You are on page 1of 10

ISET GOLDEN JUBILEE SYMPOSIUM

Indian Society of Earthquake Technology


Department of Earthquake Engineering Building
IIT Roorkee, Roorkee

October 20-21, 2012

Paper No. D012
SEISMIC RESPONSE ANALYSIS UNDER SIMULTANEOUS ACTION
OF TWO HORIZONTAL COMPONENTS OF GROUND MOTION

S.C Potnis
1
, R.S. Desai
2
and I.D. Gupta
3

1
Professor, Civil Engineering Dept., Rajarshi Shahu College of Engineering, Pune, India, potnissc@yahoo.co.in
2
Asst. Professor, Civil Engineering Dept., Rajarshi Shahu College of Engineering, Pune, India, prof_rsdesai@yahoo.com
3
Director, Central Water and Power Research Station, Pune, India, idgrh4@yahoo.com


ABSTRACT
The paper proposes a new response spectrumsuperposition method for estimating the critical response under
simultaneous action of two horizontal components of ground motion. The proposed method is able to provide
very reliable results vis--vis the true critical response based on exact time-history solution. The usefulness of
the method has been established and illustrated by computing a large number of numerical results for several
different structures and input ground motion.


1. INTRODUCTION
The strong ground motion during an earthquake is commonly measured in the form of acceleration
time histories of two horizontal and one vertical translational component. While analyzing a structure,
only one recorded or predicted horizontal component is applied along a structural axis. This may not,
in general, provide the maximum possible value of the response, termed as critical response in this
paper. Seismic response of a structure depends strongly on the orientation of the structure with respect
to the earthquake location. The critical response for a given earthquake occurs only for one specific
orientation of the structure. As the future earthquakes will be located randomly with respect to a
structure, it becomes necessary to estimate and design a structure for the critical response. To compute
the critical response along a structural axis, various investigators and codes have proposed to combine
the response along the various structural axes obtained by response spectrum superposition methods.
The earliest studies (e.g.; Newmark, 1975; Rosenblueth and Contreras, 1977) proposed the Percentage
Rules, in which the combined response is approximated by the sumof the 100% of the response
resulting from one component and some percentage of the responses resulting from the other two
components. Newmark suggested this contribution to be 40% and Rosenblueth and Contreras as
30%. Many other methods have been subsequently proposed by different investigators to improve
upon the Persentage Rule. For example, Smeby and Der Kiureghian (1985) proposed an extension of
the one directional CQC rule (Der Kiureghian, 1981) to multi-component excitation, which is known
as the CQC3 rule. This rule is based on the elementary concepts of stationary randomvibrations and
assumes the existence of a set of principal directions along which the components of ground motion
are uncorrelated. Sadhu and Gupta (2008) formulated a combination rule on similar lines, which is
shown to performbetter than the CQC3 rule when the building is stiffer to the ground motion.
However they have pointed out that their rule underestimates the response in some cases, and it has the
inconvenient requirement of the response spectra in principal directions. Some investigators have
suggested obtaining the critical response directly by using the response spectrum of the major
principal component of ground motion acting along the structural axis of interest. However, Salzar et
al. (2007) have shown that critical response could occur for an orientation different from the principal
components.
To improve upon the situation, a new method is proposed in this paper to get more reliable estimate of
the critical response directly. In the proposed method, the response along a structural axis is estimated
using a critical response spectrum obtained by finding the maximum value of the spectral amplitude at
each natural period by estimating the resultant time histories of two horizontal components of ground
motion along different angles of rotation. This method is shown to provide truly critical response in
majority of cases.

2. METHODS FOR COMPUTING CRITICAL RESPONSE
The results of three different methods, viz. 30% rule, SRSS method, and Principal component method,
are compared with those froma new method proposed in this paper as well as with the exact critical
response to analyze the relative performance of the various methods. All these methods are described
briefly in the following.

2.1 The 30% Rule
This rule was first proposed by Rosenblueth and Contreras (1977) and is specified in several
regulatory documents (UBC, 1997; IS: 1893, 2002; IBC, 2009 etc.). Based on the assumption that the
two horizontal components are uncorrelated Gaussian processes of equal intensity, it gives the
following equation for the critical response

{ }
0.3 ; 0.3
c x y x y
r Sup r r r r = + + (1)
Here,
x
r represents the maximum absolute response at a particular location along one major axis of the
structure excited by the horizontal x-component of a given earthquake. Similarly,
y
r denotes the
corresponding maximum absolute response at the same location along the other major axis of the
structure by the horizontal y-component of the earthquake. The
x
r and
y
r are estimated using a
suitable response spectrum superposition method with the response spectra of two horizontal
components of ground acceleration and the structural properties along its two major axes. The CQC
method due to Wilson et al. (1981) has been used for this purpose in the present analysis.

2.2 The SRSS Method
This method is specified in several regulatory documents (e.g., UBC, 1997; IBC, 2009), according to
which the critical response is given by

2 / 1 2 2
) + ( =
y x c
r r r (2)
The basic assumption in the SRSS method is that there is no correlation between the horizontal
components.

2.3 Principal Component Method
The critical response can also be obtained directly by response spectrumsuperposition analysis for
only one major structural axis of interest using the response spectrum of the major principal
component of ground motion. Penzien and Watabe (1975) have shown that the as recorded ground
acceleration components can be resolved along three principal directions, such that the three resultant
acceleration time histories are uncorrelated. The two horizontal components are characterized by
maximum and minimum covariance and the vertical component by intermediate covariance. The
component with maximum covariance is termed as the major principal component and its response
spectrumcan be used to get the critical response directly.
The major principal direction is generally found to be directed towards the epicenter, and it does not
vary much during the strong motion portion of the ground acceleration. However, the epicentral
location may not always be known or may suffer fromlarge errors. The major principal component in
the present analysis is therefore obtained by diagonalization of the following covariance matrix of the
two as recorded horizontal components of ground acceleration


yy yx
xy xx


with
( ) ( )
0
1
( ) ( ) ( ) ( )
d
T
ij i i j j
d
a t a t a t a t dt
T
=
}
(3)
Here ) (t a
i
is the mean value of time-history ) (t a
i
over duration T
d
, taken as the entire record length.
The direction of the eigen-vector for the larger-eigen value of the 2 2 correlation matrix represents
the major principal direction. If this makes an angle u with the direction of the recorded x-component,
the accelerogramalong the major principal direction can be defined as
t a t a t a
y x
sin ) ( + cos ) ( = ) ( (4)
The response spectrumof this accelerogramgives the required critical response spectrum.

2.4 Proposed New Method
The response spectrum of the major principal component is not always able to predict the truly critical
response. This paper has therefore proposed to define a critical response spectrum, the use of which is
able to provide more reliable estimate of the critical response compared to the exact time history
solution. Following Boore (2010), this has been termed as Rot100 spectrum, which is obtained as
follows:
- The two as recorded horizontal components of ground acceleration are resolved along an
angle to the recorded x-component using eqn. (4), by varying angle from0 to 180 in
steps of one degree to get 181 different time histories.
- Response spectra are computed for all the 181 time histories, and Rot100 is obtained by taking
the spectral amplitude at each natural period as the maximumof the 181 values.
The amplitudes of Rot100 response spectrumoccur at different angles of rotation for different
natural periods. This spectrum thus does not correspond to a single acceleration time-history.
Therefore, the use of a rotation independent spectrum, RotI100, has been also investigated to arrive at
the estimate of the critical response along a structural axis of interest. The RotI100 spectrum is defined
at all the natural periods for the same angle of rotation, , which corresponds to the minimum value of
the following penalty function (Boore, 2010)
( ) ( ) ( ) [ ]

1 =
2
1 100 ,
1
= penalty
N
i
i i
T Rot T SD
N
(5)
Here ( )
i
T SD , is the displacement response spectrum amplitude at natural period T
i
for the resultant
accelerogram at an angle of rotation to the recorded x-component of motion, Rot100(T
i
) is the
critical response spectrum, and N is the total number of periods used to define the spectrum.

2.5 Exact Critical Response
The exact critical response is obtained simply by computing the maximum absolute value of the
response at a selected location along a structural axis of interest by detailed time history solution using
normal mode theory for all the 181 time histories considered to define the Rot100. The maximum of
all the 181 values is taken to provide the exact critical response.

3. NUMERICAL RESULTS AND DISCUSSION
In order to study the relative performance of the various methods described in the previous section
to compute the critical response of multi-degree-of-freedom (MDOF) structures under simultaneous
action of two horizontal components of ground acceleration, numerical results are computed for a
large number of example buildings excited by several different accelerograms. Brief description of
these buildings and the input accelerograms is provided in the following.











Fig.

Fig. 1: Floor plans of the example five story (left hand figure) and ten story (right hand figure)
buildings.

Table 1: Reference floor masses and story stiffnesses for five and ten-storey example buildings.
Story From Top Floor Masses
(Tonnes)
Storey Stiffness in

x-direction (kN/m)
Storey Stiffness in y-
direction (kN/m)
Five Story Building
1 44.0 14450 17900
2 49.0 20600 28200
3 50.0 30100 44750
4 51.5 44350 70650
5 53.0 44350 70650
Ten Story Building
1 112 30610 37155
2 120 40225 48240
3 120 40225 48240
4 125 60410 82110
5 125 60410 82110
6 130 90805 123375
7 130 90805 123375
8 135 121150 144215
9 135 121150 144215
10 150 162370 192385

The buildings considered include a five and a ten storey symmetric structure with five different
proportions in storey stiffnesses to have a total of 10 example buildings. The floor plans of the two
types of buildings are shown in Fig. 1 and their floor masses and reference story stiffnesses in x- and
y-directions are given in Table 1. Five different values equal to 4, 2, 1, 0.5 and 0.25 of the scaling
factor () for the reference story stiffnesses in both x- and y- directions are considered to get five
different examples of both the five and the ten story buildings. All the example buildings are assumed
to have rigid floor masses supported by massless inextensible columns and a classical damping of 5%
in all the modes of vibration. The stiffness scaling factor to generate the various example buildings
and their fundamental periods T
X
and T
Y
in the x- and y- directions are listed in Table 2, which are seen
to cover a very wide range of 0.389 s to 2.937 s.
Table 2: Fundamental periods of the example buildings generated by using different stiffness
scaling factors.
Building#
Scaling
Factor,
Five Storey Buildings Ten Storey Buildings

T
X
, sec
T
Y
, sec T
X
, sec T
Y
, sec
1 4 0.410 0.337 0.77 0.58
2 2 0.581 0.477 1.09 0.86
3 1 0.821 0.674 1.54 1.26
4 0.5 1.161 0.954 2.08 1.83
5 0.25 1.645 1.348 2.94 2.63

The critical base shear response for each of the ten example buildings is estimated using two-recorded
horizontal components of five different accelerograms as detailed in Table 3. The peak ground
acceleration (PGA) and the predominant period, T
p
, as obtained from the Fourier amplitude spectrum,
are also given for the five selected accelerograms. It is seen that these accelerograms represent very
widely varying ground motion characteristics.

Table 3: Details of the input accelerograms used to compute example results.
Acc.
#
Earthquake Recording Site
x-Component y-Component
Comp.
PGA,
g
T
p
,
sec
Comp.
PGA,
g
T
p
,
sec
1. Borrego Mtn.
EQ. 1968

Engineering Bldg.,
Santa Ana, Orange
County, California
S04E

0.012 5.4 S86W

0.01 2.1
2. Imperial
Valley
EQ. 1940

El Centro Site,
California
S00E

0.33 0.67 S90W

0.21 0.53
3. Kern County
EQ. 1952

Taft Lincoln
School Tunnel,
California
N21E

0.19 0.63 S69E

0.16 0.45
4. Parkfield
EQ. 1966

Array No. 5,
Cholame, Shandon,
California
N05W

0.40 0.34 N85E

0.37 0.31
5. San Fernando
EQ. 1971
Utilities Building,
215 W. Broadway,
Long Beach,
California
N90E

0.027 5.77 N00E

0.018 4.95

The displacement response spectra of the two horizontal components of each accelerogram have been
used to estimate the critical response by 30% rule and the SRSS method. For this purpose, the x and y-
components of ground motion are used to get the maximum response values along the X
S
and Y
S
axes
of each structure by CQC modal combination method. Also, the response spectra of the principal
component, and Rot100 and RotI100 spectra are estimated for each of the five earthquake records and
applied along X
s
and Y
S
axes of each structure to get the critical response estimates directly along the
respective axes without considering the response along the other axis. Finally, the exact critical
responses along X
s
and Y
S
axes are obtained by using the time histories generated by finding the
resultant accelerograms along 181 different directions at one-degree interval.


1 2 3 4 5
-10
-5
0
5
10
15
20
25
30
B u i l d i n g N o.
P

e

r

c

e

n

t

a

g

e


E

r

r

o

r
I m p e r i a l V a l l e y E a r t h q u a k e ( 1 9 4 0 )



1 2 3 4 5
-5
0
5
10
15
20
25
30
35
B u i l d i n g N o.
P

e

r

c

e

n

t

a

g

e


E

r

r

o

r
B o r r e g o M o u n t a i n E a r t h q u a k e ( 1 9 6 8 )




1 2 3 4 5
-10
-5
0
5
10
15
20
25
B u i l d i n g N o.
P

e

r

c

e

n

t

a

g

e


E

r

r

o

r
K e r n C o u n t y E a r t h q u a k e ( 1 9 5 2 )


30% Rule
SRSS
RotI100
Rot100
Pri.Comp.


Fig. 2: Comparison of errors in critical base shear response along X
S
axis of the
various five-story buildings as determined by different methods.

1 2 3 4 5
-10
-5
0
5
10
15
20
B u i l d i n g N o.
P

e

r

c

e

n

t

a

g

e


E

r

r

o

r
I m p e r i a l V a l l e y E a r t h q u a k e ( 1 9 4 0 )




1 2 3 4 5
-15
-10
-5
0
5
10
15
20
25
30
35
B u i l d i n g N o.
P

e

r

c

e

n

t

a

g

e


E

r

r

o

r
B o r r e g o M o u n t a i n E a r t h q u a k e ( 1 9 6 8 )




1 1.5 2 2.5 3 3.5 4 4.5 5
-10
-5
0
5
10
15
20
25
30
B u i l d i n g N o.
P

e

r

c

e

n

t

a

g

e


E

r

r

o

r
K e r n C o u n t y E a r t h q u a k e ( 1 9 5 2 )


30% Rule
SRSS
RotI100
Rot100
Pri.Comp.


Fig. 3: Comparison of errors in critical base shear response along X
S
axis of the
various ten-story buildings as determined by different methods.


Typical examples of the percentage errors in the critical response amplitudes obtained by various
methods with respect to the exact method are shown in Figs. 2 and 3 for the five and ten story
buildings, respectively. It is seen that 30% rule and SRSS methods are over conservative for majority
of structures and input excitations considered. But the conservatismis highly variable, and in some
cases the 30% rule leads to unsafe results also. The results of the principal component method and that
from the use of RotI100 spectrumare seen to be quite close and are characterized by very small errors.
But, both these methods result in unsafe critical response estimates for all the structures and all the
input excitations considered. The method based on the Rot100 response spectrum is also found to be in
good agreement with the exact critical response and it always provides somewhat conservative
estimate. The use of Rot100 is therefore recommended as a new approach to obtain the most
reasonable and optimumestimate of the critical response under simultaneous action of two horizontal
components of ground acceleration. Also, for regular and symmetric buildings, this method is able to
provide directly the critical response along a structural axis without considering the response along the
other axis.
To get an overall idea about the relative performance of various methods for estimation of the critical
response, Table 4 summarizes the results on maximumvalues of the errors on both positive and
negative sides in the results obtained along two major axes for all the 10 buildings and 5 excitations. It
can be noticed that the SRSS method overestimates the critical response by as much as 36.54%, and
the underestimation by this method is insignificant. The errors in 30% rule are significant on both
positive and negative side with overestimation up to 27.96% and underestimation up to 10.28%. But,
as seen from Figs. 2 and 3, the underestimation occurs in a few cases only. The principal component
method and the use of RotI100 spectrum always result in underestimation of the critical response, with
maximum of the order of 18.54 % and 13.26% respectively. The proposed Rot100 method never
underestimates the critical response and is not too conservative also, with the largest overestimation as
9.36 % only.

Table 4: Summary of the maximum errors in the critical response obtained along the two axes of the
10 example buildings and five different earthquake excitations by different methods.

Structural
Axis
Maximum % Error in Different Methods
30% Rule
SRSS
Method
Principal
Component
RotI100
Method
Proposed
Method

X
S

27.96 36.54 0 0 9.12
- 9.33 - 0.04 -17.19 -12.89 0

Y
S


25.74 35.86 0 0 9.36
- 10.28 - 0.03 -18.54 -13.26 0


4. CONCLUSIONS
The paper has investigated in detail the various issues related to the estimation of the maximum
response, termed as critical response, of buildings under simultaneous action of two horizontal
components of earthquake ground motion. The responses estimated by various methods in vogue are
compared with the exact critical response obtained by detailed time history solution. Similar analysis
is also carried out for the first time using Rot100 and RotI100 response spectra defined in the paper.
This comparative study indicated that the 30% rule being used in several codes, including the Indian
Standards IS: 1893-1984, is not always able to ensure the desired conservatism. The SRSS method is
in general able to provide conservative estimates of the critical response, but overestimation may
sometimes be too large. These observations are contrary to some past studies (e.g., Wilson et al., 1995,
Salazar et al, 2004), which have indicated the 30% and SRSS methods to be non-conservative. Menun
and DerKiureghian (1998) compared the results of the CQC3 rule with those of the SRSS, the 30%
and the 40% rules and examined the shortcomings of these three rules. The research works of Davila
et al. (2000) and Zaghlool et al. (2000) have yielded evidence that percent rules and the SRSS rules are
inappropriate to take into account the effects of horizontal orthogonal components of ground motion.
The use of both the principal component and the RotI100 spectrumunderestimates the response in all
the cases, with the largest underestimation reaching up to 18.54% and 13.26%, respectively. The use
of the Rot100 spectrum is found to ensure the desired conservatismwith reasonable overestimation
within 10% only.
In view of the present investigations, the use of Rot100 response spectrumis proposed as a new
method to obtain directly the critical response along any of the major axes of a building, obviating the
need of estimating the response along all the major axes of the building and using a combination rule.
The existing practice of using the response spectrumof the major principal component for this purpose
is found to be exceeded by the exact critical response in more than 75 % of the cases considered in the
present analysis. Rot100 method is therefore recommended as a new approach to obtain the most
reasonable and optimumestimate of the critical response under simultaneous action of two horizontal
components of earthquake ground motion.


REFERENCES
Boore, D.M. (2010). Orientation-independent, Nongeometric-mean Measures of Seismic Intensity
from two Horizontal Components of Motion, Bull. Seism. Soc. Am., 100(4), 18301835.
DerKiureghian, A.D. (1981). A Response Spectrum Method for Random Vibration Analysis of
MDOF Systems, Earthq. Engg. & Struct. Dyn., 9, 419-436.
Dvila, F., Cominetti, S. and Cruz, E. (2000). Considering the Bi-directional Effects and the Seismic
Angle Variations in Building Design, Procs. 12th World Conf. Earthq. Eng., New Zealand.
Indian Standard 1893 - Part 1 (2002). Bureau of Indian Standards, New Delhi.
IBC (2009). International Building Code, International Code Council, Falls Church, VA.
Menun, C. and Der Kiureghian, A. (1998). A Replacement for the 30%, 40% and SRSS Rules
for Multi-component Seismic Analysis, Earthquake Spectra, 14(1): 153-156.
Newmark, N.M. (1975). Seismic Design Criteria for Structures and Facilities: Trans-Alaska Pipeline
System, Procs. U.S. Nat. Conf. on Earthq. Eng., Earthquake Engineering Institute, 94-103.
Penzien, J . and Watabe, M. (1975). Characteristics of 3-Dimensional Earthquake Ground Motion,
Earthq. Eng. & Struct. Dyn., 3, 365 374.
Rosemblueth, E. and Contreras, H. (1977). Approximate Design for Multi-component Earthquakes,
J our. Eng. Mech. Div., ASCE, 103, 895-911.
Sadhu, A. and Gupta, V.K. (2008). A Modal Combination Rule for Ordered Peak Response Under
Multi-component Ground Motion, ISET J ournal of Earthquake Technology, 45, 7996
Salazar, A.R., J urez-Duarte, J .A., Lpez-Barraza, A. and Velzquez-Dimas, J .I. (2004). Combined
Effect of the Horizontal Components of Earthquakes for Moment Resisting Steel Frames, Steel &
Composite Structures an International J ournal, 4(3): 89-209
Salazar, A.R., Lpez-Barraza, A., Lpez, L.A., Dimas, J.I. and Sinaloa, U.A (2007). Some Issues
Related to Multi-component Seismic Analysis of Steel Moment Frames, J our. of Engineering , 1( 1).
Smeby, W. and DerKiureghian, A. (1985). Modal Combination Rules for Multi-component
Earthquake Excitation, Earthq. Engg. & Struct. Dyn., 13, 1-12.
UBC (1997). Uniform Building Code Vol. 2, Structural Engineering, International Conference of
Building Officials (ICBO), Design Provisions, Whittier, CA, 492.
Wilson, E.L., DerKiureghian, A. and Bayo, E.P. (1981). A Replacement for the SRSS Method
in Seismic Analysis, Earthq. Eng. & Struct. Dyn., 9, 187-194.
Wilson, E.L., Suharwardy, I., and Habibullah, A. (1995). A clarification of the orthogonal effects in a
three-dimensional seismic analysis, Earthq. Spectra, 11, 659-666.
Zaghlool, B.S., A.J . Carr and P.J . Moss (2000). Inelastic behavior of three-dimensional structures
under concurrent seismic excitations, Proc. 12th World Conference on Earthquake Engineering, New
Zealand.

You might also like