CHAPTER 2

Seismic Analysis of Single Degree of Freedom

Systems (SDOFS)

2.1 EQUATION OF MOTION

Single degree of freedom systems (SDOFS) are rarely met in Civil Engineering
practice. However, the study of SDOFS is shown to be invaluable. This chapter aims
to provide an easy understanding of the complexity involved by earthquake loads.
m

c, k

ug(t)

u(t)

uabs(t)

Figure 2.1 SDOFS model

Under seismic loads, the dynamic equilibrium at a time t for a SDOFS can be written
as:
FI

FD

FE

(2.1)

where FI is the inertia force, FD is the damping force, and the FE is the elastic force.
Detailing the Equation (2.1), it is obtained, see Figure 2.1:

muabs (t ) cu(t ) ku(t ) 0

(2.2)

where m is the mass, c is the damping, and k is the stiffness of the system. In (2.2)
u(t) and u(t) are the relative displacement and relative velocity of the system. uabs(t)
is the absolute acceleration.
For understanding the seismic behavior of SDOFS, Figure 2.1 shows a
SDOFS that suffered a displacement of the base (absolute displacement) from the

initial position, at the left. The relative displacement of the mass is giving the final,
deformed, position - at the right. Therefore Equation (2.2) could be rewritten
m[u(t ) ug (t )] cu (t ) ku(t ) 0

(2.3)

mu(t ) cu (t ) ku(t )

(2.4)

or
mug (t )

Acc. (gal)

where ug (t ) is the ground acceleration.

As an example, Figure 2.2 shows the ground acceleration records on three
directions (East/West, North/South, and Up/Down) for the Romanian Vrancea 1977,
March 4, earthquake. The last record, up-down accelerations, presents early higher
values compared to horizontal action marking the arrival of the P-waves.
Similar example, for Kobe 1995 Japanese Earthquake is presented later by
Figure 2.9.

0
min=-126.1 at 6.16

-200

Acc. (gal)

Vrancea 1977 E/W N/S U/D records

max=162.3 at 5.62

200

10

20

30
Time (s)

200

50

max=177.6 at 6.88

min=-194.9 at 6.14

-200
0

Acc. (gal)

40

10

20

30
Time (s)

100

40

50

max=66.41 at 5.3

0
-100
0

10

min=-105.8 at 5.02
20
30
40
50
Time (s)

Figure 2.2 E/W, N/S, and U/D time-history records of accelerations for Vrancea 1977 earthquake
Dividing all the Equation (2.4) by the value of the mass, another form of the equation
of motion is obtained
u(t ) 2

u (t )

u (t )

ug (t )

(2.5)

where is the damping ratio (the fraction from the critical damping) and is natural
(undamped) circular frequency.
A solution for a second order differential equation like (2.5) may be written
u(t ) ugen (t ) u part(t )

(2.6)

where ugen(t) is the general solution (the solution of the homogeneous equation,
corresponding to the free vibration). upart(t) is the particular solution (corresponding to
the load).
Appendix A, at the end of this book shows how to solve the homogeneous
equation. However, for the case of seismic action, it is usually assumed (and proved

by practice) that the influence of the free vibration is very small and limited only at
the very beginning of the motion.
For a seismic action, which is a random action, the particular response is
obtained from the superposition (integration) of responses to impulses at each
moment

u (t )

t
0

ug ( )e

(t

sin

(t

(2.7)

)d

Deriving the Equation (2.7), the velocity response is obtained

t

u (t )
D
t
0

ug ( )e

ug ( )e

(t

(t

sin

cos

(t

(t

)d

ug ( )e

(t

cos

(t

)d
(2.8)

)d

The last approximation is based on small values for damping, as it is the case for most
civil engineering structures (i.e. < 0.10).
Vrancea 1977 E/W N/S U/D

Accel.Response Spectrum (gal)

10

E/W
N/S
U/D

10

10

E/W max=417.7 at 0.788

N/S max=615.6 at 1.156
U/D max=226.7 at 0.164
0

10
-2
10

-1

10

10

10

10

Period (s)

Figure 2.3 E/W, N/S, and U/D acceleration response spectra for Vrancea 1977 earthquake
The acceleration response is obtained from the equation of motion (2.5)
2

uabs (t )

u (t )

u (t )

1 2

t
0

ug ( )e

(t

cos

(t

ug ( )e

(t

sin

(t

)d

(2.9)

)d

2.2 SPECTRAL RESPONSE OF SDOFS

The spectral displacement, Sd, spectral velocity, Sv, and spectral absolute
acceleration, Spa, are defined as follows in Equations (2.10). From the mentioned
equations it can be observed that they are the absolute maximum responses of a
SDOFS with the damping and the circular frequency .

Figure 2.3 exemplifies on response spectra using Vrancea 1977 earthquake

E/W, N/S, and U/D accelerations record components, and compares them. The
assumed damping is 5%. For the same Romanian major earthquake, Figures 2.11,
2.12, and 2.13 at the end of this chapter, present comparisons of displacement,
velocity, and acceleration response spectra depending on the damping.
Sd ( , )

u (t ) max

Sv ( , )

u (t ) max

Sa ( , )

uabs (t ) max

(2.10)

Similar to Figure 2.3, in Figure 2.10 the response spectra using Kobe 1995 earthquake
E/W, N/S, and U/D accelerations record components is shown.
Other comparison of acceleration response spectra is shown by Figure 2.4.
The two compared earthquakes are Kobe NS 1995 and El-Centro NS 1940. Responses
were calculated for a damping of 5%, too.
Acceleration Response Spectra 5%

Acceleration Response (m/s 2)

10

Kobe NS
El-Centro NS
1

10

10

Kobe NS max=26.55 at 0.3526

-1

10

El-Centro NS max=12.14 at 0.1577

-2

10 -2
10

-1

10

10

10

10

Period (s)
Figure 2.4 Comparison between Kobe NS and El-Centro acceleration response spectra
The spectral responses are useful for determining the maximum elastic force,
maximum kinetic energy, and maximum inertia forces

FEmax

k u (t ) max

E max

1
2
m u (t ) max
2
m uabs (t ) max

FImax

kS d ( , )
1
mSv2 ( , )
2
mS a ( , )

(2.11)

It is used in Earthquake Engineering to work with the so-called pseudo-velocity

spectral response defined by the next equation
S pv ( , )

t
0

ug ( )e

(t

sin

(t

)d

(2.12)
max

In the case of the structures with 0 < < 0.20 and excepting that of very flexible
structures (i.e. with very small values of the natural circular frequency, ), the values
of the pseudo-velocity response spectra, Spv( , ), are very close to those of the
velocity response spectra, Sv( , )

(2.13)

Sv ( , ) S pv ( , )

The spectral displacement can be directly linked to the pseudo-velocity response

spectrum
Sd ( , )

S pv ( , )

(2.14)

Under the same conditions shown above for damping, the acceleration response
spectrum is linked to the pseudo-velocity response spectrum
Sa ( , )

(2.15)

S pv ( , ) S pa ( , )

where Spa( , ) is the pseudo-acceleration response spectrum.

From Equations (2.14) and (2.15) it is easy to demonstrate that for a given
earthquake all three spectra (i.e. Sd, Spv, and Spa) can be represented on the same
logarithmic diagram.
1000 cm/s2

Sd,max = 90.1 cm
for T=14.5 sec
Spv,max=137.4 cm/s for T=1.70 sec
Spa,max=615.6 cm/s2 for T=1.16 sec

500 cm/s2

100 cm/s2
50 cm/s2
20 cm/s2

10

100 cm

10

50 cm

Ps
io

tS
pe
c

at

tr

er
xi

ac
pl
is

A
s

5 cm

um

Vrancea NS,
March 4, 1977
5% damping

tr

em

ec

en

Sp

10

10 cm

20 cm

el

um

cc

-A

xi

do

eu

Pseudo-Velocity Spectrum (cm/s)

10 cm/s2

2 cm
1 cm

-1

10 -2
10

-1

10

10
Period (sec)

10

Figure 2.5 Four-way response spectra for Vrancea NS 1977 earthquake

Such representation has the circular frequency, (or the frequency, f, or the period of
vibration, T), on the horizontal linear (or logarithmic) scale axis.
The vertical logarithmic axis is used for Spv. The other spectral terms, Sd and
Spa, are measured on logarithmic scales placed 45 at the right and 45 at the left
respectively.

Figure 2.5 is showing this kind of four-way representation for the case of
Vrancea March 4, 1977 earthquake, NS component, for a damping value of 5%.
2.3 TIME-HISTORY RESPONSE OF SDOFS
2.3.1 Direct integration method (Piecewise exact method)
The analytical solution of the time-dependant second order linear differential equation
mu( ) cu ( ) ku( )

mug ( )

(2.16)

that characterizes the SDOFS submitted to earthquake loading is the sum of the
general (homogeneous) equation plus the particular solution
u( ) ugen ( ) u part( )

(2.17)

An example of the input (Vrancea NS, March 4, 1977) together with the timeresponses (displacement, velocity, and acceleration) for a SDOFS with the period of
vibration one second and five percent damping is shown in Figure 2.6. It proves that
there is a difference in time for the peaks of response and the peaks of the input. The
characteristics of the output (responses) are different from the input characteristics,
too.
Acc. (gal)

INPUT: Vrancea NS 1977

200

max=177.6 at 6.88

0
-200
0

min=-194.9 at 6.14
5

10

15

20
Time (s)

25

30

35

40

35

40

35

40

35

40

Disp.(cm)

OUTPUT: Displacement Response for T=1s and damping=5%

20
max=14.2 at 7
0
min=-10.11 at 7.58
-20
0

10

15

20
Time (s)

25

30

Vel.(cm/s)

OUTPUT: Velocity Response for T=1s and damping=5%

max=68.05 at 6.74

50
0
-50
0

min=-63.89 at 7.24
5

10

15

20
Time (s)

25

30

Acc.(gal)

OUTPUT: Acceleration Response for T=1s and damping=5%

500

max=401.3 at 7.56

0
-500
0

min=-562.7 at 6.98
5

10

15

20
Time (s)

25

30

Figure 2.6 Acceleration input and time-responses example

It might be noted that the response from Equation (2.17) is almost impossible to be
obtained through analytical integration because ug ( ) could hardly be characterized
through simple functions.

A time interval,
that from Figure 2.7.

t, of digitally recorded ground acceleration may look like

ug ,i

ug ( )

ug ,i

t
Figure 2.7 Acceleration function
It is common to consider a linear variation for the ground acceleration between two
recorded values, i.e.
ug ( ) ug ,i

Adopting the next two notations,

ug

ug ,i

ug ,i

ug ,i

ug ,i

(2.18)

ug
, then Equation
t

and

(2.18) becomes

ug ( ) ug ,i

(2.19)

For the time interval t, the equation of motion (2.16) is

mu( ) cu ( ) ku( )

m ug ,i

ui

u( )

ui

ui

u ( )

ui

ui

u( )

ui

(2.20)

t
Figure 2.8 Response functions
The Equation (2.20) has the next particular solution
u part( )

m
ug ,i
k
1
ug ,i
2

m c
k2
2

1
2

ug ( )

(2.21)

Then, the whole response is

u( )

A cos

B sin

1
D

ug ,i

(2.22)

where the constants A and B are established from the initial conditions u i and u i , i.e.
the displacement and the velocity of the system at the beginning of the interval. It is
easy to show that
1

A ui

ug , i

(2.23)
ui

Deriving the Equation (2.23), the velocity and the acceleration are obtained:
u ( )
u( )

e
e

B sin
A sin

B cos

B cos

2
D

1 2

D
2

A cos

(2.24)
D

B sin

The above calculations are performed for each time-step, t, and the interval's final
values become initial values for the next step.
2.3.2 Integration methods
Supposing that the response in terms of displacement, velocity, and acceleration is
divided into time-steps, t, an image of such an interval is shown by Figure 2.8.

Acc. (gal)

Kobe JMA E/W N/S U/D t1=28, dt=25

500

max=617.3 at 5.48

min=-476.9 at 5.1

-500
0

10

15

20

25

Acc. (gal)

Time (s)

max=579 at 7.94

500
0
-500

min=-818 at 5.56
0

10

15

20

25

Acc. (gal)

Time (s)

max=332.2 at 4.76

200
0
-200
-400

min=-315.1 at 5.24
0

10

15

20

25

Time (s)

Figure 2.9 E/W, N/S, and U/D time-history record of accelerations for Kobe 1995 earthquake
The integration methods are step-by-step numerical approaches for dynamic response
analysis. They are based on the next integral expressions for the structural responses:

u i
ui

u i

0
t

ui

u( )d

(2.25)
u ( )d

2.3.3 Euler-Gauss Method

This method assumes constant response acceleration for the system along the interval
t. It can be shown that, for this case, the next relations are available:
ui
ui

ui
ui

ui t

ui

ui t ui

ui

(2.26)
t
2

The Euler-Gauss method is a particular case for Newmark

the steps for getting the response are presented next.

Accel.Response Spectrum (gal)

10

10

10

methods and, as a result,

Kobe JMA E/W N/S U/D

E/W
N/S
U/D

E/W max=2281 at 0.3806

10

N/S max=2655 at 0.3526

U/D max=1274 at 0.2497

10 -2
10

10

-1

10
Period (s)

10

10

Figure 2.10 E/W, N/S, and U/D acceleration response spectra for Kobe 1995 earthquake
2.3.4 Newmark

Methods

These methods propose more general relations for the final velocity and displacement
of the structure at the end of the considered interval:

ui

ui

(1

ui

ui

ui t

)ui t
1
2

ui

ui

t
t

ui

(2.27)

where
is showing the type of damping and
is depending of the response
acceleration shape. For the most common SDOFS, =1/2.
For , the next cases are more used:

=0, the acceleration is constant along the interval t and equal to the initial value
(the case of Euler Method)
=1/4, the acceleration is constant along the interval t and equal to the average
between the initial and final values
=1/6, the acceleration is linearly varying from the initial value to the final value
along the interval t
=1/8, the acceleration is constant and equal to the initial value along the first half
of the interval t, and the acceleration is constant and equal to the final value
along the second half of the interval t.
Vrancea NS, March 4, 1977

Relative Displacement Spectrum (cm)

90
80

damping

70

0%
1%
2%
5%
10%

60
50
40
30
20
10
0

2
3
Period (sec)

Figure 2.11 Displacement spectra for different values of damping for Vrancea NS 1977 earthquake
Vrancea NS, March 4, 1977

Relative Velocity Spectrum (cm/s)

350
300
damping
0%
1%
2%
5%
10%

250
200
150
100
50
0

2
3
Period (sec)

Figure 2.12 Velocity spectra for different values of damping for Vrancea NS 1977 earthquake
As an example for Newmark methods, when =1/2 and =1/6, the Equations (2.27)
become as it follows, Equations (2.28).

Vrancea NS, March 4, 1977

Relative Acceleration Spectrum (cm/s/s)

1400
1200
damping

1000

0%
1%
2%
5%
10%

800
600
400
200
0

2
3
Period (sec)

Figure 2.13 Acceleration spectra for different values of damping for Vrancea NS 1977 earthquake
ui
ui

ui

ui

1
ui
2

ui

ui t

1
ui
3

t
t

1
ui
6

(2.28)
1

The Newmark methods are following the next steps for each interval, i, of time t:
1. based on the values, ui , ui , ui , previously obtained, and using Equations (2.28)
the velocity and displacement response, ui 1, ui 1 , at the end of the interval, are
calculated.
Initial conditions, u0 , u0 , u0 , usually zero, should be used for the first interval.
2. introducing the ui 1 and ui 1 in the equation of motion (2.4), the acceleration at
the end of the interval is obtained

ui

mug ,i
1

1
c tui cui kui
2
1
1
2
c t
k t
2
6

(2.29)

For the first time interval, u1 is obtained using u0 , u0 , u0 , in Equation (2.29).

An incremental, generalizing approach for Newmark methods is shown in Chapter 3.
2.4 DESIGN SPECTRA
The use of spectra of the type shown in Figures 2.3, 2.4, 2.5, 2.10, 2.11, 2.12, or 2.13,
is not always simple because the shapes are very complicated. Other problem is that a
spectrum is showing the behavior of SDOFS under a certain earthquake and for fixed
soil conditions. However, even for only one location, the earthquakes' characteristics
may differ substantially from a seismic event to the other.

For an observation time (years), it can be established a general and simpler

form of earthquake's spectra, based on probabilistic approaches. Therefore, to
determine the seismic forces in design, the designer will use coefficients calculated
from design spectra diagrams.
Figure 2.14 shows an example of a spectral value, depending on the ground
conditions, used in design of highway bridges in Japan. The same Japanese
regulations are recommending the standard design spectra from Figure 2.15. This
diagram was established through multiple regression analyses and based on passed
earthquakes' characteristic damages.
2

Natural period factor

1.5

0.7
Ground Type I
Ground Type II
Ground Type III
0.4
0.04

0.1

0.5
1
Fundamental natural period T (sec)

4 5

Figure 2.14 Example of diagram for a spectral coefficient

In Romanian P100-92 Earthquake Engineering Code, commented in Chapter 4 of this
book, Equations (4.2) and (4.3), the spectral value is denoted by . Figure 4.7 is
showing the diagram of this coefficient.
500

Acceleration (gal)

300

100
Ground Type I
Ground Type II
Ground Type III

30
0.04

0.1

0.5
1
2
Fundamental natural period T (sec)

4 5

Figure 2.15 Example of standard spectra

It is easy to demonstrate that the design spectral values for Romanian seismic code are
due to the relation existent for the coefficients involved in Equations (4.2) and (4.3),
i.e.

Sa

ks

(2.30)

In Equation (2.30) the coefficients other than are constant. Therefore the shape of
the acceleration design spectrum in P100-92 is identical with diagram and only the
values are changed through the use of constants ks, , and g.
2.5 DESIGN EARTHQUAKES
Because of the reasons shown in the above paragraphs, the use of one ore more
specific earthquakes in design could lead to some overestimation or underestimation
in getting the seismic forces.
Ground Type I
150

Acceleration (gal)

100

50

-50

-100

-150
0

10

15

20

25

Time (sec)

Figure 2.16 Example of a design acceleration record

The idea of creating design spectra that tries to cover a range of possible effects for a
seismic activity lead to another idea: to generate artificial earthquake records for
complementing the design techniques.
Ground Type I

Acceleration response spectrum (gal)

1000
700
500
300
200

100

h=1%
h=2%

70

h=5%

50

h=10%
h=20%

30
0.1

0.2

0.3

0.5

Natural period T (sec)

Figure 2.17 Spectra for design acceleration record in Figure 2.16

These earthquakes are generated based on the reverse problem of acceleration spectra:
to get earthquakes records that will have the spectra almost identical to the design
spectra. In this case, the spectra are named target spectra. As a consequence, a design
earthquake is a very useful tool in time-domain response analysis.
In Figure 2.17 a design acceleration record is shown. This record is established
starting from a real acceleration record and transformed to fit the spectrum for ground
type I in Figure 2.15. The result of this transformation is the spectra in Figure 2.16
where, for 5% damping, the shape is much closed to that in Figure 2.14.