You are on page 1of 10

Engineering Structures 24 (2002) 99–108

www.elsevier.com/locate/engstruct

Analysis of long-distance earthquake tremors and base shear


demand for buildings in Singapore
a,*
T. Balendra , N.T.K. Lam b, J.L. Wilson b, K.H. Kong a

a
Department of Civil Engineering, National University of Singapore, 10 Kent Ridge Crescent, 119260, Singapore
b
Department of Civil and Environmental Engineering, University of Melbourne, Melbourne, Australia

Received 30 November 2000; received in revised form 31 May 2001; accepted 2 June 2001

Abstract

The seismic faults which cause long distance earthquakes for Singapore have been analysed and the maximum possible earthquake
in each of these faults have been identified. A seismic hazard predictive model developed from geophysical principles, known as
the component attenuation model (CAM), has been used to estimate the bedrock motion in Singapore. CAM has been validated
by comparison of the estimated ground motion parameters with historical attenuation data. Earthquakes considered in the study
originated from the Indonesian Arc and the Burmese Arc in the south-western part of the Eurasian plate spanning between 90°E
and 105°E, and between ⫺5°S and 25°N. An earthquake of 7.6 in Richter Magnitude, from a 400 km distance has been shown to
be critical for buildings up to 25 storeys. The elastic base shear demand corresponding to this bedrock motion when accounted for
amplification by soft soil is found to be below 10% of the weight of the building.  2001 Elsevier Science Ltd. All rights reserved.

Keywords: Response spectum; Component attenuation model; Seismological model; Distant earthquakes; Base shear demand; Soil amplification; Sin-
gapore

1. Introduction ground motion can develop further resonance with build-


ings that possess natural periods similar to the site natu-
Seismic waves originating from major seismic fault ral period. Hence, tall buildings located on soft soil sites
sources in the region travel long distances before they tend to suffer the most severe vibration due to multiple
reach Singapore. The high frequency earthquake waves resonances developed in both the building and the soil.
are typically damped out rapidly as they propagate away The potential hazards of distant earthquakes and their
from the source. However, low frequency or long period effects on buildings in Singapore have been assessed in
waves are more robust to energy dissipation and hence this investigation which comprises the following steps:
can transmit more energy over much longer distances.
Thus, seismic waves reaching Singapore bedrock are 1. Identify major potential seismic fault sources in the
often rich in long period waves. Consequently, ground region that can generate significant bedrock motions
motions generated by distant earthquakes have very low at the site of interest.
peak ground acceleration but can be very destructive due 2. Adapt the component attenuation model (CAM) to
to its large displacement properties. In addition, seismic model the attenuation of seismic waves transmitted
waves propagating through soil layers can be signifi- from the identified fault sources. (Historical intensity
cantly amplified by resonance if the soil layers possess data has been used to verify the accuracy of CAM.)
natural periods that are similar to the dominant period of 3. Determine the magnitude–distance combinations of
the seismic waves transmitted from the bedrock. Hence the Design Earthquakes (DE) based on information
earthquake ground motions are significantly affected by obtained in Steps (1) and (2). The bedrock spectra of
the natural period of the soil layer. Such amplified the DE was then determined by CAM whilst synthetic
random phase-angle accelerograms generated by
stochastic simulations based on the targeted pre-
* Corresponding author. defined time independent Fourier amplitude spectrum.

0141-0296/02/$ - see front matter  2001 Elsevier Science Ltd. All rights reserved.
PII: S 0 1 4 1 - 0 2 9 6 ( 0 1 ) 0 0 0 6 5 - 7
100 T. Balendra et al. / Engineering Structures 24 (2002) 99–108

4. Conduct one-dimensional shear wave analysis of on the rate of convergence and the age of the subduction
selected soft soil sites using the program shake91 [1] plates. The earthquake ground motions considered in this
and the input excitations developed from Step (3). study is accordingly based on a MW=7.8 earthquake at
The design elastic base shear for buildings were then a distance of 400 km and a MW=8.9 earthquake at a dis-
determined in accordance with the soil response spec- tance of 600 km.
tra.

3. Attenuation modelling and introduction to


“CAM”
2. Seismic faults in the region surrounding
Singapore In regions possessing an abundance of indigenous
strong motion records, statistical regression analyses
Singapore is on a stable part of the Eurasian plate. may be used to develop empirical predictive attenuation
The closest fault is the Sumatra fault which is ⬎1500 models which relate the ground motion parameters (e.g.
km long and runs through the entire length of Sumatra peak ground acceleration and peak ground velocity) to
(Fig. 1). At the closest point this fault is about 400 km magnitudes and distances [4–7]. For the subject region
from Singapore. It is a strike slip fault and hence the under consideration, earthquakes are largely originated
energy is stored through shear deformation of the rock from the south-western part of the Sunda Arc, viz.
when the plates on either side of the fault become inter- Indonesian arc and Burmese arc. Since only limited
locked while in motion. It is considered that only a lim- strong motion accelerogram records are available in this
ited amount of energy can be stored by the shear inter- region, it is not feasible to adopt the conventional
lock, since fracture can occur at a weak point in the rock empirical modelling approach.
at a relatively low stress level. Thus, the energy that An alternative approach is to adopt attenuation models
could be released from this fault will not exceed the developed from analogue regions which are considered
Richter magnitude (ML) of 7.5–7.6 (MW=7.8). There is to possess similar seismo-tectonics and geological con-
also a subduction zone at the Java trench where the Indo- ditions. Traditionally, Western North America (WNA)
Australian plate subducts under the Eurasian plate at a was a popular candidate of an “analogue” region, only
rate of about 67 mm per year. The displacement between because similar regions from which extensive data have
the two plates is partly accommodated by sudden move- been collected and widely published were rare. In recent
ments which cause large earthquakes. This subduction years, seismological attenuation models have been
zone is about 700–800 km from Singapore. The histori- developed for the central and eastern regions of North
cal record for the last 300 years reveals that two great America (CENA). Since the seismo-tectonic and geo-
earthquakes have occurred, viz. moment magnitude logical conditions of CENA are in great contrast to those
(MW) of 8.75 in 1833 and moment magnitude of 8.25– of WNA, the available choice of a representative attenu-
8.5 in 1861 [2]. The maximum possible earthquake from ation model seems to have broadened considerably.
this subduction zone was estimated by Heaton and Kana- However, the simplistic notion of applying ENA models
mori [3] to be of magnitude ML=8.0–8.5 (MW=8.9) based to low seismicity regions, and WNA models to high seis-
micity regions, is not recommended as it has no sound
rational basis. The attenuation of seismic waves is larg-
ely a function of the regional crustal conditions (in parti-
cular regard to energy dissipation) rather than the
regional seismicity level. Further, the effects of attenu-
ation accumulate with distance, and hence the accurate
representation of the regional crustal properties is parti-
cularly important in the modelling of distant earth-
quakes. Thus, the already developed North American
attenuation models are still very restrictive if they are to
be applied to other parts of the globe.
Seismological monitoring studies can be used to deter-
mine regional attenuation properties of the earth crust,
including the Quality Factor which defines energy losses
during wave transmission. Since such studies involve
only very low intensity measurements, valuable infor-
mation on attenuation has been established for many
Fig. 1. Sumatra fault and subduction of the Indian–Australian Plate areas worldwide, and is not restricted to high activity
into Eurasian Plate. areas. Other important achievements are the establish-
T. Balendra et al. / Engineering Structures 24 (2002) 99–108 101

ment of the crustal wave-guide model by Somerville [8] 4. Proposed attenuation model for Singapore
and the upper-crust model by Boore and Joyner [9].
These and other seismological achievements have The subject region to be considered in this investi-
resolved many uncertainties concerned with the trans- gation is on the south-western part of the Eurasian plate,
mission and crustal modifications of seismic waves. spanning between 90°E and 105°E in longitude, and
These crustal models are complimentary to the concur- between ⫺5°S and 25° N in latitude. Singapore and
rent development of the generic source models of Atkin- neighbouring countries including Malaysia, Sumatra,
son [10] and Atkinson and Boore [11]. There is now Thailand and Burma, and the south-western part of the
evidence that the average characteristics of seismic body Sundra Arc lie within this region. The F-crust attenu-
waves generated at the source of the earthquake (i.e. all ation model developed within the latest framework of
crustal modification effects removed) is rather generic, CAM [20] contains expressions which predict peak
and the regional averages varies only moderately response spectral accelerations, velocities and displace-
between CENA and WNA [11], with the CENA source ments. These parameters can be translated into the peak
model being more conservative. Thus, the CENA source acceleration on bedrock (PRA) which is defined by the
model can be applied generically in different regions following expressions:
when combined with suitable crustal models. It was PRA(g)⫽aGbg (1)
demonstrated by Lam et al. [12] that the existing empiri-
cal response spectrum models of California are remark- a⫽0.192g[0.40⫹0.60(MW⫺5)1.5] (2)

冉冊
ably similar to the averaged response spectra derived 0.5
from the stochastic simulations of the generic source 2 75
G⫽ ; R⬎75 km (3)
model of CENA in combination with the crustal model 3 R

冉冊
of WNA (mainly California), for earthquakes exceeding C
30
magnitude 7. This finding supports the independent b⫽ (4)
observation by Somerville [13] that variations amongst R
large magnitude earthquakes are mainly due to path C⫽0.009[9.9⫺1.28(MW⫺6)]R0.48 (5)
(crustal) and site effects.
The recently developed seismological Fourier spec- g⫽0.585 (6)
trum models have been transformed by Lam et al. [14] where PRA(g)=peak acceleration on bedrock (in units of
into a set of response spectral attenuation relationships gravitational acceleration); a=source factor; G=cylindr-
to develop the CAM. The transformation involves: (i) ical attenuation factor which accounts for the crustal-
stochastic simulations of the Fourier spectra to generate wave guide effects; R=epicentral distance in km; b=anel-
large number of random phase angles synthetic accelero- astic attenuation factor which accounts for energy dissi-
grams [15] (see Appendix A for further details and a pation; and g=factor combining the effects of the mid-
specimen of the generated accelerogram); (ii) response crust and upper-crust modifications.
spectral analysis of the generated accelerograms; and The accuracy of Eqs. (1)–(6) been verified by the
(iii) mathematical curve-fitting and modelling of the comparison with the peak ground accelerations recorded
ensemble averaged response spectra [12,14]. from five earthquakes generated in the subduction region
CAM is made up of a number of source and crustal of the Indonesian Arc and the Burmese Arc. The
factors which can be modified to suit the crustal proper- locations of these earthquakes are shown in Fig. 2.
ties of the region of interest. Regional crustal models of Fig. 3 shows the comparison of the PGAs calculated
CAM has been applied in Australia [16,17], the Coastal by Eqs. (1)–(6) with that indicated by the respective iso-
Region of South China [18–20], Shanghai [21] and Viet- seismal maps of the Burmese earthquakes of: (i) Manda-
nam [22]. Significantly, good agreement between lay (May 1912, MW=8); and (ii) Pegu (May 1930,
CAM’s ground motion predictions and historical data MW=7.2). In addition, Table 1 shows similar compari-
has been found. For example, ground motion predictions sons for the three Sumatra earthquakes of: (i) 1926,
by CAM for Taiwan (F-crust) are supported by field rec- ML=6.75; (ii) April 1998, ML=6.8 and (iii) June 2000,
ordings of the 1999 Chi-Chi earthquake up to some 700 ML=7.9. The validity of the model from 250 to 700 km
km away from the earthquake source [20]. New attenu- is further exemplified in Table 1 by including some field
ation relationships developed within the CAM frame- data from the Burmese earthquake. The application of
work provide suitable choices of an analogue model for the CAM F-crust model [20] in generating accelero-
the neighbouring regions. In this investigation, the F- grams for Singapore conditions is further supported by
crust model is used and verified as the analogue attenu- Fig. 4 which compares two accelerograms recorded in
ation model for the region surrounding Singapore as orthogonal directions from the MW=8 earthquake
described in the following section. (originated in Sumatra in June 2000 at a distance of 700
km from Singapore) with the ensemble averaged
102 T. Balendra et al. / Engineering Structures 24 (2002) 99–108

Fig. 2. Location and year of earthquakes generated in the subduction


region of the Indonesian Arc and the Burmese Arc. A, Outer Burmese
Arc; B, Inner Burmese Arc; C, Indonesian Arc. Burmese earthquakes:
(i) 1912, earthquake at Mandalay, MW=8; (ii) 1930, earthquake at Pegu,
MW=7.2. Sumatra earthquakes: (i) 1926, ML=6.75; (ii) 1998, ML=6.8;
(iii) 2000, ML=7.9.
Fig. 3. Comparison of the peak ground acceleration’s calculated by
the CAM with that of iso-seismial maps of the Burmese earthquakes of
(a) Pegu (May 1930, MW=7.2) and (b) Mandalay (May 1912, MW=8).
response spectrum of six synthetic accelerograms simu-
lating the earthquake. The response spectra are presented
for 1% damping. R=400 km. Whereas, the very flexible sites possessing
Next, the response spectra obtained from the ensemble a higher period (⬎1.2 s) would respond more severely
average of six synthetic accelerograms associated with to the very large magnitude “mega” earthquakes (e.g.
the F-crust model for different combinations of distance MW=8.9) originating from the subduction zone at
and magnitude are shown in Fig. 5. Earthquake of mag- R=600 km.
nitude MW of 8.9 occurring at 600 km has been included The natural period of 15–25 storey buildings (typical
to identify the worst scenario. These bedrock response in Singapore) have been found to be in the range of
spectra form the basis of soil response predictions as approximately between 0.7 and 1.6s. A recently conduc-
described in the next section. ted field study [23] revealed that buildings are often
much stiffer than that predicted by the computer analysis
of the skeletal frame due to the participation of infill
5. Base shear in buildings on soft soil brick walls which maintain their integrity during low
excitations caused by earth tremors. Thus, both earth-
Sedimentary soil layers or reclamation can signifi- quake scenarios described above have been considered
cantly amplify seismic waves as they propagate upward in the study of the soil and building responses.
through the soil layers. The seismic actions on buildings Three sites in the Kallang formation on the eastern
founded on soft soil would depend on the soil response part of Singapore have been selected in accordance with
and the site natural period. The soil response in turn the predominant period range of the seismic bedrock
depends on the site period in relation to the location of motions that have been predicted for the design earth-
the peaks in the bedrock response spectrum. The bedrock quake scenarios. The selected sites are located at Marine
spectra in Fig. 5 show that sites possessing a period of Parade (MP), Katong Park (KAP) and the Katong area
between 0.4 and 0.8 s would respond very severely to a (KAT). The borehole data for each of these sites are
ML=7.6 earthquake generated from a strike–slip fault at given in Tables 2–4. The site periods are 0.7, 1.1 and
T. Balendra et al. / Engineering Structures 24 (2002) 99–108 103

Table 1
Prediction of peak rock motion

Event Magnitude Distance (km) Peak rock motion (gals) (computed) Peak rock motion (gals) (measured)

ML MW

Burma, 1912 8.0 400 5.5 5–10


Burma, 1912 8.0 300 12.5 11–14
Burma, 1930 7.2 400 2.4 1–4
Burma, 1930 7.2 300 6 2–10
Sumatra, 1926 6.75 250 11.6 7–14
Sumatra, 1998 6.8 600 0.42 0.31
Sumatra, 2000 7.9 700 0.69 0.36

Fig. 4. Comparison of the simulated velocity response spectrum with the recorded spectra of the June 2000 Sumatra earthquake.

Fig. 5. Response spectra from and ensemble of six synthetic accelograms for different combinations of earthquake magnitude and distance.

1.6 s for the MP, KAP and KAT sites, respectively, KAP sites, and in Fig. 6(b) for the KAT site. Assuming
which are comparable to the period range estimated for one-dimensional shear wave propagation in the vertical
the 15–25 storeys buildings. Ensembles of synthetic direction, the surface motions are computed using pro-
accelerograms stochastically simulated [15] for the gram shake91 [1].
design earthquake scenarios (M–R combinations) have Soil shear modulus was computed using the
been used as bedrock excitations for input to program expression given by Hardin and Drnevich [24] with the
shake91 [1] for computing the soil surface motions. The reference shear strain proposed by Lam and Wilson [25].
response spectrum of the input motion used in the Soil damping was computed using the expression given
shake91 analysis is depicted in Fig. 6(a) for the MP and by Poulos [26] with the similar reference shear strain as
104 T. Balendra et al. / Engineering Structures 24 (2002) 99–108

Table 2
Soil data for the marine parade (MP) site

Undrained shear
Thickness SPT N-value Relative Unit weight
Layer No Description strength
(m) (blow/30 cm) density (%) 2 (kN/m3)
(kN/m )

Backfill — very loose light yellowish brown clayey


1 6 2.5 35 – 19.7
SAND. Ground water is 3.0 m below ground
2 Soft grey Marine CLAY 3 0 – 14 16.3
3 Loose brownish-yellowish and whitish grey silty SAND 3 8 45 – 18.0
4 Medium dense light whitish grey silty SAND 3 19 60 – 19.1
5 Very loose yellowish brown silty SAND 3 4 35 – 18.0
6 Stiff light bluish grey sandy SILT 3 12 – 50 17.7
7 Very stiff dark brown and bluish grey silty CLAY 9 21 – 87.5 20.3
8 Hard bluish grey silty CLAY 3 46 – 191.7 20.3
9 Hard reddish brown and blueish grey silty CLAY 6.45 79 – 330.6 22.0
10 Bedrock – 100 22.0
Total soil depth 39.45

Table 3
Soil data for Katong Park (KAP) site

Undrained shear
Layer Thickness SPT N-value Relative Unit weight
Description strength
No. (m) (blow/30 cm) density (%) 2 (kN/m3)
(kN/m )

Fill — brownish grey. Loose clayey silt with some fine


1 to coarse grain sand. Below ground level of 1.5 m 6.5 11 35 – 17.3
material becomes loose SAND
Upper marine CLAY — blueish grey. Very soft, high
2 plasticity and high water content. Ground water table is 13 0 – 21.6 15.6
at this level
CLAY — stiff brown, low water content, moderate
3 2 15 – 59.9 18.7
plasticity
Lower marine CLAY — blueish grey. Very soft, high
4 9 0 – 23.9 16.2
plasticity and high sensitive silt clay
Organic CLAY — very soft, very sensitive,high water
5 4 2 – 26.3 16.2
content, lots of organic material
Stiff to very stiff silty CLAY— low water content,
6 5 13 – 52 18.7
moderate plasticity
7 Slightly weathered silty SAND — very dense and stiff 5.9 75 100 – 21.1
8 Slightly weathered SILT — very stiff 1.9 100 – 400 22.0
9 Bedrock – 100 – 22.0
Total soil depth 47.3

Table 4
Soil data for Katong (KAT) site

Undrained shear
Layer Thickness SPT N-value Relative Unit weight
Description strength
No. (m) (blow/30 cm) density (%) 2 (kN/m3)
(kN/m )

1 Fill — loose fine SAND 7.9 5 35 – 17.1


2 Marine CLAY — very soft to medium stiff 27.9 0 – 18.7 16.2
CLAY — reddish brown, medium stiff with organic
3 13.2 10 – 40.2 20.3
matter
4 CLAY — medium stiff to stiff 4.2 50 – 200.0 20.3
5 Bedrock 100 – 21.1
Total soil depth 53.2
T. Balendra et al. / Engineering Structures 24 (2002) 99–108 105

Fig. 6. (a) Response spectra from 18 synthetic accelograms for


design earthquake of magnitude, MW=7.8 and distance, R=400 km. (b)
Response spectra from 18 synthetic accelograms for design earthquake
of magnitude, MW=8.9 and distance, R=600 km.

above and using recommendations of initial and


maximum soil damping presented in Ref. [25]. For clay,
a 50% plasticity index was used to input into the equa-
Fig. 7. (a) Shear modulus/shear modulus at low strain 0.001%
tions to plot the shear modulus and soil damping curve. (G/Gmax) vs shear strain (%) for clay and sand. (b) Soil damping ratio
This value is reasonable because in the Kallang forma- versus shear strain (%) for clay and sand.
tion, clay plasticity index can range from around 40 to
70% (Tan and Lee [27]). For sand, a plasticity index of
0% was used. The variation of shear modulus and damp- demand of about 10% of the weight of the building (i.e.
ing with respect to strain in the soil, used in shake91 0.10 W) as shown by the soil response spectra. (Similar
is depicted in Figs 7(a) and 7(b) for clay and sand, soil amplification effects were observed in 1986 at Mex-
respectively. ico City which was some 400 km away from the epi-
The ensemble average of the acceleration response centre of a magnitude 8 earthquake [28].)
spectra of the surface response for 5% structural damp- This soil amplification factor estimate is conservative
ing are depicted in Fig. 8(a) for the MP, KAP and KAT as all shake analyses conducted in the study assumed a
sites. The corresponding soil amplification factor which high impedance contrast between the soil and the bed-
is defined as the ratio of peak acceleration response spec- rock that possesses a shear wave velocity of ⬎1500 m/s
tra at surface to bedrock is 12.2, 9.5 and 10.3 for MP, (e.g. site class A “hard rock” as defined in IBC2000
KAP and KAT sites, respectively [Fig. 8(b)]. Such a [29]). It can be shown that the amplification can be
high site amplification factor is seldom observed in reduced significantly for a more moderate impedance
strong motion responses, but is not unreasonable in dis- contrast at the soil–rock interface. A parametric study
tant earthquakes in view of the highly periodic nature of undertaken recently by the authors [30] (in collaboration
the seismic waves generated from the very far-field. The with Adrian Chandler of Hong Kong University) shows
soil response results in a maximum elastic base shear a 20–30% reduction of the soil amplification factor in
106 T. Balendra et al. / Engineering Structures 24 (2002) 99–108

0.08 W if the bedrock shear wave velocity has been con-


firmed to be in the “class C” (also known as “soft
rock”) category.
The design elastic base shear demand may also be
reduced for sites which have a soil depth significantly
shallower than the three soil sites analysed in the study.
Site specific soil response analysis would be required to
quantify such reductions.
In addition, buildings generally possess a reasonable
amount of over-strength and ductility (with exceptions)
and it is common practice in building codes to reduce
the elastic strength demand by a structural response fac-
tor (or load reduction factor). For example, the appropri-
ate factor for reinforced concrete moment resisting build-
ings designed according to BS8110 is 5 (Balendra et al.
[31]). The corresponding value for a typical shear wall-
frame building is found to be around 8–10 (Balendra et
al. [32]). The latter value is more appropriate for tall
buildings as they are invariably built with a dual system.
Investigations are currently being undertaken to ascer-
tain the extent in which these established structural
response factors can be applied in conditions associated
with distant earthquakes.

6. Conclusion

The CAM developed from geophysical principles has


been employed in this study to estimate the properties
Fig. 8. (a) Spectral acceleration at the surface of MP, KAP and KAT of bedrock motion generated from large magnitude
sites for structural damping of 5% due to design earthquakes at bed-
earthquakes in the south western part of the Eurasian
rock. (b) Soil amplification (ratio of surface to bedrock spectral
acceleration) for MP, KAP and KAT sites for structural damping of Plate, bounded by 90°E to 105°E and ⫺5°S to 25°N.
5% due to design earthquakes at bedrock. The F-crust model developed within the framework of
CAM has been validated for applications in Singapore,
based on comparison with historical data of earthquakes
soft bedrock which possesses a relatively low shear wave occurring in the Indonesian Arc and the Burmese Arc.
velocity (e.g. Vs=350–750 m/s as defined for class C sites Analyses using CAM show that the critical earthquake
in IBC2000). A 20% reduction (reduction factor=0.8) for buildings up to 25 storeys is of Richter magnitude
was recommended in Ref. [30]. 7.6 occurring at an epicentral distance of 400 km. When
Incidentally, the standard penetration test (SPT) blow amplification of the corresponding bedrock motion by
counts of 100 as recorded on the bedrock surface at the the soft soil has been accounted for, the maximum elastic
selected soil sites (Tables 2–4) were consistent with the base shear demand due to this critical earthquake is esti-
range of shear wave velocity defined for site class C mated to be within 10% of the weight of the building
[25,29]. A comparison was made between the soil sur- (0.10 W). This base shear estimate may be lowered to
face spectra recorded at the KAP site from the ML=6.8 0.07–0.08 W depending on the shear wave velocity
(April 1998 Sumatra) event with the spectra computed (which measures the hardness) of the bedrock underlying
by shake based on the synchronised recording of the reclamation or soil sediments.
bedrock exictation taken at the bottom of a nearby bore-
hole. Interestingly, the recorded soil spectra show an
amplification factor some 20–30% lower than the initial Acknowledgements
estimate by shake which was based on the high imped-
ance contrast of a hard bedrock (Vs⬎1500 m/s) (see The work reported in this paper was supported by the
Appendix B). This observed modifications by the soft National University of Singapore (NUS) and the Univer-
rock was consistent with the analytical findings of Ref. sity of Melbourne (MU). The financial support by “The
[30]. Consequently, the elastic base shear demand may International Exchange Agreements Collaborative
be lowered from the initial estimate of 0.10 W to 0.07– Research Awards” (IEACRA) of MU in funding this
T. Balendra et al. / Engineering Structures 24 (2002) 99–108 107

international collaborative research is acknowledged.


The assistance provided by T.K. Lim of the Meteoro-
logical Service Singapore and Michael Cheng of MU in
the preparation of Appendix B are gratefully acknowl-
edged. Contributions by Professor C.G. Koh at NUS and
Professor A.M. Chandler and his research team at Hong
Kong University in related investigations are also
acknowledged.

Appendix A. Synthetic accelerogram sample

This appendix shows a sample synthetic random


phase-angle accelerogram which was generated by
stochastic simulations based on a targeted pre-defined
time independent Fourier amplitude spectrum. Full
details of the accelerogram generation procedure have
been presented and reviewed in Ref. [15] which also
demonstrates that the ensemble averaged Fourier ampli- Fig. 10. Comparison of recorded and computed soil surface response
tude spectrum of the generated accelerograms would spectra for KAP site (NS direction).
always be reasonably consistent with the targeted Four-
ier amplitude spectrum. The duration of the synthetic with the response spectrum computed by shake based
accelerograms is dependent on both the moment magni- on the synchronised recording of the bedrock excitations
tude and distance, and was determined in accordance taken at the bottom of a nearby borehole. The recorded
with the recommendations by the seismological litera- spectrum show an amplification factor some 20–30%
ture [11]. The recommended duration accounts for both lower than the initial estimate by shake which was
the time required for the fault to rupture and the time based on the high impedance contrast of a hard bedrock
for wave transmission through the earth crust. The (Vs⬎1500 m/s). This 20–30% reduction is reasonably
adopted time–domain intensity envelope was defined in consistent with the reduction factor of 0.8 recommended
accordance with the recommendations of Ref. [15] in Ref. [30] for soil amplification over soft bedrock
(Fig. 9). (Fig. 10).

Appendix B. Comparison of shake analysis result References


with field measurements
[1] Idriss IM, Sun JI. User’s manual for shake91 — A computer
This appendix shows the comparison between the soil program for conducting equivalent linear seismic response analy-
surface response spectrum (NS direction) recorded at the ses of horizontally layered soil deposits — program modified
based on the original shake program published in December
KAP site from the ML=6.8 (April 1998 Sumatra) event 1972 by Schnabel, Lysmer and Seed. Centre for Geotechnical
Modelling, Department of Civil and Environmental Engineering,
University of California, Davis (California). November 1992.
[2] Newcomb KR, McCann WR. Seismic history and seismotecton-
ics of the Sunda Arc. Journal of Geophysical Research
1987;92:421–39.
[3] Heaton K. Seismic potential associated with subduction in the
northwestern United States. Bulletin of the Seismological Society
of America 1984;74(3):933–41.
[4] Ambraseys NN, Simpson KA, Bommer JJ. Prediction of horizon-
tal response spectra in Europe. Earthquake Engineering and
Structural Dynamics 1996;25:371–400.
[5] Boore DM, Joyner WB, Fumal TE. Equation for estimating hori-
zontal response spectra and peak acceleration from Western
North American earthquakes: a summary of recent work. Seismo-
logical Research Letters 1997;68:128–53.
[6] Sadigh K, Chang CY, Egan JA, Makdisi F, Youngs RR. Attenu-
ation relationship for shallow crustal earthquake based on Cali-
fornian strong motion data. Seismological Research Letters
Fig. 9. One of the six synthetic accelerograms simulating June 2000 1997;68:180–9.
Sumatra event (MW=8, R=700). [7] Balendra T, Koh CG, Wang MJ, Lim TK. Peak Bedrock acceler-
108 T. Balendra et al. / Engineering Structures 24 (2002) 99–108

ation and soil amplification factor for buildings in Singapore. In: [20] Lam NTK, Chandler AM, Wilson JL. The component attenuation
Proceedings Seventh East Asia–Pacific Conference on Structural model for predicting earthquake ground motions affecting Hong
Engineering and Construction, Japan; 1999. Kong from the far-field. In: Proceedings of the Mini-workshop
[8] Somerville P. Recent advances in strong ground motion predic- on Design ground motion for East-Asia Region, The University
tion. In: Proceedings of the 8th Canadian conference on Earth- of Tokyo, 4th March 2000:74–88.
quake Engineering, Vancouver; 1999:7–28. [21] Luo QF. Estimation of ground motions affecting Shanghai by
[9] Boore DM, Joyner WB. Site amplification for generic rock sites. long distance earthquake. Proceedings of the Advanced Structural
Bulletin of the Seismological Society of America 1997;87:327– Dynamics Conference Hong Kong 2000;1:225–32.
41. [22] Ngo T, Koo R, Lam NTK, Wilson JL. Estimates of seismic load-
[10] Atkinson G. Earthquake source spectra in Eastern North America. ing for concrete structures in Hanoi, Vietnam. In: Proceedings of
Bulletin of the Seismological Society of America the International Conference on Advanced Technologies in
1993;83:1778–98. Design, Construction and Maintenance of Concrete Structure,
[11] Atkinson G, Boore DM. Evaluation of models for earthquake Hanoi, Vietnam, March 2001:160–166.
source spectra in Eastern North America. Bulletin of the Seismo- [23] Balendra T, Tan CL. Full-scale measurement of wind induced
logical Society of America 1998;88:917–34. response of a high-rise building. In: Conspectus Technical Jour-
[12] Lam NTK, Wilson JL, Chandler AM, Hutchinson GL. Response nal 2000 of the Structural Engineering Department. Singapore:
spectrum modelling for rock sites in low and moderate seismicity Housing Development Board, 2000:61–8.
regions combining velocity, displacement and acceleration pre- [24] Hardin BO, Drnevich VP. Shear modulus and damping in soils:
dictions. Earthquake Engineering and Structural Dynamics measurement and parameter effects. ASCE Journal of Soil Mech-
2000;29(10):1491–526. anics and Foundations 1972;98(SM6):602–24.
[13] Somerville P. Seismic hazard evaluation (state-of-the-art paper). [25] Lam NTK, Wilson JL. Estimation of the Site Natural Period from
In: Proceedings of the 12th World Conference of Earthquake borehole records. Australian Journal of Structural Engineering
Engineering, Auckland, New Zealand; 2000, paper no. 2833. 1999;SE1(3):179–99.
[14] Lam NTK, Wilson JL, Chandler AM, Hutchinson GL. Response [26] Poulos HG. Relationship between local soil conditions and struc-
spectral relationship for rock sites derived from the component tural damage in the 1989 Newcastle earthquake. Australian Civil
attenuation model. Earthquake Engineering and Structural Engineering Transactions, Institution of Engineers Australia
Dynamics 2000;29:1457–89. 1991;CE33(3):181–8.
[15] Lam NTK, Wilson JL, Hutchinson GL. Generation of synthetic [27] Tan SB, Lee KW. Engineering geology of the marine member
earthquake accelerograms using seismological modelling: a of the Kallang Formation in Singapore. In: Proceedings of the
review. Journal of Earthquake Engineering 2000;4(3):321–54. International Symposium on Soft Clay, July 1975:75-88.
[16] Koo R, Cheng M, Lam NTK, Wilson JL, Hutchinson GL, Griffith [28] EFFIT Earthquake Engineering Field Investigation Team. The
MC. Modelling of the earthquake ground motions generated by Mexican earthquake of 19 September 1985, a field report. Lon-
the Newcastle earthquake. In: Proceedings of a conference held don: EEFIT 1986, Institution of Structural Engineers, 1986.
by the Australian Earthquake Engineering Society, Sydney, New [29] IBC 2000 International Code Council. International building
South Wales; 1999, paper no. 5. code. USA: IBC 2000 International Code Council, 2000.
[17] Koo R, Brown A, Lam NTK, Wilson JL, Gibson G. A full range [30] Lam NTK, Wilson JL, Chandler AM. Seismic displacement
response spectrum model for rock sites in the Melbourne Metro- response spectrum estimated from the frame analogy soil ampli-
politan Area. In: Proceedings of the Australian Earthquake Engin- fication model. Journal of Engineering Structures
eering Society Annual Conference, Hobart, Tasmania; 2000, 2001;23:1437–52.
paper no. 16. [31] Balendra T, Tan KH, Kong SK. Vulnerability of reinforced con-
[18] Lam NTK, Chandler AM, Wilson JL, Hutchinson GL. Seismic crete frames in low seismic region when designed according to
hazard determination for the coastal region of South China (I) BS 8110. Earthquake Engineering and Structural Dynamics
generic crustal modelling. International Journal of Seismology 1999;28:1361–81.
and Earthquake Engineering 1999;2(1):1–18. [32] Balendra T, Tan KH, Kong SK. Ultimate strength of a reinforced
[19] Lam NTK, Chandler AM, Wilson JL, Hutchinson GL. Seismic concrete frame-wall structure designed according to BS8110. In:
hazard determination for the coastal region of South China (II) ICCMC/IBST International Conference on Advanced Techno-
regional crustal modelling. International Journal of Seismology logies in Design, Construction and Maintenance of Concrete
and Earthquake Engineering 1999;2(2):1–16. Structures, Hanoi, Vietnam:66-72.

You might also like