Chapter 3

Analyzing motion of systems of particles

In this chapter, we shall discuss
1. The concept of a particle
2. Position/velocity/acceleration relations for a particle
3. Newton’s laws of motion for a particle
4. How to use Newton’s laws to calculate the forces needed to make a particle move in a particular
way
5. How to use Newton’s laws to derive `equations of motion’ for a system of particles
6. How to solve equations of motion for particles by hand or using a computer.

The focus of this chapter is on setting up and solving equations of motion – we will not discuss in detail
the behavior of the various examples that are solved.

3.1 Equations of motion for a particle

We start with some basic definitions and physical laws.

3.1.1 Definition of a particle

A `Particle’ is a point mass at some position in space. It can move about, but has no characteristic
orientation or rotational inertia. It is characterized by its mass.

Examples of applications where you might choose to idealize part of a system as a particle include:
1. Calculating the orbit of a satellite – for this application, you don’t need to know the orientation of
the satellite, and you know that the satellite is very small compared with the dimensions of its
orbit.
2. A molecular dynamic simulation, where you wish to calculate the motion of individual atoms in a
material. Most of the mass of an atom is usually concentrated in a very small region (the nucleus)
in comparison to inter-atomic spacing. It has negligible rotational inertia. This approach is also
sometimes used to model entire molecules, but rotational inertia can be important in this case.

Obviously, if you choose to idealize an object as a particle, you will only be able to calculate its position.
Its orientation or rotation cannot be computed.

3.1.2 Position, velocity, acceleration relations for a particle (Cartesian coordinates)

In most practical applications we are interested in the position or the

velocity (or speed) of the particle as a function of time. But Newton’s k
laws will only tell us its acceleration. We therefore need equations
that relate the position, velocity and acceleration. P
Position vector: In most of the problems we solve in this course, we r
O j
will specify the position of a particle using the Cartesian components x z
of its position vector with respect to a convenient origin. This means
i
y
 1. We choose three, mutually perpendicular, fixed directions in space. The three directions are
described by unit vectors i, j, k
2. We choose a convenient point to use as origin.
3. The position vector (relative to the origin) is then specified by the three distances (x,y,z) shown in
the figure.
r  x(t )i  y(t ) j  z(t )k

In dynamics problems, all three components can be functions of time.

Velocity vector: By definition, the velocity is the derivative of the position vector with respect to time
(following the usual machinery of calculus)
r (t   t )  r(t )
v  lim
 t 0 t
Velocity is a vector, and can therefore be expressed in terms of its Cartesian components
v  vx i  v y j  vz k
You can visualize a velocity vector as follows
 The direction of the vector is parallel to the direction of motion
 The magnitude of the vector v  v  vx2  v 2y  vz2 is the speed of the particle (in meters/sec, for
example).

When both position and velocity vectors are expressed in terms Cartesian components, it is simple to
calculate the velocity from the position vector. For this case, the basis vectors i, j, k are constant
(independent of time) and so
d dx dy dz
vx i  v y j  v z k   x(t )i  y(t ) j  z (t )k   i  j  k
dt dt dt dt
This is really three equations – one for each velocity component, i.e.
dx dy dz
vx  vy  vz 
dt dt dt

Acceleration vector: The acceleration is the derivative of the velocity vector with respect to time; or,
equivalently, the second derivative of the position vector with respect to time.
v(t   t )  v(t )
a  lim
 t 0 t
The acceleration is a vector, with Cartesian representation a  ax i  a y j  az k .

Like velocity, acceleration has magnitude and direction. Sometimes it may be possible to visualize an
acceleration vector – for example, if you know your particle is moving in a straight line, the acceleration
vector must be parallel to the direction of motion; or if the particle moves around a circle at constant
speed, its acceleration is towards the center of the circle. But sometimes you can’t trust your intuition
regarding the magnitude and direction of acceleration, and it can be best to simply work through the math.

The relations between Cartesian components of position, velocity and acceleration are
 dvx d 2 x dv y d2y dvz d 2 z
ax   ay   az  
dt dt 2 dt dt 2 dt dt 2

3.1.3 Examples using position-velocity-acceleration relations

It is important for you to be comfortable with calculating velocity and acceleration from the position
vector of a particle. You will need to do this in nearly every problem we solve. In this section we
provide a few examples. Each example gives a set of formulas that will be useful in practical
applications.

Example 1: Constant acceleration along a straight line. There are many examples where an object
moves along a straight line, with constant acceleration. Examples include free fall near the surface of a
planet (without air resistance), the initial stages of the acceleration of a car, or and aircraft during takeoff
roll, or a spacecraft during blastoff.

Suppose that
The particle moves parallel to a unit vector i
The particle has constant acceleration, with magnitude a
At time t  t0 the particle has speed v0
At time t  t0 the particle has position vector r  x0i
The position, velocity acceleration vectors are then
 1 
r   x0  v0 (t  t0 )  a(t  t0 ) 2  i
 2 
v   v0  at  i
a  ai

Verify for yourself that the position, velocity and acceleration (i) have the correct values at t=0 and (ii)
are related by the correct expressions (i.e. differentiate the position and show that you get the correct
expression for the velocity, and differentiate the velocity to show that you get the correct expression for
the acceleration).

HEALTH WARNING: These results can only be used if the acceleration is constant. In many
problems acceleration is a function of time, or position – in this case these formulas cannot be used.
People who have taken high school physics classes have used these formulas to solve so many problems
that they automatically apply them to everything – this works for high school problems but not always in
real life!

Example 2: Simple Harmonic Motion: The vibration of a very simple spring-

mass system is an example of simple harmonic motion. x(t) j
i
In simple harmonic motion (i) the particle moves along a straight line; and (ii) m
the position, velocity and acceleration are all trigonometric functions of time. k,L0

For example, the position vector of the mass might be given by

 r  x(t )i   X 0  X sin(2 t / T )  i
Here X 0 is the average length of the spring, X 0  X is the maximum length of the spring, and T is the
time for the mass to complete one complete cycle of oscillation (this is called the `period’ of oscillation).

Harmonic vibrations are also often characterized by the frequency of vibration:

 The frequency in cycles per second (or Hertz) is related to the period by f=1/T
 The angular frequency is related to the period by   2 / T

The motion is plotted in the figure on the right.

The velocity and acceleration can be calculated by

differentiating the position, as follows
dx(t )  2X 
v i cos(2 t / T )  i
dt  T 
d 2 x(t )  4 2 X 
a i   sin(2 t / T )  i
dt 2  T2 
 

Note that:
 The velocity and acceleration are also harmonic, and
have the same period and frequency as the
displacement.
 If you know the frequency, and amplitude and of
either the displacement, velocity, or acceleration, you can immediately calculate the amplitudes
of the other two. For example, if X , V , A denote the amplitudes of the displacement,
velocity and acceleration, we have that
2
2  2  2
V  X A    X  V
T  T  T

Example 3: Motion at constant speed around a circular path

Circular motion is also very common – examples include any sin 
rotating machinery, vehicles traveling around a circular path, and
so on.
t cos
R
The simplest way to make an object move at constant speed along j n
 t Rsin 
a circular path is to attach it to the end of a shaft (see the figure),
and then rotate the shaft at a constant angular rate. Then, notice
that i
 The angle  increases at constant rate. We can write
  t , where  is the (constant) angular speed of the Rcos
shaft, in radians/seconds.
 The speed of the particle is related to  by V  R . To
see this, notice that the circumferential distance traveled by the particle is s  R . Therefore,
V  ds / dt  Rd / dt  R .

For this example the position vector is

r  R cos i  R sin  j
The velocity can be calculated by differentiating the position vector.
dr d d
v  R sin  i  R cos j  R ( sin  i  cos j)
dt dt dt
Here, we have used the chain rule of differentiation, and noted that d / dt   .

The acceleration vector follows as

dv d d
a  R ( cos i  sin  j)   R 2 (cos i  sin  j)
dt dt dt
Note that
(i) The magnitude of the velocity is V  R , and its direction is (obviously!) tangent to the path
(to see this, visualize (using trig) the direction of the unit vector t  ( sin  i  cos j)
(ii) The magnitude of the acceleration is R 2 and its direction is towards the center of the circle.
To see this, visualize (using trig) the direction of the unit vector n  (cos i  sin  j)

We can write these mathematically as

dr dv V2
v  R t  Vt a  R 2 n  n
dt dt R

Example 4: More general motion around a circular path sin 

We next look at more general circular motion, where the particle
t cos
still moves around a circular path, but does not move at constant R
speed. The angle  is now a general function of time.
j n
We can write down some useful scalar relations:  Rsin 
d
 Angular rate:   i
dt

 Angular acceleration  
d  d 2

Rcos
dt dt 2
d
 Speed V  R  R
dt
dV d 2 d
 Rate of change of speed R R  R
dt 2 dt
dt

We can now calculate vector velocities and accelerations

r  R cos i  R sin  j
The velocity can be calculated by differentiating the position vector.
dr d d
v  R sin  i  R cos j  R ( sin  i  cos j)
dt dt dt
The acceleration vector follows as
dv d d d
a R ( sin  i  cos  j)  R ( cos  i  sin  j)
dt dt dt dt
 R ( sin  i  cos  j)  R 2 (cos  i  sin  j)
It is often more convenient to re-write these in terms of the unit vectors n and t normal and tangent to the
circular path, noting that t  ( sin  i  cos j) , n  (cos i  sin  j) . Then
dV V2
v  R t  Vt a  R t  R 2 n  t n
dt R

These are the famous circular motion formulas that you might have seen in physics class.

Using MAPLE to differentiate position-velocity-acceleration relations

If you find that your calculus is a bit rusty you can use MAPLE to do the tedious work for you. You
already know how to differentiate and integrate in MAPLE – the only thing you may not know is how to
tell MAPLE that a variable is a function of time. Here’s how this works. To differentiate the vector
r  x(t )i  y(t ) j  z(t )k
you would type

It is essential to type in the (t) after x,y,and z – if you don’t do this, Mupad assumes that these variables
are constants, and takes their derivative to be zero. You must enter (t) after _any_ variable that changes
with time.

Here’s how you would do the circular motion calculation if you only know that the angle  is some
arbitrary function of time, but don’t know what the function is

As you’ve already seen in EN3, Matlab can make very long and complicated calculations fairly painless.
It is a godsend to engineers, who generally find that every real-world problem they need to solve is long
and complicated. But of course it’s important to know what the program is doing – so keep taking those
math classes…

3.1.4 Velocity and acceleration in normal-tangential and cylindrical polar coordinates.

In some cases it is helpful to use special basis vectors to write down velocity and acceleration vectors,
instead of a fixed {i,j,k} basis. If you see that this approach can be used to quickly solve a problem – go
ahead and use it. If not, just use Cartesian coordinates – this will always work, and with MAPLE is not
very hard. The only benefit of using the special coordinate systems is to save a couple of lines of rather
tedious trigonometric algebra – which can be extremely helpful when solving an exam question, but is
generally insignificant when solving a real problem.

Normal-tangential coordinates for particles moving along a prescribed planar path

In some problems, you might know the particle speed, and the x,y coordinates of the
path (a car traveling along a road is a good example). In this case it is often easiest to t
use normal-tangential coordinates to describe forces and motion. For this purpose we
 Introduce two unit vectors n and t, with t pointing tangent to the path and n R
pointing normal to the path, towards the center of curvature
 Introduce the radius of curvature of the path R. n
If you happen to know the parametric equation of the path (i.e. the x,y coordinates are
t
known in terms of some variable  ), then
 dx dy   dy dx 
 i j  i j
1 dr  d d   d d 
r  x ( )i  y ( ) j t   n
dr d  2 2 2 2
 dx   dy   dx   dy 
d          
 d   d   d   d 
The sign of n should be selected so that
  d2x d2 y 
 i j  n  0
 d2 d  2 

The radius of curvature can be computed from
dx d 2 y dy d 2 x

1 d d2 d d2

R  2 3/2
 dy  
2
 dx 
    
 d    d   
The radius of curvature is always positive.

The direction of the velocity vector of a particle is tangent to its path. The magnitude of the velocity
vector is equal to the speed.

The acceleration vector can be constructed by adding two components:

 the component of acceleration tangent to the particle’s path is equal dV / dt
 The component of acceleration perpendicular to the path (towards the center of curvature) is
equal to V 2 / R .
dV V2
Mathematically v  Vt a t n
dt R

Example: Design speed limit for a curvy road: As a consulting firm

specializing in highway design, we have been asked to develop a j
design formula that can be used to calculate the speed limit for cars
that travel along a curvy road.  V i
The following procedure will be used:
 The curvy road will be approximated as a sine wave
y  A sin(2 x / L) as shown in the figure – for a given road,
engineers will measure values of A and L that fit the path.
 Vehicles will be assumed to travel at constant speed V around L
the path – your mission is to calculate the value of V
 For safety, the magnitude of the acceleration of the car at any point along the path must be less
than 0.2g, where g is the gravitational acceleration. (Again, note that constant speed does not
mean constant acceleration, because the car’s direction is changing with time).

Our goal, then, is to calculate a formula for the magnitude of the acceleration in terms of V, A and L. The
result can be used to deduce a formula for the speed limit.

Calcluation:

We can solve this problem quickly using normal-tangential coordinates. Since the speed is constant, the
acceleration vector is
V2
a n
R
The position vector is r  xi  A sin(2 x / L) j , so we can calculate the radius of curvature from the
formula
 dx d 2 y dy d 2 x

1 d d2 d d2

R  2 3/2
 dy  
2
 dx 
    
 d    d   
Note that x acts as the parameter  for this problem, and y  A sin(2 x / L) , so
d2 y
2
1 dx A(2 / L) 2 sin(2 x / L)
 
R  2 3/2 1  (2 A / L) cos(2 x / L)3/2
  dy  
1    
  dx  
and the acceleration is
A(2 V / L)2 sin(2 x / L)
a n
1  (2 A / L) cos(2 x / L)
3/2

We are interested in the magnitude of the acceleration…

A(2 V / L) 2 sin(2 x / L)
a 
1  (2 A / L) cos(2 x / L)
3/2

We see from this that the car has the biggest acceleration when x  L / 2 . The maximum acceleration
follows as
amax  A(2 V / L)2
The formula for the speed limit is therefore V  ( L / 2 ) 0.2 g / A

Now we send in a bill for a big consulting fee…

Polar coordinates for particles moving in a plane

When solving problems involving central forces (forces that

attract particles towards a fixed point) it is often convenient to e
describe motion using polar coordinates. er
Polar coordinates are related to x,y coordinates through
r
r  x2  y 2   tan 1 ( y / x)
j

Suppose that the position of a particle is specified by its ‘polar
coordinates’ (r , ) relative to a fixed origin, as shown in the i
figure. Let e r be a unit vector pointing in the radial direction,
and let e be a unit vector pointing in the tangential direction, i.e
er  cos i  sin  j
e   sin  i  cos j
The velocity and acceleration of the particle can then be expressed as
 dr d
v er  r e
dt dt
 d 2r  d  
2  d 2 dr d 
a  r   er   r 2  e
 dt 2  dt    dt 2 dt dt 
 
Deriving these results takes some tedious algebra, but it’s conceptually simple – here’s what we do:
1. Write down the position vector in terms of (r , ) in a fixed (i,j,k) coordinate system
2. Take the time derivatives to find acceleration and velocity in the (i,j,k) coordinate system
3. Convert the results to the {er , e , e z } coordinate system. To do this remember that the
component of v parallel to e r can be found using a dot product: vr  v  er . Similarly v  v  e

Here are the details with Mupad taking care of the tedious algebra.

These are the answers stated.

Example The robotic manipulator shown in the

figure rotates with constant angular speed  L
about the k axis. Find a formula for the k e er
maximum allowable (constant) rate of extension
O
dL / dt if the acceleration of the gripper may not j j
exceed g.  L
i 
We can simply write down the acceleration
O i
vector, using polar coordinates. We identify
  d / dt and r=L, so that

 
2
 dL  2  dL  dL 1
a   L 2 er   2   e  a  L2 4  4     g 2   g 2 /  2  L2 2
 dt   dt  dt 4
Other examples using polar coordinates can be found in sections below.

3.1.5 Measuring position, velocity and acceleration

If you are designing a control system, you will need some way to detect the
motion of the system you are trying to control. A vast array of different
sensors is available for you to choose from: see for example the list at
http://www.sensorland.com/HowPage001.html . A very short list of
common sensors is given below
1. GPS – determines position on the earth’s surface by measuring the
time for electromagnetic waves to travel from satellites in known
positions in space to the sensor. Can be accurate down to cm
distances, but the sensor needs to be left in position for a long time for
this kind of accuracy. A few m is more common.
2. Optical or radio frequency position sensing – measure position by (a) monitoring deflection of laser
beams off a target; or measuring the time for signals to travel from a set
of radio emitters with known positions to the sensor. Precision can
vary from cm accuracy down to light wavelengths.
3. Capacitative displacement sensing – determine position by measuring
the capacitance between two parallel plates. The device needs to be
physically connected to the object you are tracking and a reference
point. Can only measure distances of mm or less, but precision can be
down to micron accuracy.
4. Electromagnetic displacement sensing – measures position by detecting
electromagnetic fields between conducting coils, or coil/magnet
combinations within the sensor. Needs to be physically connected to
the object you are tracking and a reference point. Measures displacements of order cm down to
microns.
5. Radar velocity sensing – measures velocity by detecting the change in frequency of electromagnetic
waves reflected off the traveling object.
6. Inertial accelerometers: measure accelerations by detecting the deflection of a spring acting on a
mass.

Accelerometers are also often used to construct an ‘inertial platform,’ which uses gyroscopes to maintain
a fixed orientation in space, and has three accelerometers that can detect motion in three mutually
perpendicular directions. These accelerations can then be integrated to determine the position. They are
used in aircraft, marine applications, and space vehicles where GPS cannot be used.

3.1.6 Newton’s laws of motion for a particle

Newton’s laws for a particle are very simple. Let

1. m denote the mass of the particle
2. F denote the resultant force acting on the particle (as a vector)
3. a denote the acceleration of the particle (again, as a vector). Then

F  ma
Occasionally, we use a particle idealization to model systems which, strictly speaking, are not particles.
These are:
1. A large mass, which moves without rotation (e.g. a car moving along a straight line)
2. A single particle which is attached to a rigid frame with negligible mass (e.g. a person on a
bicycle)

In these cases it may be necessary to consider the moments acting on the mass (or frame) in order to
calculate unknown reaction forces.
1. For a large mass which moves without rotation, the resultant moment of external forces about the
center of mass must vanish.
2. For a particle attached to a massless frame, the resultant moment of external forces acting on the
frame about the particle must vanish.
MC  0

It is very important to take moments about the correct point in dynamics problems! Forgetting this is
the most common reason to screw up a dynamics problem…

If you need to solve a problem where more than one particle is attached to a massless frame, you have to
draw a separate free body diagram for each particle, and for the frame. The particles must obey
Newton’s laws F  ma . The forces acting on the frame must obey F  0 and MC  0 , (because the
frame has no mass).

The Newtonian Inertial Frame.

Newton’s laws are very familiar, and it is easy to write them down without much thought. They do have
a flaw, however.

When we use Newton’s laws, we assume that we can identify a convenient origin somewhere that we
regard as `fixed’. In addition, to write down an acceleration vector, we need to be able to choose a set of
fixed directions in space.

For engineering calculations, this usually poses no difficulty. If we are solving problems involving
terrestrial motion over short distances compared with the earth’s radius, we simply take a point on the
earth’s surface as fixed, and take three directions relative to the earth’s surface to be fixed. If we are
solving problems involving motion in space near the earth, or modeling weather, we take the center of the
earth as a fixed point, (or for more complex calculations the center of the sun); and choose axes to have a
fixed direction relative to nearby stars.

But in reality, an unambiguous inertial frame does not exist. We can only describe the relative motion of
the mass in the universe, not its absolute motion. The general theory of relativity addresses this problem
– and in doing so explains many small but noticeable discrepancies between the predictions of Newton’s
laws and experiment.

It would be fun to cover the general theory of relativity in this course – but regrettably the mathematics
needed to solve any realistic problem is horrendous. As engineers, we always have to solve realistic
problems, and we usually can’t afford to spend a long time doing complicated calculations, so we use the
simplest theory that will allow us to make the correct design decisions. Newton’s laws are fine for us…
3.2 Calculating forces required to cause prescribed motion of a particle

Newton’s laws of motion can be used to calculate the forces required to make a particle move in a
particular way.

We use the following general procedure to solve problems like this

(1) Decide how to idealize the system (what are the particles?)
(2) Draw a free body diagram showing the forces acting on each particle
(3) Consider the kinematics of the problem. The goal is to calculate the acceleration of each
particle in the system – you may be able to start by writing down the position vector and
differentiating it, or you may be able to relate the accelerations of two particles (eg if two
particles move together, their accelerations must be equal).
(4) Write down F=ma for each particle.
(5) If you are solving a problem involving a massless frames (see, e.g. Example 3, involving a
bicycle with negligible mass) you also need to write down MC  0 about the particle.
(5) Solve the resulting equations for any unknown components of force or acceleration (this is
just like a statics problem, except the right hand side is not zero).

It is best to show how this is done by means of examples.

Example 1: Estimate the minimum thrust that must be produced

by the engines of an aircraft in order to take off from the deck of
an aircraft carrier (the figure is from
www.lakehurst.navy.mil/NLWeb/media-library.asp)

We will estimate the acceleration required to reach takeoff speed,

assuming the aircraft accelerates from zero speed to takeoff speed
along the deck of the carrier, and then use Newton’s laws to
deduce the force.

Data/ Assumptions:
1. The flight deck of a Nimitz class aircraft carrier is about 300m long (http://www.naval-
technology.com/projects/nimitz/) but only a fraction of this is used for takeoff (the angled runway
is used for landing). We will take the length of the runway to be d=200m
2. We will assume that the acceleration during takeoff roll is constant.
3. We will assume that the aircraft carrier is not moving (this is wrong – actually the aircraft carrier
always moves at high speed during takeoff. We neglect motion to make the calculation simpler)
4. The FA18 Super Hornet is a typical aircraft used on a carrier – it has max catapult weight of
m=15000kg http://www.boeing.com/defense-space/military/fa18ef/docs/EF_overview.pdf
5. The manufacturers are somewhat reticent about performance specifications for the Hornet but
vt  150 knots (77 m/s) is a reasonable guess for a minimum controllable airspeed for this
aircraft.

Calculations:
1. Idealization: We will idealize the aircraft as a particle. We can do this because the aircraft is not
rotating during takeoff.
 2. FBD: The figure shows a free body diagram. FT represents the (unknown) force exerted on the
aircraft due to its engines.

3. Kinematics: We must calculate the acceleration required to reach takeoff speed. We are given (i)
the distance to takeoff d, (ii) the takeoff speed vt and (iii) the aircraft is at rest at the start of the
takeoff roll. We can therefore write down the position vector r and velocity v of the aircraft at
takeoff, and use the straight line motion formulas for r and v to calculate the time t to reach
takeoff speed and the acceleration a. Taking the origin at the initial position of the aircraft, we
have that, at the instant of takeoff
1
r  di  at 2i v  vt i  ati
2
This gives two scalar equations which can be solved for a and t
1 v2 2d
d  at 2 vt  at  a t t
2 2d vt

4. EOM: The vector equation of motion for this problem is

v2
FT i  mai  m t i
2d

5. Solution: The i component of the equation of motion gives an equation for the unknown force in
terms of known quantities
v2
FT  m t
2d
Substituting numbers gives the magnitude of the force as
F=222 kN. This is very close, but slightly greater than, the
200kN (44000lb) thrust quoted on the spec sheet for the j FT
Hornet. Using a catapult to accelerate the aircraft, speeding up i
the aircraft carrier, and increasing thrust using an afterburner NA mg NB
buys a margin of safety.

Example 2: Mechanics of Magic! You have no doubt seen the

simple `tablecloth trick’ in which a tablecloth is whipped out Tablecloth
from underneath a fully set table (if not, you can watch it at
a
http://wm.kusa.gannett.edgestreams.net/news/1132187192333-
11-16-05-spangler-2p.wmv) Table

In this problem we shall estimate the critical acceleration that

must be imposed on the tablecloth to pull it from underneath the objects placed upon it.

We wish to determine conditions for the tablecloth to slip out from under the glass. We can do this by
calculating the reaction forces acting between the glass and the tablecloth, and see whether or not slip will
occur. It is best to calculate the forces required to make the glass move with the tablecloth (i.e. to
prevent slip), and see if these forces are big enough to cause slip.
 1. Idealization: We will assume that the glass behaves like a particle (again, we can do this because
the glass does not rotate)

2. FBD. The figure shows a free body diagram for the glass. The forces
include (i) the weight; and (ii) the normal and tangential components
of reaction at the contact between the tablecloth and the glass. The j
normal and tangential forces must act somewhere inside the contact mg
area, but their position is unknown. For a more detailed discussion of i
contact forces see Sects 2.4 and 2.5. T
N
3. Kinematics We are assuming that the glass has the same acceleration
as the tablecloth. The table cloth is moving in the i direction, and has magnitude a. The
acceleration vector is therefore a  ai .
4. EOM. Newton’s laws of motion yield
F  ma  Ti  ( N  mg ) j  mai
5. Solution: The i and j components of the vector equation must each be satisfied (just as when you
solve a statics problem), so that
T  ma N  mg  0  N  mg
Finally, we must use the friction law to decide whether or not the tablecloth will slip from under
the glass. Recall that, for no slip, the friction force must satisfy
T  N
where  is the friction coefficient. Substituting for T and N from (5) shows that for no slip
a  g
To do the trick, therefore, the acceleration must exceed  g . For a friction coefficient of order
0.1, this gives an acceleration of order 1m / s 2 . There is a special trick to pulling the tablecloth
with a large acceleration – but that’s a secret.

Example 3: Bicycle Safety. If a bike rider brakes too hard on the

front wheel, his or her bike will tip over (the figure is from
http://www.thosefunnypictures.com/picture/7658/bike-flip.html). In
this example we investigate the conditions that will lead the bike to
capsize, and identify design variables that can influence these
conditions.

If the bike tips over, the rear wheel leaves the ground. If this happens,
the reaction force acting on the wheel must be zero – so we can detect
the point where the bike is just on the verge of tipping over by calculating the reaction forces, and finding
the conditions where the reaction force on the rear wheel is zero.

1. Idealization:
a. We will idealize the rider as a particle
(apologies to bike racers – but that’s how
we think of you…). The particle is located
j
at the center of mass of the rider. The h
i
figure shows the most important design
parameters- these are the height of the

A d B

L
 rider’s COM, the wheelbase L and the distance of the COM from the rear wheel.
b. We assume that the bike is a massless frame. The wheels are also assumed to have no mass.
This means that the forces acting on the wheels must satisfy F  0 and M  0 - and can be
analyzed using methods of statics. If you’ve forgotten how to think about statics of wheels,
you should re-read the notes on this topic – in particular, make sure you understand the nature
of the forces acting on a freely rotating wheel (Section 2.4.6 of the reference notes).
c. We assume that the rider brakes so hard that the front wheel is prevented from rotating. It
must therefore skid over the ground. Friction will resist this sliding. We denote the friction
coefficient at the contact point B by  .
d. The rear wheel is assumed to rotate freely.
e. We neglect air resistance.

2. FBD. The figure shows a free body diagram for the

rider and for the bike together. Note that
a. A normal and tangential force acts at the
contact point on the front wheel (in general, j
both normal and tangential forces always i
act at contact points, unless the contact
happens to be frictionless). Because the
contact is slipping it is essential to draw the
friction force in the correct direction – the
force must resist the motion of the bike; NA TB NB
W
b. Only a normal force acts at the contact
point on the rear wheel because it is freely
rotating, and behaves like a 2-force member.

3. Kinematics The bike is moving in the i direction. As a vector, its acceleration is therefore a  ai ,
where a is unknown.

4. EOM: Because this problem includes a massless frame, we must use two equations of motion
( F  ma and MC  0 ). It is essential to take moments about the particle (i.e. the rider’s COM).

F  ma gives TB i  ( N A  N B  W ) j  mai

It’s very simple to do the moment calculation by hand, but for those of you who find such
calculations unbearable here’s a Mupad script to do it. The script simply writes out the position
vectors of points A and B relative to the center of mass as 3D vectors, writes down the reactions at A
and B as 3D vectors, and calculates the resultant moment (we don’t bother including the weight,
because it acts at the origin and so exerts zero moment)

The two nonzero components of F  ma and the one nonzero component of MC  0 give us three
scalar equations
 TB  ma
(N A  NB W )  0
N B ( L  d )  N A d  TB h  0
We have four unknowns – the reaction components N A , N B , TB and the acceleration a so we need
another equation. The missing equation is the friction law
TB   N B

5. Solution: Here’s the solution with Mupad. It’s easy to get the same answer by hand as well.

We are interested in finding what makes the reaction force at A go to zero (that’s when the bike is about
to tip). So
h  (L  d )
NA  W  0    (L  d ) / h
h  L

This tells us that the bike will tip if the friction

coefficient exceeds a critical magnitude, which
depends on the geometry of the bike. The simplest
way to design a tip-resistant bike is to make the
height of the center of mass h small, and the distance
(L-d) between the front wheel and the COM as large
as possible.

A `recumbent’ bike is one way to achieve this – the

figure (from
http://en.wikipedia.org/wiki/Recumbent_bicycle)
shows an example. The recumbent design offers
many other significant advantages over the classic
bicycle besides tipping resistance.
Example 4: A stupid problem that you might find in the FE professional
engineering exam. The purpose of this problem is to show what you need to
do to solve problems involving more than one particle. j
i
Two weights of mass m A and mB are connected by a cable passing over two
freely rotating pulleys as shown. They are released, and the system begins to mA mB
move. Find an expression for the tension in the cable connecting the two
weights.

1. Idealization – The masses will be idealized as particles; the cable is inextensible and the mass of
the pulleys is neglected. This means the internal forces in the cable, and the forces acting
between cables/pulleys must satisfy F  0 and M  0 , and we can T
treat them as though they were in static equilibrium. T j
i
2. FBD – we have to draw a separate FBD for each particle. Since the
pulleys and cable are massless, the tension T in the cable is constant. mA mB
3. Kinematics We know that both masses must move in the j direction.
We also know that the masses always move at the same speed but in mAg mBg
opposite directions. Therefore, their accelerations must be equal and
opposite. We can express this mathematically as
a A j  aB j

4. EOM: We must write down two equations of motion, as there are two masses
(T  mA g ) j  mAa A j
(T  mB g ) j  mB aB j
We now have three equations for three unknowns (the unknowns are a A , aB and T).

5. Solution: As paid up members of ALE (the Academy of Lazy Engineers) we use Mupad to solve
the equations

So the tension in the cable is

2mAmB
T g
mA  mB
We pass!
Example 5: Another stupid FE exam problem: The
figure shows a small block on a rotating bar. The contact
between the block and the bar has friction coefficient  .
The bar rotates at constant angular speed  . Find the r
critical angular velocity that will just make the block start j
to slip when   0 . Which way does the block slide?
 i
The general approach to this problem is the same as for the 
Magic trick example – we will calculate the reaction force
exerted by the bar on the block, and see when the forces are large enough to cause slip at the contact. We
analyze the motion assuming the slip does not occur, and then find out the conditions where this can no
longer be the case.

1. Idealization – We will idealize the block as a particle. This is dangerous, because the block is
clearly rotating. We hope that because it rotates at constant rate, the rotation will not have a
significant effect – but we can only check this once we know how to
deal with rotational motion.
2. FBD: The figure shows a free body diagram for the block. The block is j T
subjected to a vertical gravitational force, and reaction forces at the
i 
contact with the bar. Since we have assumed that the contact is not N
slipping, we can choose the direction of the tangential component of the mg
reaction force arbitrarily. The resultant force on the block is
F  T cos  N sin   i  ( N cos  T sin   mg ) j
3. Kinematics We can use the circular motion formula to write down the acceleration of tbe block
(see section 3.1.3)
a  r 2 (cos i  sin  j)
4. EOM: The equation of motion is
T cos  N sin   i  ( N cos  mg ) j  mr 2 (cos i  sin  j)
5. Solution: The i and j components of the equation of motion can be solved for N and T – Mupad
makes this painless

To find the point where the block just starts to slip, we use the friction law. Recall that, at the
point of slip
T  N
For the block to slip with   0
r 2   g

so the critical angular velocity is    g / r . Since the tangential traction T is negative, and the
friction force must oppose sliding, the block must slide outwards, i.e. r is increasing during slip.
Alternative method of solution using normal-tangential
coordinates t
n
We will solve this problem again, but this time we’ll use r
the short-cuts described in Section 3.1.4 to write down the j
acceleration vector, and we’ll write down the vectors in
Newton’s laws of motion in terms of the unit vectors n and  i
t normal and tangent to the object’s path. 
(i) Acceleration vector If the block does not slip, it moves
with speed V  r around a circular arc with radius r. Its acceleration vector has magnitude V 2 / r and
direction parallel to the unit vector n. t
(ii) The force vector can be resolved into components parallel to n and t. Simple T
trig on the free body diagram shows that  N
n
F   N  mg cos  t   mg sin   T  n mg
(iii) Newton’s laws then give
F  ma   N  mg cos  t   mg sin   T  n  m 2rn
The components of this vector equation parallel to t and n yield two equations, with solution
N  mg cos T  mg sin   m 2r
This is the same solution as before. The short-cut makes the calculation slightly more straightforward.
This is the main purpose of using normal-tangential components.

Example 6: Window blinds. Have you ever wondered how

window shades work? You give the shade a little downward
jerk, let it go, and it winds itself up. If you pull the shade down
slowly, it stays down.

The figure shows the mechanism (which probably only costs a Stationary
few cents to manufacture) that achieves this remarkable feat of Hub
engineering. It’s called an `inertial latch’ – the same principle is Rotating
used in the inertia reels on the seatbelts in your car. Drum Window
shade
The picture shows an enlarged end view of the window shade.
The hub, shown in brown, is fixed to the bracket supporting the
shade and cannot rotate. The drum, shown in peach, rotates as
the shade is pulled up or down. The drum is attached to a
torsional spring, which tends to cause the drum to rotate
counterclockwise, so winding up the shade. The rotation is
prevented by the small dogs, shown in red, which engage with
the teeth on the hub. You can pull the shade downwards freely, Stationary
since the dogs allow the drum to rotate counterclockwise. Hub

To raise the shade, you need to give the end of the shade a jerk
downwards, and then release it. When the drum rotates
sufficiently quickly (we will calculate how quickly shortly) the
dogs open up, as shown on the right. They remain open until the drum slows down, at which point the
topmost dog drops and engages with the teeth on the hub, thereby locking up the shade once more.

We will estimate the critical rotation rate required to free the rotating drum.

1. Idealization – We will idealize the topmost dog as a particle on the

end of a massless, inextensible rod, as shown in the figure.
a. We will assume that the drum rotates at constant angular
rate  . Our goal is to calculate the critical speed where the
dog is just on the point of dropping down to engage with R 
the hub.
b. When the drum spins fast, the particle is contacts the outer
rim of the drum – a normal force acts at the contact. When
the dog is on the point of dropping this contact force goes to
zero. So our goal is to calculate the contact force, and then
to find the critical rotation rate where the force will drop to
zero.
c. We neglect friction.

2. FBD. The figure shows a free body diagram for the particle. The
particle is subjected to: (i) a reaction force N where it contacts the rim; N
(ii) a tension T in the link, and (iii) gravity. The resultant force is T
F   T cos(   )  N cos  i  ( N sin   T sin(   )  mg ) j 
j
3. Kinematics We can use the circular motion formula to write down the mg
acceleration of the particle(see section 3.1.3) 
a   R 2 (cos i  sin  j) i
4. EOM: The equation of motion is
 T cos(   )  N cos  i  ( N sin   T sin(   )  mg ) j  a  R 2 (cos i  sin  j)
5. Solution: The i and j components of the equation of motion can be solved for N and T – Mupad
makes this painless

normal reaction force is therefore

N  mg cos(   ) / sin   mR 2

We are looking for the point where this can first become zero or negative. Note that max{cos(   )}  1
at the point where    =0. The smallest value of N therefore occurs at this point, and has magnitude
Nmin  mg / sin   mR 2
The critical speed where N=0 follows as
  g / ( R sin  )
Changing the angle  and the radius R gives a convenient way to control the critical speed in designing
an inertial latch.

Alternative solution using polar coordinates

We’ll work through the same problem again, but this time handle the vectors using polar
coordinates.
e er
1. FBD. The figure shows a free body diagram for the particle.
The particle is subjected to: (i) a reaction force N where it N
contacts the rim; (ii) a tension T in the link, and (iii) gravity. T
The resultant force is 
F  ( N  T cos   mg sin  )er  (T sin   mg cos  )e
j
mg
2. Kinematics The acceleration vector is now 
a   R 2 er i
3. EOM: The equation of motion is
( N  T cos   mg sin  )er  (T sin   mg cos  )e   R 2 er
4. Solution: The er , e components of the equation of motion can be solved for N and T – again, we
can use Mupad for this

The normal reaction force is therefore

N  mg cos(   ) / sin   mR 2

We are looking for the point where this can first become zero or negative. Note that max{cos(   )}  1
at the point where    =0. The smallest value of N therefore occurs at this point, and has magnitude
Nmin  mg / sin   mR 2
The critical speed where N=0 follows as
  g / ( R sin  )
Changing the angle  and the radius R gives a convenient way to control the critical speed in designing
an inertial latch.
Example 7: Aircraft Dynamics Aircraft performing certain
instrument approach procedures (such as holding patterns or 
procedure turns) are required to make all turns at a standard k
rate, so that a complete 360 degree turn takes 2 minutes. All j
turns must be made at constant altitude and constant speed, V.  R
People who design instrument approach procedures need to i
know the radius of the resulting turn, to make sure the aircraft
won’t hit anything. Engineers designing the aircraft are interested in the forces needed to complete the
turn – specifically, the load factor, which is the ratio of the lift force on the aircraft to its weight.

In this problem we will calculate the radius of the turn R and the bank angle required, as well as the load
factor caused by the maneuver, as a function of the aircraft speed V.

Before starting the calculation, it is helpful to understand what makes an aircraft travel in a circular path.
Recall that
1. If an object travels at constant speed around a circle, its acceleration vector has constant
magnitude, and has direction towards the center of the circle
2. A force must act on the aircraft to produce this acceleration – i.e. the resultant force on the
aircraft must act towards the center of the circle. The necessary force comes from the horizontal
component of the lift force – the pilot banks the wings, so that the lift acts at an angle to the
vertical.

With this insight, we expect to be able to use the equations of motion to calculate the forces.

1. Idealization – The aircraft is idealized as a particle – it’s not obvious that this is accurate, because the
aircraft clearly rotates as it travels around the curve. However, the forces we wish to calculate turn
out to be fully determined by F=ma and are not influenced by the rotational motion.

2. FBD. The figure shows a free body diagram for the aircraft. It is subjected to (i) a gravitational force
(mg); (ii) a thrust from the engines FT , (iii) a drag force FD , acting perpendicular to the direction of
motion, and (iv) a lift force FL , acting perpendicular to the plane of the wings.

The resultant force is

F 
 T  FD )cos  FL sin  sin  i  (FD  FT )sin  FL sin  cos  j   FL cos  mg  k L
( F
k FT
j
(you may find the components of the lift force difficult to  R
visualize – to see where these come from, note that the lift
force can be projected onto components along OR and the
O i mg FD
k direction as FL  FL sin  RO  FL cos k . Then note
that RO   sin  i  cos j .)

3. Kinematics
a. The aircraft moves at constant speed around a circle, so the angle   t , where  is the
(constant) angular speed of the line OP. Since the aircraft completes a turn in two minutes,
we know that   2 / (2  60)   / 60 rad/sec
b. The position vector of the plane is
r  R sin ti  R cos tj
 We can differentiate this expression with respect to time to find the velocity
v  R(cos ti  sin t ) j
c. The magnitude of the velocity is V  R , so if the aircraft flies at speed V, the radius of the
turn must be R  V / 
d. Differentiating the velocity gives the acceleration
a   R 2 (sin ti  cos t ) j

4. EOM: The equation of motion is

( FT  FD )cos  FL sin  sin   i  ( FD  FT )sin   FL sin  cos  j   FL cos  mg  k
 mR 2 (sin  i  cos ) j
 mV  (sin  i  cos ) j

5. Solution: The i j and k components of the equation of motion give three equations that can be solved
for FT , FL and  . We assume that the drag force is known, since this is a function of the aircraft’s
speed.

This solution is correct, but you need a PhD to understand what it means (other symbolic manipulation
programs like Maple and Mathematica give a comprehensible solution). Fortunately, I happen to have a
PhD… The solution can be simplified to
  tan 1(V  / g ) FL  mg 1  V 2 2 / g 2 FT  FD
As we will see below, if you choose to solve this problem in normal-tangential coordinates, you don’t
need a PhD to

We can calculate values of  , R  V /  and the load factor FL / mg for a few aircraft
a. Cessna 150 – V=70knots (36 m/s) :   110 R=690m, FL / mg  1.02
 b. Boeing 747: V=200 knots (102 m/s)   280 R=1950m, FL / mg  1.14
c. F111 V=300 knots (154 m/s)   390 R=2950m, FL / mg  1.3

Alternative solution using normal-tangential coordinates  k

k
This problem can also be solved rather more quickly using j
normal and tangential basis vectors.  R n
t
(i) Acceleration vector. The aircraft travels around a circular i
path at constant speed, so its acceleration is
V2
a n  V n
R FL 
where n is a unit vector pointing towards the center of the circle. k FT
j
(ii) Force vector. The force vector can be written in terms of the  R
unit vectors n,t,k as
F  ( FT  FD )t  FL sin  n  ( FL cos   mg )k O i mg FD

(iii) Newton’s law F  ( FT  FD )t  FL sin  n  ( FL cos  mg )k  mV n

The n, t and k components of this equation give three equations that can be solved for FT , FL and  .
This time it is easy to solve the equations by hand…
  tan 1(V  / g ) FL  mg 1  V 2 2 / g 2 FT  FD

This example again shows why normal-tangential coordinates are useful – describing forces, and solving
the resulting force-acceleration relations are much simpler than working with a fixed coordinate system.

3.3 Deriving and solving equations of motion for systems of particles

We next turn to the more difficult problem of predicting the motion of a system that is subjected to a set
of forces.

3.3.1 General procedure for deriving and solving equations of motion for systems of particles

It is very straightforward to analyze the motion of systems of particles. You should always use the
following procedure
1. Introduce a set of variables that can describe the motion of the system. Don’t worry if this sounds
vague – it will be clear what this means when we solve specific examples.
2. Write down the position vector of each particle in the system in terms of these variables
3. Differentiate the position vector(s), to calculate the velocity and acceleration of each particle in
terms of your variables;
4. Draw a free body diagram showing the forces acting on each particle. You may need to introduce
variables to describe reaction forces. Write down the resultant force vector.
5. Write down Newton’s law F  ma for each particle. This will generate up to 3 equations of
motion (one for each vector component) for each particle.
 6. If you wish, you can eliminate any unknown reaction forces from Newton’s laws. If you are
trying to solve the equations by hand, you should always do this; of you are using MATLAB, it’s
not usually necessary – you can have MATLAB calculate the reactions for you. The result will be
a set of differential equations for the variables defined in step (1)
7. If you find you have fewer equations than unknown variables, you should look for any
constraints that restrict the motion of the particles. The constraints must be expressed in terms of
the unknown accelerations.
8. Identify the initial conditions for the variables defined in (1). These are usually the values of the
unknown variables, their time derivatives, at time t=0. If you happen to know the values of the
variables at some other instant in time, you can use that too. If you don’t know their values at
all, you should just introduce new (unknown) variables to denote the initial conditions.
9. Solve the differential equations, subject to the initial conditions.

Steps (3) (6) and (8) can usually be done on the computer, so you don’t actually have to do much calculus
or math.

Sometimes, you can avoid solving the equations of motion completely, by using conservation laws –
conservation of energy, or conservation of momentum – to calculate quantities of interest. These short-
cuts will be discussed in the next chapter.

3.2.2 Simple examples of equations of motion and their solutions

The general process described in the preceding section can be illustrated using simple examples. In this
section, we derive equations of motion for a number of simple systems, and find their solutions.

The purpose of these examples is to illustrate the straightforward, step-by-step procedure for analyzing
motion in a system. Although we solve several problems of practical interest, we will simply set up and
solve the equations of motion with some arbitrary values for system parameter, and won’t attempt to
explore their behavior in detail. More detailed discussions of the behavior of dynamical systems will
follow in later chapters.

Example 1: Trajectory of a particle near the earth’s surface (no air resistance)

At time t=0, a projectile with mass m is launched from a V0

position X0  Xi  Yj  Zk with initial velocity vector
V0  Vx i  Vy j  Vz k . Calculate its trajectory as a function k
X0
of time. j

i
1. Introduce variables to describe the motion: We can
simply use the Cartesian coordinates of the particle ( x(t ), y(t ), z (t ))

2. Write down the position vector in terms of these variables: r  x(t )i  y(t ) j  z(t )k

3. Differentiate the position vector with respect to time to find the acceleration. For this example, this is
trivial
 dx dy dz d 2x d 2 y d 2z
v i  j k a i  j  k
dt dt dt dt 2 dt 2 dt 2

4. Draw a free body diagram. The only force acting on the particle is gravity – the k
free body diagram is shown in the figure. The force vector follows as F  mgk . j
mg
i
5. Write down Newton’s laws of motion. This is easy
 d 2x d 2 y d 2z 
F  ma  mgk  m  2 i  2 j  2 k 
 dt dt dt 
The vector equation actually represents three separate differential equations of motion
d 2x d2y d 2z
0 0  g
dt 2 dt 2 dt 2

6. Eliminate reactions – this is not needed in this example.

7. Identify initial conditions. The initial conditions were given in this problem – we have that
 dx   dy   dz 
x  X 0  Vx   y  Y0  Vy   z  Z 0  Vz 
 dt   dt   dt 

8. Solve the equations of motion. In general we will use MAPLE or matlab to do the rather tedious
algebra necessary to solve the equations of motion. Here, however, we will integrate the equations by
hand, just to show that there is no magic in MAPLE.

The equations of motion are

d 2x d2y d 2z
0 0  g
dt 2 dt 2 dt 2
It is a bit easier to see how to solve these if we define
dx dy dz
 vx  vy  vz
dt dt dt
The equation of motion can be re-written in terms of (vx , v y , vz ) as
dvx dv y dvz
0 0  g
dt dt dt
We can separate variables and integrate, using the initial conditions as limits of integration
vx t vy t vz t

 dv   0dt
Vx
x
0
 dv   0dt
Vy
x
0
 dv    gdt
Vz
z
0

vx  Vx v y  Vy vz  Vz  gt
Now we can re-write the velocity components in terms of (x,y,z) as
dx dy dz
 Vx  Vy  Vz  gt
dt dy dt
Again, we can separate variables and integrate
 x t y t z t


X0
dx   Vx dt
0
 dy   Vy dt
Y0 0
 dz   Vz  gt  dt
Z0 0

1
x  X 0  Vxt y  Y0  Vy t z  Z 0  Vz t  gt 2
2
so the position and velocity vectors are
 
  1
r   X  Vxt  i  Y  V y t j   Z  Vz t  gt 2  k
 2 
v  Vx i  V y j  Vz  gt  k

Here’s how to integrate the equations of motion using Mupad

Applications of trajectory problems: It is traditional in elementary physics and dynamics courses to

solve vast numbers of problems involving particle trajectories. These invariably involve being given
some information about the trajectory, which you must then use to work out something else. These
problems are all somewhat tedious, but we will show a couple of examples to uphold the fine traditions of
a 19th century education.

Estimate how far you could throw a stone from the top
of the Kremlin palace.
v0
Note that 
1. The horizontal and vertical components of
velocity at time t=0 follow as
Vx  v0 cos Vy  0 Vz  v0 sin  H
2. The components of the position of the particle
at time t=0 are X  0,Y  0, Z  H k
3. The trajectory of the particle follows as
 1 
r   v0 cos t  i   H  v0 sin  t  gt 2  k i D
 2 
4. When the particle hits the ground, its position vector is r  Di . This must be on the trajectory, so

 v0 cos tI  i   H  v0 sin  tI  gtI2  k  Di

1
 2 
where t I is the time of impact.
5. The two components of this vector equation gives us two equations for the two unknowns
{t I , D} , which can be solved
The RootOf in MAPLE is scary – any time that MAPLE gives you a scary result, look in the help and see
if there is a `convert’ function that might make it less scary.

For a rough estimate of the distance we can use the following numbers
1. Height of Kremlin palace – 71m
2. Throwing velocity – maybe 25mph? (pretty pathetic, I know - you can probably do better,
especially if you are on the baseball team).
3. Throwing angle – 45 degrees.
Substituting numbers gives 36m (118ft).
If you want to go wild, you can maximize D with respect to  , but this won’t improve your estimate
much.

Silicon nanoparticles with radius 20nm are in thermal motion near a flat surface. At the surface,
they have an average velocity 2kT / m , where m is their mass, T is the
temperature and k=1.3806503 × 10-23 is the Boltzmann constant. Estimate the
maximum height above the surface that a typical particle can reach during its
thermal motion, assuming that the only force acting on the particles is gravity

1. The particle will reach its maximum height if it happens to be traveling

vertically, and does not collide with any other particles.
2. At time t=0 such a particle has position X  0, Y  0, Z  0 and velocity
Vx  0 Vy  0 Vz  2kT / m
3. For time t>0 the position vector of the particle follows as


 1
r   2kT / m t  gt 2  k
2


Its velocity is
v  
2kT / m  gt k
4. When the particle reaches its maximum height, its velocity must be equal to zero (if you don’t see
this by visualizing the motion of the particle, you can use the mathematical statement that if
r  yk is a maximum, then dr / dt  v  (dy / dt )k  0 ). Therefore, at the instant of maximum
height v   
2kT / m  gtmax k  0


5. This shows that the instant of max height occurs at time tmax  2kT / m / g 
6. Substituting this time back into the position vector shows that the position vector at max height is
  2kT 1 2kT  kT
r   k  k
 mg 2 mg  mg
7. Si has a density of about 2330 kg/m^3. At room temperature (293K) we find that the distance is
surprisingly large: 10mm or so. Gravity is a very weak force at the nano-scale – surface forces
acting between the particles, and the particles and the surface, are much larger.

Example 2: Free vibration of a suspension system.

A vehicle suspension can be idealized as a mass m supported by a spring. The j

m b
spring has stiffness k and un-stretched length L0 . To test the suspension, the i
vehicle is constrained to move vertically, as shown in the figure. It is set in
motion by stretching the spring to a length L and then releasing it (from rest). k,L0 x(t)
Find an expression for the motion of the vehicle after it is released. O
As an aside, it is worth noting that a particle idealization is usually too crude to
model a vehicle – a rigid body approximation is much better. In this case, however, we assume that the
vehicle does not rotate. Under these conditions the equations of motion for a rigid body reduce to
F  ma and M  0 , and we shall find that we can analyze the system as if it were a particle.

1. Introduce variables to describe the motion: The length of the spring x(t ) is a convenient way to
describe motion.

2. Write down the position vector in terms of these variables: We can take the origin at O as shown in
the figure. The position vector of the center of mass of the block is then
 b
r   x(t )   j
 2

3. Differentiate the position vector with respect to time to find the acceleration. For this example, this is
trivial
dx d 2x
v j a j
dt dt 2

4. Draw a free body diagram. The free body diagram is shown in the figure: RAx mg
the mass is subjected to the following forces
j
 Gravity, acting at the center of mass of the vehicle i
 The force due to the spring RBx
 Reaction forces at each of the rollers that force the vehicle to move Fs
vertically.
Recall the spring force law, which says that the forces exerted by a spring act parallel to its length, tend to
shorten the spring, and are proportional to the difference between the length of the spring and its un-
stretched length.
5. Write down Newton’s laws of motion. This is easy
d 2x
F  ma  ( RAx  RBx )i   mg  k ( x  L0 )  j  m j
dt 2
The i and j components give two scalar equations of motion
( RAx  RBx )  0
d 2x
 k 
   g  ( x  L0 ) 
dt 2  m 
6. Eliminate reactions – this is not needed in this example.

7. Identify initial conditions. The initial conditions were given in this problem – at time t=0, we know
that x  L and dx / dt  0

8. Solve the equations of motion. Again, we will first integrate the equations of motion by hand, and then
repeat the calculation with MAPLE. The equation of motion is
d 2x  k 
   g  ( x  L0 ) 
dt 2  m 
We can re-write this in terms of
dx
 vx
dt
This gives
dvx dvx dx dv  k 
  vx x    g  ( x  L0 ) 
dt dx dt dx  m 
We can separate variables and integrate
vx x
 k 
 vx dvx     g  m ( x  L0 ) dx
0 L
1
 vx2   g ( x  L) 
2
k
2m
k

x 2  L2  L0 ( x  L)
m
k  mg  
2 2
mg  
 vx   L  L0     x  L0   
m  k   k  
 
Don’t worry if the last line looks mysterious – writing the solution in this form just makes the algebra a
bit simpler. We can now integrate the velocity to find x
k  mg  
2 2
dx mg  
vx    L  L0    x  L0   
dt m  k   k  
 
x t
dx
   dt
k 
2 2
L mg   mg  0
 L  L0     x  L0   
m  k   k  

The integral on the left can be evaluated using the substitution
  x  L0  mg / k 
 cos
 L  L0  mg / k 
so that
0
 sin  d
t
 x  L0  mg / k 
   dt 0  cos 1
0
k
1  cos 2   0
 L  L0  mg / k 
m 
k
  0  t
m
 x  L0  mg / k   k 
  cos  t 
 L  L0  mg / k   m 
 k 
 x   L  L0  mg / k  cos  t   L0  mg / k
 m 
Here’s the MAPLE solution

Note that it’s important to include the assume() statements, otherwise Mupad gives the solution in the
form of an exponential of a complex number. The solution in this other form is also correct, but is
difficult to visualize.

The solution is plotted in the figure. The behavior of vibrating

systems will be discussed in more detail later in this course, but it
is worth noting some features of the solution:
1. The average position of the mass is x  L0  mg / k .
Here, mg/k is the static deflection of the spring i.e. the
deflection of the spring due to the weight of the vehicle
(without motion). This means that the car vibrates
symmetrically about its static deflection.
2. The amplitude of vibration is L  L0  mg / k . This
corresponds to the distance of the mass above its average
position at time t=0.
3. The period of oscillation (the time taken for one
complete cycle of vibration) is T  2 m / k
1 k
4. The frequency of oscillation (the number of cycles per second) is f  (note f=1/T).
2 m
Frequency is also sometimes quoted as angular frequency, which is related to f by
  2 f  k / m . Angular frequency is in radians per second.
An interesting feature of these results is that the static deflection is related to the frequency of oscillation -
so if you measure the static deflection   mg / k , you can calculate the (angular) vibration frequency as
  g /

Example 3: Silly FE exam problem

This example shows how polar coordinates can be used to

analyze motion.
L e er
The rod shown in the picture rotates at constant angular speed
in the horizontal plane. The interface between block and rod
has friction coefficient  . The rod pushes a block of mass m, r j
which starts at r=0 with radial speed V. Find an expression
for r(t).   i

1. Introduce variables to describe the motion – the polar coordinates r ,  work for this problem

2. Write down the position vector and differentiate to find acceleration – we don’t need to do
this – we can just write down the standard result for polar coordinates
 d 2r  dr
a  r 2  er  2 e
 dt 2  dt
 
T
e
3. Draw a free body diagram – shown in the figure – note that it is N
important to draw the friction force in the correct direction. The block er
will slide radially outwards, and friction opposes the slip.

4. Write down Newton’s law


 d r
2  dr 

Ter  Ne  m   r 2  er  2 e 
 2  dt

 dt  


5. Eliminate reactions

F=ma gives two equations for N and T. A third one comes from the friction law T   N

The third solution can be rearranged into an equation of motion for r

 d 2r dr
 2   r 2  0
2 dt
dt

6. Identify initial conditions Here, r=0 dr/dt=V at time t=0.

7. Solve the equation: If you’ve taken AM33 you will know how to solve this equation… But if
not, or you are lazy, you can use MAPLE to solve it for you.

This can be simplified slightly by hand:

e t sinh( 1   2 t )
V
r (t ) 
2 1   2

Example 4: Motion of a pendulum (R-rated version)

j
i
A pendulum is a ubiquitous engineering system. You are, of course, familiar with
how a pendulum can be used to measure time. But it’s used for a variety of other
scientific applications. For example, Professor Crisco’s lab uses pendulum to
l
measure properties of human joints, see
Crisco Joseph J; Fujita Lindsey; Spenciner David B, ‘The dynamic flexion/extension 
properties of the lumbar spine in vitro using a novel pendulum system.’ Journal of
biomechanics 2007;40(12):2767-73

In this example, we will work through the basic problem of deriving and solving the equations of motion
for a pendulum, neglecting air resistance.

1. Introduce variables to describe motion: The angle  (t ) shown in the figure is a convenient
variable.

2. Write down the position vector as a function of the variables We introduce a Cartesian coordinate
system with origin at O, as shown in the picture.

Elementary geometry shows that r  l sin  i  l cos j

Note that we have assumed that the cable remains straight – this will be true as long as the internal force
in the cable is tensile. If calculations predict that the internal force is compressive, this assumption is
wrong. But there is no way to check the assumption at this point so we simply proceed, and check the
answer at the end

3. Differentiate the position vector to find the acceleration: The computer makes this painless.

4. The free body diagram is shown in the figure. The force exerted by the cable on
R
the particle is introduced as an unknown reaction force. The force vector is
F   R sin  i   R cos  mg  j 
j
5. Newton’s laws of motion can be expressed as i
mg
F  ma
  d 
2
d 2    d 
2
d 2 
 R sin  i   R cos  mg  j  m  l sin     l cos 2  i  m l cos    l sin  2  j
  dt  dt    dt  dt 

Equating the i and j components gives two equations for the two unknowns
  d 
2
d 2 
 R sin   m  l sin     l cos  
  dt  dt 2 
  d 
2
d 2 
R cos  mg  m l cos    l sin  
  dt  dt 2 

6. Eliminate the reaction forces. – In this problem, it is helpful to eliminate the unknown reaction force
R. You can do this on the computer if you like, but in this case it is simpler to do this by hand. You can
simply multiply the first equation by cos and the second equation by sin  and then add them. This
yields


mg sin   m sin 2   cos 2 
d 2
dt 2

d 2 g
  sin   0
2 l
dt

7. Identify initial conditions. Some calculations are necessary to determine the initial conditions in this
problem. We are given that   0 at time t=0, and the horizontal velocity is V0 at time t=0, but to solve
the equation of motion, we need the value of d / dt  0 . We can find the relationship we need by
differentiating the position vector to find the velocity
d
v  l cos i  l sin  j
dt
Setting v  V0i and   0 at t=0 shows that
d
V0i  li
dt
so d / dt  V0 / l

8. Solve the equations of motion This equation of motion is too difficult for MAPLE but actually the
solution does exist and is very well known – this is a classic problem in mathematical physics. With
initial conditions   0, d / dt  V0 / l t  0 the solution is
  V  g V0   V0
 2sin 1  0 sn  t ,  1
 gl  
  gl  l gl
 (t )  
  V0 gl  V0
2am  t ,  1
  l V0  gl
The first solution describes swinging motion of the pendulum, while the second solution describes the
motion that occurs if you push the pendulum so hard that it whirls around on the pivot. The equations
may look scary, but you can simply use MAPLE to calculate and plot them.
1. In the first equation, sn(x, k ) is a special function called the `sin amplitude.’ You can think of it
as a sort of trig function for adults – in fact for k=0, sn(x,0)  sin( x) and we recover the PG
version. You can compute it in Mupad using jacobiSN(x,k)
2. Similarly, am(x, k ) is a function called the `Amplitude.’ You can calculate it in Mupad using
jacobiAM(x,k). In Mupad, the am function has range - <am(x, k )   , so the solution
predicts that as the pendulum whirls around the pivot, the angle  increases from 0 to 2 , then
jumps to 2 , increases to 2 again, and so on.

You might have solved the pendulum problem already in an elementary physics course, and might
remember a different solution. This is because you probably only derived an approximate solution, by
assuming that the angle  remains small. This occurs when the initial velocity satisfies V / gl  1 , in
which case the solution can be approximated by
V  g
 (t )  sin  t 
gl  l 

3.3.3 Numerical solutions to equations of motion using MATLAB

In the preceding section, we were able to solve all our equations of motion exactly, and hence to find
formulas that describe the motion of the system. This should give you a warm and fuzzy feeling – it
appears that with very little work, you can predict everything about the motion of the system. You may
even have visions of running a consulting business from your yacht in the Caribbean, with nothing more
than your chef, your masseur (or masseuse) and a laptop with a copy of MAPLE.
Unfortunately real life is not so simple. Equations of motion for most engineering systems cannot be
solved exactly. Even very simple problems, such as calculating the effects of air resistance on the
trajectory of a particle, cannot be solved exactly.

For nearly all practical problems, the equations of motion need to be solved numerically, by using a
computer to calculate values for the position, velocity and acceleration of the system as functions of time.
Vast numbers of computer programs have been written for this purpose – some focus on very specialized
applications, such as calculating orbits for spacecraft (STK); calculating motion of atoms in a material
(LAMMPS); solving fluid flow problems (e.g. fluent, CFDRC); or analyzing deformation in solids (e.g.
ABAQUS, ANSYS, NASTRAN, DYNA); others are more general purpose equation solving programs.

In this course we will use MATLAB, which is widely used in all engineering applications. You should
complete the MATLAB tutorial before proceeding any further.

In the remainder of this section, we provide a number of examples that illustrate how MATLAB can be
used to solve dynamics problems. Each example illustrates one or more important technique for setting
up or solving equations of motion.

Example 1: Trajectory of a particle near the earth’s surface (with air resistance)

As a simple example we set up MATLAB to solve the

particle trajectory problem discussed in the preceding
section. We will extend the calculation to account for the V0
effects of air resistance, however. We will assume that our
FD
projectile is spherical, with diameter D, and we will assume k
that there is no wind. You may find it helpful to review the X0
j
discussion of aerodynamic drag forces in Section 2.1.7
before proceeding with this example. mg
i
1. Introduce variables to describe the motion: We can
simply use the Cartesian coordinates of the particle ( x(t ), y(t ), z (t ))

2. Write down the position vector in terms of these variables: r  x(t )i  y(t ) j  z(t )k

3. Differentiate the position vector with respect to time to find the acceleration. Simple calculus gives

dx dy dz d 2x d 2 y d 2z
v i  j k a i  j  k
dt dt dt dt 2 dt 2 dt 2

4. Draw a free body diagram. The particle is now subjected to two forces, as shown in the picture.

Gravity – as always we have Fg  mgk .

Air resistance.

1
The magnitude of the air drag force is given by FD   CD DV 2 , where
2
   is the air density,
 CD is the drag coefficient,
 D is the projectile’s diameter, and
 V is the magnitude of the projectile’s velocity relative to the air. Since we assumed the air is
stationary, V is simply the magnitude of the particle’s velocity, i.e.
2 2 2
 dx   dy   dz 
V       
 dt   dt   dt 

The Direction of the air drag force is always opposite to the direction of motion of the projectile
through the air. In this case the air is stationary, so the drag force is simply opposite to the direction
of the particle’s velocity. Note that v / V is a unit vector parallel to the particle’s velocity. The drag
force vector is therefore
1  D2 2 v 1  D2  dx dy dz 
FD    CD V    CD V  i  j k
2 4 V 2 4  dt dt dt 

The total force vector is therefore

1  D2  dx dy dz 
F  mgk   CD V  i  j k
2 4  dt dt dt 

5. Write down Newton’s laws of motion.

1  D 2  dx dy dz   d 2x d 2 y d 2z 
F  ma  mgk  CD V  i  j k   m 2 i  2 j 2 k 
2 4  dt dt dt   dt dt dt 

It is helpful to simplify the equation by defining a specific drag coefficient c   D 2CD , so that
8m
 dx dy dz   d x d y d z 
2 2 2
 gk  cV  i  j  k    2 i  2 j  2 k 
 dt dt dt   dt dt dt 
The vector equation actually represents three separate differential equations of motion
d 2x dx d2y dy d 2z dz
2
 cV 2
 cV 2
  g  cV
dt dt dt dt dt dt

6. Eliminate reactions – this is not needed in this example.

7. Identify initial conditions. The initial conditions were given in this problem - we have that
 dx   dy   dz 
x  X 0  Vx   y  Y0  Vy   z  Z 0  Vz 
 dt   dt   dt 

8. Solve the equations of motion. We can’t use the magic ‘dsolve’ command in MAPLE to solve this
equation – it has no known exact solution. So instead, we use MATLAB to generate a numerical
solution.
This takes two steps. First, we must turn the equations of motion into a form that MATLAB can use.
This means we must convert the equations into first-order vector valued differential equation of the
dy
general form  f (t , y ) . Then, we must write a MATLAB script to integrate the equations of motion.
dt

Converting the equations of motion: We can’t solve directly for (x,y,z), because these variables get
differentiated more than once with respect to time. To fix this, we introduce the time derivatives of
(x,y,z) as new unknown variables. In other words, we will solve for ( x, y, z, vx , v y , vz ) , where
dx dx dx
vx  vy  vz 
dt dt dt
These definitions are three new equations of motion relating our unknown variables. In addition, we can
re-write our original equations of motion as
dvx dv y dvz
 cVvx  cVv y   g  cVvz
dt dt dt
So, expressed as a vector valued differential equation, our equations of motion are
x  vx 
y  vy 
   
d  z   vz 
  
dt  vx   cVvx 
v y   cVv 
   y 
 vz    g  cVvz 

MATLAB script. The procedure for solving these equations is discussed in the MATLAB tutorial. A
basic MATLAB script is listed below.

function trajectory_3d
% Function to plot trajectory of a projectile
% launched from position X0 with velocity V0
% with specific air drag coefficient c
% stop_time specifies the end of the calculation

g = 9.81; % gravitational accel

c=0.5; % The constant c
X0=0; Y0=0; Z0=0; % The initial position
VX0=10; VY0=10; VZ0=20;
stop_time = 5;

initial_w = [X0,Y0,Z0,VX0,VY0,VZ0]; % The solution at t=0

[times,sols] = ode45(@eom,[0,stop_time],initial_w);

plot3(sols(:,1),sols(:,2),sols(:,3)) % Plot the trajectory

function dwdt = eom(t,w)

% Variables stored as follows w = [x,y,z,vx,vy,vz]
% i.e. x = w(1), y=w(2), z=w(3), etc
x = w(1); y=w(2); z=w(3);
vx = w(4); vy = w(5); vz = w(6);
vmag = sqrt(vx^2+vy^2+vz^2);
 dxdt = vx; dydt = vy; dzdt = vz;
dvxdt = -c*vmag*vx;
dvydt = -c*vmag*vy;
dvzdt = -c*vmag*vz-g;
dwdt = [dxdt;dydt;dzdt;dvxdt;dvydt;dvzdt];
end

end

This produces a plot that looks like this (the plot’s been edited to add the grid,etc)

Example 2: Simple satellite orbit calculation

j
The figure shows satellite with mass m orbiting a planet with
mass M. At time t=0 the satellite has position vector r  Ri
V
and velocity vector v  Vj . The planet’s motion may be R i
neglected (this is accurate as long as M>>m). Calculate and
plot the orbit of the satellite. m
M
1. Introduce variables to describe the motion: We will use
the (x,y) coordinates of the satellite.

2. Write down the position vector in terms of these variables: r  xi  yj

3. Differentiate the position vector with respect to time to find the acceleration.
dx dy d 2x d 2 y
v i j a 2 i 2 j
dt dt dt dt

j
Fg

i
4. Draw a free body diagram. The satellite is subjected to a gravitational force.
GMm
The magnitude of the force is Fg  , where
r2
 G is the gravitational constant, and
 r  x 2  y 2 is the distance between the planet and the satellite

The direction of the force is always towards the origin: r / r is therefore a unit vector parallel to the
direction of the force. The total force acting on the satellite is therefore
GMm r GMm
F
2 r
  xi  yj
r r3

5. Write down Newton’s laws of motion.

GMm  d 2x d 2 y 
F  ma    xi  yj   m  2 i  2 j
r3  dt dt 
The vector equation represents two separate differential equations of motion
d 2x GM d2y GM
2
 3 x 2
 3 y
dt r dt r

6. Eliminate reactions – this is not needed in this example.

7. Identify initial conditions. The initial conditions were given in this problem - we have that
 dx   dy 
x  R 0 y  0 V 
 dt   dt 

8. Solve the equations of motion. We follow the usual procedure: (i) convert the equations into
MATLAB form; and (ii) code a MATLAB script to solve them.

Converting the equations of motion: We introduce the time derivatives of (x,y) as new unknown
variables. In other words, we will solve for ( x, y, vx , v y ) , where
dx dx
vx  vy 
dt dt
These definitions are new equations of motion relating our unknown variables. In addition, we can re-
write our original equations of motion as
dvx GM dv y GM
 x  y
3
dt r dt r3
So, expressed as a vector valued differential equation, our equations of motion are
x  vx 
y  vy 
d    

dt  vx   GMx / r  3
 
v y   GMy / r 3 

Matlab script: Here’s a simple script to solve these equations.

 function satellite_orbit
% Function to plot orbit of a satellite
% launched from position (R,0) with velocity (0,V)

GM=1;
R=1;
V=1;
Time=100;
w0 = [R,0,0,V]; % Initial conditions

[t_values,w_values] = ode45(@odefunc,[0,time],w0);

plot(w_values(:,1),w_values(:,2))

function dwdt = odefunc(t,w)

x=w(1); y=w(2); vx=w(3); vy=w(4);
r = sqrt(x^2+y^2);
dwdt = [vx;vy;-GM*x/r^3;-GM*y/r^3];
end

end

Running the script produces the result shown (the plot was annotated by
hand)

Do we believe this result? It is a bit surprising – the satellite seems to be

spiraling in towards the planet. Most satellites don’t do this – so the
result is a bit suspicious. The First Law of Scientific Computing states
that ` if a computer simulation predicts a result that surprises you, it is
probably wrong.’

So how can we test our computation? There are two good tests:
1. Look for any features in the simulation that you can predict
without computation, and compare your predictions with those of the computer.
2. Try to find a special choice of system parameters for which you can derive an exact solution to
your problem, and compare your result with the computer
We can use both these checks here.

1. Conserved quantities For this particular problem, we know that (i) the total energy of the system
should be constant; and (ii) the angular momentum of the system about the planet should be constant
(these conservation laws will be discussed in the next chapter – for now, just take this as given). The total
energy of the system consists of the potential energy and kinetic energy of the satellite, and can be
calculated from the formula
E
GMm 1 2
r
 mv  
2
GMm 1
r 2

 m vx2  v 2y 
 
E
m
GM 1 2
r

 vx  v 2y
2

The total angular momentum of the satellite (about the origin) can be calculated from the formula
  
H  r  mv   xi  yj  m(vx i  v y j)  m xv y  yvx k

 
H
  xv y  yvx
m
(If you don’t know these formulas, don’t panic – we will discuss energy and angular momentum in the
next part of the course)

We can have MATLAB plot E/m and H / m , and see if these are really conserved. The energy and
momentum can be calculated by adding these lines to the MATLAB script

for i =1:length(t)
r = sqrt(w_values(i,1)^2 + w_values(i,2)^2)
vmag = sqrt(w_values(i,1)^2 + w_values(i,2)^2)
energy(i) = -GM/r + vmag^2/2;
angularm(i) = w_values(i,1)*w_values(i,4)-w_values(i,2)*w_values(i,3);
end

You can then plot the results (e.g. plot(t_values,energy)). The results are shown below.

These results look really bad – neither energy, nor angular momentum, are conserved in the simulation.
Something is clearly very badly wrong.

Comparison to exact solution: It is not always possible to find a

j
convenient exact solution, but in this case, we might guess that some m
special initial conditions could set the satellite moving on a circular path.
A circular path might be simple enough to analyze by hand. So let’s
R
assume that the path is circular, and try to find the necessary initial 
conditions. If you still remember the circular motion formulas, you could i
use them to do this. But only morons use formulas – here we will derive M
the solution from scratch. Note that, for a circular path
(a) the particle’s radius r=constant. In fact, we know r=R, from the
position at time t=0.
(b) The satellite must move at constant speed, and the angle  must increase linearly with time,
i.e.   t where    is a constant (see section 3.1.3 to review motion at constant speed
around a circle).
With this information we can solve the equations of motion. Recall that the position, velocity and
acceleration vectors for a particle traveling at constant speed around a circle are
 r  R cos  (t )i  R sin  (t ) j
v  R ( sin  (t )i  cos  (t ) j)
a   R 2  cos  (t )i  R sin  (t ) j
We know that v  V from the initial conditions, and v is constant. This tells us that
V  R
Finally, we can substitute this into Newton’s law
GMm r GMm V2
F  ma    ma   (cos i  sin  j)  m (cos i  sin  j)
R2 R R2 R
Both components of the equation of motion are satisfied if we choose
GM V2

R2 R
So, if we choose initial values of GM ,V , R satisfying this equation, the orbit will be circular. In fact, our
original choice, GM  1,V  1, R  1 should have given a circular orbit. It did not. Again, this means our
computer generated solution is totally wrong.

Fixing the problem: In general, when computer predictions are

suspect, we need to check the following
1. Is there an error in our MATLAB program? This is nearly
always the cause of the problem. In this case, however, the
program is correct (it’s too simple to get wrong, even for me).
2. There may be something wrong with our equations of motion
(because we made a mistake in the derivation). This would not
explain the discrepancy between the circular orbit we predict
and the simulation, since we used the same equations in both
cases.
3. Is the MATLAB solution sufficiently accurate? Remember that
by default the ODE solver tries to give a solution that has 0.1%
error. This may not be good enough. So we can try solving the
problem again, but with better accuracy. We can do this by modifying the MATLAB call to the
equation solver as follows
options = odeset('RelTol',0.00001);
[t_values,w_vlues] = ode45(@odefunc,[0,time],w0,options);
4. Is there some feature of the equation of motion that makes them especially difficult to solve? In
this case we might have to try a different equation solver, or try a different way to set up the
problem.

The figure on the right shows the orbit predicted with the better accuracy. You can see there is no longer
any problem – the orbit is perfectly circular. The figures below plot the energy and angular momentum
predicted by the computer.
There is a small change in energy and angular momentum but the rate of change has been reduced
dramatically. We can make the error smaller still by using improving the tolerances further, if this is
needed. But the changes in energy and angular momentum are only of order 0.01% over a large number
of orbits: this would be sufficiently accurate for most practical applications.

Most ODE solvers are purposely designed to lose a small amount of energy as the simulation proceeds,
because this helps to make them more stable. But in some applications this is unacceptable – for example
in a molecular dynamic simulation where we are trying to predict the entropic response of a polymer, or a
free vibration problem where we need to run the simulation for an extended period of time. There are
special ODE solvers that will conserve energy exactly.

Example 3: Earthquake response of a 2-storey building

The figure shows a very simple idealization of a 2-storey building. The

roof and second floor are idealized as blocks with mass m. They are m
supported by structural columns, which can be idealized as springs with
stiffness k and unstretched length L. At time t=0 the floors are at rest l2 k k y2
and the columns have lengths l1  L  mg / k l2  L  mg / 2k (can
m
you show this?). We will neglect the thickness of the floors themselves,
to keep things simple. k k
l1 y1
For time t>0, an earthquake makes the ground vibrate vertically. The
ground motion can be described using the equation d  d0 sin t .
d0sint
Horizontal motion may be neglected. Our goal is to calculate the motion of the first and second floor of
the building.

It is worth noting a few points about this problem:

1. You may be skeptical that the floor of a building can be idealized as a particle (then again, maybe
you couldn’t care less…). If so, you are right – it certainly is not a `small’ object. However,
because the floors move vertically without rotation, the rigid body equations of motion simply
reduce to F  ma and M=0, where the moments are taken about the center of mass of the block.
The floors behave as though they are particles, even though they are very large.
2. Real earthquakes involve predominantly horizontal, not vertical motion of the ground. In
addition, structural columns resist extensional loading much more strongly than transverse
loading. So we should really be analyzing horizontal motion of the building rather than vertical
 motion. However, the free body diagrams for horizontal motion are messy (see if you can draw
them) and the equations of motion for vertical and horizontal motion turn out to be the same, so
we consider vertical motion to keep things simple.
3. This problem could be solved analytically (e.g. using the `dsolve’ feature of MAPLE) – a
numerical solution is not necessary. Try this for yourself.

1. Introduce variables to describe the motion: We will use the height of each floor ( y1 , y2 ) as the
variables.

2. Write down the position vector in terms of these variables: We now have to worry about two masses,
and must write down the position vector of both
r1  y1 j r2  y2 j
Note that we must measure the position of each mass from a fixed point.

3. Differentiate the position vector with respect to time to find the acceleration.
 dy   dy 2 
v1   1  j a1   1  j
 dt   dt 2 
 
 dy   dy 
2
v 2   2  j a2   2  j
 dt   dt 2 
 

4. Draw a free body diagram. We must draw a free body diagram for each
mass. The resultant force acting on the bottom and top masses, respectively, are mg
F1  mg  2FS1  2FS 2  j F2  mg  FS 2 j
where FS1, FS 2 are the forces in the two springs (note that we assume that all FS2 FS2
the springs are in tension – this makes the calculation easier).

The spring forces are equal to the stiffness multiplied by the increase in length mg
of the springs
FS1  k (l1  L) FS 2  k (l2  L) FS1 FS1
We will have to find the spring lengths l1 , l2 in terms of our coordinates y1 , y2 to solve the problem.
Geometry shows that
l2  y2  y1 l1  y1  d0 sin t
So, finally
F1  mg  2k ( y1  d0 sin t  L)  2k ( y2  y1  L) j
F2  mg  2k ( y2  y1  L) j

5. Write down Newton’s laws of motion. F=ma for each mass gives
d 2 y1
mg  2k (l1  L)  2k (l2  L) j  m j
dt 2
d 2 y2
mg  2k (l2  L) j  m j
dt 2
This is two equations of motion – we can substitute for l1 , l2 and rearrange them as
 dy12  k k 
   g  2 ( y1  d0 sin t  L)  2 ( y2  y1  L) 
dt2  m m 
dy22  k 
   g  2 ( y2  y1  L) 
dt2  m 

6. Eliminate reactions – this is not needed in this example.

7. Identify initial conditions. We know that, at time t=0

dy1 dy2
y1  L  mg / k 0 y2  2L  3mg / 2k 0
dt dt

8. Solve the equations of motion. We need to (i) reduce the equations to the standard MATLAB form and
(ii) write a MATLAB script to solve them.

Converting the equations. We now need to do two things: (a) remove the second derivatives with respect
to time, by introducing new variables; and (b) rearrange the equations into the form dy / dt  f (t , y) . We
dl dl
remove the derivatives by introducing v1  1 v2  2 as additional unknown variables, in the usual
dt dt
way. Our equations of motion can then be expressed as
dy1
 v1
dt
dy2
 v2
dt
dv1  k k 
   g  2 ( y1  d0 sin  t  L)  2 ( y2  y1  L) 
2  m m 
dt
dv2  k 
  g  2 ( y2  y1  L) 
2  m 
dt

We can now code MATLAB to solve these equations directly for dy/dt. A script (which plots the
position of each floor as a function of time) is shown below.
function building
%
k=100;
m=1;
omega=9;
d=0.1;
L=10;
time=20;
g = 9.81;
w0 = [L-m*g/k,2*L-3*m*g/(2*k),0,0];
[t_values,w_values] = ode45(@eom,[0,time],w0);
plot(t_values,w_values(:,1:2));

function dwdt = eom(t,w)

y1=w(1);
y2=w(2);
v1=w(3);
v2=w(4);
 dwdt = [v1;v2;...
-2*k*(y1-d*sin(omega*t)-L)/m+2*k*(y2-y1-L)/m;...
-g-2*k*(y2-y1-L)/m]
end
end

The figures below plot the height of each floor as a function of time, for various earthquake frequencies.
For special earthquake frequencies (near the two resonant frequencies of the structure) the building
vibrations are very severe. As long as the structure is designed so that its resonant frequencies are well
away from the frequency of a typical earthquake, it will be safe.

We will discuss vibrations in much more detail later in this course.

This problem illustrates a shortcoming of solving problems on the computer without thinking too much
about them. The way we set up the problem, it looks as though the solution depends on g, and the
unstretched spring length L, but in fact this is not the case. Not being aware of this makes the equations
of motion much more complicated than they really are, and makes it harder to interpret the results. Of
course we could learn by trial and error that the solution is independent of L and g, but a better approach
is to eliminate these variables from the equations of motion altogether.

There is a standard way to do this – instead of solving for the lengths of the springs y, we solve for the
deflection of the masses from their static equilibrium positions. We will discuss this in more general
terms when we discuss vibration problems later in the course. But we’ll work through the process here,
because it’s useful to use the same approach in the mass launcher design project.
The figure illustrates the idea. The figure
on the left shows the masses at their static m
equilibrium positions. Here the springs
have lengths z2
l1  L  mg / k , l2  L  mg / 2k . We are m
now going to describe the motion by the
y2
L-mg/2k k k m
displacement of each mass from its static
equilibrium position, z2 , z2 . We simply m z1
work through the derivations again with y1
these new variables. L-mg/k k k
d0sint
The accelerations of the masses don’t
depend on what we use for the origin
(provided the origin is fixed, of course), so
 dz   dz 2 
v1   1  j a1   1  j
 dt   dt 2 
 
 dz   dz 
2
v 2   2  j a2   2  j
 dt   dt 2 
 

The free body diagram doesn’t change either, and the forces in the springs are mg
still given by
FS1  k (l1  L) FS 2  k (l2  L)
FS2 FS2
But now we have to find the spring lengths l1 , l2 in terms of our coordinates
z1, z2 to solve the problem. Geometry shows that
mg
l1  L  mg / k  z1  d0 sin t
l2  L  mg / 2k  z2  z1
FS1 FS1
Newton’s laws of motion now become
d 2 z1
mg  2k (l1  L)  2k (l2  L) j  m j
dt 2
d 2 z2
mg  2k (l2  L) j  m j
dt 2
Finally substituting for l1, l2 and simplifying we find that lots of terms magically cancel, and
d 2 z1 2k
 (d0 sin t  z1 )
2 m
dt
d 2 z2 2k
 ( z1  z2 )
2 m
dt
These equations don’t involve L or g. Furthermore the initial conditions are simply
dz dz
z1  z2  0; 1  2  0
dt dt
so the solution for z1, z2 must be independent of L and g.
Example 4: The dreaded pendulum revisited (apologies…)

You may have lost interest in pendulum problems by now. Bear with us, however- it i
is a convenient example that illustrates how to solve problems with constraints.
l
So we re-visit the problem shown in the figure. This time, we will describe the
system using (x,y) coordinates of the mass instead of the inclination of the
j 
cable.

1. Introduce variables to describe the motion: We will use the position of the mouse relative to point O
( x, y) as the variables.

2. Write down the position vector in terms of these variables:

r  xi  yj
Note that we’ve chosen j to point vertically downwards

3. Differentiate the position vector with respect to time to find the acceleration.
dx dy d 2x d2y
v i j a i j
dt dt dt 2 dt 2

4. Draw a free body diagram. We can use the FBD we drew earlier. The force must R
now be expressed in terms of x and y instead of  , however. Simple trig shows that
y x
cos  sin   
x2  y 2 x2  y 2
The resultant force is therefore j i
mg
x y
F   R sin  i  R cos j  mgj   R iR j  mgj
x2  y 2 x2  y 2

5. Write down Newton’s laws of motion. F=ma gives

x y  d 2x d2y 
F  R iR j  mgj  m  i j
 dt 2 dt 2 
x2  y 2 x2  y 2 
This is two equations of motion – we can rearrange them as
d 2x R x d2y R y
  g
2 m x2  y 2 2 m x2  y 2
dt dt

6. Eliminate reactions –We could eliminate R if we wanted – but this time we won’t bother. Instead, we
will carry R along as an additional unknown, and use MATLAB to calculate it.

7. If there are more unknown variables than equations of motion, look for constraint equations. We
now have three unknowns (x,y,R) but only two equations of motion (eliminating R in step 6 won’t help –
in this case we will have two unknowns but only one equation of motion). So we need to look for an
additional equation somewhere.
The reason we’re missing an equation is that we took x,y to be arbitrary – but of course the mouse has to
remain attached to the cable at all times. This means that his distance from O is fixed –i.e.
x2  y 2  l
This is the missing equation.

We could, if we wanted, use this equation to eliminate one of our unknown variables (doing the algebra
by hand). Instead, however, we will use MATLAB to do this for us. For this purpose, we need to turn
the equation into a constraint on the accelerations, instead of the position of the particle. To get such an
equation, we can differentiate both sides of the constraint with respect to time
d d x dx y dy
x2  y 2  l   0
dt dt x 2  y 2 dt x 2  y 2 dt
dx dy
x  y 0
dt dt
This is now a constraint on the velocity. Differentiating again gives
2 2
 dx  d 2 x  dy  d2y
   x     y 0
 dt  dt 2  dt  dt 2
2 2
d 2x d2y dx   dy 
x y      
dt 2
dt 2  dt   dt 
Again – if you have trouble doing the derivatives, just use MAPLE (don’t forget to specify that x, y are
functions of time, i.e. enter them as x(t), y(t) when typing the constraint formula into MAPLE).

[Aside – when I first coded this problem I tried to constrain the velocity of the particle, instead of the
acceleration. This doesn’t work (as I should have known), and produces some rather interesting
MATLAB errors – if you are curious, try it and see what happens. If you are even more curious, you
might like to think about why you need to constrain accelerations and not velocities.]

7. Identify initial conditions. We know that, at time t=0

dx dy
x0  V0 y  L 0
dt dt

8. Solve the equations of motion. We need to (i) reduce the equations to the standard MATLAB form and
(ii) write a MATLAB script to solve them.

In the usual way, we introduce vx  dx / dt , v y  dy / dt as additional unknowns. Our set of unknown

variables is x, y, vx , v y , R . The equations of motion, together with the constraint equation, can be
expressed as
 dx
 vx
dt
dy
 vy
dt
dvx R x
 0
dt m x2  y 2
d 2v y R y
 g
dt m x2  y 2
d 2x d2y
x y  vx2  v 2y
dt 2 dt 2
This can be expressed as a matrix equation for an unknown vector [dy / dt , R]
 dx 
1 0 0 0 0  
0   dt 
 1 0 0 0   dy   vx 
   v 
x/m 
  dt   
y
0 0 1 0
 x2  y 2   dvx    0 
   dt   
0 y/m   g 
 
  y  v 2  v 2 
0 0 1
 x2  y 2
dv
   dt   x y 
0 0 x y 0   
 R 
We can now use MATLAB to solve the equations for the unknown time derivatives dy / dt , together with
the reaction force R. Here’s a MATLAB script that integrates the equations of motion and plots (x,y) as a
function of time. Notice that, because we don’t need to integrate R with respect to time, the function
‘eom’ only needs to return dy / dt .
function pendulum

g = 9.81;
l=1;
V0=0.1;
time=20;
w0 = [0,l,V0,0]; % Initial conditions
[t_values,w_values] = ode45(@eom,[0,time],w0);
plot(t_values,w_values(:,1:2));

function dwdt = eom(t,w)

% The vector w contains [x,y,vx,vy]
x=w(1);y=w(2);vx=w(3);vy=w(4);
M = eye(5); % This sets up a 5x5 matrix with 1s on all the diags
M(3,5) = (x/m)/sqrt(x^2+y^2);
M(4,5) = (y/m)/sqrt(x^2+y^2);
M(5,3) = x;
M(5,4) = y;
M(5,5) = 0;
f = [vx;vy;0;g;-vx^2-vy^2];
sol = M\f; % This solves the matrix equation M*[dydt,R]=f for [dwdt,R]
dwdt = sol(1:4); % only need to return time derivatives dw/dt
 end
end

Final remarks: In general, it is best to avoid using constraint equations – it makes the problem harder to
set up, and harder for MATLAB to solve. Sometimes they are unavoidable, however – in some cases
you might not see how to identify a suitable set of independent coordinates; and there are some systems (a
rolling wheel is the most common example) for which a set of independent coordinates do not exist.
These are given the fancy name `non-holonomic systems’ – mentioning that you happen to be an expert
on non-holonomic systems during a romantic candle-light dinner is sure to impress your dates.

3.3.4 Case Study - Simple model of a vehicle

As a representative application of the methods outlined in the preceding section, we will set up and solve
the equations of motion for a very simple idealization of a vehicle. This happens to be an example of a
non-holonomic system (sorry we aren’t at a romantic dinner). Don’t worry if the model looks rather scary
– this calculation is more advanced than anything you would be expected to do at this stage of your
career… Our goal is to illustrate how the method we’ve developed in this section can be applied to a real
engineering system of interest. You should be able to follow and understand the procedure.

The figure shows how the vehicle is idealized. Here are a few L/2
remarks: L/2
1. We consider only 2D planar motion of the vehicle
2. For simplicity we assume the vehicle has only two y 
wheels, one at the front and one at the rear. (It’s not
a motorcycle, however, because we won’t account x  Motion
for the vehicle leaning around corners) Steering
3. The car is idealized as a particle with mass m angle
supported on a massless chassis with wheelbase L.
4. Aerodynamic drag forces are assumed to act directly
on the particle.
5. The most complicated and important part of a vehicle dynamics model is the description of how
the wheels interact with the road. Here, we will just assume that
a. The front and rear of the vehicle have to move in a direction perpendicular to each
wheel’s axle. Reaction forces must act on each wheel parallel to the axle in order to
enforce this constraint (see the FBD).
b. In addition, we assume that the vehicle has rear-wheel drive. This is modeled as a
prescribed force P(t ) exerted by the ground on the rear wheel, acting parallel to the
rolling direction of the wheel (see the FBD).
c. The front wheel is assumed to roll freely and have negligible mass – this means that the
contact force acting on the wheel must be perpendicular to its rolling direction (see the
FBD). The vehicle is steered by rotating the front wheel through an angle  (t ) with
respect to the chassis.
This is a vastly over-simplified model of wheel/road interaction – for example, it predicts that the
car can never skid. If you are interested in extending this calculation to a more realistic model,
ask us… We’d be happy to give you better equations!
1. Introduce variables to describe the motion: We will use the (x,y) coordinates of car and its orientation
 as the variables.

2. Write down the position vector in terms of these variables: r  xi  yj

3. Differentiate the position vector with respect to time to find the acceleration.
dx dy d 2x d 2 y
v i j a i 2 j
dt dt dt 2 dt

4. Draw a free body diagram. The vehicle is subjected to (i)

P RR
a thrust force P and a lateral reaction force RR acting on the
rear wheel, (ii) a lateral reaction force RF acting on the front
 
wheel, and a drag force FD acting at the center of mass.
j RF
FD
The drag force is related to the vehicle’s velocity by i
 dx dy  
FD  cvv  cv  i  j 
 dt dt 
where c is a drag coefficient. For a more detailed discussion of drag forces see the first example in this
section.

The resultant force on the vehicle is

dx
F  ( P(t ) cos   RR sin   RF sin(   )  cv )i
dt
dx
 ( P(t ) sin   RR cos   RF cos(   )  cv ) j
dt
Because we are modeling the motion of the vehicle’s chassis, which can rotate, we must also consider the
moments acting on the chassis. You should be able to verify for yourself that the resultant moment of all
the forces about the particle is
L
M ( RF cos   RR )k
2

5. Write down Newton’s laws of motion. F=ma gives

dx
( P(t ) cos   RR sin   RF sin(   )  cv )i
dt
dy
 ( P(t ) sin   RR cos   RF cos(   )  cv )j
dt
 d 2x d 2 y 
 m  2 i  2 j
 dt dt 
This gives two equations of motion; the condition M=0 for the chassis gives one more.

6. Eliminate reactions – It’s easier to eliminate them with MATLAB.

7. If there are more unknown variables than equations of motion, look for constraint equations. This is
the hard part of this application. We have two unknown reaction forces, so we need to find two
constraint equations that will determine them. These are (i) that the rear of the vehicle must move
perpendicular to the axle of the rear wheel; and (ii) the front of the vehicle must move perpendicular to
the axle of the front wheel. Consider the rear wheel:
 L   L 
1. Note that the position vector of the rear wheel is r   x  cos   i   y  sin   j
 2   2 
 dx L d   dy L d 
2. The velocity follows as v    sin   i    cos  j
 dt 2 dt   dt 2 dt 
3. Note that n  sin  i  cos j is a unit vector parallel to the axle of the rear wheel.
4. For the rear of the vehicle to move perpendicular to the rear axle, we must have v  n  0 . This
shows that
dx dy L d
sin   cos   0
dt dt 2 dt
Similarly, for the front wheel, we can show that
dx dy L d
sin(   )  cos(   )  (sin  sin(   )  cos  cos(   )) 0
dt dt 2 dt
dx dy L d
 sin(   )  cos(   )  cos  0
dt dt 2 dt
where we have used the trig formula cos( A  B)  cos A cos B  sin A sin B .

We need to re-write these equations as constraints on the acceleration of the vehicle. To do this, we
differentiate both constraints with respect to time, to see that
d dx dy L d 
 sin   cos  0
dt  dt dt 2 dt 
dx d d 2x dy d d 2 y L d 2
 cos  sin   sin   cos  0
dt dt dt 2 dt dt dt 2 2 dt 2
d dx dy L d 
 sin(   )  cos(   )  cos  0
dt  dt dt 2 dt 
dx  d d  d 2x dy  d d  d2y
 cos(   )     sin(   ) 2  sin(   )     cos(   ) 2
dt  dt dt  dt dt  dt dt  dt
L d d  L d 2
 sin   cos  0
2 dt dt 2 dt 2

7. Identify initial conditions: We will assume that the vehicle starts at rest, with
dx dy d
x  y     0
dt dt dt

7. Solve the equations of motion. We need to write the equations of motion in a suitable matrix form for
MATLAB. We need to eliminate all the second derivatives with respect to time from the equations of
motion, by introducing new variables. To do this, we define
dx dy d
vx  vy  
dt dt dt
as new variables, and then solve for [ x, y, , vx , v y , ] . We also need to eliminate the unknown reactions.
It is not hard to show that the equations of motion, in matrix form, are
1 0 0 0 0 0 0 0   dx / dt 
0 1 0  
 0 0 0 0 0   dy / dt 
0 0 1 0 0 0 0 0   d / dt 
  
0 0 0 m 0 0  sin(   )  sin    dvx / dt 
f
0 0 0 0 m 0  cos(   )  cos    dv y / dt 
  
0 0 0 0 0 0 cos  1   d  / dt 
0 0 0 sin  cos  L/2 0 0   RF 
  
0 0 0 sin(   ) cos     ( L / 2)cos  0 0   RR 
where
 vx 
 v 
 y 
  
 
 P(t ) cos   cvvx 

f   P (t )sin   cvv 
y 
 0 
 
  cos vx  sin  v y 
 
  cos(   )v    d   sin(   )v    d   sin  L  d 
 x  y 
 dt   dt  2 dt 
Finally, we can type these into MATLAB – here’s a simple script that solves the equations of motion and
plots the (x,y) coordinates of the car to show its path, and also plots the speed of the car as a function of
time. The example simulates a drunk-driver, who steers with steering angle   0.1  0.2sin(t ) , and
drives with his or her foot to the floor with P=constant.

function drivemycar

L=1;
m=1;
c=0.1;
time=120;
y0 = [0,0,0,0,0,0];
options = odeset('RelTol',0.00001);
[t_vals,w_vals] = ode45(@eom,[0,time],y0,options);
plot(w_vals(:,1),w_vals(:,2));

function [alpha,dadt,P]=driver(t)
% This function behaves like the driver of the car -
% at time t it returns the steering angle alpha
% dalpha/dt and the driving force P

alpha = 0.1+0.2*sin(t);
dadt = 0.2*cos(t);
P = 0.2;
end
function dwdt = eom(t,w)
% Equations of motion for the vehicle
%

[alpha,dadt,P] = driver(t); % Find out what the driver is doing

M = zeros(8); % This sets up a 8x8 matrix full of zeros
M(1,1) = 1;
M(2,2) = 1;
M(3,3) = 1;
M(4,4) = m;
M(4,7)=-sin(w(3)+alpha);
M(4,8)=-sin(w(3));
M(5,5)=m;
M(5,7)=-cos(w(3)+alpha);
M(5,8)=-cos(w(3));
M(6,7)=cos(alpha);
M(6,8)=-1;
M(7,4)=sin(w(3));
M(7,5)=cos(w(3));
M(7,6)=L/2;
M(8,4)=sin(w(3)+alpha);
M(8,5)=cos(w(3)+alpha);
M(8,6)=-L*cos(alpha)/2;
v = sqrt(w(4)^2+w(5)^2);
f = [w(4);
w(5);
w(6);
P*cos(w(3))-c*v*w(4);
-P*sin(w(3))-c*v*w(5);
0;
(-cos(w(3))*w(4)+sin(w(3))*w(5))*w(6);
(-cos(w(3)+alpha)*w(4)+sin(w(3)+alpha)*w(5))*(w(6)+dadt)-
sin(alpha)*L*w(6)*dadt/2];
sol = M\f; % This solves the matrix equation M*[dwdt,R]=f for [dwdt,R]
dwdt = sol(1:6); % only need to return time derivatives dw/dt
end
end

3.4 Review of concepts for Chapter 3

3.4.1 Concept checklist

Here are the skills that you should develop based on the material in this chapter:
 Be able to idealize an engineering design as a set of particles, and know when this idealization
will give accurate results
 Choose an appropriate set of geometric variables to describe the motion of a system of particles
(eg components in a fixed coordinate system; components in a polar coordinate system, etc)
 Be able to differentiate position vectors (with proper use of the chain rule!) to determine velocity
and acceleration; and be able to integrate acceleration or velocity to determine position vector.
 Be familiar with simple harmonic motion and definitions of amplitude, frequency and period
  Be able to describe motion in normal-tangential and polar coordinates (eg be able to write down
vector components of velocity and acceleration in terms of speed, radius of curvature of path, or
coordinates in the cylindrical-polar system).
 Be able to convert between Cartesian to normal-tangential or polar coordinate descriptions of
motion
 Be able to draw a correct free body diagram showing forces acting on system idealized as
particles
 Understand and be able to describe mathematically the forces exerted by springs and dampers,
and draw forces exerted by springs/dampers on a free body diagram.
 Be able to write down Newton’s laws of motion in rectangular, normal-tangential, and polar
coordinate systems
 Be able to obtain an additional moment balance equation for a rigid body moving without rotation
or rotating about a fixed axis at constant rate.
 Be able to use Newton’s laws of motion to solve for unknown accelerations or forces in a system
of particles
 Use Newton’s laws of motion to derive differential equations governing the motion of a system of
particles
 Solve the differential equations of motion analytically using Mupad, for cases where analytical
solutions are available
 Be able to re-write second order differential equations as a pair of first-order differential
equations in a form that MATLAB can solve
 Write a MATLAB script to solve differential equations governing the motion of a system

Of course, these are all just tricks of the trade. They are supposed to help you design a system that does
something useful; or to understand (and ultimately to predict) the behavior of some physical or biological
system.

3.4.2 Summary of main equations and definitions

Position-velocity-acceleration relations in a Cartesian j

Frame
v(t) dt
r (t )  x(t )i  y (t ) j  z (t )k r(t) path of particle
dx dy dz
v(t )  vx (t )i  v y (t ) j  vz (t )k  i  j  k r(t+dt)
dt dt dt
O
dv dv y dv d x k d2y
2
d 2z i
a(t )  ax (t )i  a y (t ) j  az (t )k  x i  j z k  i j k
dt dt dt 2 2 2
dt dt dt

The direction of the velocity vector is tangent to its path.

The magnitude of the velocity vector vx2  v 2y  vz2 is the distance traveled along the path per
unit time (speed).

A unit vector tangent to the path can be found as

 vx (t )i  v y (t ) j  vz (t )k
t
vx2  v 2y  vz2

Straight line motion with constant acceleration

 1 
r   X 0  V0t  at 2  i v  V0  at  i a  ai
 2 
Here, a is the (constant) acceleration; X 0 ,V0 are the position and speed at time t=0.

Straight line motion with time/position dependent acceleration

 t   t 
Acceleration given as a function of time: r   X 0   v(t )dt  i v   V0   a(t )dt  i
   
 0   0 
v (t ) x (t )
Acceleration given as a function of position  vdv   a( x)dx
V0 0

Separation of variables for one-dimensional motion

v t
dv g (t )
a 
dt f (v)
  f (v)dv   g (t )dt
V0 0
x (t ) t
dx g (t )
v 
dt f (v)
  f ( x)dv   v(t )dt
X0 0

Simple Harmonic Motion

r   X 0  X sin(2 t / T )  i
v  V cos(2 t / T )i
a   A sin(2 t / T )i
2X 2 V 4 2 X
V A 
T T T2
 sin 
Circular Motion at Constant Speed t cos
R
r  R  cos i  sin  j j n
v   R   sin  i  cos j  Vt
 t Rsin 
i
V2
a   R(cos i  sin  j)   Rn 
2 2
n Rcos
R

General Circular Motion

  d / dt   d / dt  d 2 / dt 2
sin 
s  R V  ds / dt  R t cos
R
r  R  cos i  sin  j j n
v   R   sin  i  cos j  Vt  t Rsin 
i
a  R ( sin  i  cos j)  R 2 (cos  i  sin  j)
Rcos
dV V2
  Rt   Rn 
2
t n
dt R

Note that the straight-line motion relations can be used to relate  , , , by exchanging
x   , v  , a  

Motion along an arbitrary path in normal-tangential coordinates

v  Vt
t
dV V2
a t n
dt R R

For a path with r  x( )i  y( ) j

n
t
dx d 2 y dy d 2 x

1 d d2 d d2

R  2 3/2
 dx   dy  
2
    
 d    d   
Position-velocity-acceleration relations in polar-coordinates

er  cos i  sin  j e   sin  i  cos j

i  cos er  sin  e j  sin  er  cos e
e
er
x  r cos y  r sin 

r  x2  y 2   tan 1 y / x r
j

r  rer i
dr d
v  er  r e
dt dt
 d 2r  d  
2  d 2 dr d 
a  r   e 
r  r 2  e
 dt 2  dt    dt
2 dt dt 


Newton’s laws

For a particle F  ma

For a rigid body moving without rotation or

rotating at fixed angular rate about a fixed axis
MC  0 (you must take moments about the j
center of mass) i

NA TB NB
W
Drawing free body diagrams:

1. Decide which part of a system you will idealize as a particle (you may need more than one
particle)
2. Draw the part of the system you have idealized as a particle by itself (very important!). It is
important to make sure that your particle is isolated – it can’t be touching something else.
You may need more than one drawing if you have more than one particle in your system.
3. Draw on any of the following external forces that apply. Make sure you draw them acting in
the correct direction, acting on the correct part of the body:
a. gravity (at the COM);
b. air resistance or lift forces (and sometimes moments) – various conventions are used
to locate these forces but in this course we usually put them at the COM and neglect
moments;
c. Buoyancy forces (act at the COM of the displaced fluid)
d. Electrostatic or electromagnetic forces
 4. Draw the forces exerted by springs attached to the particle. It is best to assume that springs
always pull on the point they are connected to, and that the magnitude of the force in the
spring is Fs  k (l  l0 ) , where l is the length of the spring, and l0 is its unstretched length.
5. Draw the forces exerted by dashpots or dampers (like springs, assume they pull on the object
they are connected to, and exert a force magnitude Fd   dl / dt where l is the length of the
dashpot.
6. Draw forces exerted by cables. Cables always pull, and exert a force parallel to the direction
of the cable. The magnitude of the force has to be left as an unknown.
7. Draw any unknown reaction forces, with the following rules:
a. Reaction forces must act at any point on any point of the body that is touching
something outside the particle (i.e. a part of your system that you did not include in
your drawing in step 2).
b. If the connection between the two touching objects prevents them from rotating with
respect to one another (or, like a motor, makes them rotate with some controllable
angular speed), you will need to draw both reaction forces and moments. (Reaction
moments do sometimes come up in dynamics problems, but they are not very
common, so think carefully before including them).
c. If friction acts at the contact point, and you don’t know whether the two objects slide
at the contact (or you know they do not slide), draw both a normal and a tangential
force with unknown magnitudes N,T (or some suitable variable). The direction of the
friction force is not important. DO NOT assume T   N .
d. If friction acts at the contact point, and you know the contact slips, draw both a
tangential and a normal force. You must draw the tangential force so that it opposes
the direction of sliding (ask a faculty member or TA if you don’t understand this). If
slip occurs you can assume T   N .
e. If the contact point is frictionless, draw only a normal force.
f. If your particle is being touched by a two-force member (no, this is not a gender and
sexuality class… a two force member is a massless rod, connected through freely
rotating hinges at both ends. A massless freely rotating wheel can also be idealized
as a two-force member) you can assume the reaction force acts parallel to the two-
force member.
g. If you have more than one particle in your system, make sure that any forces exerted
by one particle on the other have equal and opposite reactions.

Calculating unknown forces or accelerations using Newton’s laws:

1. Decide how to idealize the system (what are the particles?)

2. Draw a free body diagram showing the forces acting on each particle
3. Consider the kinematics of the problem. The goal is to calculate the acceleration of each particle
in the system – you may be able to start by writing down the position vector and differentiating it,
or you may be able to relate the accelerations of two particles (eg if two particles move together,
their accelerations must be equal).
4. Write down F=ma for each particle.
5. If you are solving a problem involving a massless frames (see, e.g. Example 3, involving a
bicycle with negligible mass) you also need to write down MC  0 about the particle.
6. Solve the resulting equations for any unknown components of force or acceleration (this is just
like a statics problem, except the right hand side is not zero).
Problems like this will usually ask you to make some design prediction at the end, which might involve
calculating critical conditions for something to slip, tip, break, etc.
 At the onset of slip at a contact T   N
 At the critical point where an object tips over, a reaction force somewhere will go to zero. You
will have to identify where this point is, find the reaction force, and set it to zero.

Deriving equations of motion for a system of particles

1. Introduce a set of variables that can describe the motion of the system. Don’t worry if this sounds
vague – it will be clear what this means when we solve specific examples.
2. Write down the position vector of each particle in the system in terms of these variables
3. Differentiate the position vector(s), to calculate the velocity and acceleration of each particle in
terms of your variables;
4. Draw a free body diagram showing the forces acting on each particle. You may need to introduce
variables to describe reaction forces. Write down the resultant force vector.
5. Write down Newton’s law F  ma for each particle. This will generate up to 3 equations of
motion (one for each vector component) for each particle.
6. If you wish, you can eliminate any unknown reaction forces from Newton’s laws. If you are
trying to solve the equations by hand, you should always do this; of you are using MATLAB, it’s
not usually necessary – you can have MATLAB calculate the reactions for you. The result will be
a set of differential equations for the variables defined in step (1)
7. If you find you have fewer equations than unknown variables, you should look for any
constraints that restrict the motion of the particles. The constraints must be expressed in terms of
the unknown accelerations.
8. Identify the initial conditions for the variables defined in (1). These are usually the values of the
unknown variables, their time derivatives, at time t=0. If you happen to know the values of the
variables at some other instant in time, you can use that too. If you don’t know their values at
all, you should just introduce new (unknown) variables to denote the initial conditions.
9. Solve the differential equations, subject to the initial conditions.

Trajectory equations for particle moving near earth’s

surface with no air resistance

V0
r  X 0i  Y0 j  Z 0k  k

dr t  0 X0
 Vx0i  V y 0 j  Vz 0k  j
dt 
i
 
  1
r   X 0  Vx0t  i  Y0  V y 0t j   Z 0  Vz 0t  gt 2  k
 2 
 
v  Vx 0  i  V y 0 j  Vz 0  gt  k
a   gk
Solving differential equations with Mupad:

Example: to solve
d2y dy
 2n  n2 y  0
dt 2 dt
dy
with initial conditions y  y0  v0 at time t=0
dt

Re-writing a second-order differential equation as a pair of first-order equations for

MATLAB

Example: to solve
d2y dy
 2n  n2 y  0
2 dt
dt
we introduce v  dy / dt as an additional variable. This new equation, together with the original
ODE can then be written in the following form
d  y  v 
   
dt  v   2n v  n2 y 
This is now in the form
dw  y
 f (t , w ) w 
dt v 
as required.