The Thirtieth KKHTCNN Symposium on Civil Engineering

November 2-4, 2017, Taiwan

Application of CMS ground motions in nonlinear dynamic analysis of tall

buildings
* Kimleng Khy1 and Chatpan Chintanapakdee2
1, 2
Department of Civil Engineering, Chulalongkorn University, Bangkok 10330, Thailand
E-mail: 1khy_kimleng@yahoo.com, 2chatpan.c@chula.ac.th

ABSTRACT

The most commonly used target spectrum in selecting and scaling of records to be used in nonlinear response history analysis
(NLRHA) is uniform hazard spectrum (UHS) and conditional mean spectrum (CMS). When using CMS with tall buildings,
higher mode periods should also be considered. The application of CMS ground motions in nonlinear dynamic analysis of tall
buildings is investigated in this study. Four existing tall buildings subjected to earthquake excitations in Bangkok were
employed. NLRHA was conducted using multiple sets of CMS ground motions where period of interest T * were considered
at the first-three translational mode periods of the building in each direction. It was found that CMS ground motions
conditioned at higher mode periods provide larger force demands than CMS conditioned at the fundamental period for some
locations along the height of the building. Envelope of demands from using multiple sets of CMS ground motions conditioned
at different periods should be used in design, which requires significant computational effort. Using the UHS spectral matching
ground motions can provide results close to such envelope and reduce computational effort significantly.

INTRODUCTION

Nonlinear dynamic analysis known as nonlinear response history analysis (NLRHA) becomes more popular analysis
method in evaluation of existing buildings and design of new buildings. The NLRHA is performed by using ground motion
records that are selected and scaled to match with a defined target spectrum. The most commonly used target spectrum is
uniform hazard spectrum (UHS) and conditional mean spectrum (CMS). UHS has the same probability of exceedence at all
periods from consideration of many scenarios of earthquakes. The spectral shape of UHS may not be resembled to that of a real
ground motion. It is unlikely for the ground motion from earthquake scenario contributing to UHS at the period of interest to
have as large spectral value as the UHS at all other periods. Using UHS spectral matching ground motions would be overly
conservative to estimate responses to an earthquake scenario or the develop a fragility function (Baker and Cornell 2006, Baker
2011). On the other hand, CMS approach proposed by Baker (2011) is attractive as it can preserve realistic spectral shape of
earthquakes. CMS will match the ordinate of UHS only at the period of interest T * . However, the choice of period of interest
T * when using CMS with tall buildings is not definite as higher modes could be as important as the fundamental mode. Higher
modes could be more dominant for some response parameters. For tall buildings, PEER TBI (2010) recommends to use
multiple sets of CMS ground motions conditioned at the first-three translational periods of the building in each direction
considered. In this case, the design demand values are obtained by enveloping response results from using CMS conditioned at
different periods. Therefore, significant computational effort will be required when using CMS as target spectrum for tall
buildings. When using spectral matching ground motions and UHS as target spectrum, only one set of analysis is required. This
study attempts to compare the seismic demands of tall buildings computed from NLRHA using envelope results of multiple
CMS at different periods and using spectral matching ground motions with UHS as target spectrum.

CASE STUDY BUILDINGS

Four existing tall RC shear wall buildings located in Bangkok, capital city of Thailand, were employed. These 15-, 20-, 31-,
and 39-story buildings are denoted as BL1, BL2, BL3, and BL4, respectively. BL2, BL3, and BL4 have podium at the first few
stories and one tower continuing up to the top roof, which is typical style of tall buildings in Bangkok. The primary lateral force
resisting system consists of RC core walls and shear walls. The gravity load carrying system is RC columns with post-tensioned
flat slabs. Responses to excitation in only X-direction of the building are presented in this paper. Typical floor plans and three
dimensional models of the studied buildings are shown in Fig. 1. Basic characteristics of them are summarized in Table 1.

1
Graduate Student
2
Assistant professor
 9th to 39th floor
1st to 15th floor

4th to 20th floor 7th to 31th floor

15-story (BL1) 20-story (BL2) 31-story (BL3) 39-story (BL4)

Fig. 1. Plans and three dimensional models of case study buildings.
Table 1. Basic characteristics of case-study buildings.
Building BL1 BL2 BL3 BL4
Total height (m) 55.40 54.50 89.95 125.55
Typical story height (m) 3.20 2.75 2.85 3.20
RC wall section area / floor area 0.012 0.022 0.012 0.015
RC column section area/ floor area 0.012 N/A 0.013 0.013
Wall thickness (m) 0.25 0.20 0.30 0.35
Max column size (m x m) 1.2x0.6 N/A 1.8x0.5 1.8x0.8
T1 2.76 1.38 4.29 4.63
Natural period of the first-three
T2 0.53 0.32 1.12 1.06
translational modes in X direction (sec)
T3 0.21 0.14 0.51 0.48

EARTHQUAKE GROUND MOTIONS

The maximum considered earthquake (MCE) ground motions were employed. CMS ground motions were obtained for six
periods of interest ( T * ): 0.2, 0.5, 1, 1.5, 2, and 3 sec. Since Bangkok is located on soft soil layers, three pairs of ground motions
were selected and scaled for target spectrum on rock site, and then they were simulated to propagate through soft soil layers
underlying downtown area of Bangkok by software SHAKE2000 (Ordonez 2012) for each of the six periods of interest. The
six sets of output ground motions are considered as CMS soft soil ground motions. Detailed information of those ground
motions and soil properties can be found in Poovarodom et al. (2017). The average spectrum of six records in each set of period
of interest represents UHS spectral ordinate at that period of interest as CMS usually matches UHS at the period of interest.
UHS and CMS for 2.5% damping ratio of soft soil ground motions are shown in Fig. 2. When using CMS in NLRHA, CMS at
conditioned periods closest to the periods of the first-three translational modes of the building in the direction considered were
used. CMS ground motions in the set for period of interest of 3 sec were modified by SeismoMatch program (SeismoSoft
2016) to have spectral shape matched with UHS and they are so-called UHS spectral matching ground motions. The individual
matched spectra, and the mean matched spectrum to the UHS for 2.5% damping ratio are shown in Fig. 3. For tall buildings,
damping ratio of 2.5% was used in this study (PEER TBI 2010).

Fig. 2. UHS and CMS for 2.5% damping ratio of soft-soil Fig.3. Matched spectra and target spectrum for 2.5%
ground motions. damping.
ANALYTICAL MODEL AND ANALYSIS CONSIDERATION

Nonlinear structural models were created in PERFORM-3D program (CSI 2011) for conducting NLRHA. RC walls were
modeled using inelastic fiber shear wall elements representing concrete and steel material. Nonlinear fiber modeling was
incorporated over the entire height because flexural yielding may occur at any location due to higher-mode effects. The material
stress-strain relationship for concrete proposed by Mander et al. (1988) was adopted. Steel fiber was defined by bilinear
inelastic behavior. The expected material strength was used to account for the fact that actual material strengths are generally
greater than nominal strengths specified by designers. RC columns were modeled by combination of linear elastic element with
nonlinear plastic hinge zone at both ends. Plastic zones were modeled by inelastic fiber element in similar manner to RC wall.
RC beams included middle elastic portion and rotational plastic hinge elements at both ends. Plastic hinge properties were
defined by tri-linear moment-hinge rotation. Joints between members were considered to be rigid connections. Slabs were
assumed to be elastic and modeled by elastic shell elements. Rigid floor diaphragm was assigned to each floor level.
NLRHA was conducted using two different types of ground motions. The first type consists of three sets of CMS ground
motions corresponding to conditioned periods closest to periods of the first-three translational modes of the building in the
direction considered. The second type is the set of six UHS spectral matching ground motions as shown in Fig. 3.

RESULTS

The results presented below are the mean values of responses considered: story shear force normalized by weight of the
building (Vstory /W), story overturning moment (Mstory), story displacement and story drift. Only results of 15-story and 39-story
buildings are presented. CMS conditioned at short periods (higher mode periods) resulted in larger force demands than CMS
conditioned at long period (first-mode period) for some locations along the height of the building. For instance, CMS
conditioned at short periods cause significantly larger shear force at the base region, and they cause larger overturning moment
at upper stories compared to CMS conditioned at long period as shown in Fig. 4. Therefore, design demand values should
conservatively consider the envelope of force demands from NLRHA using CMS ground motions conditioned at multiple
periods. For story displacement and story drift demands, the CMS at the fundamental period is sufficient for conducting the
evaluation as it provides larger results than the CMS at higher mode periods throughout the height of the building as shown in
Fig. 4. This is because story displacement and story drift are dominantly contributed by the first-mode response.
Fig. 5 compares the enveloped results from NLRHA using three sets of CMS ground motions conditioned at three different
periods (Envelope CMS in the Figure) and results from NLRHA using UHS spectral matching ground motions (UHS in the
Figure). It was found that the using UHS spectral matching ground motions provides similar force demands and slightly larger
story displacement and drift demands compared to the enveloped results from NLRHA using CMS conditioned at multiple
periods. Therefore, using the UHS as target spectrum to select and scale ground motions can take care of the enveloping task
when using CMS; hence, it can significantly reduce computation efforts because only one set of analysis is required as
compared to multiple sets of analysis required when using the CMS as target spectrum.

a) 15-story building

b) 39-story building
Fig. 4. Comparison of results from NLRHA using three sets of CMS ground motions conditioned at three different periods.
 a) 15-story building

b) 39-story building
Fig. 5. Comparison of NLRHA results from the envelope of CMS and from UHS spectrum matching ground motions.

CONCLUSIONS

Four real tall buildings were used in this study to compare seismic responses computed by NLRHA using CMS and UHS
as target spectrum in selecting and scaling of ground motions. When using CMS as target spectrum, higher mode periods
should also be considered for response parameters dominated by higher modes such as base shear force, and mid-height
overturning moment. Enveloping of force demands computed from NLRHA using CMS ground motions conditioned at
multiple periods should be undertaken before used as design demand values. For response quantities dominated by the
fundamental mode such as story displacement and story drift ratio, CMS ground motions conditioned at the fundamental period
is sufficient to be used in the analysis. Using UHS as target spectrum can avoid enveloping of results and significantly reduce
computation effort as required when using multiple sets of CMS ground motions conditioned at multiple periods.

ACKNOWLEDGEMENT

The financial supports of Thailand Research Fund, Chulalongkorn University through Center of Excellence in Earthquake
Engineering and Vibration, and JICA through the ASEAN University Network/Southeast Asia Engineering Education
Development Network (AUN/SEED-Net) program are gratefully acknowledged.

REFERENCES

Baker, J. W. (2011), "Conditional mean spectrum: Tool for ground-motion selection", Journal of Structural Engineering, Vol.
137(3), 322-331.
Baker, J. W. and Cornell, A. C. (2006), "Spectral shape, epsilon and record selection", Earthquake Engineering & Structural
Dynamics, Vol. 35(9), 1077-1095.
Ordonez, G. A. (2012), SHAKE2000-A computer program for the 1-D analysis of geotechnical earthquake engineering
problems. GeoMotion, LLC; Lacey, Washington, USA.
PEER TBI (2010), Guidelines for Performance-Based Seismic Design of Tall Buildings, PEER-2010/05, Pacific Earthquake
Engineering Research Center, University of California, Berkeley, CA.
Poovarodom, N., Jirasakjamroonsri, A. and Warnitchai, P. (2017), "Development of new design spectral accelerations for
Bangkok considering deep basin effects", 16th World Conference on Earthquake Engineering, Santiago, Chile.
SeismoSoft (2016), SeismoMatch, A computer application capable of adjusting earthquake accelerograms to match a specific
target response spectrum, Version 2016. http://www.seismosoft.com.