 . At greater values of the dead zone width
(d > θ0 /3) only steady-state oscillations with sticking are possible.
Similar peculiarities are characteristic of forced oscillations with dry fric-
tion, sinusoidally excited at other sub-resonant frequencies of even orders ω =
ω0 /(2n). In particular, for ω = ω0 /4 a continuum of non-sticking asymmetric
steady-state motions exists for the same values of the system parameters, if the
1
width d of the dead zone does not exceed 15 θ0 , where θ0 is the drive ampli-
tude. The full angular excursion between extreme elongations equals 32 15 θ0 . Each
of these periodic motions occurs without a transient, if the initial velocity equals
17
zero, and the initial displacement lies in the interval between θ0 + d and 15 θ0 − d.

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

330 CHAPTER 11. PENDULUM WITH DRY AND VISCOUS DAMPING

0.50 .
ϕ (t )
θ ( t)
0 0

-0.50 o
-99.0 0 99.0o ϕ (t )
Drive frequency 0.25w 0,
o
drive amplitude 90.0,
o
dead zone 3.0 , 0
initial angle 93.0,o
θ ( t)
initial ang. velocity 0.0w 0 ,
o
max. elongation 93.0 ,
o o
min. elongation -99.0 -99.0
0 1 2T

Figure 11.15: Phase diagram and time-dependent graphs of ϕ̇(t) and ϕ(t) with
their harmonics for non-sticking asymmetric steady-state oscillations at ω = ω0 /4
and small width of the dead zone (d = 3.0◦ ).

An example of such asymmetric steady-state oscillation is shown in Figure 11.15.

The stability of asymmetric non-sticking solutions at ω = ω0 /(2n) was dis-
cussed in [64]. The authors claim that in the parameters domain in which these
solutions exist they are marginally stable in the third-order approximation.
At sub-resonant drive frequencies of odd orders ω = ω0 /(2n+1) non-sticking
solutions for an oscillator with dry friction do not exist: At least twice during each
cycle of the steady-state motion velocity turns to zero for finite time intervals.
An example of steady-state forced oscillation at ω = ω0 /3 is shown in Fig-
ure 11.16. These time-dependent graphs of ϕ̇(t) and ϕ(t) with their harmonics
and the phase orbit are obtained in a computer simulation. For the drive ampli-
tude 80.0◦ and dead zone width 10.0◦ oscillations are symmetric with maximum
elongation 78.0◦ . The spectrum contains harmonics only of odd orders: The first
harmonic has the amplitude 91.8◦, the third 15.6◦, the fifth 1.4◦ . These values are
practically the same in cases where the dead zone width lies in the interval 5.0◦ –
25.0◦ . Such motions consisting of two sliding phases and two sticking phases of a
finite duration during each cycle of sinusoidal excitation in a dry friction oscillator
are investigated in [71].

11.4 Concluding Remarks

In this chapter we concentrated on peculiarities in behavior of a simple mechanical
system — a torsion spring oscillator with dry and viscous friction. The intensity
of dry friction is characterized by the width d of the dead zone. The amplitude of
non-forced oscillations reduces under dry friction during each cycle by the same

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

11.4. CONCLUDING REMARKS 331

0.83 .
ϕ (t )

θ ( t)
-0.83 o o
-78.0 0 92.0
ϕ (t )
Drive frequency w 0 /3,
drive amplitude 80.0o,
dead zone 10.0 o, 0
o
maximal deflection 77.9 ,
initial deflection 80.0 o,
initial angular velocity 0.0
o
- 92.0
0 1 2 T0

Figure 11.16: Phase diagram and time-dependent graphs of ϕ̇(t) and ϕ(t) with
their harmonics for symmetric steady-state oscillations at ω = ω0 /3 with two
sliding and two sticking phases during each cycle (θ0 = 80.0◦ , d = 10.0◦ ).

amount 4d, proportional to the dead zone width, and the oscillator stops dead
after a finite time. Under sinusoidal forcing, dry friction cannot restrict the growth
of resonant oscillations: At ω = ω0 the amplitude of forced oscillations grows
indefinitely, increasing in each cycle through πθ0 − 4d, if the drive amplitude θ0
exceeds the threshold value 4d/π.
In non-resonant cases (ω 6= ω0 ) of harmonic excitation, after a transient of
a finite duration, steady-state oscillations of a finite amplitude can establish due
to dry friction even in the absence of viscous friction. Generally, such periodic
steady-state motion consists of two symmetric non-sticking phases of equal du-
ration T /2, if, for the given width d of the dead zone, the drive amplitude θ0 is
large enough to prevent sticking. The steady-state amplitude of these symmetric
oscillations for given ω uniquely depends on θ0 and d.
At sub-resonant frequencies of excitation ω = ω0 /n (n = 2, 4, . . .) certain
peculiarities of forced oscillations reveal themselves. In particular, in the absence
of viscous friction a continuum of different non-sticking steady-state oscillations
can exist for the same values of the system parameters θ0 and d. Such oscillations
are generally asymmetric: The angular elongation to one side is greater than to the
other, though the total angular excursion between the extreme (turnaround) points
is the same for a given value of the drive amplitude θ0 (and is independent of the
dead zone width d).
The asymmetry of a certain steady-state regime of this continuum depends
on the initial conditions. Among each continuum of such solutions coexisting at
given values of θ0 and d, there is a single symmetric oscillation. If even a small
amount of viscous friction is present in the system, these asymmetric regimes
can be observed only at the initial stage: After a long transient the contribution

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

332 CHAPTER 11. PENDULUM WITH DRY AND VISCOUS DAMPING

of natural (second) harmonic that causes asymmetry dies out, and eventually the
oscillations become symmetric.

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

Bibliography

[1] G. L. Baker, J. A. Blackburn, The Pendulum: A Case Study in Physics. New

York: Oxford University Press, 2005.
[2] Ch. Kittel, W. D. Knight, M. A. Ruderman, Mechanics. Berkeley Physics
Course, New York: McGraw-Hill Book Company, 1965.
[3] A. B. Pippard, The physics of vibration (The simple classical vibrator). New
York: Cambridge University Press, 1978.
[4] L. D. Landau and E. M. Lifschitz, Mechanics, Moscow: Nauka Publishers,
(in Russian), 1976, Mechanics, New York: Pergamon, 1988.
[5] K. Johannessen, An approximate solution to the equation of motion for large-
angle oscillations of the simple pendulum with initial velocity Eur. J. Phys.
31, 511–18, 2010.

[6] A. Belendez, J. J. Rodes, A. Hernandez, T. Belendez, Approximation for a

large-angle simple pendulum period, Eur. J. Phys. 30, L25–8, 2009.
[7] C. G. Carvalhaes, P. Suppes, Approximations for the period of the simple
pendulum based on the arithmetic-geometric mean, Am. J. Phys. 76, 1150–
4, 2008.
[8] F. M. S. Lima, Simple ‘log formulae’ for the pendulum motion valid for any
amplitude, Eur. J. Phys. 29, 1091–8, 2009.
[9] F. M. S. Lima, P. Arun, An accurate formula for the period of a simple pen-
dulum oscillating beyond the small angle regime, Am. J. Phys. 74, 892–5,
2006.
[10] Y. Q. Xin and D. Pei, Comment on Approximation for a large-angle simple
pendulum period, Eur. J. Phys. 30, L79–82, 2009.
[11] E. I. Butikov, The rigid pendulum—an antique but evergreen physical model,
Eur. J. Phys. 20, 429–441, 1999.
[12] G. L. Baker, J. P. Gollub Chaotic Dynamics. An Introduction, New York:
Cambridge University Press, 1990.

333

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

334 BIBLIOGRAPHY

[13] J. W. Miles, Resonance and symmetry breaking for the pendulum, Physica,
D31, 252–268, 1988.
[14] H. Heng, R. Doerner, B. Hübinger, W. Martienssen, Approaching nonlinear
dynamics by studying the motion of a pendulum. I. Observing trajectories in
phase space, Int. J. Bifurcation and Chaos, 4, 751–760, 1994.
[15] R. Doerner, B. Hübinger, H. Heng, W. Martienssen, Approaching nonlin-
ear dynamics by studying the motion of a pendulum. II. Analyzing chaotic
motion, Int. J. Bifurcation and Chaos, 4, 761–771, 1994.
[16] H. W. Broer, I. Hoveijn, M. van Noort, C. Simó, G. Vegter, The Parametri-
cally Forced Pendulum: A Case Study in 1 21 Degree of Freedom, J. Dynam-
ics and Differential Equations, 16, 897–947, 2004.
[17] E. I. Butikov, On the dynamic stabilization of an inverted pendulum, Am. J.
Phys. 69, 755–68, 2001.
[18] R. D. Peters, Resonance response of a moderately driven rigid planar pendu-
lum, Am. J. Phys. 64, 170–173, 1996.
[19] E. I. Butikov, Subharmonic Resonances of the Parametrically Driven Pendu-
lum, J. Phys. A: Math. and Gen. 35, 6209–6231, 2002.
[20] E. I. Butikov, Regular and Chaotic Motions of the Parametrically Forced
Pendulum: Theory and Simulations, Computational Science – ICCS 2002,
Springer Verlag, LNCS 2331, 1154–1169, 2002.

[21] E. I. Butikov, An improved criterion for Kapitza’s pendulum stability, J.

Phys. A: Math. and Theor. 44, 295202 (16 pp), 2011.
[22] S. M. Curry, How children swing, Am. J. Phys. 44, 924–926, 1976.
[23] P. L. Tea Jr., H. Falk, Pumping on a swing, Am. J. Phys. 36, 1165–1166,
1968.
[24] W. Case, The pumping of a swing from the standing position, Am. J. Phys.
64, 215–220, 1996.
[25] M. A. Pinsky, A. A. Zevin, Oscillations of a pendulum with a periodically
varying length and a model of swing, Int. J. Non-Linear Mech. 34, 105–109,
1999.
[26] D. Stilling, W. Szyskowski, Controlling angular oscillations through mass
reconfiguration: a variable length pendulum case, Int. J. Non-Linear Mech.
37, 89–99, 2002.
[27] E. I. Butikov, Pendulum with a square-wave modulated length (simulation
program), http://butikov.faculty.ifmo.ru/Applets/PendParSquare.html, 2013.

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

BIBLIOGRAPHY 335

[28] E. I. Butikov, Parametric excitation of a linear oscillator, Eur. J. Phys. 25,

535–554, 2004.
[29] E. I. Butikov, Parametric resonance in a linear oscillator at square-wave mod-
ulation, Eur. J. Phys. 26, 157–174, 2005.
[30] J. B. McLaughlin, Period-doubling bifurcations and chaotic motion for a
parametrically forced pendulum, J. Stat. Physics 24, 375–388, 1981.
[31] B. P. Koch, R. W. Leven, B. Pompe, C. Wilke, Experimental evidence for
chaotic behavior of a parametrically forced pendulum, Phys. Lett. A 96, 219–
224, 1983.
[32] R. W. Leven, B. Pompe, C. Wilke, B. P. Koch, Experiments on periodic and
chaotic motions of a parametrically forced pendulum, Physica D 16, 371–
384, 1985.
[33] W. van de Water, M. Hoppenbrouwers, Unstable periodic orbits in the para-
metrically excited pendulum, Phys. Rev. A 44, 6388–6398, 1991.
[34] J. Starrett, R. Tagg, Control of a chaotic parametrically driven pendulum,
Phys. Rev. Lett. 74, 1974–1977, 1995.
[35] S.-Y. Kim, B. Hu, Bifurcations and transitions to chaos in an inverted pen-
dulum, Phys. Rev. E 58, 3028–3035, 1998.
[36] A. Stephenson, On an induced stability, Phil. Mag. 15, 233–236, 1908.
[37] P. L. Kapitza, Dynamic stability of the pendulum with vibrating suspension
point, Soviet Physics – JETP 21, 588–597, 1951 (in Russian), see also Col-
lected papers of P. L. Kapitza, edited by D. Ter Haar, London: Pergamon, v.
2, pp. 714–726, 1965.
[38] P. L. Kapitza, Pendulum with an oscillating pivot Sov. Phys. Uspekhi, 44, pp.
7–20, 1951.
[39] L. D. Landau and E. M. Lifschitz, Mechanics, Moscow: Nauka Publishers,
1988 (in Russian), Mechanics, New York: Pergamon, 1976, pp. 93–95.
[40] F. M. Phelps, III, J. H. Hunter, Jr. An analytical solution of the inverted
pendulum, Am. J. Phys. 33, 285–295, 1965, 34, 533–535, 1966.
[41] D. J. Ness, Small oscillations of a stabilized, inverted pendulum, Am. J.
Phys. 35, 964–967, 1967.
[42] H. P. Kalmus, The inverted pendulum, Am. J. Phys. 38, 874–878, 1970.
[43] M. M. Michaelis, Stroboscopic study of the inverted pendulum, Am. J. Phys.
53, 1079–1083, 1985.

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

336 BIBLIOGRAPHY

[44] J. A. Blackburn, H. J. T. Smith, N. Groenbech-Jensen, Stability and Hopf

bifurcations in an inverted pendulum, Am. J. Phys. 60, 903–908, 1992.
[45] H. J. T. Smith, J. A. Blackburn, Experimental study of an inverted pendulum,
Am. J. Phys. 60, 909–911, 1992.
[46] M. J. Moloney, Inverted pendulum motion and the principle of equivalence,
Am. J. Phys. 64, 1431, 1996.
[47] W. T. Grandy, Jr., M. Schöck, Simulations of nolinear pivot-driven pendula,
Am. J. Phys. 65, 376–381, 1997.
[48] J. G. Fenn, D. A. Bayne, B. D. Sinclair, Experimental investigation of the
‘effective potential’ of an inverted pendulum, Am. J. Phys. 66, 981–984,
1998.
[49] E. I. Butikov, On the dynamic stabilization of an inverted pendulum, Am. J.
Phys. 69, 755–768, 2001.
[50] P. S. Landa, Nonlinear Oscillations and Waves in Dynamical Systems. Dor-
drecht: Kluwer Academic Publishers, 1996.
[51] D. J. Acheson, Multiple-nodding oscillations of a driven inverted pendulum,
Proc. Roy. Soc. London A 448, 89–95, 1995.
[52] E. I. Butikov, Subharmonic Resonances of the Parametrically Driven Pendu-
lum, J. Phys. A: Mathematical and General, 35, 6209-6231, 2002.
[53] B. Feeny, A. Guran, N. Hinrichs, K. Popp, A historical review on dry friction
and stickslip phenomena, ASME Applied Mechanics Reviews, 51, 321–341,
1998.
[54] E. I. Butikov, Physics of Oscillations, User’s Manual. New York: American
Institute of Physics, Physics Academic Software, 1996.
[55] L. M. Burko, A Piecewise-Conserved Constant of Motion for a Dissipative
System, Eur. J. Phys. 20, 281–288, 1999.
[56] A. Marchewka, D. S. Abbott, R. J. Beichner, Oscillator damped by a con-
stantmagnitude friction force. Am. J. Phys. 72, 477–483, 2004.
[57] J. P. Den Hartog, Forced vibrations with combined Coulomb and viscous
damping, Transactions of the American Society of Mechanical Engineers,
53, 107–115, 1930.
[58] E. I. Butikov, Torsion Pendulum with Dry and Viscous Friction. Web page
with an embedded simulation program (Java applet), see on the web at
http://butikov.faculty.ifmo.ru/Applets/DryViscOsc.html, 2013.
[59] S. W. Shaw, On the dynamic response of a system with dry friction, J. Sound
Vib. 108, 305–325, 1986.

✐ ✐

✐ ✐
 ✐ ✐

✐ ✐
Free ebooks ==> www.Ebook777.com

BIBLIOGRAPHY 337

[60] H.-K. Hong, C.-S. Liu, Non-sticking oscillation formulae for Coulomb fric-
tion under harmonic loading, J. Sound Vib. 244, 883–898, 2001.
[61] J. W. Liang, B. F. Feeny, Identifying Coulomb and Viscous Friction in Forced
Dual-Damped Oscillators, Journal of Vibration and Acoustics, 126, 118–
125, 2004.
[62] A. C. J. Luo, B. C. Gegg, Stick and non-stick periodic motions in periodi-
cally forced oscillators with dry friction, J. Sound Vib., 291, 132–168, 2006.
[63] G. Csernák, G. Stépán, On the periodic response of a harmonically excited
dry friction oscillator. J. Sound Vib., 295, 649–658, 2006.
[64] G. Csernák, G. Stépán, S. W. Shaw, Sub-harmonic resonant solutions of a
harmonically excited dry friction oscillator, Nonlinear Dynamics, 53, 93–
109, 2007.
[65] S. Chatterjee, Resonant locking in viscous and dry friction damper kinemat-
ically driving mechanical oscillators, J. Sound Vib., 332, 3499–3516, 2013.
[66] E. Pratt, A. Léger, X. Zhang, Study of a transition in the qualitative behavior
of a simple oscillator with Coulomb friction, Nonlinear Dynamics, 74, 517–
531, 2013.
[67] A. C. J. Luo, Regularity and Complexity in Dynamical Systems, Springer,
2011.
[68] F. J. Elmer, Nonlinear dynamics of dry friction, J. Phys. A: Math. Gen. 30,
6057–6063, 1997.
[69] J. Awrejcewicz, P. Olejnik, Analysis of dynamic systems with various fric-
tion laws, Applied Mechanics Reviews, 58, 389–411, 2005.
[70] A. F. Filippov, Differential Equations with Discontinuous Righthand Sides,
Mathematics and its Applications (Soviet Series), v. 18, Dordrecht: Kluwer,
1988.

[71] S. P. Yang, S. Q. Guo, Two-stop-two-slip motions of a dry friction oscillator,

Sci. China Tech. Sci. 53, 623–632, 2010.

✐
www.Ebook777.com ✐

✐ ✐
 ✐ ✐

✐ ✐

✐ ✐

✐ ✐
 Free ebooks ==> www.Ebook777.com

Accessing the E-book edition

Using the VitalSource® ebook DOWNLOAD AND READ OFFLINE
Access to the VitalBookTM ebook accompanying this book is To use your ebook offline, download BookShelf to your PC,
via VitalSource® Bookshelf – an ebook reader which allows Mac, iOS device, Android device or Kindle Fire, and log in to
you to make and share notes and highlights on your ebooks your Bookshelf account to access your ebook:
and search across all of the ebooks that you hold on your
On your PC/Mac
VitalSource Bookshelf. You can access the ebook online or
Go to http://bookshelf.vitalsource.com/ and follow the
offline on your smartphone, tablet or PC/Mac and your notes
instructions to download the free VitalSource Bookshelf
and highlights will automatically stay in sync no matter where
app to your PC or Mac and log into your Bookshelf account.
you make them.
On your iPhone/iPod Touch/iPad
1. Create a VitalSource Bookshelf account at Download the free VitalSource Bookshelf App available
https://online.vitalsource.com/user/new or log into via the iTunes App Store and log into your Bookshelf
your existing account if you already have one. account. You can find more information at https://support.
vitalsource.com/hc/en-us/categories/200134217-
2. Redeem the code provided in the panel below Bookshelf-for-iOS
to get online access to the ebook. Log in to
Bookshelf and click the Account menu at the top right On your Android™ smartphone or tablet
of the screen. Select Redeem and enter the redemption Download the free VitalSource Bookshelf App available
code shown on the scratch-off panel below in the Code via Google Play and log into your Bookshelf account. You can
To Redeem box. Press Redeem. Once the code has find more information at https://support.vitalsource.com/
been redeemed your ebook will download and appear in hc/en-us/categories/200139976-Bookshelf-for-Android-
your library. and-Kindle-Fire

On your Kindle Fire

Download the free VitalSource Bookshelf App available
from Amazon and log into your Bookshelf account. You can
find more information at https://support.vitalsource.com/
hc/en-us/categories/200139976-Bookshelf-for-Android-
and-Kindle-Fire

N.B. The code in the scratch-off panel can only be used once.
When you have created a Bookshelf account and redeemed
the code you will be able to access the ebook online or offline
on your smartphone, tablet or PC/Mac.

SUPPORT
If you have any questions about downloading Bookshelf,
creating your account, or accessing and using your ebook
edition, please visit http://support.vitalsource.com/

www.Ebook777.com
 i i

i i

i i

i i
 Free ebooks ==> www.Ebook777.com

Simulations of
Physics

Butikov
Simulations of Oscillatory Systems: with Award-Winning Software,

Oscillatory Systems
Physics of Oscillations provides a hands-on way of visualizing and un-
derstanding the fundamental concepts of the physics of oscillations. Both
the book and software are designed as exploration-oriented supplements for
courses in general physics and the theory of oscillations.

Simulations of Oscillatory Systems

The book is conveniently structured according to mathematical complexity. with Award-Winning Software,
Each chapter in Part I contains activities, questions, exercises, and problems
of varying levels of difficulty, from straightforward to quite challenging. Physics of Oscillations
Part II presents more sophisticated, highly mathematical material that
delves into the serious theoretical background for the computer-aided study
of oscillations.

The software package allows you to observe the motion of linear and
nonlinear mechanical oscillatory systems and to obtain plots of the
variables that describe the systems along with phase diagrams and plots
of energy transformations. These computer simulations provide clear, vivid
illustrations of oscillations in various physical systems, bringing to life many
abstract concepts, developing your physical intuition, and complementing
the analytical study of the subject.

Features
• Provides interactive software that serves as a desktop laboratory for
exploring simulated systems and replicating the experiments
• Enables you to perform interesting mini-research projects involving
the physics of oscillations
• Develops your physical intuition and theoretical foundations
• Includes various examples of the behavior of simulated systems
• Requires no previous knowledge of algorithmic languages or
programming

K24904

Eugene I. Butikov
6000 Broken Sound Parkway, NW
Suite 300, Boca Raton, FL 33487 ISBN: 978-1-4987-0768-8
711 Third Avenue 90000
New York, NY 10017
an informa business 2 Park Square, Milton Park
www.crcpress.com Abingdon, Oxon OX14 4RN, UK
9 781498 707688
w w w.crcpress.com
www.Ebook777.com

K24904 cvr mech.indd 1 1/19/15 11:32 AM