University of Technology, Sydney

Structural Dynamics and

Earthquake Engineering
49134
Lecture Notes

Faculty of Engineering and Information Technology,

UTS

Prepared by Prof Bijan Samali and Prof Jianchun LI

Preview
Structural vibrations arise from normal human activity and from the operation of
mechanical equipment within buildings, from external traffic, and from wind storms
and earthquakes. Structural vibrations under conditions of normal use were not
usually a problem when working stress methods were used to design conventional
floor and framing systems with traditional construction. This can be inferred from the
minimum guidance provided in existing standards for controlling vibrations in
ordinary buildings. However, structural systems are becoming lighter and more
flexible and have lower damping than before. There has been a reduction in the use of
non structural members and partitions which, in the past, provided additional
stiffening and damping. Methods of structural analysis and design are growing more
refined, the systems are better integrated, and the use of high strength construction
materials and welded or bolted joints is advantages and versatility in multi-purpose
buildings. Limit states design, with its emphasis on structural behaviour, also results,
in some instances, in more flexible structural systems.
These recent developments in design and construction practice may lead to structural
systems in which the motion is perceptible and objectionable unless it is controlled by
proper design. Serviceability requirements in current standards and specifications,
which are little more than rules of thumb based on experience with traditional
construction, often are not sufficient for minimising objectionable motion in flexible
modern structures.

Introduction
Structural dynamics is the discipline concerned with the study of structural response
to time-dependent loads. The conventional methods of analysis and design by which
dynamic loads are replaced by static loads is proven to be unsatisfactory for numerous
types of structures. Although such methods may be adequate from strength point of
view, they are definitely inadequate for serviceability considerations. Structures such
as tall and slender buildings, foot bridges, suspension bridges, lighting towers, flag
poles as well as structural components like thin slabs, etc are all susceptible to
excessive vibrations under certain dynamic loads. Depending upon the structural as
well as loading characteristics, the resulting response could be a few times larger than
those predicted by static analyses.
To determine the dynamic response of a structure to a given dynamic load one must
perform a comprehensive dynamic analysis which is usually complicated, tedious and
time consuming. In addition it is necessary to master the subject and usually there are
no short cuts. The purpose of this subject is to discuss the fundamental characteristics

of a dynamic problem and without getting involved with tedious details provide some
guidelines by which the engineer could determine whether or not his structure is
likely to suffer from excessive vibrations.
Some remedial methods to alleviate excessive motions are briefly discussed as well.

Table of Contents
CHAPTER 1 .................................................................................................................. 4
Essential Characteristics of a Dynamic Problem ....................................................... 4
Equations of Motion .................................................................................................. 6
Types of Loading ....................................................................................................... 7
Solution of the Equation of Motion ........................................................................... 8
Undamped Free-Vibrations ........................................................................................ 9
Damped Free-Vibrations .......................................................................................... 11
Critical Damping ...................................................................................................... 11
Underdamped Systems............................................................................................. 12
Over Damped Systems ............................................................................................. 16
CHAPTER 2 ................................................................................................................ 19
Response to Harmonic Loading ............................................................................... 19
Resonant Response .................................................................................................. 24
Forced Vibration Test to determine Damping ......................................................... 26
Vibration Isolation ................................................................................................... 27
CHAPTER 3 ................................................................................................................ 34
Response to Impulsive Loads .................................................................................. 34
Rectangular Impulse ................................................................................................ 37
Triangular Response ................................................................................................ 38
Response Spectra ..................................................................................................... 39
Approximate Analysis of Impulsive-Load Response .............................................. 41
CHAPTER 4 ................................................................................................................ 44
Response to General Dynamic Loading .................................................................. 44
Duhamel Integral for an Undamped System ............................................................ 44
Numerical Evaluation of the Duhamel Integral for an Undamped System ............. 45
Response of Damped Systems ................................................................................. 49
CHAPTER 5 ................................................................................................................ 51
Analysis of Nonlinear Structural Responses............................................................ 51
Incremental Equation of Equilibrium ...................................................................... 51

CHAPTER 1

Essential Characteristics of a Dynamic Problem

A structural-dynamic problem differs from its static-loading counterpart in two
important respects. The first difference to be noted, by definition, is the time varying
nature of the dynamic problem. Because the load and the response vary with time, it is
evident that a dynamic problem does not have a single solution, as a static problem
does; instead the engineer must establish a succession of solutions corresponding to
all times of interest in the response history. Thus a dynamic analysis is clearly more
complex and time consuming than a static analysis. However, a more fundamental
distinction between static and dynamic problems is illustrated in Fig 1.1. If a simple
beam is subjected to a static load P, as shown in Fig 1.1(a), its internal moments and
shears and deflected shape depend directly upon the given load and can be computed
from P by established principles of force equilibrium. On the other hand, if load p(t) is
applied dynamically, as shown in Fig 1.1(b), the resulting displacements of the beam
are associated with accelerations which produce inertia forces resisting the
accelerations. Thus the internal moments and shears in the beam in Fig 1.1(b) must
equilibrate not only the externally applied load but also the inertia forces resulting
from the accelerations of the beam.

Fig 1.1 Basic difference between static and dynamic load: (a) static loading; (b)
dynamic loading.
Inertia forces which resist accelerations of the structure in this way are the most
important distinguishing characteristic of a structural dynamic problem. In general, if
the inertia forces represent a significant portion of the total load equilibrated by the
internal elastic forces of the structure, then the dynamic character of the problem must
be accounted for in its solution. On the other hand, if the motions are so slow that the
inertia forces are negligibly small, the analysis for any desired instant of time may be
made by static structural analysis procedures even thought the load and response may
be time-varying.
The above mentioned characteristics of a dynamic system can be illustrated by the
governing equation of motion and is best explained by considering the simplest
possible structural system. For this purpose we consider the one-storey structure
idealised as shown in Fig 1.2(a).

The essential physical properties of any linearly elastic structural system subjected to
dynamic loads include its mass, its elastic properties (flexibility or stiffness), its
energy-loss mechanism, or damping, and the external source of excitation or loading.
In the idealisation shown in Fig 1.2(a), we assume that the columns supporting the
block are mass-less and the entire mass of the structure is included in that block; the
block is rigid whereas the columns are flexible to lateral deformation but rigid in the
vertical direction. The structure is assumed to be supported on rigid ground. An
energy absorbing element is also introduced. The viscous damper included in the
structure of Fig 1.2(a) is the most commonly used element of this type.
The basic concepts can be most conveniently developed by studying the dynamics of
the structure of Fig 1.2(a).

a) Idealised one-storey structure, b) Free-body diagram

c) Elastic force-deformation relation, d) Damping force-velocity relation

Fig 1.2 (a); (b); (c); (d)

Equations of Motion
The motion of the idealised one-storey structure due to dynamic excitation will be
governed by an ordinary differential equation. The governing equation, or equation of
motion, is derived here for externally applied dynamic loads.
Figure 1.2(a) shows a linear structure of mass m, lateral stiffness k, and viscous
damping c subjected to an externally applied dynamic force p(t). This notation
indicates that the force p varies with time t. Under the influence of such a force, the
rigid block displaces in the lateral direction by an amount u(t). Because the force p
varies with time, so does the displacement u.
The various forces acting on the mass at some instant of time are shown in a freebody diagram of the mass (Fig 1.2(b)). These include the external force p(t), the
elastic resisting force fs, the damping force fD, and the inertia force fI. The elastic and
damping forces act to the left because they resist the deformation and velocity,
respectively, which are taken as positive to the right. The inertia force also acts to the
left, opposite to the direction of positive acceleration. At each instant of time, the
mass is in equilibrium under the action of these forces at that time. From the free body
diagram, this condition of dynamic equilibrium is
f I f D f S p(t )

(1.1)

The inertia, damping, and elastic forces are next expressed in terms of u(t) and related
quantities. For a linear structure, the elastic force is

f S ku

(1.2)
(Fig 1.2c)

where k is the lateral stiffness of the structure and u is the interstorey displacement.
The damping force is
f D cu

(1.3)
(Fig 1.2d)

Where c is the damping coefficient for the structure and u is the interstorey velocity.
As shown in Figs 1.2(c) and 1.2(d), the elastic force is proportional to the relative
displacement and the damping force to the relative velocity. The inertial force
associated with the mass m undergoing an acceleration is
f I mu

(1.4)

Substitution of Eqs 1.2, 1.3, and 1.4 into Eq 1.1 results in

mu cu ku p(t )

(1.5)

This is the equation of motion governing the deformation u(t) of the idealised
structure of Fig 1.2(a) subjected to an external dynamic force p(t).

Under gradual application of load p(t), the mass undergoes no acceleration or velocity
and therefore the equation of motion (Eq 1.5) reduces to
ku p(t )

(1.6)

which is the familiar stiffness-displacement-loading relationship. From Eq (1.6) one

can easily calculate the structural response u from
u

p( t )
k

(1.7)

However, to calculate the response u from Eq (1.5) one is required to solve an

ordinary 2nd order differential equation for a given set of initial conditions with the
solution u(t). For dynamically sensitive structures (where inertia and damping forces
are not negligible) the solution obtained from Eq (1.5) will differ significantly from
the solution obtained from Eq (1.7).

Types of Loading
In the equation of motion (Eq 1.5), p(t) is a dynamic (time-varying) load, i.e., its
magnitude, direction, or position varies with time.
Two basically different approaches are available for evaluation structural response to
dynamic loads: deterministic and non deterministic. The choice of method to be used
in any given case depends upon how the loading is defined. If the time variation of
loading is fully known, even though it may be highly oscillatory or irregular in
character, it will be referred to as a prescribed dynamic loading; and the analysis of
the response of any specified structural system to a prescribed dynamic loading is
defined as a deterministic analysis. On the other hand, if the time variation is not
completely known but can be defined in a statistical sense, the loading is termed a
random dynamic loading; a non deterministic analysis correspondingly is the analysis
of response to a random dynamic loading. Analysis of structures due to random
loading is beyond the scope of this subject. Prescribed dynamic loads are briefly
discussed here. From an analytical standpoint, it is convenient to divide prescribed or
deterministic loadings into two basic categories, periodic and non-periodic. Some
typical forms of prescribed loadings and examples of situations in which such
loadings might be developed are shown in Fig 1.3.
As is indicated in Figs 1.3(a) and 1.3(b), periodic loadings are repetitive loads which
exhibit the same time variation successively for a large number of cycles. The
simplest periodic loading is the sinusoidal variation shown in Fig 1.3(a), which is
termed simple harmonic; such loadings are characteristic of unbalanced-mass effects
in rotating machinery. Other forms of periodic loading; eg, those caused by
hydrodynamic pressures generated by a propeller at the stern of a ship are more
complex. However, by means of Fourier analysis any periodic loading can be
represented as the sum of a series of simple harmonic components; thus, in principle,
the analysis of response to any periodic loading follows the same general procedure.

Non periodic loadings may be either short-duration impulsive loadings or longduration general forms of loads. A blast or explosion is a typical source of impulsive
load. An earthquake or wind loading is a typical long-duration loading.

Characteristics and sources of typical dynamic loadings; (a) simple harmonic; (b)
complex; (c) impulsive; (d) long-duration.
Figure 1.3
Solution of the Equation of Motion
The solution of Eq (1.5) will be obtained by considering first the homogeneous
equation with the right side set equal to zero:

mu cu ku 0

(1.8)

Motions taking place with the applied force set equal to zero are called Free
Vibrations, and it is the free vibration response of the system which we now wish to
examine.

The solution of Eq (1.8) is of the form

u( t ) G est

(1.9)

substituting Eq (1.9) into Eq (1.8) leads to

(ms 2 cs k)Gest 0

(1.10)

After dividing by mGest and introducing the notation

2

k
m

(1.11)

Eq (1.10) becomes
s2

c
s 2 0
m

(1.12)

The value of s which can be derived from this expression depends on the value of c;
thus the type of motion represented by Eq (1.9) will depend on the damping in the
system.

Undamped Free-Vibrations
If the system is undamped; i.e. c = 0, it is evident that the value of s given by Eq
(1.12) is

s i

(1.13)

Thus the response given by Eq (1.9) is

u(t ) G1e it G 2 e it

(1.14)

In which G1 and G2 are arbitrary constants. Eq (1.14) can be put into a more
convenient form by introducing Eulers equation
e it cos t i sin t

(1.15)

The result may be written as

u(t ) A sin t B cos t

(1.16)

In which constants A and B may be expressed in terms of the initial conditions, i.e.
the displacement u(0) and velocity u (0) at time t = 0, which initiated the free

vibration of the system. It is easily seen that u(0) = B and u (0) =A; thus Eq (1.16)
becomes

u (t )

u (0)

sin t u (0) cos t

(1.17)

This solution represents a simple harmonic motion (SHM) and is illustrated

graphically in Fig 1.4. The quantity is the circular frequency or angular velocity of
the motion; it is measured in radians per unit of time. The cyclic frequency f, which is
usually referred to merely as the frequency of the motion, is given by
f

(1.18)

and its reciprocal is called the period T,

1
f

(1.19)

The motion represented by Eq(1.17) also can be expressed in the form

u(t ) cost

(1.20)

where is the amplitude of the motion and is given by

u(0)2 u(0)

(1.21)

is the phase angle and is given by

u (0)

tan 1

u
(
0
)

(1.22)

10

Fig 1.4
Damped Free-Vibrations
If damping is present in the system, the solution of Eq (1.12) which defines the
response is
2

c
c
2


2m
2m

(1.23)

Three types of motions are represented by this expression, according to the quantity
under the square-root sign is either positive, negative, or zero. It is convenient to
discuss first the limiting case, when the radical vanishes. This is called the criticaldamping condition.

Critical Damping
If the radical in Eq (1.23) is set equal to zero, it is evident that c/2m = ; thus the
critical damping value Cc is

C c 2m

(1.24)

Then the value of s in Eq (1.23) is

s

c

2m

(1.25)

and the response given by Eq (1.9) is

11

u(t ) (G1 G 2 t )e t

(1.26)

in which the 2nd term is multiplied by t because only a single value of s is available in
the solution, Eq (1.25).
Introducing the initial conditions in Eq (1.26) leads to the final form of the critically
damped response equation.
u(t ) u(0)(1 t ) u (0)t e t

(1.27)

Eq (1.27) is illustrated graphically in Fig 5. It should be noted that the free response of
a critically damped system does not include oscillation about the zero-deflection
position; instead the displacement returns to zero in accordance with the exponential
decay term of Eq (1.27). One useful definition of the critically damped condition is
that it is the smallest amount of damping for which no oscillation occurs in the free
response.

Fig1.5 Free-Vibration Response with Critical Damping

Underdamped Systems
If the damping is less than critical, it is evident from Eq (1.24) that c < 2m and thus
the radical in Eq (1.23) must be negative. To evaluate the free-vibration response in
this case, it is convenient to express the damping as a ratio to the critical damping
value; thus
c
c
(1.28)

C c 2m
In which is called the damping ratio. Introducing Eq (1.28) into Eq (1.23) leads to

2 2

(1.29)

or changing the sign of the radical and introducing a new symbol D gives
S iD

(1.30)

where
12

D 1 2

(1.31)

The quantity D is called the damped vibration frequency. One can define TD, damped
vibration period, fD, damped vibration cyclic frequency in Hz, with the following
relationships
fD

D
2
1
, TD

2
D f D

(1.32)

For most structures damping ratio is less than 10%. In this range the damped
frequency D differs very little from the undamped frequency , as may be noted in
Eq (1.31). To examine the influence of damping on frequency, it is convenient to
remember that a plot of the ration of damped to undamped frequency D/ versus the
damping ratio is a circle of unit radius, as shown in Fig 1.6.
The free-vibration response of an underdamped system can be evaluated by
substituting Eq (1.30) into Eq (1.9); thus

u(t ) G1e t iD t G 2 e t iD t e t G1e iD t G 2 e iD t

(1.33a)

The term in brackets represents simple harmonic motion (compare with Eq (1.14))
thus Eq (1.33a) can be written more conveniently as
u(t ) e t A sin D t B cos D t

(1.33b)

Finally, when the initial conditions u(0) and u (0) are introduced, the constants of Eq
(1.33b) can be evaluated as follows

Fig 1.6

13

u(0) e0 (0 B) B u(0)
u (t ) e t A sin D t B cos D t e t AD cos D t BD sin D t
u (0) B AD u (0) u(0) AD
u (0) u (0)
A
D

Substituting for A and B in Eq (1.33b), we obtain

u (0) u (0)

u ( t ) e t
sin D t u (0) cos D t
D

(1.34)

Alternatively, the response expression can be written in the form

u(t ) e t cosD t

(1.35)

in which
2

u (0) u (0)
2

u (0)

(1.36)

u (0) u (0)
D u (0)

(1.37)

tan 1

A plot of the response of an underdamped system to an initial displacement u(0) and

initial velocity u (0) is shown in Fig 1.7. Let us examine from Fig 1.7 how one can
estimate the structural damping .
Consider any two successive positive peaks shown in Fig 1.7 that is, un and un+1. From
Eq (1.35), the ratio of these two successive values are given by

un

exp 2
u n 1

(1.38)

Taking natural logarithm (ln) of both sides of Eq (1.38) gives

ln

un

2
u n 1
D

(1.39)

Using Eq (1.31) we get

14

(1.40)

1 2

Fig 1.7 Effect of Damping on Free Vibration

is called the logarithmic decrement.

For low damping Eq (1.40) can be approximated by
2

(1.41)

In such cases, Eq (1.38) can be written as a series expansion

un
2 ...
e e 2 1 2
u n 1
2!
2

(1.42)

For low values of sufficient accuracy can be obtained by retaining only the first two
terms in the series, in which case

u n u n 1
2u n 1

(1.43)

To illustrate the accuracy of Eq (1.43), the ratio of the exact value of as given by
Eq (1.39) to the approximate value as given by Eq (1.43) is plotted versus the
approximate value in Fig 1.8.

15

Fig 1.8
For lightly damped systems, greater accuracy can be obtained in evaluating the
damping ratio by considering response peaks which are several cycles apart. Say m
cycles, then
ln

un

2m
u n m
D

(1.44)

which can be simplified for very low damping to the approximate relation

u n u n m
2mu n m

(1.45)

When one is observing damped free-vibrations experimentally, a convenient method

for estimating the damping ratio is to count the number of cycles required to give a
50% reduction in amplitude. The relationship to be used in this case is presented
graphically in Fig 1.9. As a quick rule of thumb it is convenient to remember that for
10% critical damping, the amplitude is reduced by 50% in one cycle.
Over Damped Systems
Although structural systems having greater than critical damping are not encountered
in normal conditions, we consider it for completeness. For > 1, Eq (1.23) may be
written as
S 2 1

(1.46)

in which = 2 1
Substituting Eq (1.46) into Eq (1.9) leads to
t B cosh
t
u(t ) e t A sinh

(1.47)

A and B could be evaluated by considering the initial conditions. It is noted from Eq

(1.47) that the response of an overdamped system is not oscillatory; it is similar to the
motion of the critically damped system of Fig 1.5, but the return towards the neutral
position is slowed as the damping ratio is increased.

16

Fig 1.9 Number of Cycles Required to Reduce the Free Vibration Amplitude by 50%
plotted as a function of Damping Ratio

Example:
A one-storey building is idealised as a rigid girder supported by weightless columns,
as shown in Fig 1.10. In order to evaluate the dynamic properties of this structure, a
free-vibration test is made, in which the roof system (rigid girder) is displaced
laterally by a hydraulic jack and then displace the girder 5mm. During the jacking
operation, it is observed that a force of 90 kN is required to displace the girder 5 mm.
After the instantaneous release of this initial displacement, the maximum
displacement on the return swing is only 4mm and the period of this displacement
cycle is T = 1.40 seconds. From these data, the following dynamic behaviour
properties are to be determined:
1. Effective Weight of the Girder:
2
m
W
T
2
2

k
kg
3
T
1.4 90 10
W kg
9.81 8,766734 N
2
2 0.005
8,766.734 kN
2

2. Frequency of Vibration:
1
1

0.714 Hz
T 1.40
rad
2f 4.48
s

17

3. Damping Properties:
5
0.223
4

0.223
Damping ratio:

0.0355 3.55 %
2
2
Damping coefficient: C Cc 2m

Logarithmic decrement: ln

kN
8766.734
C 0.0355 2
4.48 284.37
m/s
9.81

Damped frequency: D 1 2 0.999

Fig 1.10

18

CHAPTER 2

Response to Harmonic Loading

Undamped System
a) Complementary Solution
We now assume that the system of Fig 2.1a is subjected to a harmonically varying
load p(t) of amplitude p0 and circular frequency . In this case the differential
equation of motion becomes
mu(t ) cu ( t ) ku (t ) p0 sin t
(2.1)
We examine the undamped system first. The governing equation, Eq (2.1), reduces to
mu(t ) ku(t ) p0 sin t

(2.2)

The complementary solution of this equation is the free-vibration response of Eq

(1.16), i.e.
u(t ) A sin t B cos t

(2.3)

b) Particular Solution
The general solution includes also the particular solution, i.e. the specific behaviour
generated by the form of the dynamic loading. The response to the harmonic loading
can be assumed to be harmonic and in phase with the loading; thus
u p (t ) G sin t

(2.4)

G is an arbitrary constant. Substituting Eq (2.4) into Eq (2.2) leads to

m 2 G sin t kG sin t p0 sin t

(2.5)

sin t 0

or

m 2 G kG p 0

or

or

p0
2 p0
k
G1 2
G
2
k
1 2

p
m 2
GG 0
k
k

(2.6)

G is the amplitude of the response.

19

p0
k
Eq (2.6) can be rewritten as G
(2.7)
(1 2 )
in which represents the ratio of the applied load frequency to the natural freevibration frequency; i.e.

(2.8)

c) General Solution
The general solution to the harmonic excitation of the undamped system is then given
by the combination of the complementary solution and the particular solution, in
which the value of G is given by Eq (2.7); thus

u ( t ) u c ( t ) u p ( t ) A sin t B cos t

p0 1
sin t
k 1 2

(2.9)

in Eq (2.9) constants A and B depend upon the initial conditions. For a system starting
at rest I.e. u(0) = u (0) = 0, we get

p 0 1
k 1 2

B0

(2.10)

Then the response given by Eq (2.9) becomes

u(t )

p0 1
sin t sin t
k 1 2

(2.11)

where p0/k = ust = static displacement, i.e. displacement which would be produced by
the load p0 applied statically.
1/(1 2)

= magnification factor (MF), representing

amplification effect of harmonically applied load.

sin t

= response component at frequency of the applied load =

steady-state response, directly related to the load.

sin t

= response component at natural vibration frequency = freevibration effect induced by the initial conditions.

dynamic

Since in a practical case, damping will cause the last term to vanish eventually, it is
called the transient response. For this hypothetical, undamped system, of course,
this term would not vanish and continues indefinitely.
Response Ratio, R(t)

20

A convenient measure of the influence of the dynamic character of the loading is

provided by the ratio R(t) of the dynamic response to the displacement that would be
produced by the static application of the same load:
u(t) u(t)
(2.12)

p0
u st
k
The response ratio resulting from harmonic loading of an undamped system is
(starting from rest)
R (t )

R (t )

1
sin t sin t
1 2

(2.13)

R(t) is portrayed graphically in Fig 2.1 for = 2/3

Fig 2.1 Response to Harmonic Load from at-rest initial conditions: (a) steady state;
(b) transient; (c) total R(t)

21

Damped System
Returning to equation of motion including damping, Eq (2.1), dividing by m, and
noting that c/m = 2 leads to
u( t ) 2 u ( t ) 2 u ( t )

p0
sin t
m

(2.14)

The complementary solution of this equation is the damped free-vibration response

given by Eq (1.33b) (assuming that the structure is less than critically damped, as is
the case for all practical structures)
(2.15)
u c (t ) e t A sin D t B cos D t
The particular solution to this harmonic loading is of the form
u p (t ) G1 sin t G 2 cos t

(2.16)

in which the 2nd term is required because, in general, the response of a damped system
is not in phase with the loading.
Substitution of Eq (2.16) into Eq (2.14) and separating the multiples of sin t from
the multiples of cos t leads to

p0

2
2
sin t
G1 G 2 (2) G1 sin t
m

G 2 G (2) G 2 cos t 0

2
1
2

(2.17)

which leads to (by dividing by 2):

p0

2
G1 1 G 2 2
k

2
G 1 G 2 0
1
2

(2.18)

Solving these equations (Eq (2.18)) simultaneously leads to

G1

p0
1 2
,
k 1 2 2 2 2

G2

p0
2
k 1 2 2 2 2

(2.19)

Introducing these expressions into the particular solution Eq (2.16) and combining
with the complementary solution finally yields the general solution:

u ( t ) e t A sin D t B cos D t

p0
1
1 2 sin t 2 cos t
2
2
2
k 1 2
(2.20)

22

The first term in Eq (2.20) represents the transient response to the applied loading.
The constants A and B could be evaluated for any given initial conditions, but this
term damps out quickly and generally is of little interest. The 2nd term in Eq (2.20) is
the steady-state response, at the frequency of the applied loading but out-of-phase
with it. This steady-state response can be written in the following form
u(t ) sint

(2.21)

where

p0
k

(2.22)

1 2
2 2

represents the amplitude of the steady-state response. The phase angle by which the
response lags behind the applied load is given by

tan 1

2
1 2

(2.23)

The ratio of the resultant response amplitude to the static displacement which would
be produced by the force p0 will be called the dynamic magnification factor or
response factor, D, thus
D

p0
k

1 2
2 2

(2.24)

Response factor, D, varies with the frequency ratio and the damping ratio . Plots
of these relationships are shown in Fig 2.2.

Fig 2.2 Response factor for a one-storey structure subjected to harmonic forces

23

Resonant Response
From Fig 2.2 it may be noted that the peak steady-state response occurs at a frequency
ratio near unity for lightly damped systems. The condition when the frequency ratio is
unity ( = 1), i.e. when the frequency of the applied load equals the natural vibration
frequency, is called RESONANCE. From Eq (2.13) it is apparent that the steadystate response of an undamped system tends toward infinity at resonance (see Fig
2.3a). A more general result may be obtained from Eq (2.24), which shows that for
resonance ( = 1) the dynamic magnification factor D is inversely proportional to the
damping ratio:

1
(2.25)
2
However, although it is close to the maximum, this does not represent the maximum
response for any damped system; the frequency ratio for maximum response may be
found by differentiating Eq (2.24) with respect to and equating to zero. For
D1

practical structures having damping ratios < 1/ 2 , the peak-response frequency is

found to be

peak 1 2 2

(2.26)

and the corresponding peak response is

D max

(2.27)

2 1 2

For reasonable amounts of damping, the difference between Eq (2.27) and Eq (2.25)
is negligible.
For a more complete understanding of the nature of the resonant response of a
structure to harmonic loading, it is necessary to consider the general response
equation, Eq (2.20). At the resonant exciting frequency ( = 1) this equation becomes

u ( t ) e t A sin D t B cos D t

p 0 cos t
k 2

(2.28)

Assuming that the system starts from rest [u(0) = u (0) = 0], the constants are

p0
p
1
0
,
k 2D
k 2 1 2

p0 1
k 2

(2.29)

Thus Eq (2.28) becomes

24

1 p 0 t

e
u(t)
sin D t cos D t cos t
1 2

2 k

(2.30)

For the amounts of damping to be expected in a structural system, the sine term in this
equation will contribute little to the response amplitude, moreover, the damped
frequency is nearly equal to the undamped frequency. Thus the response ratio R(t) in
this case is approximately
R (t)

u ( t ) 1 t

e
1 cos t
p0
2
k

(2.31)

Eq (2.31) is illustrated graphically in Fig 2.3b.

Fig 2.3: Response ratio versus time

We can now summarise our findings (refer to Fig 2.2) as follows:

For close to zero, umax is about the same as ust; (D = 1), that is, the dynamic
effects are negligible if the forcing frequency is much small than the natural
frequency of the structure. For small values of , the maximum displacement

25

is controlled by the stiffness of the system with little effect of mass or

damping.

When = 1, i.e. at resonance frequency, D = 1/2 , that is, the response factor
is inversely proportional to the damping ratio of the forcing frequency is the
same as the natural frequency of the structure. For = 1 and close to 1, the
response factor is controlled by the damping ratio with negligible influence
of mass or stiffness.

The response factor is essentially independent of damping and approaches

zero as the forcing frequency becomes much higher than the natural
frequency of the structure. At high forcing frequencies the maximum
displacement depends primarily on the mass.

At 1 2 , D is maximum and is equal to 1/ 2 1 2

Forced Vibration Test to determine Damping

If one plots / versus D (by varying k / m and measuring umax and

computing ust), a graph similar to Fig 2.4 will be obtained. The maximum
amplitude 2 (i.e. at D = Dmax/ 2 ) is called the half power point. The width of
the plot at this point is equal to 2 . By measuring this width and dividing it by 2, one
can estimate the damping ratio. Smaller damping ratios are associated with sharp
peaks of D versus around 1 and larger damping ratios are associated with wide
and flat peaks around 1

26

Fig 2.4 Evaluation of damping from forced vibration tests

Vibration Isolation
Two different classes of problem may be identified in which vibration isolation may
be necessary:
1. Operation equipment may generate oscillatory forces which could produce
harmful vibrations in the supporting structure.
2. Sensitive instruments may be supported by a structure which is vibrating
appreciably.
The first situation is illustrated in Fig 25. A rotating machine produces an oscillatory
vertical force p0 sin t due to unbalance in its rotating parts. If the machine is mounted
on a SDOF spring-damper support system, as shown, its steady-state displacement
response is given by
u(t)

p0
D sin t
k

(2.32)

where D is defined by Eq (2.24)

The force exerted against the base by the spring supports is

27

f s ku(t ) p0 D sint

(2.33)

At the same time, the velocity of the motion relative to the base is
u ( t )

p0
D cos t
k

(2.34)

which leads to a damping force on the base:

f D cu (t )

cp 0 D
cos t 2 p0 D cos t
k

(2.35)

Since the damping force is 90 out of phase with the spring force, it is evident that the
amplitude of the base force f is
f max f s2,max f D2,max p 0 D 1 2

(2.36)

The ratio of the maximum base force to the applied-force amplitude, which is known
as transmissibility (TR) of the support system, thus is given by
TR

f max
2
D 1 2
p0

(2.37)

Fig 2.5 SDOF vibration isolation system (applied loading)

A plot of the transmissibility as a function of the frequency ratio and damping ratio is
shown in Fig 2.6. Although it is similar to Fig 2.2, all curves pass through the same
point at a frequency ratio of 2 . This difference from Fig 2.2 is due, of course, to
the influence of the damping force. Because of this characteristic, it is evident that
damping tends to reduce the effectiveness of a vibration-isolation system for
frequencies greater than this critical ratio.

28

Fig 2.6 Vibration transmissibility ratio (applied load or displacement)

The second type of situation in which vibration isolation is important is illustrated in

Fig 2.7. The mass m to be isolated is supported by a spring-damper system on a
foundation slab which is subjected to harmonic vertical motions. The displacement of
the mass relative to the base is given by

u(t ) u g0 2 D sint

(2.38)

But the total motion of the mass is

u t (t ) u g0 1 2 D sin t
2

(2.39)

where ug0 is the amplitude of support displacement.

Fig 2.7 SDOF vibration isolation system (support excitation)

The transmissibility (TR) in this situation is defined as the ratio of the amplitude of
motion of the mass to the base-motion amplitude, it can be seen that the expression
for transmissibility is the same as that given by Eq (2.37). This can be expressed

29

mathematically as TR

t
u max
2
D 1 2
u g0

(2.40)

and Fig 2.6 serves to define the 3effectiveness of vibration isolation systems for both
basic SDOF isolation situations.
For the design of a vibration-isolation system, it is convenient to express the
behaviour of the system in terms of its isolation effectiveness rather than the
transmissibility, where the effectiveness is defined as 1-TR. Also, when it is noted in
Fig 2.6 that an isolation system is effective only for frequency ratios 2 and that
damping is undesirable in this range, it is evident that the isolation mounting should
have very little damping. Thus it is acceptable to use the transmissibility expression
for zero damping.

TR

1
,
2
1

1 TR

2 2
2 1

(2.41)

in which it is understood that 2 . Finally it may be noted that

W

2
m
2 2 2 st
2

k
kg
g

(2.42)

where g is the acceleration due to gravity and st = w/k is the deflection that the
weight of the system to be isolated will produce on the vibration-mounting devices.
Thus, it is evident that the effectiveness of the mounting system can be expressed in
terms of the frequency of the input motion and this static deflection value st.
Solving Eq (2.41) for the frequency ratio in terms of the isolation effectiveness leads
to

2 1 TR
1 1 TR

(2.43)

Eq (2.43) can now be expressed in terms of the input frequency ( f 2 ) and the
static deflection st.

f 3.13

1 2 1 TR
st 1 1 TR

(2.44)

where f is in Hz and st is in inches. A plot of Eq (2.44) is presented in Fig 2.8.

Knowing the frequency of the impressed excitation, one can determine directly from
this graph the support-pad deflection st required to achieve any desired level of
vibration-isolation, assuming that the isolators have little damping.

30

Fig 2.8 Vibration Isolation Design Chart

Example 1:
Deflections sometimes develop in concrete bridge girders due to creep, and if the
bridge consists of a long series of identical spans, these deformations will cause a
harmonic excitation in a vehicle travelling over the bridge at constant speed. Of
course, the springs and shock absorbers of the car are intended to provide a vibrationisolation system which will limit the vertical motions transmitted from the road to the
occupants. Above figure shows a highly idealised model of this type of system, in
which the vehicle weight is 18kN and its spring stiffness is defined by a test which
showed that adding 500N caused a deflection of 2mm. The bridge profile is
represented by a sine curve having a wavelength (girder span) of 12m and a single
amplitude of 30mm. From these data it is desired to predict the steady-state vertical
motions in the car when it is travelling at a speed of 65km/h, assuming that damping
is 40% of critical.

31

The transmissibility for this case is given by Eq (2.40), hence the amplitude of vertical
motion is

u g0 D 1 2 u g0
2

t
max

1 2

1 2
2 2

when the car is travelling at 65km/hr = 18m/sec, the excitation period is

Tp

12
0.665 s
18

while the natural period of the vehicle is

2
W
2
2

kg

18,000
0.538 s
500
9.81
0.002

Hence / T / T p = 0.538/0.665 = 0.81 and with 0.4 , the response

amplitude is

1 2 0.4 0.81

t
max

30

1 0.81 2 0.4 0.81

2 2

30 1.624 48.7 mm

It is of interest to note that if there was no damping in the vehicle, 0 , the

amplitude would be
t
u max
u g0

1
30

87.2 mm
2
1
1 0.656

This demonstrates the important function of the shock absorbers in limiting the
motions resulting from waviness of the road surface.

Example 2:
A rotating machine weighing 100kN is known to develop vertically oriented harmonic
forces having an amplitude of 2500N at its operating speed of 40rad/sec. In order to
limit the vibrations induced in the building in which this machine is to be installed, it
is to be supported by a spring at each corner of its rectangular base. The designer
wants to know what support spring stiffness will be required to limit to 400N the total
harmonic force transmitted from the machine to the building.
The transmissibility in this case is 400/2500 = 0.16, Utilising Eq (2.41);
1
2 1 6.25
TR

32

from which

2 7.25 2

W
kg

or

k 2

W
kN
2250
7.25g
m

Thus the stiffness of each of the four support springs is

k 2250
kN

562.5
4
4
m

It is of interest to note that the static deflection caused by the weight of the machine
on these spring supports is
st

100
0.044 m 44 mm
2250

33

CHAPTER 3
Response to Impulsive Loads
Impulsive loads are generally short in duration. Impulsive or shock loads are of great
importance in the design of certain classes of structural systems eg vehicles. Damping
has much less importance in controlling the maximum response of a structure to
impulsive loads than for periodic and harmonic loads. The maximum response to an
impulsive load will be reached in a very short time, before the damping forces can
absorb much energy from the structure. For this reason we will only consider the
undamped response of a SDOF system to impulsive loads.
Impulsive loads may be of different shapes and durations. An arbitrary impulsive load
is shown below (Fig 3.1a). Some impulsive loads can be expressed by simple
analytical functions such as a Sine-Wave Impulse. For such loads, closed-form
solutions of the equations of motion can be obtained. A sine-wave impulse is shown
below (Fig 3.1b). As shown in Fig 19b, the response will be divided into two phases,
corresponding to the interval during which the load acts, followed by the freevibration phase.
Phase I: During this phase the structure is subjected to harmonic loading, starting
from rest. The undamped response, including the transient as well as the steady-state
term, is given by Eq (2.11)
0 t t1

u(t )

p0 1
sin t sin t
k 1 2

(3.1)

Phase II: The free-vibration motion which occurs during this phase depends on the
displacement u(t1) and velocity u ( t1) existing at the end of phase I, and may be
expressed as follows
For

t t t1 0

u(t )

u ( t1 )
sin t u ( t1 ) cos t

(3.2)

in which the new time variable t t t1 has been introduced for convenience. The
magnitude of dynamic response which results from this impulsive load depends on the
ratio of the load duration to the period of vibration of the structure (t1/T). The
response ratio R(t) = u(t)/P0/k for t1/T = is shown below in Fig 3.2. In general, the
maximum response produced by the impulsive load rather than the complete history is
of most interest to the structural engineer. The time when the peak response occurs
can be determined by differentiating Eq (3.1) with respect to time and equating to
zero; thus

p
du ( t )
1
cos t cos t
0 0
dt
k 1 2

(3.3)

cos t cos t

(3.4)

34

t 2n t

(3.5)

n 0,1,2,3,...

(a)

(b)
Fig 3.1
This expression is valid, of course, only so long as t , that is if the maximum
response occurs while the impulsive load is acting.
For the most interesting loading condition, when the load frequency approaches the
free-vibration frequency, i.e. where , the time of maximum response is
t

2

1

(3.6)

35

Fig 3.2
The maximum-response amplitude can then be obtained by introducing Eq (3.6) into
Eq (3.1); the result is valid only if t , which will be the case if 1 , that is

u max

p0 1
2

sin
sin
2

k 1
1

2


1

(3.7)

For > 1 ( ) the maximum response occurs during the free-vibration phase
(phase II). The initial velocity and displacement for this phase are given by
introducing t1 = into Eq (3.1)

p0

cos

u ( t 1 )
k 1 2

u ( t ) p 0 1 0 sin

1
k 1 2

The amplitude of this free-vibration motion is given by

(3.8)

p0
2

u ( t1 )
2

u ( t1 ) k 2 2 2 cos

(3.9)

Hence the dynamic magnification factor for this condition is

u
2

1 ; t t1
D max
cos
2
p0
1
2
k

(3.10)

36

Example:
As an example of the maximum-response analysis for a long-duration sine impulse,
where the maximum occurs while the load is acting, consider the case where = 2/3
2
( 2 / 3 or t1 = 3/4T). For this case Eq (3.6) gives t
4 / 5 and with
1 3/ 2
this value substituted in Eq (3.1), the dynamic magnification factor is

1 4
2
6
sin sin 1.77
4
3
5
1 5
9

A specific example of a short-duration impulse, where the peak response occurs

during the free-vibration phase, is the case where 4 / 3 ( 4 / 3 or t1+ 3/8T).
In this case the dynamic magnification factor is found from Eq (3.10) to be

4
2

3
D cos
1.31
16
4

1
2
9
3

Rectangular Impulse
A rectangular impulsive load is shown below (Fig 3.3). Again the response will be
divided into the loading phase and the subsequent free-vibration phase.
Phase I: The suddenly applied load which remains constant during phase I is called a
step loading. The general solution for this situation is given by
u(t)

p0
1 cos t
k

0 t t1

Phase II: The free-vibration during phase II is again given by Eq (3.2)

u ( t1 )
u(t )
sin t u ( t1 ) cos t
t t t1 0

(3.11)

(3.12)

For this rectangular impulse, the maximum response will always occur in phase I if t 1
T/2 and the dynamic magnification factor D is 2 for this case. For shorter-duration
loadings, the maximum response will occur during the free-vibration of phase II, and
the response amplitude is given by
2

u max

u ( t1 )
2
u ( t 1 )

With u (t )

(3.13)

P0
sin t and 2 / T this becomes
k

37

u max

p0
p
2
2
2
t1 cos 2
t1 sin 2
t1 0
1 2 cos
k
T
T
T
k

21 cos
t1
T

(3.14)

From which
D

u max
t
2 sin 1
p0
T
k

t1 1

T 2

(3.15)

Fig 3.3 Rectangular Impulse

Triangular Response
The last impulse loading to be analysed here is the decreasing triangular response
shown below (Fig 3.4)

Fig 3.4 Triangular Impulse

38

Phase I: The loading during this phase is P0 (1 t/t1). The general solution to this type
of loading is given by

u(t)

p0
k

sin t

cos t 1
t1
t1

(3.16)

P0
t
(1 ) to the
k
t1
homogeneous solution uh(t) = A sin t + B cos t and evaluating constants A and B
to satisfy zero initial conditions (i.e. u (0) = u (0) = 0).

which is obtained by adding the particular solution up(t) =

Phase II: Evaluating Eq (3.16) and its first derivative at the end of phase I (t = t 1)
gives

u(t1 )

p0
k

sin t1

p cos t1
t

cos t1 , u ( t1 ) 0
sin t1
k t1
t1
t1

(3.17)

which can be substituted into Eq (3.2) to obtain the free-vibration response in phase
II. The maximum values of these response functions are found, as for the other
examples, by evaluating them for the times at which the zero velocity condition is
achieved. For loadings of very short duration (t1/T < 0.4) the maximum response
occurs during the free-vibrations of phase II; otherwise it occurs during the loading
u max
interval (phase I). Values of the dynamic magnification factor D =
computed
( P0 / k )
for various loading durations are presented in the following table
Dynamic Magnification Factor for Triangular Impulse Loading
t1/T
D

0.20
0.60

0.40
1.05

0.50
1.19

0.75
1.38

1.00
1.53

1.50
1.68

2.00
1.76

Response Spectra
In the expressions derived above the maximum response produced in an undamped
SDOF structure by a given form of impulsive loading depends only on the ratio of the
impulse duration to the natural period of the structure t1/T. Thus it is convenient to
plot the dynamic magnification factor D as a function of t1/T for various forms of
impulsive loading Fig 3.5). These are known as the displacement-response spectra, or
merely the response spectra, of the impulsive loads. These are used to predict the
maximum effect to be expected from a given type of impulsive loading acting on a
simple structure.

39

Example:
As an example of the use of response spectra in evaluating the maximum response of
a SDOF structure to an impulsive load, consider the system shown in Fig 3.6, which
represents a single-storey building subjected to a blast load. For the given weight and
column stiffness of this structure, the period of vibration is
T

2
W
2600
2
2
0.077 s

kg
1.75 106 9.81

Fig 3.5 Displacement-response spectra (shock spectra) for three types of impulse

Fig 3.6

40

The impulse-length ratio thus is t1/T = 0.05/0.077 = 0.65. From response spectra (Fig
3.5), the dynamic magnification factor D is 1.35. Thus the maximum displacement
will be
u max D

P0
4000
1.35
0.00308 m 3.1 mm
k
1.75 106

and the maximum elastic forces which will develop are

f s,max ku max 1.75 106 0.00308 5390 kN

If the blast impulse had been only one-tenth as long (t1 = 0.005s), the dynamic
magnification factor for this impulse-length ratio (t1/T = 0.065) would have been only
D = 0.24 and hence the elastic resisting forces would have been only fs = 958kN. Thus
for an impulse of very short duration, a large part of the applied load is resisted by the
inertia of the structure, and the stresses produced are much smaller than those due to
the longer loading.

Approximate Analysis of Impulsive-Load Response

From the study of the response spectra presented earlier, two general conclusions may
be drawn concerning the response of structures to impulsive loadings:
1. For long-duration loading, for example t1/T > 1, the dynamic magnification
factor depends principally on the rate of increase of the load to its maximum
value. A step loading of sufficient duration produces a magnification factor of
2; a very gradual increase causes a magnification factor of 1.
2. For short-duration loads, for example t1/T < , the maximum displacement
amplitude umax depends principally upon the magnitude of the applied impulse
t1

I=

p(t )dt

and is not strongly influenced by the form of the loading impulse.

The dynamic magnification factor D, however is quite dependent upon the

form of loading because it is proportional to the ratio of impulse area to peakload amplitude, as may be noted by comparing curves of response spectra
graph in the short-period range. Thus umax is the more significant measure of
response.
A convenient approximate procedure for evaluating the maximum response to a shortduration impulsive load, which represents a mathematical expression of this second
conclusion, may be derived as follows: the impulse-momentum relationship for the
mass m may be written
t1

mu p( t ) ku ( t )dt

(3.18)

41

in which u represents the change of velocity produced by the loading. In this

expression it may be observed that for small values of t1 the displacement developed
during the loading, u(t1) is of the order of (t1)2 while the velocity change u is of the
order of (t1). Thus since the impulse is also of the order of t1, the elastic force term
ku(t) vanishes from the expression as t1 approaches zero and is negligibly small for
short-duration loadings. On this basis, the approximate relationship may be used is
t1

mu p( t )dt

(3.19)

1 1
p( t )dt
m 0

(3.20)

The response after the termination of the loading is a free-vibration

u ( t1 )
u(t )
sin t u ( t1 ) cos t

(3.21)

in which t = t t1. But since the displacement term u(t1) is negligibly small and the
velocity u (t1 ) u , the following approximate relationship may be used

u(t)

1 1
sin t
p
(
t
)
dt

m 0

(3.22)

Example:
As an example of the use of this approximate formula, consider the response of the
structure shown in Fig 3.7 to the impulsive loading indicated. In this case,

kg

3.13 rad/sec and

W

t1

p(t )dt 40kN sec . The response is then approximately

0

40(9.81)
sin t . The maximum response results when sin t = 1, that is
(3.13)(900)
umax = 13.9 mm. The maximum elastic force developed in the spring, which is of
major concern to the structural engineer is fmax = k umax = 9000 x 0.0139 = 125.1 kN.
2
Since the period of vibration of this system is T =
2 sec , for this short-duration

loading (t1/T = 0.15) the approximate analysis may be expected to be quite reliable. In
fact, the maximum response determined by direct integration of the equation of
motion is very close to 13.9 mm and the error of the approximate method is less than
2%.
u (t )

42

Fig 3.7

43

CHAPTER 4
Response to General Dynamic Loading
Duhamel Integral for an Undamped System
The procedure described earlier for approximating the response of a structure to a
short-duration impulse may be used as the basis for developing a formula for
evaluating response to a general dynamic loading. Consider the arbitrary general
dynamic loading p(t) shown below, specifically the intensity of loading p ( ) acting at
time t = . This loading acting during the short interval of time d produces a shortduration impulse p ( ) d on the structure, and Eq (3.22) can be used to evaluate the
response to this impulse. It should be noted carefully that although the procedure is
only approximate for impulses of finite duration, it becomes exact as the duration of
loading approaches zero. Thus for the differential time interval d , the response
produced by the loading p ( ) is exactly (for t > ).

du ( t )

p()d
sin ( t )
m

(4.1)

In Eq (4.1), the term du(t) represents the differential response to the differential
impulse over the entire response history for t > ; it is not the change of u during a
time interval dt.
The entire loading history may be considered to consist of a succession of such short
impulses, each producing its own differential response of the form of Eq (4.1). For a
linearly elastic system, then, the total response can be obtained by summing all the
differential responses developed during the loading history, that is, by integrating Eq
(4.1) as follows
t

1
u(t)
p() sin ( t )d
m 0

(4.2)

Eq (4.2) is generally known as the Duhamel integral for an undamped system. It may
be used to evaluate the response of an undamped SDOF system to any form of
dynamic loading p(t), although in the case of arbitrary loadings the evaluation will
have to be performed numerically.

Eq (4.2) may also be expressed in the form

44

u ( t ) p() h ( t )d

(4.3)

where the new h symbol has the definition

h ( t )

1
sin ( t )
m

(4.4)

Eq (4.3) is called the convolution integral; computing the response of a structure to an

arbitrary loading using this integral is known as obtaining the response through the
time domain. The function h (t - ) is generally referred to as the unit-impulse
response, because it expresses the response of the system to an impulse of unit
magnitude applied at time t = . In Eq (4.2) it has been assumed that the loading was
initiated at time t = 0 and that the structure was at rest at that time. For any other
specified initial conditions, (i.e. u(0) 0, u (0) 0 ) an additional free-vibration
response must be added to this solution; thus in general

u(t)

t
u (0)
1
sin t u (0) cos t
p() sin( t )d

m 0

(4.5)

Numerical Evaluation of the Duhamel Integral for an Undamped System

If the applied-loading function is integrable, the dynamic response of the structure can
be evaluated by the formal integration of Eq (4.2) or (4.5). In many practical cases,
however, the loading is known only from experimental data, and the response must be
evaluated by numerical processes. For such analyses it is useful to note the
trigonometric identity,
sin(t ) sin t cos cos t sin

(4.6)

We can write Eq (4.2) in the form (zero initial conditions being assumed)
t

1
1
u ( t ) sin t
p() cos d cos t
p() sin d

m 0
m 0
or

where

u(t ) A(t ) sin t B(t ) cos t

A ( t )

B( t )

1
p() cos d

m 0

t
1
p() sin d

m 0

(4.7)

(4.8)

45

The numerical integration of Duhamel integral thus requires the evaluation of the
integrals A(t) and B(t) numerically. Consider for example A(t). The function to be
integrated is shown graphically below. For convenience of numerical calculation, the
function has been evaluated at equal time increments , successive values of the
function being identified by appropriate subscripts
t

1
y()d
m 0
y() p() cos

A( t )

(4.9)
(4.10)

y()d area under y( ) vs.

curve. One could use either simple summation or

trapezoidal rule or Simpsons rule for numerical integration.

a) Simple Summation:
A( t )

1
y 0 y1 y 2 ... y N1
m

(4.11)

b) Trapezoidal rule:
A( t )

1
y0 2y1 2y 2 ... 2y N1 y N
m 2

(4.12)

46

c) Simpsons rule:
A( t )

1
y0 4y1 2y 2 ... 4y N1 y N
m 3

(4.13)

where N = t/ must be an even number for Simpsons rule.

Using any of the summation processes of Eqs (4.11)-(4.13) leads to an approximation

of the integral for the specific time t under consideration. Generally, however, the
entire history of response is required rather than merely the displacement at some
specific time; in other words, the response must be evaluated successively at a
sequence of times t1, t2, , where the interval between these times is (or 2 if
Simpsons rule is used). The evaluation of the term B(t) can be carried out in exactly
the same way, that is,
t

1
B( t )
y()d
m 0

where y() p() sin

(4.14)

Example:
The dynamic response of a water tower subjected to a blast loading has been
calculated to illustrate the numerical evaluation of the Duhamel integral. The
idealisation of the structure and of the blast loading is shown in the figure below.
For this system, the vibration frequency and period are

kg
36700 9.81
rad

30
,
W
400
s

2 2

0.209 s
30

Using Trapezoidal rule with = 0.01sec, one can present the hand solution of the
first 6 steps in tabular form:

47

Evaluation of A

,s
(1)
0.00
0.01
0.02
0.03
0.04
0.05
0.06
0.07

p ( )
kN
(2)

0.00
200.00
400.00
600.00
400.00
200.00
0.00
0.00

sin

cos

(3)

(4)

0.00
0.295
0.564
0.783
0.932
0.997
0.974
0.863

1.00
0.955
0.826
0.622
0.363
0.072
-0.227
-0.505

Evaluation of B
(col.2 x
col.3)
(y)
(9)

(10)

B
(col.10 x
col.6)
(11)

Col. 2xcol. 4
(y)
(5)
0.0
191.0
330.4
373.2
145.2
14.4
0.0
0.0

Asin
(col.8 x
col.3)

(12)

2m

(6)

(7)
-6

4.09 x 10

B cos
(col.11 x
col.4)

0.0
191.0
712.4
1416.0
1934.4
2094.0
2108.4
2108.4

A (col. 6 x col.
7)
(8)
0
0.781 x 10-3
2.91 x 10-3
5.79 x 10-3
7.91 x 10-3
8.56 x 10-3
8.62 x 10-3
8.62 x 10-3

(col.12 col.13)

Fs(t) =
ku(t) = k
x col. 14

(14)

kN

(13)

48

0.00
59.0
225.6
469.8
372.8
199.4
0.0
0.0

0.0
59.0
343.6
1039.0
1881.6
2453.8
2653.2
2653.2

0.0
0.241 x 10-3
1.40 x 10-3
4.25 x 10-3
7.69 x 10-3
10.04 x 10-3
10.85 x 10-3
10.85 x 10-3

0.0
0.23 x 10-3
1.64 x 10-3
4.53 x 10-3
7.37 x 10-3
8.53 x 10-3
8.39 x 10-3
7.44 x 10-3

0.0
0.230 x 10-3
1.16 x 10-3
2.64 x 10-3
2.79 x 10-3
0.723 x 10-3
-2.46 x 10-3
-5.48 x 10-3

0.0
0.0
0.48 x 10-3
1.89 x 10-3
4.58 x 10-3
7.81 x 10-3
10.85 x 10-3
12.92 x 10-3

0.0
0.0
17.62
69.36
168.09
286.63
398.19
474.16

Response of Damped Systems

The derivation of the Duhamel integral equation which expresses the response of a
damped system to a general dynamic loading is entirely equivalent to the undamped
analysis except that the free-vibration response initiated by the differential load
impulse p( )d is subjected to exponential decay. Thus setting u(0) = 0 and letting
u (0) [ p( )d ] / m in Eq (1.34) leads to

p()d

du ( t ) e ( t )
sin D ( t )
mD

(4.15)

in which the exponential decay begins as soon as the load is applied at time t = .
Summing these differential response terms over the entire loading interval then results
in
t

u(t)

1
p()e ( t ) sin D ( t )d

mD 0

(4.16)

which is the damped-response equivalent of Eq (4.2).

Comparing Eq (4.16) with the convolution integral of Eq (4.3) shows that the unitimpulse response for a damped system is given by
h ( t )

1 ( t )
e
sin D ( t )
mD

(4.17)

For numerical evaluation of the damped-system response, Eq (4.16) in a form similar

to Eq (4.7):
u(t ) A(t ) sin D t B(t ) cos D t

(4.18)

49

where in this case,

1
e
A
(
t
)

p
(

)
cos Dd

m D 0
e

t

e
B(t ) 1
p( ) t sin Dd

e
D 0

These integrals can be evaluated by an incremental summation process equivalent to

that used previously but taking account of the exponential decay in the process. The
accuracy of the solution to be expected from any of these numerical processes
depends, of course, upon the length of the time interval d . In general, this must be
selected short enough for both the load function and trigonometric functions to be
well defined; T /10 is a common rule of thumb which usually provides
satisfactory results. Of the three summation procedures presented, Simpsons rule
provides the best accuracy.
A plot of the elastic-force histories in the damped and undamped cases calculated by
digital computer for 46 time steps is present in the figure below. It is apparent that
damping has little effect during the first part of the response but causes a noticeable
response reductions thereafter.

50

CHAPTER 5

Analysis of Nonlinear Structural Responses

In the analysis of linear structures subjected to arbitrary dynamic loading, Duhamel
integral analysis described earlier provides the most convenient solution technique.
However, it must be emphasised that since the principle of superposition was
employed in its derivation, it may be applied only to linear systems, that is, systems
for which the properties remain constant during the response. On the other hand, there
are many important classes of structural dynamic problems which cannot be assumed
to be linear, eg the response of a building to an earthquake motion severe enough to
cause serious damage. Consequently it is necessary to develop another method of
analysis suitable for use with nonlinear systems.
Probably the most powerful technique for nonlinear analysis is the step-by-step
integration procedure. In this approach, the response is evaluated for a series of short
time increments, t , generally taken of equal length for computational convenience.
The condition of dynamic equilibrium is established at the beginning and end of each
interval, and the motion of the system during the time increment is evaluated
approximately on the basis of an assumed response mechanism (generally ignoring
the lack of equilibrium which may develop during the interval). The nonlinear nature
of the system is accounted for by calculating new properties appropriate to the current
deformed state at the beginning of each time increment. The complete response is
obtained by using the velocity and displacement computed at the end of one
computational interval as the initial conditions for the next interval; thus the process
may be continued step by step from the initiation of loading to any desired time,
approximating the nonlinear behaviour as a sequence of successively changing linear
systems.

Incremental Equation of Equilibrium

The structure to be considered is the SDOF system shown in Fig 5.1a. The forces
acting on the mass are shown in Fig 5.1b. The general nonlinear characteristics of the
spring and damping forces are shown in Figs 5.1c and d respectively, while an
arbitrary applied loading is sketched in Fig 5.1e. At any instant of time t the
equilibrium of forces acting on the mass m requires:
f I (t ) f D (t ) f S (t ) p(t )
while a short time t later the equation would be

(5.1)

f I (t t ) f D (t t ) f S (t t ) p(t t )

(5.2)

Subtracting Eq (5.1) from Eq (5.2) yields the incremental form of the equation of
motion for the time interval t:
f I (t ) f D (t ) f S (t ) p(t )

(5.3)

51

The incremental forces in this equation may be expressed as follows:

f I ( t ) f I ( t t ) f I ( t ) mu( t )
f ( t ) f ( t t ) f ( t ) c( t )u ( t )
D

D
D

f S ( t ) f S ( t t ) f S ( t ) k ( t )u ( t )
p( t ) p( t t ) p( t )

(5.4)

in which it is assumed that the mass remains constant and the terms c(t) and k(t)
represent the damping and stiffness properties corresponding to the velocity and
displacement existing during the time interval as shown in Figs 5.1c and d
respectively. In practice the tangent slopes defined at the beginning of the time
intervals are used instead of the secant slopes. This is done to avoid iteration. In other
words we use:

df
c( t ) D ,
du t

df
k(t) S
du t

(5.5)

Substituting the force expressions of Eq (5.4) into Eq (5.3) leads to the final form of
the incremental equilibrium equations for time t:
mu(t ) c(t )u (t ) k(t )u(t ) p(t )

(5.6)

Fig 5.1
52

Step-by-Step Integration
Many procedures are available for the numerical integration of Eq (5.6). The
technique employed here is simple in concept but is found to yield excellent results
with relatively little computational effort. The basic assumption of the process is that
the acceleration varies linearly during each time increment while the properties of the
system remain constant during this interval. The motion of the mass during the time
interval is indicated in graphical form in Fig 5.2, together with equations for the
assumed linear variation of the acceleration and the corresponding quadratic and
cubic variations of the velocity and displacement, respectively. Evaluating these latter
expressions at the end of the interval ( t ) leads to the following equations for the
increments of velocity and displacement.
u ( t ) u( t )t u( t )

t
,
2

u ( t ) u ( t )t u( t )

t 2
t 2
u( t )
2
6

(5.7)

It is convenient to use the incremental displacement as the basic variable of the

analysis; hence the 2nd of Eqs (5.7) is solved for the incremental acceleration, and this
expression is substituted into first of Eqs (5.7) to obtain

6
6

u( t ) t 2 u ( t ) t u ( t ) 3u( t )

3
t
u ( t )
u( t )
u ( t ) 3u ( t )
t
2

(5.8)

Substituting Eqs (5.8) into Eq (5.6) leads to the following form of the equation of
motion:
6
t
6

m 2 u ( t ) u ( t ) 3u( t ) c( t ) u ( t ) 3u ( t ) u( t ) k ( t )u ( t ) p( t )
t
2
t

(5.9)

Finally transferring all terms associated with the known initial conditions to the righthand side gives
~
k (t )u(t ) ~
p(t )

(5.10)

in which
~
6
3
k ( t ) k ( t ) 2 m c( t )
t
t

(5.11)

and

53

t
6

~
p ( t ) p( t ) m u ( t ) 3u( t ) c( t ) 3u ( t ) u( t )
2
t

(5.12)

Fig 5.2

After Eq (5.10) is solved for the displacement increment, this value is substituted into
2nd of Eqs (5.8) to obtain the incremental velocity. The initial conditions for the next
time step then result from the addition of these incremental values to the velocity and
displacement at the beginning of the time step. This numerical-analysis procedure is
based on two significant approximations: (1) that the acceleration varies linearly and
(2) that the damping and stiffness properties remain constant during the time step. In
general, neither of these assumptions is entirely correct, even thought the errors are
small if the time step is short. Therefore errors generally will arise in the incrementalequilibrium relationship which might tend to accumulate from step to step, and this
accumulation should be avoided by imposing the total-equilibrium condition at each
step of the analysis. This may be achieved conveniently by expressing the
accelerations at the beginning of the time step in terms of the total external load minus
the total damping and elastic forces.

Summary of procedure
For any given time increment, the analysis procedure consists of the following
operations:
1. Initial velocity and displacement values u (t) and u(t) are known, either from
values at the end of the preceding increment or as initial conditions of the
problem.

54

2. With these values and the specified non-linear properties of the structure, the
damping c(t) and stiffness k(t) for the interval, as well as current values of the
damping fD(t) and elastic fs(t) forces are found, eg from Figs 5.1c and 5.1d.
3. The initial acceleration is given by
1
u( t ) p( t ) f D( t ) f S ( t )
(5.13)
m
This is merely a rearrangement of the equation of equilibrium for time t.
~
4. The effective load increment ~
p (t ) and effective stiffness k (t ) are computed
from Eqs (5.11) and (5.12).

5. The displacement increment is given by Eq (5.10), and with it the velocity

increment is found from the 2nd Equation of Eqs (5.8).
6. Finally the velocity and displacement at the end of the increment are obtained
from

u ( t t ) u ( t ) u ( t )

u ( t t ) u ( t ) u ( t )

(5.14)

When step 6 has been completed, the analysis for this time increment is finished, and
the entire process may be repeated for the next time interval. The process can be
carried out consecutively for any desired number of time increments. Linear systems
also can be treated by the same process, of course; in this case the damping and
stiffness properties remain constant. As with any numerical-integration process, the
accuracy of this step-by-step method will depend on the length of the time increment
t . Three factors must be considered in the selection of this interval: (1) the rate of
variation of the applied loading p(t), (2) the complexity of the nonlinear damping and
stiffness properties, and (3) the period T of vibration of the structure. The time
increment must be short enough to permit the reliable representation of all these
factors, the last one being associated with the free-vibration behaviour of the system.
Thus, if the load history is relatively simple, the choice of the time interval will
depend essentially on the period of vibration of the structure. This linear-acceleration
method is only conditionally stable and will give a divergent solution if the time
increment is greater than about half the vibration period. However, the increment
must be considerably shorter than this to provide reasonable accuracy, so that
instability is not a practical problem. In general, an increment-period ratio of
t / T 1/10 is a good rule of thumb for obtaining reliable results. If there is any
doubt about the adequacy of a given solution, a second analysis can be made halving
the time increment; if the response is not changed appreciably in the 2nd analysis, it
may be assumed that the errors introduced by the numerical integration are negligible.

55

Example:
To demonstrate a hand-solution technique for applying the linear-acceleration stepby-step method described here, the response of the elasto-plastic SDOF frame shown
in figure below to the loading history indicated has been used for the analysis.
In this structure, the damping coefficient has been assumed to remain constant; hence
the nonlinearity results only from the change of stiffness as yielding takes place. The
effective stiffness thus may be expressed (see Eq (5.11)) as
~
6
3
k (t ) k(t )
m
c 66 k ( t )
2
0.1
0.1

where k(t) is either 5 kips/in or zero, according as the frame is elastic or yielding.
Also the effective incremental loading is given by (see Eq (5.12))
0.1
6m

~
p (t ) p(t )
3c u 3m
c u p(t ) 6.6 u 0.31 u
2
0.1

The velocity increment given by the 2nd equation of Eqs (5.8) becomes

u 30u 3u 0.05u
A convenient tabular arrangement for the hand calculations is shown in the table of
next page. For this elasto-plastic system, the response behaviour changes drastically
as the yielding starts and stops. The dynamic elasto-plastic response calculated in the
following Table is plotted in Figure 5.3, with the response during the yielding phase
shown as a dashed line. Also plotted for comparison is the linear elastic response
obtained by a similar step-by-step analysis but with k 71 and fs = 5u throughout the
calculations. The effect of the plastic yielding shows up clearly in this comparison.
Also shown to indicate the character of the loading is the static displacement p/k, that
is the deflection which would have occurred in the elastic structure if there had been
no damping and inertia effects.

56

Fig 5.3
Figures for the Example

57