Académique Documents
Professionnel Documents
Culture Documents
ABSTRACT
Ground motion records and hazard spectra used for non-linear response history analysis are the
important factors influencing the evaluation of seismic demand of tall buildings near active faults. Tall
buildings located near active faults are likely to be subjected to earthquake ground motions containing
high energy pulse, which may impose large seismic demand on the buildings. Design engineers usually
follow the ASCE 7(2005) standard to design such buildings which does not have any special provision
for selecting and scaling ground motions for near-fault buildings. However, Tall Buildings Initiative
(TBI (2010)) has developed a state-of-the-art procedure for selecting and scaling ground motions which
addresses this issue. This study compares seismic demand of a near-fault tall building using these two
different procedures for selecting and scaling ground motions. Two types of hazard spectra (Uniform
hazard spectrum & multiple Conditional mean spectrum) and different sets of ground motions matched
with those spectra have been used. A 40 story reinforced concrete core wall building has been used as
a case study building. Response spectrum analysis (RSA) and non-linear response history analysis
(NLRHA) were carried out to compare elastic and inelastic demand of the building respectively. The
findings of this study suggests that the advance ground motion selection and scaling procedure of
TBI(2010) based on better understanding of near-fault seismic hazard gives less seismic demand for
near-fault tall buildings than the ASCE 7(2005) code based procedure.
Keywords: Pulse-like ground motion, Near-fault tall buildings, Conditional mean spectrum, Uniform
hazard spectrum
INTRODUCTION
Ground motions close to ruptured fault have been found to cause severe damage to the structures located
nearby fault zone (Northridge 1994, Kobe 1995, Chi Chi 1999) than those located far away from such
zone. Near-fault zone typically comprises of the area located within a distance of 20 km from ruptured
fault. Near-fault ground motion contain special type of pulse-like motion in either acceleration, velocity
or displacement time history (Bertero, et al., 1978)(Loh, et al., 2000). The response of structures during
an earthquake depends directly on the level of ground shaking and in case of near-fault structures on the
nature of ground motion time history. Characteristic short duration impulsive nature of near-fault motion
at the beginning of the record imposes high input energy into the buildings causing devastating damage
(Chopra et al., 1976) (Bertero et al., 1978). Near-fault motions (mostly pulse-like) can have devastating
effects to medium and high rise buildings in city like Manila in Asia where the fault-line runs across the
city centre.
The response of near-fault buildings is also dependent upon the spectral shape of ground motion records.
The parameter (normalized difference between spectral acceleration of a recorded ground motion with
median predicted spectral acceleration using ground motion prediction equation) which is used in
ground motion prediction equation is an indicator of spectral shape(Baker, 2005).The near-fault records
are usually associated with high positive value of (>1). So, the selection of ground motion for the
dynamic analysis of near-fault buildings should not only consider the dominant scenario earthquake
represented by magnitude (Mw) and distance(R) but also records with similar target spectral shape or
value. Baker & Cornell (2006) developed a new kind of hazard spectrum known as conditional mean
1
spectrum (CMS) and found that we can alternatively use CMS which is representative of causal M, R,
at the site and select ground motions which has similar spectral shape as that of the constructed CMS.
CMS is a spectrum whose spectral values at different periods are conditioned on a target spectral
acceleration value(Sa) at a structural period of interest(T*) (Baker & Cornell,2006) (Baker,2011). The
Sa(T*) corresponds to a causal earthquake(M,R,) whose M,R & conditional mean value is used to
compute mean spectral values at other periods. General seismic design code like the ASCE 7(2005)
recommends the use of uniform hazard spectrum(UHS), as a target spectrum, which is an envelope
spectrum computed independently for a range of time periods for the same level of probability of
exceedance. UHS does not represent the spectrum from single scenario earthquake (M, R, ) and it is
impossible for a real ground motion to have such a spectral shape with very high positive value at all
the periods. At low probability of exceedance (say 2% in 50 years) ordinates of UHS are always higher
than median predicted response spectrum for dominant earthquake scenario (M,R). Even if a ground
motion has spectral acceleration as high as UHS, it can only be at single period but not at multiple
periods. So, a ground motion having spectral acceleration same as UHS for a range of periods must have
a lower probability of exceedance than that of UHS itself. Thus, the use of UHS as a target spectrum for
any site will be too conservative (Baker & Cornell, 2006).
In this study, focus has been given to selection process of ground motion records using the conventional
code-based approach of the ASCE 7(2005) with UHS as target spectrum. And, another approach used
is recently proposed state of the art recommendation by TBI (2010) with special provision for selection
of ground motion records for near-fault buildings using of multiple site specific CMS.
SEISMIC HAZARD
Overview of Building and Site
The location of the building used for this study is 409 Shaw, Manila, Philippines. The nearest active
fault posing substantial threat to the building is West Marikina Valley Fault(WMVF) located at a
distance of around 400m.This fault is a strike-slip fault with slip rate of 7 to 10 mm/yr. and is a major
threat to most of the tall buildings in Manila. A 40 story tall reinforced concrete building designed
following Los Angeles Tall Building Structural Design Council (LATBSDC) 2005 has been used as a
case study building for this study (Munir & Warnitchai, 2012). The case-study building is 415 feet tall
with a typical story height of 10 feet and 20 feet lobby height. It has three levels of garage parking in
the basement with a story height of 10 feet each whose foundation rests on a firm bedrock.
Hazard Spectra
The site specific uniform hazard spectrum and de-aggregation results required in this study was used
from the Probabilistic Seismic Hazard Assessment(PSHA) report of Manila,Philippines prepared by
AIT Consulting(AIT Consulting,2012).The earthquake event(M,R,) controlling MCE level hazard at
409 Shaw site for first three mean translation natural period of the building are shown in Table 1.It can
be seen that the fault controlling MCE level hazard at all three periods is the same and is located very
near to the site (400 m only).There are no distant sources(faults) controlling hazard at any periods at the
site. Thus, the building at the site will be highly affected by ground motions coming from the nearby
fault(some may be pulse-like motions).Two types of hazard spectrum has been used in this study, viz.
uniform hazard spectrum at MCE level and multiple conditional mean spectra.The uniform hazard
spectrum at MCE level corresponds to pseudo-acceleration response spectrum having 5% equivalent
viscous damping with 2% probability of exceedance in 50 years (ASCE 7, 2005).And,CMS has been
used as another hazard spectrum for multiple periods of the case study building following TBI(2010)
recommendation. Inaddition to CMS, Conditional Spectrum(CS) has also been used to account for
variability in spectral response at other periods.As the case study building is a tall MDOF inelastic
structure, it will be sensitive at multiple periods (especially to 1st, 2nd and 3rd mean mode periods) due
to higher mode effect and lengthened effective first mode period. So, multiple CMS spectra and CS were
constructed for selecting ground motion records to perform NLRHA as shown in Fig. 3(a).
selected records using different hazard spectra are compared in Fig. 3(b) to see which response spectra
has higher hazard level than UHS. It can be clearly seen that the hazard at lower periods is higher for
records matched to UHS rather than records matched to CMS. At higher periods, the hazard due to
records matched to UHS and CMS is similar.
COMPARISON OF SEISMIC DEMAND
Two separate models of the 40 story reinforced core wall building were used for elastic and inelastic
analysis of the building. A three dimensional linear elastic finite element model of the building was
made in ETABS version 9.7.4 software(ETABS (2008)) for RSA. And, a three dimensional non-linear
inelastic model of the structure system made by Munir & Warnitchai (2012) in Perform 3D version 4
software (Perform-3D(2006)) was used for NLRHA. In the non-linear model, bottom portion (level 0 to
level 5) of the core wall is expected to behave as a plastic region and the remaining portion (level 6 to
roof) is expected to remain elastic. So, the bottom portion was modeled using inelastic shear wall (fiber
model) elements and upper portion was modeled using elastic shear wall elements. Coupling beams are
modeled using elastic beam elements with plastic moment hinges at the two ends. Columns and slabs
are modeled as elastic column and shell elements respectively. The geometric non-linearity (P-) effects
has been considered. In the non-linear model, damping ratios in first six translation modes has been set
to 1%, 1.4%, 2%, 2.7%, 3.8% and 5.3% using formula i = 1.4i-1,where i is damping ratio of ith mode
(Satake, et al., 2003).In Perform 3D,these damping ratios have been applied using modal damping model.
Additionally, 0.01% of Rayleigh damping has been used to dampen the response of higher modes in the
structure.
Elastic demand of the core walls were computed using response spectrum analysis. UHS at MCE level
was used as one of the response spectrum loading and multiple CMS were used as another type of
loading. Demands were computed in both the directions without any reduction by response modification
factor. Elastic base shear demand obtained from RSA using ASCE 7(2005) approach was found to be
19% and 16.5% more than the base shear demand obtained using TBI (2010) approach along the two
principle directions of the building as shown in Fig. 4. Elastic moment demand from RSA using the
ASCE 7(2005) approach were found to be similar to that computed using TBI (2010) approach in both
the principle directions as shown in Fig. 5.
Inelastic demand of the core walls have been computed by NLRHA using ground motions selected using
two different approaches. Since, the same model has been used, ground motions selected will be the
major factor influencing the seismic demand.The inelastic base shear demand obtained from NLRHA
using ASCE 7(2005) approach was found to be 35% and 23.5% more than the base shear demand
obtained using TBI (2010) approach along the two principle directions of the building as shown in Fig.
6. Since, there is flexural yielding at the bottom portion of the core wall, inelastic story moment demand
along the mid-height of the building was checked. It was found that inelastic story moment obtained
from NLRHA using ASCE 7(2005) approach was found to be nearly equal to that obtained using TBI
(2010) approach in both principle directions of the building as shown in Fig. 7.
CONCLUSIONS
In this study, seismic demand of high rise reinforced concrete building located near an active fault has
been computed using two different approaches for selecting ground motions and hazard spectra. It was
found that base shear demand of the core wall computed using TBI(2010) approach for ground motion
selection and scaling was less than that using ASCE 7(2005) code-based approach. The base shear
demand from CMS at T2 was found to be maximum among the three CMS used. This means that the
spectral demand of near-fault tall buildings at second mode (not first mode as is typically assumed)
governs the elastic/inelastic base shear design demand. One of the drawback of using TBI(2010)
approach is additional amount of computational time and effort required to analyze the building.
However, TBI (2010) approach considers most of the parameters affecting near-fault tall buildings
explicitly unlike the ASCE 7(2005) code based approach which has no special provision for such
buildings. Hence, the results obtained by following ground motion selection procedure of TBI (2010)
can be used for the design of near-fault tall buildings with more reliability than the ASCE 7(2005) code
based approach used for general type of building.
Table 1. De-aggregation at natural time periods of the case study building from uniform hazard
spectrum at MCE level
T(s)
3.58
0.77
0.34
M
7.28
7.26
7.25
10
10
1.12
1.21
1.40
R (km)
0.40
0.40
0.40
0.1
1605
821
GM
0.01
0.01
1176
CMS (3.58s)
CS
0.1
0.1
182
170
1602
CMS (0.77S)
GM
CS
0.01
0.01
10
0.1
Period (s)
10
Period (s)
(a)
(b)
Figure 1. Ground motion records linearly scaled to CMS at (a)T1 and (b)T2
10
10
1
1119
174
1605
CMS (0.34s)
GM
CS
0.1
0.01
0.01
0.1
10
Period (s)
0.1
1.17UHS
A.M. of SRSS Sa
727
1148
1617
864
879
1111
1787
0.01
0.01
0.1
Period (s)
10
(a)
(b)
Figure 2. (a) Ground motion records linearly scaled to CMS at T3 (b) Scaled SRSS Sa of selected
ground motion records with average SRSS Sa and UHS at MCE level scaled by 1.17.
Table 2. Rotated ground motion records selected using CMS (T1) as target spectrum
NGA
#
1605
1176
821
Earthquake
Event
Duzce-Turkey
Kocaeli-Turkey
Erzican-Turkey
Year
1999
1999
1992
Scale
Factor
1.7695
1.413
2.6845
Tp (s)
Mw
-- 5.6
4.5 4.6
2.7 2.2
7.14
7.51
6.69
Rrup
(km)
6.6
4.8
4.4
Vs30
(m/s)
276
297
274.5
Duration
(s)
25.885
35.000
20.775
Table 3. Rotated ground motion records selected using CMS (T2) as target spectrum
NGA
#
170
1602
182
Earthquake
Event
Imperial Valley-06
Duzce-Turkey
Imperial Valley-06
Year
1979
1999
1979
Scale
Factor
4.1906
1.6843
1.7481
Tp (s)
Mw
6.53
7.14
6.53
Rrup
(km)
7.3
12
0.6
Vs30
(m/s)
192.1
326
210.5
Duration
(s)
39.985
55.900
36.820
Table 4. Rotated ground motion records selected using CMS (T3) as target spectrum
NGA
#
1119
174
1605
Earthquake
Event
Kobe-Japan
Imperial Valley-06
Duzce-Turkey
Year
1995
1979
1999
Scale
Factor
1.8322
2.2765
1.8316
Tp (s)
Mw
6.9
6.53
7.14
Rrup
(km)
0.3
12.4
6.6
Vs30
(m/s)
312
196.2
276
Duration
(s)
40.960
39.035
25.885
Table 5. As-recorded Ground Motion Records selected using UHS as target spectrum
NGA
#
879
1787
1111
864
1148
727
1617
Earthquake Event
Year
Landers
Hector Mine
Kobe- Japan
Landers
Kocaeli-Turkey
Superstition Hill-02
Duzce-Turkey
1992
1999
1995
1992
1999
1987
1999
Scale
Factor
3.3614
5
5
4.4477
8
8
7
Tp (s)
Mw
7.28
7.13
6.9
7.28
7.51
6.54
7.14
Rrup
(km)
2.2
11.7
7.1
11
13.5
5.6
3.9
Vs30
(m/s)
684.9
684.9
609
379.3
523
362.4
424.8
Duration
(s)
48.125
45.310
40.960
44.000
30.000
22.200
41.500
10
UHS
0.1
CMS (3.58s)
CMS (0.77s)
CMS (0.34s)
0.1
Median Spectrum
0.01
0.01
0.1
Period (s)
10
0.01
0.01
Target Spectrum
UHS
CMS1
CMS2
CMS3
T1 (3.58s)
T2 (0.77s)
T3 (0.34s)
0.1
Period (s)
10
(a)
(b)
Figure 3. (a) Conditional mean spectra at T1, T2 and T3 along with UHS and median spectrum for 409
Shaw site all having 2475 years mean return period (b) Comparison of geometric mean
spectral acceleration response spectrum of ground motions scaled to UHS and CMS (T1, T2,
T3) at MCE level.
40
35
35
CMS1
30
CMS2
25
CMS3
UHS
20
15
Level No.
Level No.
30
CMS1
CMS2
CMS3
UHS
25
20
15
10
10
5
0
0
0
20
40
60
20
40
60
80
(a)
(b)
Figure 4. Comparison of elastic story (a) shear demand of core walls using UHS and multiple CMS
along (a) 1-1 axis (b) 2-2 axis
40
40
35
35
CMS1
30
30
CMS3
20
UHS
Level No.
Level No.
CMS2
25
15
25
20
15
10
10
CMS1
CMS2
CMS3
UHS
Moment (Kip-in x
107)
10
(a)
(b)
Figure 5. Comparison of elastic story moment of core walls using UHS and multiple CMS along
(a) 1-1 axis (b) 2-2 axis
40
40
35
35
Level No.
30
25
20
15
CMS1
30
CMS3
UHS
20
15
10
10
CMS2
25
Level No.
CMS1
CMS2
CMS3
UHS
10
20
30
10
15
20
(a)
(b)
Figure 6. Comparison of inelastic story shear demand by NLRHA using UHS and multiple CMS (a)
along 1-1 axis (b) along 2-2 axis
35
CMS1
CMS2
CMS3
UHS
35
30
Story (N)
30
Story (N)
40
CMS1
CMS2
CMS3
UHS
25
20
25
20
15
15
10
10
5
0
0
0
0.5
1
1.5
Moment (Kip-in x 107)
0.5
1
1.5
2
Moment (Kip-in x 107)
2.5
(a)
(b)
Figure 7. Comparison of inelastic story moment demand by NLRHA using UHS and multiple CMS (a)
along 1-1 axis (b) along 2-2 axis.
REFERENCES
AIT Consulting. (2012). Probabilistic Seismic Hazard Assessment for Metro Manila, Manila, Philippines:
Sy^2+Associates Inc.
ASCE/SEI. (2005). ASCE Standard: Minimum Design Load for Buildings and Other Structures, Reston, Virginia:
American Society of Civil Engineers.
Baker, J. W., (2011). Conditional Mean Spectra: Tool for Ground Motion Selection. Journal of Structural
Engineering, 137(3), pp. 1-25.
Baker, J. W. & Cornell, C. A. (2005).A vector-valued ground motion intensity measure consisting of spectral
acceleration and epsilon. Earthquake Engineering and Structural Dynamics, 34(10), pp. 1193-1217.
Baker, J. W. & Cornell, C. A. (2006). Spectral Shape, Epsilon and Record Selection. Earthquake Engineering and
Structural Dynamics, 35(9), pp. 1077-1095.
Bertero, V., Mahin, S. & Herrera, R. (1978). Aseismic design implications of near-fault San Fernando earthquake
records. Earthquake Engineering & Structural Dynamics, 6(1), pp. 31-42.
Chopra, A. K., Bertero, V. V. & Mahin, S. A., (1976). Response of the Olive View Medical Center Main Building
during the San Fernando Earthquake.
ETABS. (2008). ETABS:Extended 3D Analysis of Building Systems. Berkeley: Computer and Structures,Inc.
Loh, C. H., Lee, Z. K., Wu, T. C. & Peng, S. Y., (2000). Ground motion characteristics of the Chi-Chi earthquake
of 21 September 1999. Earthquake Engineering & Structural Dynamics, 29(6), pp. 867-897.
Munir, A. & Warnitchai, P., (2012). The cause of unproportionately large higher mode contributions in the inelastic
seismic responses of high-rise core-wall buildings. Earthquake Engineering and Structural Dynamics, 41(15),
pp. 2195-2214.
NEHRP Consultants Joint Venture. (2011). Selecting and Scaling Earthquake Ground Motions for Performing
Response-History Analyses, Gaithersburg, Maryland: National Institute of Standards and Technology.
PEER
Ground
Motion
Database.
(2010).
Pacific
Earthquake
Engineering
Center.http://peer.berkeley.edu/peer_ground_motion_database. (Accessed March 2013).
Research
Perform-3D. (2006). Perform 3D: Non linear Analysis and Performance Assessment for 3D Structures. Berkeley:
Computers and Structures,Inc.
Satake, N. et al. (2003). Damping Evaluation using full scale datas of buildings in Japan. Journal of Structural
Engineering, 129(4), pp. 470-477.
TBI Guidelines Working Group. (2010). Guidelines for Performance-Based Seismic Design of Tall Buildings,
Berkeley, California: Pacific Earthquake Engineering Research Center.