Chapter 3

Dynamics of Earthquake Analysis

3.1 Introduction
Earthquake or seismic analysis is a subset of structural analysis which involves the calculation of
the response of a structure subjected to earthquake excitation. This is required for carrying out
the structural design, structural assessment and retrofitting of the structures in the regions where
earthquakes are prevalent. Various seismic data are necessary to carry out the seismic analysis of
the structures. These data are accessible into two ways viz. in deterministic form or in
probabilistic form. Data in deterministic form are used for design of structures etc whereas data
in probabilistic form are used for seismic risk analysis, study of structure subjected to random
vibration and damage assessment of structures under particular earthquake ground motion. Major
seismic input includes ground acceleration/velocity/displacement data, magnitude of earthquake,
peak ground parameters, duration etc.

In this chapter, the seismic response of the structures is investigated under earthquake excitation
expressed in the form of time history of acceleration. The response is investigated for the
structures modeled as Single Degree of Freedom (SDOF) and discrete Multi Degree of Freedom
(MDOF) System.

3.2 Equation of Motion for SDOF System

Consider a SDOF system (shown in Figure 3.1), subjected to an earthquake acceleration, ɺɺ x g (t ) .
Let m, k and c represent the mass, stiffness and damping, respectively of the SDOF system
undergoing relative displacement, velocity and acceleration of x (t ) , xɺ (t ) and ɺɺ
x (t ) , respectively.
The various forces acting on the system will be inertial force, stiffness force and damping force.

x(t)

m x (t ) + ɺɺ
m ( ɺɺ x g ( t ))

k m
c
kx (t )
cxɺ (t )
x g (t )
ɺɺ

(a) (b)
Figure 3.1 (a) SDOF system (b) Free body diagram.

60
Consider the equilibrium of the various forces acting on the mass, as shown in Figure 3.1(b), we
get,
x ( t ) + ɺɺ
m ( ɺɺ x g ( t )) + cxɺ ( t ) + kx ( t ) = 0
(3.1)
or
mxɺɺ( t ) + cxɺ ( t ) + kx ( t ) = − mxɺɺg ( t )
(3.2)

where,
x (t ) = relative displacement of mass with respect to ground

xɺ (t ) = relative velocity of mass with respect to ground

x (t ) = relative acceleration of mass with respect to ground

ɺɺ

x g (t ) = earthquake ground acceleration

ɺɺ

The equation of motion is expressed in the normalized form as

x(t ) + 2ξω0 xɺ(t ) + ω02 x(t ) = − ɺɺ
ɺɺ xg (t )
(3.3)

where, ξ and ω0 denotes the damping ratio and natural frequency of SDOF system, respectively

expressed as
k
ω0 =
m (3.4)
c
ξ =
2 mω 0 (3.5)
The damped natural frequency of SDOF system is given by

ω d = ω0 1 − ξ 2 (3.6)

The equation of motion for a linear, viscously damped SDOF system is second order differential
equation with constant coefficients. The solution of this equation for the specified earthquake
acceleration, ɺɺ
x g (t ) will provide the response of the SDOF system.

61
3.3 Response Analysis of SDOF System
For a given time history (acceleration versus time data) of earthquake ground motion, the
response of viscously damped SDOF system can be obtained either by Time Domain Analysis or
Frequency Domain Analysis.

3.3.1 Time Domain Analysis

This method helps in obtaining response of SDOF system in both linear and non linear range.
Duhamel integration and Numerical schemes such as Newmark integration, Runge-Kutta
methods are invariably accompanied for obtaining numerical solution of differential equation.

Duhamel Integral is used to obtain the response of SDOF system subjected to earthquake ground
motion. Equation of motion for a SDOF system subjected to ground motion acceleration is given
by equation (3.2). The solution of which can be split into homogeneous and particular part as
x (t ) = xh (t ) + x p (t )
(3.7)
where,
xh(t) = homogeneous solution, and xp(t) = particular solution.
Homogeneous or complimentary solution (as depicted from Figure 3.2) is the damped free-
vibration response given by equation (3.8)

xh (t ) = g (t ) x0 + h(t ) xɺ0 (3.8)

where, x0 and xɺ0 are initial displacement and velocity of the SDOF system, respectively.

x0 =1
SDOF g(t)
xɺ0 = 0

x0 = 0
SDOF h(t)
xɺ0 = 1

Figure 3.2 Response of SDOF system to initial unit displacement and velocity.

62
Putting the boundary conditions (as shown in Figure 3.2) in the solution of the homogeneous
part, g(t) and h(t) can be obtained as
 ξω0 
g (t ) = e−ξω0 t cos ωd t + sin ωd t 
 ωd  (3.9)
e−ξω0 t
h(t ) = sin ωd t
ωd (3.10)
For obtaining particular solution part of equation (3.7), it is assumed that the irregular ground
acceleration is made up of very brief impulses as shown in Figure 3.3. The vibration caused by
all the impulse are added together to obtain the total response.

xg ( t )
ɺɺ

xg ( τ )
ɺɺ

τ t-τ
dτ

Figure 3.3 Impulse from earthquake acceleration.

Consider the vibration caused by a single impulse. Newton’s second law states that the rate of
change of moment of a mass is equal to the applied force i.e.
d
(mxɺ (t )) = − m ɺɺ
xg (t )
dt (3.11)
Thus, the change in momentum over a brief interval, dτ brought by the instantaneous force
− m ɺɺ
x g ( τ ) is given by

d ( mxɺ ( t )) = − m ɺɺ
x g ( τ) d τ
(3.12)

63
It should be noted that the small changes in velocity and displacement occurring during the time
interval dτ will make a negligible contribution to the change in momentum. The change in
velocity during the interval is
dxɺ ( t ) = − ɺɺ
xg (τ) d τ (3.13)

Thus, the change in displacement at time, t caused by the impulse at τ is given by

dx p (t ) = − ɺɺ
x g ( τ ) d τ. h (t − τ)
(3.14)

Each impulse in Figure (3.3) will produce a vibration of this form. Because the system is linear,
the effect of each impulse is independent of every other impulse and the total resulting motion
can be obtained by the principle of super position.
t
x p (t ) = − ∫ ɺɺ
xg (τ) h(t − τ) d τ (3.15)
0

This integral is known as convolution or Duhamel integral. Explicit solution may be obtained
for simple forms of forcing function such as rectangular and triangular.

From equations (3.8), (3.9), (3.10) and (3.15), the total response (given in equation (3.7)) of
system can be given by
t
x g (τ )]h (t − τ )dτ
x ( t ) = x0 g (t ) + xɺ0 h (t ) − ∫ [ ɺɺ (3.16)
0

For the system with at rest condition (i.e. x0 =0 and xɺ0 =0) the response is given by
t
x (t ) = − ∫ ɺɺ
x g ( τ) h (t − τ) d τ (3.17)
0

This is known as time domain solution because the response is calculated using time as a
variable.

In order to obtain recurrence formulas for time domain analysis, consider a SDOF system with
displacement and velocity defined at initial time, ti and the response is required at, ti+1 (refer
Figure 3.4). Suppose xi and xɺi are the initial displacement and velocity of the system,
respectively,

64
 xg ( t )
ɺɺ

xgi+1
ɺɺ

ɺxɺgi

τ
Time
t
dτ
∆ti

Figure 3.4 Linear variation of ground acceleration across chosen time step ti .

The response of the system (reproducing equation (3.16)) is expressed as

t
x(t ) = g (t − ti ) xi + h (t − ti ) xɺi − ∫ ɺɺx (τ)
ti g h (t − τ) d τ (3.18)

 ξω 
g (t − ti ) = e −ξω0 ( t −ti )  0 sin ωd (t − ti ) + cos ωd (t − ti ) 
 ωd 
e −ξω0 ( t −ti ) 
h(t − ti ) = sin ωd (t − ti ) 
ωd 
i +1
 (3.19)
x
ɺɺ − x
ɺɺ i

xg ( τ) = ɺɺ
ɺɺ xgi + g g
( τ − ti ) 
∆ti

e −ξω0 ( t −τ ) 
h ( t − τ) = sin ωd (t − τ) 
ωd 

Back substituting in equation (3.18),

t  x i +1 − ɺɺ
ɺɺ xgi   e −ξω0 ( t −τ ) 
x (t ) = g (t − ti ) xi + h(t − ti ) xɺi − ∫  ɺɺ xgi + g ( τ − ti )   sin ωd (t − τ)  d τ
ti
 ∆ti   ωd  (3.20)

x(t ) = g (t − ti ) xi + h(t − ti ) xɺi + f1 ( t − ti ) ɺɺ

xgi + f 2 ( t − ti ) ɺɺ
xgi +1
(3.21)

65
Similarly, the velocity of the system at time, t is given by
xɺ (t ) = gɺ (t − ti ) xi + hɺ(t − ti ) xɺi + fɺ1 ( t − ti ) ɺɺ
xgi + fɺ2 ( t − ti ) ɺɺ
xgi +1 (3.22)

At t = ti +1 , x (t ) = xi +1 and xɺ (t ) = xɺi +1 , the displacement and velocity of the system are

expressed as

xi +1 = g (∆ti ) xi + h(∆ti ) xɺi + f1 ( ∆ti ) ɺɺ

xgi + f2 ( ∆ti ) ɺɺ
xgi +1 (3.23)

xɺi +1 = gɺ (∆ti ) xi + hɺ(∆ti ) xɺi + fɺ1 ( ∆ti ) ɺɺ

xgi + fɺ2 ( ∆ti ) ɺɺ
xgi +1 (3.24)

In the matrix form, the above equations can be re-written

 xi +1   g (∆ti ) h(∆ti )   xi   f1 ( ∆ti ) f 2 ( ∆ti )   ɺɺ

xgi 
 =    + ɺ   i +1  (3.25)
 xɺi +1   gɺ (∆ti ) h(∆ti )   xɺi   f1 ( ∆ti ) fɺ2 ( ∆ti )   ɺɺ
ɺ xg 

or

 xi +1   a11 a12   xi   b11 b12   ɺɺ

x gi 
 =    + b b   ɺɺ i +1 
(3.26)
 xɺi +1   a21 a22   xɺi   21 22   x g 
or
{xi +1} = [ A]{ xi } + [ B]{ ɺɺ
xgi } (3.27)

where,

x   ɺɺx gi  a a  b b 
{xi } =  i  , {x g } =  i +1  ,
ɺɺ i
[A] =  11 12  , and [B] =  11 12 
 xɺi   ɺɺ
x g  a21 a22  b21 b22 

The elements of matrices [A] and [B] from Nigam and Jennings (1969) are given by equations
(3.28) and (3.29)

66
  ξ  
a11 = e −ξω0∆ti  sin ωd ∆ti + cos ωd ∆ti  
 1 − ξ2  
 

e −ξω0∆ti 
a12 = sin ωd ∆ti 
ωd 
 (3.28)
ω0 −ξω0 ∆ti 
a21 = − e sin ωd ∆ti
1 − ξ2 

 ξ 
a22 = e −ξω0∆ti  cos ωd ∆ti − sin ωd ∆ti  
 1 − ξ2 
  

 2ξ2 − 1 ξ  sin ωd ∆ti  2ξ 1   

b11 = e −ξω0∆ti  2 +  + 3 + 2  cos ωd ∆ti  
 ω0 ∆ti ω0  ωd  ω0 ∆ti ω0   


  2ξ − 1  sin ωd ∆ti
2
2ξ  1 2ξ 
b12 = e −ξω0 ∆ti  2  + 3 cos ωd ∆ti  − 2 + 3 
 ω0 ∆ti  ωd ω0 ∆ti  ω0 ω0 ∆ti 

 2ξ2 − 1 ξ   ξ  
b21 = e−ξω0 ∆ti  2 +   cos ωd ∆ti − sin ωd ∆ti  
 ω0 ∆ti ω0   1 − ξ2 
 

 2ξ 1   1 
− 3 + 2  ( ωd sin ωd ∆ti + ξω0 cos ωd ∆ti )  + 2
ω ∆ ω ω ∆ 
 0 i t 0   0 ti

 2   
−ξω0 ∆ti 2ξ − 1 ξ 
b22 = e  2  cos ωd ∆ti − sin ωd ∆ti 
 ω0 ∆ti  1− ξ 2 



2ξ  1 
− 3 ( d
ω sin ω ∆ t + ξω cos ω ∆ ti ) − 
ω0 ∆ti  ω0 ∆ti
d i 0 d 2
 (3.29)

67
Finally, the acceleration response of the SDOF system can be obtained by reproducing the
equation of motion as
( − cxɺi +1 − kxi +1 − mxɺɺgi +1 )
xi +1 =
ɺɺ (3.30)
m

Hence, if the displacement and velocity of the system are known at some time ti, the state of the
system at all subsequent times, ti+1 can be computed exactly by a step-by-step application of
equation (3.27) to (3.30). The computational advantage of this approach lies in the fact that the
elements of [A] and [B] matrix depend only on ξ, ω0 and ∆ti. The value ξ and ω0 are constant

and if ∆ti is also constant, xi , xɺi and ɺɺ

xi can be evaluated by the execution of multiplication and
summation operations for each step of integration. The matrices [A] and [B], defined by rather
complicated expressions, equations (3.28) and (3.29) need to be evaluated only at the beginning
of each response evaluation. If varying time intervals are used, it is necessary, in general, to
compute [A] and [B] at each step of integration. However, by rounding the time coordinates of
the record, the number of these matrices needed during the calculation can be reduced to only a
few. These, too can be computed at the beginning of the calculation and called upon when
needed, thereby saving the computational time.

68
3.3.1.1 Numerical Methods for Seismic Analysis of SDOF System

There are number of numerical methods available for solving initial boundary value problems.
Most commonly used methods are Newmark’s Beta method (Linear acceleration method) and
Runge-Kutta method which are described here.

3.3.1.1.1 Newmark’s Beta Method

In this method, acceleration, velocity and displacement at time, t = ti+1 is obtained as a function
of acceleration, velocity and displacement at t = ti (which is always known), assuming linear
acceleration during small time step (Figure 3.5). Assume a SDOF system subjected to earthquake
ground motion, the equation of motion is given by

mxɺɺ(t ) + cxɺ (t ) + kx (t ) = − mxɺɺg (t ) (3.31)

∆ti
(a) (b)
Figure 3.5 (a) Linear ground acceleration, (b) Linear relative acceleration of SDOF system over
time step, ∆ti .

In the incremental form, equation (3.31) can be re-written as

m∆ɺɺ
xi + c∆xɺi + k ∆xi = − m∆ɺɺ
xgi (3.32)

where,

∆ɺɺ
xi = ɺɺ
xi +1 − ɺɺ
xi 
∆xɺi = xɺi +1 − xɺi 

∆xi = xi +1 − xi  (3.33)
∆ti = ti +1 − ti 

∆ɺɺ xgi +1
xgi = ɺɺ − ɺɺ
xgi 

69
Assuming linear variation of acceleration (see Figure 3.5(b))
xi +1 − ɺɺ
ɺɺ xi
x (t ) = ɺɺ
ɺɺ xi + .t (3.34)
∆ ti

On integrating,
xi +1 − ɺɺ
ɺɺ xi t 2
xɺ (t ) = xɺi + ɺɺ
xi t + . (3.35)
∆ti 2

t2 x − ɺɺ
ɺɺ x t3
x(t ) = xi + xɺi t + ɺɺ
xi + i +1 i . (3.36)
2 ∆ti 6

In equation (3.36) put t = ∆ti and express in terms of ∆ɺɺ

xi i.e.
6 6
∆ɺɺ
xi = ∆xi − xɺi − 3ɺɺ
xi (3.37)
∆ ti2
∆ ti

Similarly, from equation (3.35) put t = ∆ti and solve for ∆ɺxi
3 ∆ ti
∆xɺi = ∆xi − 3 xɺi − xi
ɺɺ (3.38)
∆ ti 2

Substituting ∆ɺɺ
xi , ∆xɺi in equation (3.32) and solve for ∆xi i.e.
peff
∆xi = (3.39)
keff

where,

 6   ∆t 
peff = − m∆ɺɺxgi +  m + 3c  xɺi +  3m + i c  ɺɺ
xi (3.40)
 ∆ti   2 
6 3
keff = m + c + k (3.41)
∆ti 2
∆ti

Knowing the ∆xi , determine ∆ɺxi from equation (3.38). At t = ti+1, displacement and velocity can
be determined as
xi +1 = xi + ∆xi 
 (3.42)
xɺi +1 = xɺi + ∆xɺi 

70
The acceleration at time ti+1 is calculated by considering equilibrium of equation (3.31) to avoid
the accumulation of the unbalanced forces i.e.
1
xi +1 =
ɺɺ  − mxɺɺgi +1 − cxɺi +1 − kxi +1  (3.43)
m

In this way, using this step-by-step numerical integration scheme, the response of SDOF system
can be obtained for given time history. Repeat the same steps to obtain response at t =ti+2 and so
on. The accuracy of output response depends upon the magnitude of time-step ‘ ∆ti ’ chosen.

Optimum values of ‘ ∆ti ’ should be chosen to obtain fastest converging results with required
precision. Time stepping methods has got limitation that error goes on accumulating with
calculation proceeds. In order to keep Newmark’s Beta method stable, the time step should be
taken such that (Chopra, 2007)
∆ti
≤ 0.551 (3.44)
T0

where, T0 = 2π / ω0 ,is the time period of the SDOF system.

This method is conditionally stable if above inequality satisfies, otherwise method will “blow-
up” giving illogical results.

3.3.1.1.2 Runge-Kutta Method

Knowing the initial conditions, response of SDOF system with time can be determined using
Runge-Kutta method. Let the equation of motion of the SDOF system be

mxɺɺ( t ) + cxɺ (t ) + kx ( t ) = − mxɺɺg (t ) (3.45)

Define a vector,

x
x =   (3.46a)
 xɺ 
On differentiating the above equation

71
  xɺ 
xɺ =   (3.46b)
 ɺɺ
x

Using Equation (3.45) and (3.46a), above equation can be reproduced as,

 0 1
 xɺ  x  0 
  =  −k −c    +  ɺɺ 
 ɺɺ
x  xɺ   − xg 
 m m 

or, xɺ = Ex + F (3.47)
 0 1 
 0 
where, E =  −k −c  ; F = 
 − ɺɺ
xg 
 m m 

x gi +1
ɺɺ
ɺxɺgi

∆ti
Figure 3.6: Time step for Runge-Kutta method

Determine the following constant vectors,

k1 = ∆ti  Exi + F i  (3.48)

  k   ∆F i 
k2 = ∆ti  E  xi + 1  +  F i +  (3.49)
  3  3  
  2k   2∆ F i 
k3 = ∆ti  E  xi + 2  +  F i +  (3.50)
  3   3  

(Note: Subscript ‘i’ refers to the value at time t = ti , ‘∆’ refers to difference in value at ti+1 and ti,
and ‘¯ ’ indicates a matrix)

72
then,
∆xi  k + 2k2 + k3
∆xi =   = 1 (3.51)
∆xɺi  4

 xi +1 
xi +1 =   = xi + ∆xi (3.52)
 xɺi +1 

From equations (3.52) and (3.45)

( − cxɺi +1 − kxi +1 − mxɺɺgi +1 )
xi +1 =
ɺɺ (3.53)
m

Repeat same steps to obtain response at time t = ti+2 and so on. A Runge-Kutta method is
conditionally stable for linear second order differential equation like equation (3.45). But still the
time step, ∆ti should be taken short enough to ensure required precision in results.

73
3.3.2 Frequency Domain Analysis

This method is used for obtaining response of linear systems subjected to irregular excitations
such as earthquake forces and it requires knowledge of complex frequency response function for
its proper application. If the stiffness ‘k’ and damping ‘c’ of the SDOF system are frequency
dependent, then this approach is much superior to the time domain.

In frequency domain analysis, the response of a SDOF system is given by

∞
1
x(t ) = ∫ Xɺɺ g (ω) H (ω)eiωt d ω (3.54)
2π −∞

x g ( t ) ] and H ( ω) is the complex frequency

where, Xɺɺ g ( ω ) is the Fourier transform of [- ɺɺ

response function.

Consider a SDOF system (Figure 3.7), subjected to the forcing function of eiωt, producing

displacement response as
x(t ) = H (ω)eiωt (3.55)

e iω t SDOF x(t ) = H (ω)eiωt

Figure 3.7: Explanation of complex frequency response function.

Consider equation of motion of SDOF system

x + 2 ξω o xɺ + ω o2 x = e iω t
ɺɺ (3.56)

Substituting equation (3.55) in equation (3.56)

H(ω)(iω) 2 e iωt + 2 ξωo H ( ω)iω e iωt + ω o2 H( ω) e iωt = e iωt (3.57)

or, H ( ω)[ −ω2 + 2 ξωωo i + ωo2 ] = 1 (3.58)

Thus, the complex frequency response function is expressed as

74
 1
H (ω) = (3.59)
(ω − ω ) + i 2ξωωo
2
o
2

x g (t ) ] is expressed as
The Fourier transform of [- ɺɺ
∞

∫ [− ɺɺx (t )]e
− iωt
Xɺɺ g (ω) = g dt
−∞ (3.60)

The response of a SDOF system can be obtained by substituting equations (3.59) and (3.60) in
equation (3.54) and solving the integral.

Note: Properties of Fourier transform are as follows,

Consider a function f (t ) which is periodic and it is absolutely integrable i.e.,

∞

∫
−∞
f (t ) dt < ∞ (3.61)

Then the Fourier transform of f (t ) exists and given by

∞
F (ω) = ∫
−∞
f (t )e −iωt dt (3.62)

From the inverse Fourier transform

∞
1
∫ F (ω)e d ω
iωt
f (t ) = (3.63)
2π −∞
The f (t ) and F ( ω) makes the Fourier Pair.

75
3.4 Numerical Examples on SDOF system
Example 3.1
xg (t ) = ɺɺ
A SDOF system is subjected to a harmonic ground motion of ɺɺ xo sin ωt . Determine the

steady state response using time and frequency domain method and considering that the system
starts from rest. The natural frequency and fraction of critical damping of SDOF system are ωo
and ξ , respectively.
Solution: Equation of motion is given by

x (t ) + 2 ξω o xɺ (t ) + ω 2o x (t ) = − ɺɺ
ɺɺ xo sin ω t

A. Time Domain Analysis (Using Duhamel Integration)

Using equation (3.17) steady state response of system in time domain is given by
t
xg (τ )h(t − τ )dτ
x(t ) = −∫ ɺɺ
0

e −ξωo ( t −τ )
x g ( τ) = ɺɺ
where, ɺɺ xo sin ωτ and h(t − τ) = sin(ωd (t − τ))
ωd

Thus,
e−ξωo (t −τ)
t
x(t ) = −∫ ( ɺɺ
xo sin ωτ) sin(ωd (t − τ))d τ
0
ωd

xo
ɺɺ 1
=− sin(ωt − θ)
ωo (1 −β )2 + (2ξβ)2
2 2

ω  2ξβ 
where, β = and θ = tan −1  2 
ωo  1− β 

B. Frequency Domain Analysis

The steady state response is given by equation (3.54) i.e.,

∞
1
x(t ) = ∫
2π −∞
Xɺɺ g (ω) H (ω)eiωt d ω

76
where,
1
H (ω) =
ω − ω + i 2ξωωo
2
o
2

and
∞
Xɺɺ g (ω) = FT [ − ɺɺ
xo sin( ωt )] = ∫ [ − ɺɺx sin(ωt )] e
o
− iωt
dt
−∞

Evaluating the integral,

∞
Xɺɺ g (ω) = FT [ − xo sin(ωt )] = − ∫ ɺɺ
xo sin(ωt )e − iωt dt
−∞

∞
 eiωt − e−iωt 
∫ ( e − e ) e dt
xo
ɺɺ
=− iωt − iωt − iωt
 Note : sin ωt = 
2i −∞  2i 

xo  ∞ i ( ω−ω)t
ɺɺ ∞

=− ∫ e dt − ∫ ei ( −ω−ω) t dt 
2i  −∞ −∞ 

xo
ɺɺ
=−  2πδ ( ω − ω) − 2πδ ( −ω − ω) 
2i 

xo [ δ(ω− ω) − δ(ω+ ω)]

= iπɺɺ

Figure 3.8 Graphical Representation of FT [ − ɺɺ

x0 sin ωt ] .

Recalling the properties of Fourier Transform,

∞ ∞
1
∫ δ(t )e dt and δ ( t ) =
− iωt
1= ∫ 1.eiωt d ω
−∞
2π −∞

77
Implying that,
∞

∫e
iωt
d ω = 2πδ(t )
−∞

Recalling the properties of Dirac Delta ‘δ(t)’ function

1
δ(kt) = δ(t) and δ(t − to ) = 0 if t ≠ to
k
∞

∫ δ(t − t )dt = 1
−∞
o

∫ δ(t − t )ψ(t )dt = ψ(t )

−∞
o o

δ(t − to ) = δ(to − t )

Using the above properties, the response of the SDOF system is expressed as
1
x(t ) = ∫ eiωt H (ω) Xɺɺ g (ω)d ω
2π
∞
 
x(t ) =
1
∫
2π −∞
eiωt  2
1
 iπɺɺ (
xo [ δ(ω − ω) − δ(ω + ω)] d ω )
 ωo − ω + i 2ξωωo 
2

ixɺɺo  ∞ eiωt
∞
eiωt 
= ∫ 2 δ( ω− ω) d ω − ∫ δ(ω + ω)d ω
2  −∞ ωo − ω + i 2ξωωo
2
ω − ω + i 2ξωωo
2
−∞ o
2


ixɺɺo  eiωt e−iωt 

=  2 − 
 ωo − ω + i 2ξωωo ωo − ω − i 2ξωωo 
2 2 2
2

ω  2ξβ 
Taking β = and θ = tan −1  
ωo  1− β
2
,

 
ixɺɺo  eiωt − iθ e−iωt 
+ iθ
= e − e 
2 
(ω ) + ( 2ξβωω )
2 2
(ω − ω2 ) + ( 2ξβωωo )
2
−ω
2 2 2 2
 o o o 

78
 ixɺɺo  1 
=  (ei ( ωt −θ) − e − i ( ωt −θ ) ) 
2ωo2  (1 − β2 ) 2 + (2ξβ)2 

ixɺɺo  1 
=   2i sin( ωt − θ)
2ωo  (1 − β )2 + (2ξβ)2
2 2 


xo
ɺɺ 1
=− sin(ωt − θ)
ωo (1 −β2 )2 + (2ξβ)2
2

Thus, the same expression for the steady state response of the SDOF system subjected to the
harmonic earthquake acceleration is obtained using the time and frequency domain approach.

Example 3.2
Show that the displacement response of an undamped SDOF system subjected earthquake
x0e− at is given by
xg (t ) = ɺɺ
acceleration, ɺɺ

x0
ɺɺ a − at 
x(t ) = −  sin ω0t − cos ω0t + e 
a + ω02
2
 ω0 

where, ω0 = natural frequency of the SDOF system; and a = parameter having the same unit as

that of ω0 .

Solution: The displacement response of the SDOF system to earthquake acceleration,

x0e− at
xg (t ) = ɺɺ
ɺɺ
is expressed in time domain analysis as

x ( t ) = − ∫ ɺɺ
xg ( τ ) h ( t − τ ) d τ
t

x0
ɺɺ
x (t ) = − sin ω0 ( t − τ ) d τ
t
− aτ

ω0 ∫e
0

On integrating by parts,

79
 t
x
ɺɺ  e − aτ  x
ɺɺ ω0 − aτ
x (t ) = − 0 sin ω0 ( t − τ )  + 0 e cos ω0 ( t − τ ) d τ
t

ω0
−
 a  0 ω0
∫
0 a

t
x0  1
ɺɺ x0  e − aτ
 ɺɺ xω
 ɺɺ
x ( t ) = −  sin ω0t  +  − cos ω0 ( t − τ )  + 0 2 0 sin ω0 ( t − τ ) d τ
t
− aτ

ω0  a  a  a 0 a ∫e
0

xg  1
ɺɺ  ɺɺ
x0 − at ɺɺ
x0 ω20
x ( t ) = −  sin ω0t  − 2 e + 2 cos ω0t − 2 x ( t )
ω0  a  a a a

 ω2  x
ɺɺ a − at 
x ( t ) 1 + 20  = − 02  sin ω0t − cos ω0t + e 
 a  a  ω0 

x0
ɺɺ a − at 
x (t ) = −  sin ω0t − cos ω0t + e 
a + ω02
2
 ω0 

Alternate Solution by Direct Solution of Differential Equation

The differential equation of motion of an SDOF system is expressed as

x(t ) + ω02 x(t ) = −ɺɺ
ɺɺ xg (t )

Let the solution of the equation be

x ( t ) = xh (t ) + x p (t )

where xh(t) is the solution of homogeneous part of differential equation and xp(t) is the particular
solution.

The homogeneous solution will take the following form

xh (t ) = A cos ω0t + Bsin ω0t

The particular solution of the given differential equation will be

xp (t ) = ce−at

xɺ p (t ) = −a c e− at

80
x p (t ) = a 2ce− at
ɺɺ

xe − at
x ( t ) + ω 02 x ( t ) = − ɺɺ
Substituting the above in the equation, ɺɺ

a 2 ce − at + ω02 ce − at = − ɺɺ
x0 e − at
x0
ɺɺ
c=−
a 2 + ω02

x0
ɺɺ
x p (t ) = − e − at
a + ω0
2 2

The response of the system will be

x (t ) = xh (t ) + x p (t )

x0
ɺɺ
x (t ) = A cos ω0t + B sin ω0t − e − at
a + ω0
2 2

The initial conditions for system with at rest are i.e. at t = 0 are x (0) = 0 and xɺ (0) = 0 .
x0
ɺɺ
x (0) = 0 A− =0
a + ω02
2

x0
ɺɺ
A=
a + ω02
2

Similarly,
axɺɺ0
xɺ (0) = 0 −ω0 A × 0 + ω0 B × 1 + e − a ×0 = 0
a + ω0
2 2

a x0
ɺɺ
B=−
ω0 a + ω20
2

Substitute the expression for A and B, the response of SDOF system simplifies to

x0
ɺɺ − a ɺɺx0 x
ɺɺ
x (t ) = cos ω0t + sin ω0t − 2 0 2 e − at
a + ω0
2 2
ω0 a + ω0
2 2
a + ω0

x0
ɺɺ a − at 
=−  ω sin ω0t − cos ω0t + e 
a + ω02
2
 0 
81
Example 3.3

Using the frequency domain approach and time domain analysis, show that the displacement of
xg ( t ) = c0δ ( t ) is given by
an undamped SDOF system subjected to earthquake acceleration, ɺɺ

c0
x (t ) = − sin ( ω0t )
ω0

where ω0 is the natural frequency of the SDOF system and δ ( t ) is the Dirac delta function.

Solution:

A. Frequency Domain Analysis

The displacement of SDOF system to earthquake excitation is given by

1 ∞ ɺɺ
x (t ) = X g ( ω) H ( ω) eiωt d ω
2π ∫−∞

where,
1
H ( ω) = (For undamped system)
ω − ω2
2
0

xg ( t ) is
And Fourier transform of ɺɺ
∞
Xɺɺ g ( ω) = ∫ − ɺɺ
xg ( t ) e−iωt dt
−∞

∞
Xɺɺ g ( ω) = −∫ c0 δ ( t ) e−iωt dt
−∞

Xɺɺg ( ω) = −c0e−iωt t =0

Xɺɺ g ( ω) = −c0

Therefore, the response of the SDOF system is given by

1 ∞ 1
x (t ) = ∫ ( − c0 ) 2 e iωt d ω
2π −∞ ω0 − ω 2

c0 ∞ e iωt
2π ∫−∞ ω02 − ω2
=− dω

82
The integral have two poles at ω = ± ω0 Applying Cauchy residual theorem for solving the
.
integral

c0  eiωt eiωt 
x (t ) = −  2 πi lim ( ω − ω ) + 2 πi lim ( ω + ω0 ) 
2π  ω→ω0 ω − ω ω→ω0 ω − ω
2 2 0 2 2
0 0 
 eiω0t e−iω0t 
x ( t ) = −ic0 − + 
 2ω0 2ω0 
ic0 iω0t −iω0t
x (t ) = e − e 
2ω0 
ic0
x (t ) = [ 2i sin ω0t ]
2 ω0

i 2c0
x (t ) = sin ω0t
ω0
c0
x (t ) = − sin ( ω0t )
ω0

B. Time Domain Approach

x ( t ) = − ∫ h ( t − τ ) ɺɺ
xg ( τ ) d τ
t

where,
sin ω0 ( t − τ )
h (t − τ) =
ω0

xg ( τ) = c0 δ ( τ)
ɺɺ

Therefore, the response of the SDOF system is given by

sin ω0 ( t − τ )
x (t ) = −∫ c0δ ( τ ) d τ
t

0 ω0

 sin ω0 ( t − τ ) 
x ( t ) = − c0 
 ω0  τ= 0

c0
x (t ) = − sin ω0t
ω0

83
Example 3.4

Show that the maximum displacement response of a damped SDOF system subjected earthquake
x g ( t ) = xɺ g 0δ (t ) is
acceleration, ɺɺ

xɺ g 0  ξ  1 − ξ2 
xmax = − exp − tan −1  
ω0  1 − ξ2  ξ 
  
where ω0 and ξ are natural frequency and damping ratio of the SDOF system, respectively; xɺ g 0 is

the increment in velocity or the magnitude of acceleration impulse and δ (t ) is the Dirac delta
function.

Solution: The general solution of SDOF system will be given by

t
xg ( τ) h(t − τ)d τ
x(t ) = −∫ ɺɺ
0
t
sin ( ωd ( t − τ) ) d τ
1 −ξω0 ( t −τ)
= −∫ xɺg 0δ ( τ) e
0
ωd
xɺ g 0
=− e−ξω0t sin ( ωd t )
ωd

For the maximum displacement of the system, xɺ (t ) = 0

xɺ g 0
i.e. − e −ξω0t [ −ξω0 sin ωd t + ωd cos ωd t ] = 0
ω0

ωd ω0 1 − ξ2 1 − ξ2
i.e. tan(ωd t ) = = =
ξω0 ξω0 ξ

From above equation, the time at which maximum displacement occurs will be

1 1 − ξ2
tm = tan −1
ωd ξ

−1
 1 − ξ2 
ω t
If d m = tan   , it implies that the sin ω d t m = 1 − ξ 2 (refer Figure 3.9)
 ξ 
 

84
 1
1

Figure 3.9

The maximum displacement will be given by

xɺ g 0
xmax = − e−ξω0tm sin ( ωd tm )
ωd

xɺ g 0  1  1 − ξ2 
=− exp −ξω0 tan −1    × 1 − ξ2
ωd  ωd  ξ 
  

Substituting for ω d = ω0 1 − ξ 2 and simplifying

xɺ g 0  ξ  1 − ξ2 
xmax = − exp − tan −1  
ω0  1 − ξ2  ξ 
  

85
Example 3.5

Find the response of SDOF system having time period as 1 sec and damping ratio as 0.02
subjected to the El-Centro, 1940 earthquake motion (refer Appendix-I for the digitized
acceleration values). Plot the displacement response of the SDOF system using (a) Exact method
of time domain analysis, (b) Newmark’s Beta method, and (c) Runge-Kutta method.

Solution:

Based on the computer program written in the FORTAN language, the response of the SDOF

system with time period as 1 sec and damping ratio as 0.02 subjected to the El-Centro, 1940

earthquake motion were obtained and is plotted in Figure 3.10. The calculated maximum

displacement of the system is found to be 0.15163m, 0.15166m and 0.15158m for exact method

of time domain analysis, Newmark’s Beta method and Runge-Kutta method, respectively. As

expected all the methods predict the same response of the system. Further, time interval taken for

numerical integration of equation of motion of the system is 0.002 sec for Newmark’s Beta and

Runge-Kutta methods.

86
Figure 3.10 Response of SDOF system of Example 3.5 by various methods.

87
3.5 Response Analysis of MDOF System

Multi degree of freedom (MDOF) systems are usually analyzed using modal superposition
analysis. A typical MDOF system with n degrees of freedom is shown in Figure (3.11). This
system when subjected to ground motion undergoes deformations in number of possible ways.
These deformed shapes are known as modes of vibration or mode shapes. Each shape is vibrating
with a particular natural frequency. Total unique modes for each MDOF system are equal to the
possible degrees of freedom of system. The equations of motion for MDOF system is given by

x (t )} + [c ]{ xɺ ( t )} + [ k ]{ x ( t )} = - [ m ]{r} ɺɺ
[m ]{ɺɺ x g (t) (3.64)

where, [m] = Mass matrix (n × n); [k] = Stiffness matix (n × n); [c] = Damping matrix (n × n);
{r} = Influence coefficient vector (n×1); { x ( t )} = relative displacement vector; { xɺ ( t )} = relative
velocity vector, {ɺɺ
x ( t )} = relative acceleration vector, and ɺɺ
x g (t ) = earthquake ground acceleration.

Xn

X2

X1

ɺxɺg (t)

Figure 3.11 MDOF system.

88
The undamped eigen values and eigen vectors of the MDOF system are found form the
characteristic equation

{[k ] − ω 2
i [m ]} φ i = 0 i = 1, 2, 3, … , n (3.65)

d et {[k ] − ω 2
i [m ]} = 0 (3.66)

where,

ωi2 = eigen values of the i mode

th

φi = eigen vector or mode shape of the ith mode

ωi = natural frequency in the ith mode.

Let the displacement response of the MDOF system is expressed as

{x(t)} = [φ] {y(t)} (3.67)

where {y(t)} represents the modal displacement vector, and [ φ] is the mode shape matrix given

by

[φ] = [ φ1 , φ2 ,......... φn ] (3.68)

Substituting { x} = [ φ]{ y} in equation (3.64) and pre-multiply by [φ] T

[φ] [ m ][ φ]{ ɺɺy (t )} + [ φ] [c ][φ]{ yɺ (t )} + [ φ] [ k ][ φ]{ y (t )} = − [φ] [ m ]{r } ɺɺxg (t )

T T T T
(3.69)

The above equation reduces to

y ( t ) } + [ C d ] { yɺ ( t ) } + [ K d ] { y ( t ) } = − [ φ ]
] { ɺɺ [m ]{r } ɺɺx g ( t )
T
[M m (3.70)

where,

[φ ] [ m ][φ ] = [ M m ] = generalized mass matrix

T

[φ ] [c ][ φ ] = [Cd ]
T
= generalized damping matrix

[φ] [ k ][φ] = [ K d ]
T
= generalized stiffness matrix

89
By virtue of the properties of the [φ], the matrices [Mm] and [Kd] are diagonal matrices. However,
for the classically damped system (i.e. if the [Cd] is also a diagonal matrix), the equation (3.70)
reduces to the following equation

y i ( t ) + 2 ξ i ω i yɺ i ( t ) + ω i2 y i ( t ) = − Γ i ɺɺ
ɺɺ x g (t ) ( i = 1, 2, 3, … , n ) (3.71)

where,
yi (t ) = modal displacement response in the ith mode,

ξi = modal damping ratio in the ith mode, and

Γ i = modal participation factor for ith mode expressed by

{φi }T [m ]{r}
Γi = (3.72)
{φi }T [m ]{φi }

The equation (3.71) represents n second order differential equations (i.e. similar to that of a
SDOF system) and the solution of which will provide the modal displacement response in the ith
mode, yi (t ) for i=1 to n. The displacement response of the MDOF system can be obtained by
equation (3.67) using the {y(t)}. The other response quantities of the structure can be obtained
from the displacement response of the system.

90
3.6 Numerical Examples on MDOF System

Example 3.6

A two-story building is modeled as 2-DOF system and rigid floors as shown in the Figure 3.12.
Determine the top floor maximum displacement and base shear due to El-Centro, 1940
earthquake ground motion. Take the inter-story stiffness, k =197.392 × 103 N/m, the floor mass,
m = 2500 kg and damping ratio as 2%.

m
x2

2m
x1

2k

Figure 3.12

Solution:

Mass of each floor, m = 2500 kg and stiffness, k = 197.392 kN/m

 3k −k 
Stiffness matrix = [k] =  
 −k k 
5000 0 
Mass matrix = [m] = 
 0 2500 

Using the equations (3.65) and (3.66), the frequencies and mode-shapes of the structures are

ω1 = 6.283 rad/sec and ω2 =12.566 rad/sec

91
 0.5  −1
{φ1} =   , and {φ2 } =  
1  1 

The modal column matrix is given by

0.5 −1
[φ] = [φ1 φ2 ] =  
 1 1

The modal participation factors are given by

{φ } [m]{r}
T

Γi = i T
{φi } [m ]{φi }
5000 0  1
[0.5 1]   
{φ } [m]{r} = 2500 1
T

Γ1 = 1 T  0 =1.333
{φ1} [m]{φ1} [0.5 1] 5000 0  0.5
 0 2500 1 


Similarly,

5000 0  1
[ −1 1]   
{φ2 } [m]{r} 2500 1
T

Γ2 = =  0
= −0.333
{φ2 } [m]{φ2 } [ −1 1] 5000 0   −1
T

 0 2500 1 


The response in the each mode of vibration is computed by solving the Equation (3.71) for the

system. The displacement and base shear response is shown in the Figures 3.13 and 3.14,

respectively. The maximum top floor displacement and base shear are found to be 0.202 m and

40.72 kN, respectively.

92
Figure 3.13 Top floor displacement response of two DOF system of Example 3.6.

93
Figure 3.14 Base shear response of two DOF system of Example 3.6.

94
Example 3.7

An industrial structure is modeled as 2-DOF system as shown in the Figure 3.15. Determine the
horizontal and vertical displacement of the free end of the structure due to El-Centro, 1940
earthquake ground motion. Take EI =80 × 103 N.m2, L= 2m, m1= 100kg and m2= 200kg. The
damping shall be considered as 2 percent.

m1=100kg m2=200kg
EI, L
x1

x2

EI, L

Figure 3.15

Solution: Given,

Mass, m1= 100kg, m2= 200kg, Length, L= 2 m and flexural rigidity, EI= 80 × 103 Nm2

6 EI  8 −3
Stiffness matrix = [k ] = ;
7 L3  −3 2 

300 0 
Mass matrix = [m] =  
 0 200 

Using equation (3.65), eigen values and eigen vectors can be obtained as

ω1 = 5.4925 rad/sec; ω2 = 16.856 rad/sec

 2.7   5.103 
{φ1} =  ; {φ2 } =  
6.25  −3.307 

95
Modal participation can be obtained by

{φi } [m]{r}
T

Γi =
{φi } [m ]{φi }
T

Γ1 = 0.081 and Γ 2 = 0.153

The displacement response in the each mode of vibration is computed by solving the Equation

(3.71) for the system. The horizontal and vertical displacement of the free end of the structure is

shown in the Figures 3.16 and 3.17, respectively. The maximum horizontal and vertical

displacement of the free end of the structure is found to be 0.039 m and 0.0699 m, respectively.

96
Figure 3.16 Horizontal displacement response of the Industrial Structure of Example 3.7.

97
Figure 3.17 Vertical displacement response of the Industrial Structure of Example 3.7.

98
3.8 Tutorial Problems

Q1. Develop general computer programs (preferably using Matlab or Scilab) to obtain the

response of a SDOF system under earthquake excitation using (a) Newmark’s Beta method,

(b) Runge-Kutta method, and (c) Exact method of time domain analysis. Compare the

results from above three methods by plotting the response of a SDOF system having time

period as 0.5 sec and damping ratio as 0.05 under the El-Centro, 1940 motion.

Q2. Derive the expressions for the elements of matrices [A] and [B] (i.e. equations (3.28) and
(3.29)) used in exact method for evaluation of the response of a SDOF system under
earthquake excitation.
Q3. Derive the expression for displacement response of an undamped SDOF system subjected to
earthquake ground motion of ɺɺx g ( t ) = ɺɺxo ( e − at − e − bt ) . Take, ω0 = natural frequency of the

SDOF system; and a = parameter having the same unit as that of ω0 .

Q4. A rigid-jointed plane frame is fixed at A and roller support at C as shown in Figure 3.18.
The members AB and BC are rigidly connected at B making a right angle and are
supporting a mass of 200 kg. Neglect the mass of frame, determine the maximum horizontal
displacement and base shear due to El-Centro, 1940 earthquake. Take the flexural rigidity,
EI = 294772.2 Nm2 and length, L = 4m for both members. Consider the damping as 2%.

B
C

Figure 3.18

99
Q5. A 2-degrees-of-freedom system (Figure 3.19) is subjected to horizontal earthquake
excitation of El-Centro, 1940 earthquake. Take the flexural rigidity, EI = 106 Nm2 and
length, L = 2m. The each lumped mass is 100 kg. Determine the maximum displacement of
the two masses. Take 2% damping in each mode of vibration.

100 kg

EI, L

100 kg

EI, L

Figure 3.19
Q6. A three-story building is modeled as 3-DOF system and rigid floors as shown in Figure
3.20. Determine the maximum top floor maximum displacement and base shear due to El-
Centro, 1940 earthquake ground motion. Take the inter-story lateral stiffness of floors i.e. k1
= k2= k3=16357.5 kN/m, the floor mass m1= m2=10000 kg and m3=5000 kg and damping
ratio as 2%.
m3
x3

k3
m2
x2

k2
m1
x1

k1

Figure 3.20

100
3.8 Answers to Tutorial Problems

Q1. Maximum Displacement = 0.057 m

Q3.

x0
ɺɺ a − at  x0
ɺɺ b − bt 
x(t ) = −  sin ω0t − cos ω0t + e  + 2  sin ω0t − cos ω0t + e 
a + ω02
2
 ω0  b + ω0
2
 ω0 

Q4. Horizontal displacement = 0.0684 m

Base shear = 2160.2 N

Q5. Displacement of lower mass = 7.76×10-3 m

Displacement of top mass = 24.02×10-3 m

Q6. Top floor displacement = 0.0234 m

Base shear = 196.4 × 103 N

101