Page 1 21 August 2003

Figure 1: Illustration of dynamic loads on a simple structure and its foundation

Dynamics of Structures and Soils
Class notes by Professor Panos D. Kiousis
Colorado School of Mines
1. Introduction
The objective of this class is to examine the effects of dynamic loads on structures and
their foundations. Dynamic loads of interest are due to earthquakes, wind, machines, and
others. Machine loads are usually defined with a reasonable degree of accuracy, and can
be treated reliably by deterministic structural dynamics. On the other hand, earthquake
and wind loads are random and are treated best by stochastic or random structural
dynamics. However, the easier approach of deterministic structural dynamics with
conservative factors of safety is often preferred even in the case of stochastic loads
such as earthquakes and winds. In this course, we will only consider the deterministic
approach. Simple illustration of dynamic loads on a structure are presented in Figure 1.
Page 2 21 August 2003
Figure 2
The following simple example demonstrates the importance of dynamic loading.
Consider the simple frame of Figure 2. The frame is made out of steel (E = 200 GPa).
The frame is loaded by a harmonic load (as shown in Fig. 1a) with amplitude 10 kN and
circular frequency of 10 rad/sec (period of 2B/10 = 0.628 sec, or frequency 1.6 Hz).
We examine the response of the frame when I
o
ranges from 500 cm
4
to 4000 cm
4
with
increments of 500 cm
4
. The values of I
o
are presented in the first column of Table 1. If
the frame is loaded statically by the amplitude of the dynamic load 10 kN, the expected
horizontal deformations of the girder are presented in the second column of Table 1. As
expected we see that as the moment of inertia increases, the frame becomes stiffer and
the horizontal deflection decreases. Note now what happens when the structure is
loaded by the harmonic load. The maximum deformation for each case is presented in the
third column of Table 1. Note that the weakest frame has the smallest deformation,
(this will not always be the case) while the frame with I
o
= 2500 cm
4
has the largest
deformation. How and why this happens is probably not clear at the moment. It has to
do with the phenomenon of resonance with which most of us are somehow familiar. The
exact how and why will become very clear a few weeks into the semester.
Table 1: Response of Frame to Static and Dynamic Loads
I
o
(cm
4
) Static Deflection (mm) Dynamic Amplitude (mm)
500 26.67 6.15
1000 13.33 8.00
1500 8.89 11.43
2000 6.67 20.00
2500 5.33 80.00
3000 4.44 40.00
3500 3.81 16.00
4000 3.33 10.00
Page 3 21 August 2003
mu cu ku F(t)
&& &
+ + =
Figure 3: Setup and dynamic equilibrium of SDOF system
&
e
Figure 4: Force development of spring
and dashpot
2. Single Degree of Freedom (SDOF) Dynamic Structures.
2.1 Definition of a SDOF Dynamic System.
Single Degree of Freedom (SDOF) dynamic systems result from simplifying models of
structures. Even though the relation of a real structure to the most common form of
SDOF (Figure 3) is not obvious, it will become clear very soon.
The SDOF system of Figure 3 consists of a mass m on frictionless wheels that is
connected to an immoveable wall with a spring of constant k and a viscus dashpot with a
constant c. The spring is an element that reacts to stretching or compression with a
force, while the dashpot (which is similar to an automobile shock absorber) is an element
that reacts to stretching or compressive velocity with force. Figure 4a demonstrates
the response of a typical spring. Note that the initial response is linear with slope k.
Upon further loading the response usually becomes nonlinear. At the early stages of this
course only the linear part of the spring response will be considered. The linear response
of spring is described by the simple equation: , where e is the change of length F k e
S
=
of the spring. The stored energy of a deformed linear spring is given by: . U
1
2
k e
2
=
Note that all stored energy of a linear spring is
released upon unloading. That is, a linear spring
does not dissipate energy. Figure 4b
demonstrates the response of a typical
dashpot. Note that a dashpot does not develop
force due to change of length. A dashpot
responds only to velocity. As such, the
dashpots function is to dissipate energy of
dynamic systems. The response of a dashpot
is: . Note that in the example of F c e
D
=
&
Figure 3, where the left end of the spring and
the dashpot is fixed, the displacement of the right end u and u0 are the same as the
Page 4 21 August 2003
F F F F(t) mu cu ku F(t)
I D S
+ + = + + = or
&& &
(1)
mu cu ku 0
&& &
+ + = (2)
mu ku 0
&&
+ = (3)
u Asin t Bcos t
n n
= +
(4)
extensions of the spring e and e0.
Consider now the SDOF system shown in Figure 3. A time dependent force F(t) is applied
on the mass which causes the mass to accelerate and move. Consider an instance in time
where the mass is moving and accelerating with positive values of displacement u, velocity
u0, and acceleration u(. Again it is pointed out that since the left wall is not moving, the
displacement u is also the stretching of the spring, and the velocity u 0 is also the velocity
of change of length of the dashpot. Thus, forces F
S
and F
D
are acting on the mass
directed as shown if Figure 3. The acceleration of the mass u ( is also generating an
inertia force F
I
in the opposite direction of the acceleration. This is also shown in Figure
3. The free body diagram of the mass generates the following dynamic equilibrium
equation:
Note that equation (1) presents a second order linear differential equation with constant
coefficients.
2.2 Solution of a SDOF Dynamic System.
2.2.1 Free Vibrations.
When no external load is applied to the mass of figure 3, we say that the mass is
subjected to Free Vibrations. The differential equation (1) becomes:
This is a homogeneous deferential equation, the solution of which is the subject of this
section.
We shall first consider the case with no damping c = 0. Equation (2) becomes:
The solution of Equation (3) may be given in the following form:
where A and B are arbitrary constants of integration that will be determined from the
Page 5 21 August 2003
u
v
sin t u cos t
o
n
n o n
= +

(5)
u u cos t
o n
=
(6)
T
4
n
7T
4
n
T
n T
2
n
3T
4
n
3T
2
n
5T
4
n
Figure 5: Harmonic free vibration of undamped SDOF system
initial conditions of the differential equation and is called the circular natural
n
2
k
m
=
frequency of the system. Note that substitution of Equation (4) into the differential
equation (3) T
n
will be discussed in more detail later.
To determine the constants A and B, we need the initial conditions of the system. These
are commonly the initial displacement and velocity of the system. That is:
for t = 0 u(0) = u
o
and u 0(0) = v
o
Substituting these conditions in Equation (4) we find that A = v
o
/T
n
and B = u
o
. Thus,
Equation (4), which is the solution of Equation (3) can be rewritten as:
Let us consider the simple case where the initial velocity v
o
= 0. Then, the system
represents the case of a mass attached to a spring, that is pulled off balance to a
distance u
o
and then is freed to vibrate. The solution is expressed by:
The plot of the displacement u(t) vs. time t is presented in Figure 5.
Significant features of this solution include a)The amplitude of the vibration does not
diminish, i.e. there is no loss of energy; b)The vibration is periodic with period T
n
; c) The
vibration period, better known as the natural period of the undamped SDOF is equal to
Page 6 21 August 2003
T
2
2
m
k
n
n
= =

(7)
f
1
T 2
1
2
k
m
n
n
n
= = =

(8)
V
12EI
H
3
=
Figure 6 SDOF Portal frame

The inverse of the natural period is commonly known as the natural frequency of the
undamped SDOF:
The natural frequency is equal to the number of complete cycles in one unit of time. In
periodic vibrations, one cycle is the main form of deformation that continuously repeats
itself. In earthquake engineering the common unit of time is one second. On the other
hand, in automobile engines the unit of time is commonly one minute. For example, we say
the an engine is revving at 3000 rpm (revolutions per minute).
We can also see that the circular natural frequency is equal to 2B times the natural
frequency. Thus, the circular natural frequency is the angle in radians that a vibration
completes in the unit of time. It is understood that a complete vibration cycle is 2B
radians or 360
o
.
It is now time to make a connection of our SDOF dynamic system with a real structural
example. Consider a portal frame as shown in Figure 6.
It is common that the girder of the frame, representing the floor or roof slab, is much
stiffer than the columns (i.e. I
girder
>> I
column
). It is also common, since the slab carries
practically all the mass of the frame, that all the mass is assumed concentrated on the
frame girder. Note that due to the high stiffness of the girder compared to the columns,
the horizontal force P(t) causes no bending on the girder, just a horizontal translation.
Thus, the columns deform with no rotation at the ends, as shown in Figure 6.
Page 7 21 August 2003
mu cu ku 0
&& &
+ + = (2)
u e
st
=
(9)
s
c
m
s
k
m
e 0
2 st
+ +
|
\

|
.
|
=
(10)
s
c
m
s
k
m
0 s
c
2m
c
2m
k
m
2
1,2
2
+ + = =
|
\

|
.
|
(11)
u Ae Be
s t s t
1 2
= + (12)
From equilibrium, the horizontal load P and the corresponding deflection ) are related as
follows: or P = k ), where . It becomes clear that this portal frame is P
24EI
H
3
= k
24EI
H
3
=
identical to the SDOF system of Figure 3 with c = 0, m = mass of slab/girder, and k is
calculated above. Thus, if the frame is pulled out of position by a displacement u
o
and
then allowed to vibrate freely, it will vibrate indefinitely (???!!!) as shown in Figure 5 with
natural period T
n
and natural frequency f
n
as calculated by Equations (7) and (8)
respectively. Of course, indefinite vibrations are not possible. Some dissipation of
energy is expected, which means that the non-dissipating model of Equation (3) is not
entirely realistic.
Let us consider again equation (2):
In this equation, if c > 0 there exists dissipation of energy. This is not an obvious fact.
However, the solution of this differential equation will make it clear.
The solution of this homogeneous linear differential equation can be written in the form:
where s is a constant to be determined later. Substituting this solution into Equation (2)
results in:
This requires that the expression in the parenthesis be zero:
Since there are two values for s (i.e. s
1
and s
2
), the solution (9) becomes the general
solution of Equation (2) as follows:
Page 8 21 August 2003
c
2m
k
m
or c c 2m 2 km
2
n
2
c n
|
\

|
.
|
= = = = = (13)
( )
| |
u (A Bt)e e
c/(2m)t t
= + = + +

u v u t
o o n o
n

(14)
s 1
1,2
2
n
=
|
\

|
.
|

(11a)
( ) ( )
( ) u Ae Be e C cosh t C sinh t
2
n
2
n
n
1 t 1 t
t
1
*
2
*
= + = +
+



( ) ( )
(17)
where A and B are arbitrary constants that depend on the initial conditions of the
problem.
Note from Equation (11) that the type of solution for s
1
and s
2
depends of the value of the
radical (expression in the square root). Three cases are of interest and will be examined
separately: The radical is a) zero, b) greater than zero, c) less than zero.
Case 1 Radical is equal to zero:
What we see here is that for a special value of c, which we call critical coefficient of
damping c
c
the radical becomes equal to zero, which results in s
1
= s
2
= s = -c/(2m). Thus,
the solution to our differential equation becomes:
where a new, very important variable > = c/c
c
has been introduced. > is known as the
damping factor. In the above solution, the initial conditions at t = 0: u = u
o
and u 0 = v
o
have
been introduced.
The solution of equation (11) can then be rewritten as:
This is achieved by using equations: and thus .

= =
c
c
c
m
cr n
2
c
m
n
2
=
Case 2 Radical is greater than zero, i.e. > > 1.
In this case both s
1
and s
2
are real numbers and the solution equation (12) is written:
where .
*
n
2
1 =
Cases 1 and 2 are of limited interest in both structural and geotechnical engineering.
They are called Critically Damped and Over-Damped cases respectively and result in
Page 9 21 August 2003
s i 1 i where 1
1,2
2
n n d d
2
n
=
|
\

|
.
|
= =
(18)
Figure 7: Free vibration of overdamped system
( )
u e Acos t Bsin t
n
t
d d
= +

(20)
motions that are not vibratory as is demonstrated in Figure 7 for an example of a dynamic
system with natural undamped frequency T
n
= 1 rad/sec, and initial displacement u
o
= 1 cm,
and initial velocity v
o
= 0. It becomes clear that the influence of the value of c and thus
indirectly the value of > is one that dissipates energy and thus attenuates the vibration
displacements. It also becomes clear that for values of > $1, there exist no oscillations.
Case 3 Radical is less than zero, i.e. > < 1.
This is the most interesting case, also known as underdamped. Since the radical is less
than zero, the solution described by equations (11) and (11a) becomes complex:
In the above equation, T
d
is the damped circular natural frequency of the dynamic system.
Substitution of Equation (16) into Equation (12) and using the identity that
results in the general solution of the system: e e b i b
a ib a +
= + (cos sin )

Page 10 21 August 2003
u = e u cos t
v u
sin t
n
t
o d
o n o
d
d

+
+ |
\

|
.
|

(
(

(21)
u Ue cos( t a)
n
t
d
=

(22)
Figure 8: Free vibration of underdamped systems.
T
2
while f
2
d
d
d
d
= =

(20)
Note in the above expression, the terms in the parentheses constitute the oscillating
terms with circular frequency T
d
. These terms involve no reduction of amplitude or loss
of energy. This is achieved by the exponential term of the Equation (17).
Using the initial conditions for t = 0: u = u
o
and u 0 = v
o
, we can rewrite Equation (17) as:
Equation (18) can also be written as:
where and . U is the amplitude of oscillation U u
v u
2
o
2 o n o
d
2
= +
+ |
\

|
.
|

tana
v u
u
o n o
d o
=
+

before the decay of amplitude (the exponential term) is applied. Similarly, a is known as a
phase angle.
Examples of free damped
vibrations for circular natural
frequency T
n
= 1 rad/sec,
initial displacement u
o
= 1 cm,
and initial velocity v
o
= 0 are
demonstrated in Figure 8 for
three different values of >.
It becomes clear, that the
smaller the value of >, the
smaller the dissipation of
energy and the loss of
amplitude with time.
The circular natural frequency of each case is equal to T
d
, whereas the corresponding
damped period and damped natural frequency are given by:
Page 11 21 August 2003
Consider again a free damped vibration with initial conditions u(0) = u
o
and u 0(0) = 0. The
solution of the SDOF system becomes a simplified version of Equation (18):
u(t) u e cos t
o
t
d
n
=

At time t= 0, we find that

u(0) = u
o
.
Similarly at some time t
n
= n T
d
where n is an integer
number we shall have
u(t ) u e
n o
nT
n d
=

Note that the displacement u(t
n
) corresponds to an
integer multiple of the period T
d
and thus is also an
amplitude displacement as it is shown in the figure.
We can easily see that:
or
u(0)
u(t )
e
n
n T
n d
=


n
n
n d
ln
u(0)
u(t )
n T =
|
\

|
.
| =
Using the definition of we find that . For small T
2 2
1
d
d
n
2
= =

n
=

2
1
2
n
damping (> < 0.2), the denominator of the expression become practically 1 and we have:
or
n
2 n =

=
|
\

|
.
|
1
2
ln
u(0)
u(t )
n
This is a very useful relation to determine experimentally the damping of a SDOF system.