Cyclic Softening of Low-Plasticity Clay and Its Effect on

Seismic Foundation Performance

Daniel B. Chu1; Jonathan P. Stewart2; Ross W. Boulanger3; and P. S. Lin4

Abstract: During the 1999 Chi-Chi Earthquake 共M w = 7.6兲, significant incidents of ground failure occurred in Wufeng, Taiwan, which
Downloaded from ascelibrary.org by Florida International University on 09/06/13. Copyright ASCE. For personal use only; all rights reserved.

experienced peak accelerations ⬃0.7 g. This paper describes the results of field investigations and analyses of a small region within
Wufeng along an E–W trending line 350 m long. The east end of the line has single-story structures for which there was no evidence of
ground failure. The west end of the line had three to six-story reinforced concrete structures that underwent differential settlement and
foundation bearing failures. No ground failure was observed in the free field. Surficial soils consist of low-plasticity silty clays that extend
to 8 – 12 m depth in the damaged area 共west side兲, and 3 – 10 m depth in the undamaged area 共east side兲. A significant fraction of the
foundation soils at the site are liquefaction susceptible based on several recently proposed criteria, but the site performance cannot be
explained by analysis in existing liquefaction frameworks. Accordingly, an alternative approach is used that accounts for the clayey nature
of the foundation soils. Field and laboratory tests are used to evaluate the monotonic and cyclic shear resistance of the soil, which is
compared to the cyclic demand placed on the soil by ground response and soil–structure interaction. Results of the analysis indicate a
potential for cyclic softening and associated strength loss in foundation soils below the six-story buildings, which contributes to bearing
capacity failures at the edges of the foundation. Similar analyses indicate high factors of safety in foundation soils below one-story
buildings as well in the free field, which is consistent with the observed field performance.
DOI: 10.1061/共ASCE兲1090-0241共2008兲134:11共1595兲
CE Database subject headings: Soil liquefaction; Seismic effects; Soil structure interaction; Shallow foundations; In situ tests;
Taiwan; Earthquakes.

Introduction foundation performance and describe the soil conditions as evalu-

ated from in situ and laboratory testing. We then examine the site
The 1999, Chi-Chi, Taiwan Earthquake caused extensive ground performance in an analysis framework that accounts for the
failure and structural damage in Wufeng, Taiwan. Liquefaction- clayey nature of the foundation soils. A brief overview of this
induced ground failure occurred in the form of sand boils, lateral work was presented by Chu et al. 共2007兲; this paper provides
spreading, and ground settlement 共Stewart, 2001兲. However, some significantly more detail on the site characterization and analysis,
of the most interesting examples of ground failure occurred in and includes analysis of the site within a liquefaction framework,
areas underlain by low plasticity clayey soils, in which ground which has not been previously discussed.
failure was generally not manifest in the free field or beneath
low-rise buildings, but only beneath relatively tall three to six-
story reinforced concrete frame structures with shallow founda- Case History: Wufeng Site A
tions 共mats and footings兲. As will be shown in this paper, these
case histories push the limits of liquefaction analysis tools that The subject of this paper is a region referred to as Site A located
form the current standards of practice. in the southern part of Wufeng, Taiwan. As shown in Fig. 1, Site
In this paper we document case histories of good and poor A consists of a series of buildings along an E–W trending line
共A – A⬘兲 350 m in length. The east end of line A – A⬘ is a residen-
1
Chief Geotechnical Engineer, Ninyo & Moore Geotechnical & Envi-
tial area with single-story buildings, which sustained neither
ronmental Sciences Consultants, 475 Goddard, Ste. 200, Irvine, CA structural damage nor ground failure. This single-story housing
92618. community is located approximately 200 m from the fault rup-
2
Professor and Vice Chair, Civil and Environmental Engineering ture. The west side of line A – A⬘ consisted of three to six-story
Dept. 5731 Boelter Hall, Univ. of California, Los Angeles, CA 90095 reinforced concrete buildings. Based on our field reconnaissance
共corresponding author兲. E-mail: [email protected] and discussions with local residents, out of 18 midrise buildings
3
Professor, Dept. of Civil and Environmental Engineering, Univ. of on the west side, 11 sustained severe damage and were demol-
California, Davis, CA 95616. ished within 3 months of the earthquake. Typical building dam-
4
Professor, Dept. of Civil Engineering, National Chung-Hsing Univ. age involved column failure at the soft first-floor level. These
250 Kuo Kuang Rd., Taichung, Taiwan.
buildings also sustained settlement on the order of 10– 30 cm
Note. Discussion open until April 1, 2009. Separate discussions must
be submitted for individual papers. The manuscript for this paper was with accompanying floor slopes as high as 6–8% 共based on
submitted for review and possible publication on August 18, 2007; ap- postearthquake surveys of foundation slabs兲. For most structures,
proved on March 24, 2008. This paper is part of the Journal of Geotech- the settlement pattern was irregular across the building envelopes
nical and Geoenvironmental Engineering, Vol. 134, No. 11, November and was strongly dependent on foundation details. A common
1, 2008. ©ASCE, ISSN 1090-0241/2008/11-1595–1608/$25.00. foundation detail involved a mat foundation at the rear of the

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / NOVEMBER 2008 / 1595

J. Geotech. Geoenviron. Eng. 2008.134:1595-1608.

Downloaded from ascelibrary.org by Florida International University on 09/06/13. Copyright ASCE. For personal use only; all rights reserved.

Fig. 1. Locations of borings, CPT, and in situ vane shear tests at Wufeng Site A

building 共away from the street兲 and forward columns 共along the etration testing 共SPT兲 and in situ vane shear tests, four backhole
street兲 supported by footings with intermediate slabs on grade. As test pits for in situ vane shear tests, and nine cone penetration test
shown in Fig. 2, the largest settlement typically occurred in the 共CPT兲 soundings. The locations of borings, CPT soundings, and
forward columns and the largest floor slopes were in the deformed test pits are shown in Fig. 1. A summary of the field work is
slabs on grade. Rear mat foundations underwent much less defor- summarized in Chu et al. 共2004兲 and all results are reported at
mation, although there was tilting in some cases, indicating some the following Web site: http://peer.berkeley.edu/lifelines/researchគ
foundation settlement occurred. projects/3A02. Fig. 3 presents an east-west cross section, which
shows that a surficial low plasticity silt and clay layer extends
across the site, being approximately 3 – 10 m thick on the east
Field Exploration side and approximately 8 – 12 m thick on the west side. Index
tests in the clay layer demonstrate plasticity index values ranging
A subsurface exploration program was performed in 2001 and from about null to 16.
2002 consisting of four rotary wash borings with standard pen- In each boring, standard penetration tests were performed in
accordance with ASTM D 6066-96 at selected intervals 共typically
1 – 1.5 m兲. In addition to SPT, stationary piston samplers were
utilized to sample the soft clay deposits. The stationary piston
sampler is a stainless steel thin-wall tube sampler with a piston,
piston rod, and a modified sampler head. The tube is 90 cm in
length, 7.5 cm outside diameter, and 7.1 cm inside diameter.
In situ vane shear tests were performed in borings WAS-3 and
WAS-4 at selected intervals in fine-grained soils. Vane shear tests
were also performed in test pits excavated in soil layers just
below foundation elements. The locations of these test pits are
indicated in Fig. 1 as VST-1 to VST-4. The vane shear test kit,
which conforms to ASTM D 2573-01, has a height of 15 cm and
a tapered width with diameter ranging from 9.1 to 6.4 cm. During
testing, the vane was pushed approximately 65 cm into the clay
layer either by a hydraulic feed from the drill rig or by a backhoe.
After approximately 5 min, the vane was rotated at a rate of
6 ° / min with measurements of torque. It usually took approxi-
Fig. 2. Typical foundation failure 关punched footings and intermediate mately 2 – 4 min to reach the peak torque, after which the re-
slab heaving, photo taken by R. Seed 共1999兲, with permission兴 molded shear strength was established by quickly rotating the

1596 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / NOVEMBER 2008

J. Geotech. Geoenviron. Eng. 2008.134:1595-1608.

Downloaded from ascelibrary.org by Florida International University on 09/06/13. Copyright ASCE. For personal use only; all rights reserved.

Fig. 3. Geotechnical cross section A – A’ at Wufeng Site A

vane for five to seven revolutions and then repeating the test again given change of overburden pressure was limited to the amount
using the same shearing rate of 6 °/ min to record the maximum associated with primary consolidation 共i.e., secondary compres-
torque. Resulting peak and residual undrained shear strengths sion effects were removed兲.
ranged from 23 to 144 and 14 to 58 kPa, respectively. The sensi- Even though every effort was made to minimize sample dis-
tivity ranged from 1.0 to 4.7, with an average of 2.1, indicating a turbance effects, the test results still indicated a broad transition
medium degree of sensitivity 共Mitchell and Soga 2005兲. Results zone between recompression and virgin compression 共indicating
of each individual vane shear tests are presented by Chu 共2006兲 substantial sample disturbance兲. To estimate vertical preconsoli-
and at the aforementioned web site. dation pressure 共␴⬘p兲, generally three applications of the Casa-
grande construction were performed to provide a lower bound,
upper bound, and best estimate.
Laboratory Testing Profiles of ␴⬘p at the west side of the site along with profiles of
in situ vertical effective stress are shown in Figs. 4 and 5. The
Under the direction of the fourth writer, the laboratory at National dashed lines in the stress profiles indicate the vertical stresses at
Chung-Hsing Univ. in Taiwan conducted on materials from the the time of drilling, whereas the solid lines indicate the stresses at
site soil index tests, consolidation tests, isotropically consolidated the time of the earthquake, which include overburden stresses
undrained 共ICU兲 monotonic triaxial compression tests, and isotro- from the structure. Overconsolidation ratios 共OCR兲 were calcu-
pically and anisotropically consolidated undrained 共ICU and lated as indicated in Figs. 4 and 5 to represent the OCR at the
ACU兲 cyclic triaxial tests. Test procedures and results are de- time of the earthquake. The data show that the clays at the site are
scribed in the following sections. overconsolidated near the ground surface, and become nearly nor-
mally consolidated at depths beyond about 8 m. The likely causes
of the near-surface overconsolidation are desiccation from water
Consolidation Tests table fluctuations.
Consolidation testing was performed to evaluate preconsolidation
pressures, which provide insight into the stress history of soil
Isotropically Consolidated Undrained Triaxial
deposits. Eight consolidation tests were performed on samples
Compression Tests
from Wufeng Site A, west area 共from Borings WAS-3 and
WAS-4兲 in accordance ASTM D2435. During both loading and Eighteen ICU monotonic triaxial compression tests were per-
unloading stages of the consolidation tests, loads were applied to formed to evaluate the undrained shear strength of the clayey
the specimen for a minimum of 24 h or until primary consolida- soils at the site. Test specimens were first cut to 15.5 cm tall by
tion was complete 共whichever came last兲. For the purpose of plot- 7.1 cm diameter cylinders to maintain a height to diameter ratio
ting consolidation curves, the change in void ratio attributed to a of at least two. Specimens were then saturated under a back pres-

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / NOVEMBER 2008 / 1597

J. Geotech. Geoenviron. Eng. 2008.134:1595-1608.

Downloaded from ascelibrary.org by Florida International University on 09/06/13. Copyright ASCE. For personal use only; all rights reserved.

Fig. 4. Soil index tests, consolidation history, and undrained shear strength profile for west side of Site A 共WAS-3 and WAC-8兲

Fig. 5. Soil index tests, consolidation history, and undrained shear strength profile for west side of Site A 共WAS-4 and WAC-9兲

1598 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / NOVEMBER 2008

J. Geotech. Geoenviron. Eng. 2008.134:1595-1608.

 function of the number of cycles of shearing 共N兲. Eight cyclic
triaxial tests were performed, which focused on the most preva-
lent conditions in the field.
A critical boundary condition in these tests is the ratio of static
shear stress to effective consolidation stress on the eventual fail-
ure plane 关e.g., taken as a plane at 45+ ␾ / 2 degrees from hori-
zontal in a triaxial test 共where ␾ = effective stress friction angle兲,
and most commonly as a horizontal plane in the field兴. This ratio
of static shear stress to effective consolidation stress is termed ␣.
Elastic solutions for vertical normal stress and horizontal shear
stress 共Poulos and Davis 1974兲 were used to evaluate ␣ in the
foundation soils, and a representative value of ␣ = 0.12 was se-
Downloaded from ascelibrary.org by Florida International University on 09/06/13. Copyright ASCE. For personal use only; all rights reserved.

lected and used in the testing program. Another critical issue is

the OCR of the tested soil, which we varied from about 2 to 3 to
match field conditions 共Figs. 4 and 5兲. The testing program con-
Fig. 6. Shear strength ratio 共normalized by the consolidation stress兲 sisted of six tests 共three for OCR= 2 and three for OCR= 3兲 with
versus OCR from ICU triaxial compression tests ␣ = 0.12. Two additional tests were performed for OCR= 1 and
␣ = 0 to facilitate comparisons to results of previous tests by oth-
ers. The tests were carried out to determine the number of loading
cycles required to cause large strains for ␶cyc / su = 0.6– 1.0. All
of the tests were performed using a MTS-810 cyclic triaxial
sure up to 200 kPa to a minimum B value 共Skempton 1954兲 of machine.
0.95. All specimens were consolidated to an isotropic confining Undrained cyclic triaxial tests were run under load-control
stress 共␴⬘c 兲 of approximately twice the preconsolidation stress 共␴⬘p conditions using uniform cyclic deviatoric loading at a frequency
to ensure that they reached the virgin compression line. After the of 1 Hz. Pore pressure was measured from the bottom of the
completion of primary consolidation, some specimens were then specimen. Zergoun and Vaid 共1994兲 demonstrated that pore pres-
unloaded to produce overconsolidated conditions. Target overcon- sure measurements were generally not reliable during cyclic load-
solidation ratios were OCR= 1, 2, and 4. The loading rate during ing tests conducted rapidly 共e.g., 1 Hz兲 because of the inability of
the triaxial compression tests was 0.75 mm/ min, which provided water pressure to equilibrate throughout the soil specimen and
a time to failure of approximately 20– 30 min 共typical of mono- measurement system. However, for the present study, a 1 Hz
tonic tests兲. loading rate was adopted to be representative of earthquake con-
In Fig. 6, shear strengths obtained from the ICU triaxial com- ditions and we accepted the fact that pore pressure measurements
pression tests are presented as strength ratios 共strength normalized during fast cyclic loading may be unreliable. Tests were per-
by preshearing consolidation stress兲 plotted versus OCR. Also formed according to the following steps:
included in Fig. 6 is the range of typical strength ratios for clays 1. Saturation: Specimens were confined by compressed air in
in undrained direct simple shear 共DSS兲, as reported by Ladd the triaxial cell and back-pressure saturated to remove air
共1991兲. The ICU triaxial compression shear strength ratios exceed bubbles in the soil and in the drainage tubes. Back-pressures
the typical DSS values by approximately 50%, which is consis- saturation was continued to progressively higher values until
tent with typical differences between strengths obtained in these a minimum B value of 0.95 was obtained. Typical back pres-
two test types 共e.g., Kulhawy and Mayne 1990兲. The differences sures needed to reach this B value were 200 kPa.
are attributed primarily to soil anisotropy. Summaries of the soil 2. Consolidation: Specimens were consolidated using the same
shear strength profiles at WAS-3 and WAS-4 共west side兲 are pre- procedures as for the monotonic undrained compression
sented in Figs. 4 and 5, respectively. The undrained shear tests. The effect of the building loads was accounted for
strengths 共su兲 include the estimated su prior to the earthquake when calculating the vertical consolidation stresses, horizon-
共including the effective stress associated with the presence of the tal static shear stresses, and corresponding ␣ values that were
building兲 and two estimates of su at the time of site exploration: to be simulated in the triaxial tests. The calculation of these
共1兲 from in situ vane shear tests and 共2兲 from correlations with stresses is detailed in Chu 共2006兲.
CPT tip resistance using Nk = 20 共based on site-specific calibra- 3. Cyclic loading: Sinusoidal cyclic loading was applied to the
tion兲. The shear strengths from the vane shear and CPT soundings specimen using cyclic deviatoric stresses that produce
tend to be smaller than those estimated from the laboratory shear ␶cyc / su = 0.6 on the shear plane. The specimen was cycled at
strength ratios, which may be due to: 共1兲 the effective consolida- this load until it experienced either 3% axial strain 共single
tion stresses being lower at the time of site exploration than under amplitude兲 or 50 cycles of loading. If a specimen withstood
the buildings and 共2兲 the approximations involved in estimating 50 cycles of loading with less than 3% axial strain, the speci-
Nk for the CPT or in converting shear strengths from triaxial men was immediately subjected to additional cyclic devia-
compression to DSS loading conditions. toric loading producing ␶cyc / su = 0.8. If the axial strain still
did not reach 3% after 50 cycles of loading, cyclic deviatoric
loading was continued with ␶cyc / su = 1.0 until the specimen
Cyclic Undrained Triaxial Compression Tests
reached 3% axial strain.
Cyclic degradation of saturated clays is known to reduce the shear An example set of test results is shown in Fig. 7, which shows
resistance that can be mobilized during repeated cyclic loading plots of normalized deviatoric stress 共q / p⬘c , where q = deviatoric
共termed ␶cyc兲 from the monotonic undrained shear strength 共su兲. stress and p⬘c = two-dimensional mean effective consolidation
To evaluate these effects, we performed ICU and ACU cyclic stress= 关␴⬘1c + ␴⬘3c兴 / 2兲 versus mean effective stress 共p⬘ / p⬘c , where
triaxial testing directed toward characterizing the ratio ␶cyc / su as a p⬘ = two-dimensional mean effective stress= 关␴⬘1 + ␴⬘3兴 / 2兲, normal-

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / NOVEMBER 2008 / 1599

J. Geotech. Geoenviron. Eng. 2008.134:1595-1608.

Downloaded from ascelibrary.org by Florida International University on 09/06/13. Copyright ASCE. For personal use only; all rights reserved.

Fig. 7. Example of cyclic triaxial undrained test results on sample with PI= 15 from Boring WAS-4 at 7.3– 7.45 m depth. Applied conditions are
OCR= 2, ␣ = 0.12, and CSR= 0.6– 0.8.

ized deviatoric stress 共q / p⬘c 兲 versus axial strain, excess pore pres- The postcyclic undrained shear strengths are normalized by
sure ratio 共ru = ⌬u / p⬘c , where ⌬u = excess pore pressure兲 versus the effective consolidation stress and plotted versus OCR in Fig.
number of cycles and axial strain versus number of cycles. The 9. Also shown for reference is the ICU shear strength ratio origi-
test results in Fig. 7 indicate fat hysteresis loops of the type that is
typical of clayey soil behavior. Moreover, the maximum accumu-
lated pore pressure ratio is relatively modest 共ru = 0.6兲 共again con-
sistent with clay behavior兲, although the measured pore pressures
likely contain errors due to the fast loading rate. Note that the
specimen for which results are shown in Fig. 7 was tested at two
stress amplitudes before failure occurred. The results are con-
verted to an equivalent number of cycles at a single stress ampli-
tude so as to facilitate the preparation of ␶cyc / su – N plots, which
are shown in Fig. 8. The conversions to equivalent numbers of
cycles were performed according to procedures described in Liu
et al. 共2001兲.

Postcyclic Monotonic Undrained Triaxial Compression

Tests
Postcyclic monotonic undrained triaxial compression tests were
performed immediately after the cyclic tests to evaluate the post-
cyclic reconsolidated shear strength of the clayey soils. The tests
were conducted by first dissipating excess pore pressures built up
during cyclic loading, and then monotonically shearing the speci- Fig. 8. Stress ratio 共␶cyc / su兲 versus N curves resulting from cyclic
mens to failure under undrained, strain-controlled conditions. triaxial undrained tests

1600 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / NOVEMBER 2008

J. Geotech. Geoenviron. Eng. 2008.134:1595-1608.

Downloaded from ascelibrary.org by Florida International University on 09/06/13. Copyright ASCE. For personal use only; all rights reserved.

Fig. 9. Comparison of the undrained shear strength ratio between the

precyclic monotonic and postcyclic monotonic tests
Fig. 10. Soil index test results 共LL, PI, and wn兲 of the silty clays at
Wufeng Site A and the liquefaction susceptibility criteria of Bray and
Sancio 共2006兲
nally presented in Fig. 6. As can be seen from Fig. 9 for OCR
= 2, the strength ratios for precyclic and postcyclic conditions are
0.73 and 0.5, respectively. For OCR= 3, the strength ratios for
precyclic and postcyclic conditions are 1.13 and 0.52, respec- ari, Turkey, which suffered substantial ground failure during the
tively. This produces a degradation ratio 共the ratio of the precyclic 1999 Kocaeli Earthquake. Bray and Sancio recommended that
strength ratio divided by the postcyclic strength ratio兲 of 1.5 and soils with PI⬍ 12 and water contents 共wn兲 greater than 0.85 LL
2.2 for OCR of 2 and 3, respectively. These strengths include the 共where LL= liquid limit兲 were susceptible to liquefaction,
beneficial effect of reconsolidation after cyclic loading, whereas whereas soils with 12⬍ PI⬍ 18 共where PI⫽plasticity index兲 and
the postcyclic undrained strengths without reconsolidation would 0.80 LL⬍ wn ⬍ 0.85LL were “systematically more resistant to liq-
have been smaller. The degradation ratio, even with reconsolida- uefaction but still susceptible to cyclic mobility.” Fig. 10 shows
tion, is similar to the sensitivity measured from in situ vane shear index test results 共LL, PI, and wn; originally presented in Figs. 4
tests 共which averaged 2.1兲. and 5兲 for surficial fine-grained soils at the site along with the
Bray and Sancio recommended boundaries. Approximately 60%
of the near-surface materials would be considered susceptible or
Evaluation of Site Performance Using Liquefaction moderately susceptible to liquefaction according to these criteria.
Framework Based on this result, many engineers would reasonably conclude
that ground failure potential for Wufeng Site A could be evaluated
An evaluation of seismic ground failure potential for fine-grained, using liquefaction triggering procedures, as described in the fol-
low-plasticity soils such as those at Wufeng Site A is typically lowing section.
performed as a three-phase process. In the first phase, often Boulanger and Idriss 共2006兲 also recently proposed liquefac-
termed a liquefaction susceptibility analysis, a decision is made as tion susceptibility criteria. Whereas the Chinese criteria distin-
to whether the soil behavior is more “sandlike” or “claylike.” If guish “liquefiable” from “nonliquefiable” materials, Boulanger
the latter is chosen 共claylike behavior兲, many engineers mistak- and Idriss 共2006兲 distinguish between sandlike and claylike be-
enly assume that no ground failure potential exists. This case is havior on the basis of critical behavioral metrics 共undrained
explored further subsequently in the paper. If the former is chosen strength normalization, shape of stress–strain loops, etc.兲, with the
共sandlike behavior兲, subsequent analysis occurs within a liquefac- distinction having a direct correspondence to the type of engineer-
tion framework. In this case the second phase is undertaken, in ing procedures that are best suited to evaluating their seismic
which the potential for liquefaction triggering is evaluated based behavior. They recommended that fine-grained soils with PI⬎ 7
on correlations with soil penetration resistance. Assuming lique- should be considered claylike, and therefore evaluated using in-
faction is triggered somewhere in the profile, the third phase of formation from in situ testing, laboratory testing, and empirical
analysis is intended to identify the effects of liquefaction, which correlations that are similar to those used for clays. They recom-
often includes an evaluation of whether or not liquefaction effects mended that fine-grained soils not meeting the above-mentioned
are likely to be manifest at the ground surface. Each phase of criteria should be considered as sandlike 共and hence evaluated
analysis is discussed in the following. using SPT- and CPT-based liquefaction correlations兲, unless
shown otherwise through detailed in situ and laboratory testing.
Fig. 11 shows the Wufeng Site A data relative to the Boulanger-
Liquefaction Susceptibility
Idriss criteria, which indicates about 50% of the samples are in
It is standard practice in a liquefaction analysis to use soil index the transition zone or sandlike behavior zone. Again, many engi-
test results to judge whether a soil is liquefiable or is too clay rich neers would reasonably conclude from these results that the site
to liquefy. The so-called Chinese criteria have been widely used could be analyzed using a liquefaction framework.
for this purpose since the early 1980s 共Seed and Idriss 1982; Both Bray and Sancio 共2006兲 and Boulanger and Idriss 共2006兲
Youd et al. 2001兲. Bray and Sancio 共2006兲 recently proposed emphasize the value of material-specific laboratory testing for
alternative liquefaction susceptibility criteria based largely on marginal cases such as the Wufeng soils. This approach is applied
field observations and laboratory testing of samples from Adapaz- subsequently in this paper.

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / NOVEMBER 2008 / 1601

J. Geotech. Geoenviron. Eng. 2008.134:1595-1608.

 techniques are applied here to the Site A soils. As described ear-
lier the application of liquefaction triggering procedures to these
soil materials is generally consistent with the Bray and Sancio
共2006兲 and Boulanger and Idriss 共2006兲 liquefaction susceptibility
criteria. The former is very similar to the susceptibility criteria
used by Cetin et al. 共2004兲 and Moss et al. 共2006兲 in the prepa-
ration of the databases that underlie the correlation relationships.
Liquefaction triggering analysis was performed for both the
near-surface fine-grained soil layers 共within intervals deemed sus-
ceptible to liquefaction兲 and deeper sand layers. Fig. 12 shows
results for both materials. The symbols labeled CL reflect the
shallow, fine-grained soils that meet susceptibility criteria. The
Downloaded from ascelibrary.org by Florida International University on 09/06/13. Copyright ASCE. For personal use only; all rights reserved.

symbols labeled SM correspond to the deeper sands. Liquefaction

is predicted in both layers using both the SPT and CPT ap-
Fig. 11. Soil index test results 共LL, PI兲 of the silty clays at Wufeng proaches. The ranges of the cyclic stress ratio 共CSR兲 and the
Site A and the liquefaction susceptibility criteria of Boulanger and corresponding penetration resistance for liquefiable soil layers at
Idriss 共2006兲 each boring or CPT sounding are shown by the vertical and hori-
zontal bars in Fig. 12. The range of penetration resistance indi-
cated for each point represents the upper and lower bound of field
Liquefaction Triggering measurements within the layer. In addition, the range of the cor-
The purpose of a liquefaction triggering analysis is to evaluate responding CSR indicates the upper and lower bound computed
whether the amplitude and number of cycles of shear loading in within the layer, and results principally from the variation with
sandy soil is sufficient to generate significant pore water pressure depth of stress reduction factors, rd.
and strength loss. Liquefaction triggering analyses are generally
performed using empirical correlations that link the in situ pen-
Surface Manifestation of Liquefaction
etration resistance of soil to the cyclic shear demand that is re-
quired for visual evidence of soil liquefaction to be apparent from Given that liquefaction could reasonably be expected at the site
field case histories. The procedure presented in Youd et al. 共2001兲 given the results of the susceptibility and triggering analyses, the
represents the standard of practice; examples of more recent tech- next question concerns whether or not liquefaction effects would
niques based on larger databases are Cetin et al. 共2004兲 共SPT- be expected to manifest at the ground surface. Ishihara 共1985兲
based approach兲 and Moss et al. 共2006兲 共CPT approach兲. Those found through detailed analysis of field case history data that the

Fig. 12. Results of liquefaction triggering analysis for Wufeng Site A east side 共WAS-1 and WAC-2兲 and west side 共WAS-3, WAS-4, WAC-8, and
WAC-9兲

1602 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / NOVEMBER 2008

J. Geotech. Geoenviron. Eng. 2008.134:1595-1608.

Downloaded from ascelibrary.org by Florida International University on 09/06/13. Copyright ASCE. For personal use only; all rights reserved.

Fig. 13. Response spectra for recorded motions and computed motions at Wufeng Site A

occurrence of liquefaction in some layer of a deposit is not nec- Analysis of Cyclic Softening of Clays
essarily associated with damage to structures and disruption of the
ground surface. He investigated the conditions under which liq- Cyclic softening is a term used to describe liquefaction-like be-
uefaction effects are manifest at the ground surface in terms of the havior in clays. It involves cyclic pore pressure induced softening
thickness of liquefiable strata and overlying nonliquefiable strata. and strength loss. However, peak pore pressure ratios in clays are
Youd and Garris 共1995兲 reexamined the Ishihara criteria and in- typically smaller than those in sands, and accordingly the charac-
dicated that the criteria do not apply under conditions that lead to teristics of the cyclic stress–strain responses differ. Boulanger and
lateral spreading or ground oscillation 共i.e., the criteria could pre- Idriss 共2007兲 describe a procedure to evaluate the potential for
dict no manifestation, when in fact surface disruption would cyclic softening of clays by comparing the seismic shear demand
likely occur兲. For Wufeng Site A, the thickness of the near- to the shear stresses required to induce large strains 共taken as
surface, nonliquefiable layer is mostly dictated by depth to water ⫾3% cyclic shear strains兲. That procedure is applied to the
table, which is approximately 1 m. Because the underlying “liq- Wufeng Site A conditions in the following.
uefiable” strata are relatively thick, liquefaction effects would be
expected to manifest at the surface at Site A.
Seismic Demand
Summary of Liquefaction-Based Analysis Seismic demand is represented by a CSR, defined as the ratio of a
When interpreting the results of the analyses discussed in the representative cyclic shear stress acting on a horizontal plane to
preceding sections, it should be recalled that the field perfor- the preearthquake vertical effective stress. It is necessary to con-
mance during the earthquake consisted of bearing failures of tall sider demand from both free-field ground response and from soil–
buildings, no distress to single-story buildings, and no evidence of structure interaction effects.
ground failure in the free field 共cracking, sand boils, etc.兲. The Free-field seismic demand is characterized using shear stresses
above-presented methods of analysis essentially apply for free- calculated from wave propagation analyses. Those analyses in-
field conditions, where no ground failure was observed at Wufeng volved the deconvolution to rock of a recording on relatively firm
Site A. ground at Recording Station TCU065 共located 15– 300 m from
Arguably the most critical step in the liquefaction evaluation is Site A in Wufeng, see Fig. 1兲. The deconvolution was performed
the susceptibility analysis. Using established methods, about 50– using the soil properties at TCU065 and equivalent-linear ground
60% of the Site A materials are judged “susceptible” or to be response analysis according to the procedure given in Silva
within a transition zone, which many engineers would take as 共1986兲. Those rock motions were then propagated upwards from
liquefiable so as to err on the conservative side. The triggering the rock through site profiles for the west and east sides of Site A
analysis subsequently indicates the soils would be expected to using both equivalent-linear and nonlinear analysis techniques.
liquefy, and the effects would be expected to manifest at the The nonlinear codes and parameter selection protocols are taken
ground surface. Hence, this method of analysis is unable to ex- from Kwok et al. 共2007兲. Details on the site dynamic properties
plain the observed field performance. In a design situation, the and additional details on the ground response analyses are given
engineer may choose between designing for the predicted effects in Chu 共2006兲. Fig. 13 shows geometric mean 5%-damped re-
of liquefaction or undertaking a program of site-specific material sponse spectral accelerations from the TCU065 recording 共firm
testing to better evaluate the cyclic behavior characteristics of the soil兲, the deconvolved rock motion, and the range of predicted
soils 共e.g., Bray and Sancio, 2006; Boulanger and Idriss, 2006兲. motions for the west and east sides of Site A. The calculated
This is the approach taken subsequently in this paper. motions at Site A differ from TCU065 principally at high frequen-

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / NOVEMBER 2008 / 1603

J. Geotech. Geoenviron. Eng. 2008.134:1595-1608.

 Using base shear and moment histories applied to the founda-
tions of the east-side and west-side structures as described earlier,
horizontal shear stresses and vertical stresses were evaluated in
the foundation soils using elastic solutions for flexible founda-
tions 共Poulos and Davis 1974兲. In effect, the base shear and mo-
ment acting on a foundation at a given point in time was assumed
to instantaneously impart the shear and vertical stresses in the
ground estimated by the elastic solutions, and this is repeated for
each time step to define histories of shear stress change and ver-
tical stress change in the foundation soil from seismic soil–
structure interaction. Relatively sophisticated dynamic finite
element analyses of a two-dimensional elastic foundation-soil
Downloaded from ascelibrary.org by Florida International University on 09/06/13. Copyright ASCE. For personal use only; all rights reserved.

system were performed to verify that the time lag between foun-
dation loading 共with shear and vertical loads兲 and corresponding
stress changes in foundation soils was negligible 共Coe and Bran-
denberg, personal communication, 2007兲. That work confirmed
that the assumption of coinciding the stress changes with the base
shears and moments is reasonable.
Fig. 14. Pushover curve representing relationship between total lat- Peak stresses induced by soil–structure interaction below the
eral forces applied to building and roof displacement single-story and six-story buildings are summarized in Fig. 15.
CSRs for the east-side single-story building are generally small.
For the six-story building, shallow CSRs are more severe below
cies. Both the east and west sides have predicted peak accelera- the footings than below the mat. The highest CSRs occur near the
tions of about 0.65 g, which is slightly smaller than the recorded edges of the building; applied stresses are relatively low below
geometric mean value of 0.7 g. The west-side motions 共for both the middle of the building, and the corners represent an interme-
the equivalent linear and nonlinear models兲 are similar to diate case. The stress changes across the base of the footing and
TCU065 at midperiods 共near 1.0 s兲, with no appreciable differ- mat foundations reported in Fig. 15 reflect the moment loading of
ence at longer periods. those foundation elements and not localized stress concentrations
Seismic demand is induced in the foundation soils by the in- associated with foundation rigidity, which is known to concen-
ertial forces associated with the vibrating building both because trate stresses beneath the edges of foundations 共e.g., ATC 1996兲.
of base shear and rocking. Because of the relatively small size 共in Accordingly, the edge stresses shown in Fig. 15 may be on the
plan兲 of the structures at the site, kinematic soil–structure inter- low side, especially at shallow depths.
action effects from base slab averaging were found to be negli- The final characterization of cumulative seismic demand in-
gible using the model of Kim and Stewart 共2003兲. Inertial volves summing horizontal shear stress histories from free-field
interaction effects were evaluated for two generic building types. ground response 共␶ff兲, soil–foundation–structure interaction asso-
For the east side of Site A, the building was single story and of ciated with base shear 共␶V兲, and soil–foundation–structure inter-
wood-frame construction with shallow strip foundations. Because action associated with base rocking 共␶ M 兲
these structures had no observable damage, they were assumed to
respond essentially elastically, and the base shear and moment
were estimated using the response spectral ordinate 关from Fig. ␶cum = ␶ff + ␶V + ␶ M 共1兲
13共b兲兴 at the first mode period of 0.07 s.
For the west side of Site A, a six-story reinforced concrete The maximum value of the cumulative shear stress is then
frame with masonry infill walls was used. The foundation consists evaluated
of a combination of spread footings supporting four columns on
the street side of the building and a mat supporting twelve addi-
tional columns. These structures are known to have responded 共␶cum兲max = max共␶ff + ␶V + ␶ M 兲 共2兲
inelastically during the Chi Chi Earthquake 共many collapsed兲. A
nonlinear structural model was developed that included elastic For most design applications, the use of Eq. 共2兲 is inconvenient
springs to simulate soil–foundation flexibility in translation and because it requires evaluation of each of the three stress histories.
rocking, elastic beam elements with elastic-perfectly plastic Evaluation of peak values of each quantity is easier. Accordingly,
hinges at their ends 共the plastic moment capacity was evaluated an issue of practical concern is the degree of overestimation that
from reinforced concrete section properties兲, and elastic column would be associated with calculating the maximum value of the
elements with end hinges characterized by postyield hardening cumulative shear stress as the sum of the maximums 关i.e.,
and eventual collapse 共per FEMA-356 component models兲. A 共␶ff兲max + 共␶V兲max + 共␶ M 兲max兴. Beneath the footing foundations in the
backbone curve for this structural model was developed using front of the six-story building, the degree of overestimation varies
pushover analysis in SAP2000 共Computers and Structures, Inc. from about 15% in the upper 4 m depth to 20% at depths of
2000兲, with the result shown in Fig. 14. A trilinear fit to the 10– 15 m. Beneath the mat foundation, the error is significantly
backbone curve was developed 共as shown in Fig. 14兲, which was larger, ranging from 50% in the upper 4 m to 30% from
then used to define the properties of an equivalent nonlinear 10 to 15 m depth. However, for the present study 共␶cum兲max was
single degree-of-freedom oscillator 共elastic period= 0.42 s兲. The used per Eq. 共2兲.
oscillator response to the west-side ground motions 关Fig. 13共a兲兴 The CSR is evaluated from the maximum shear stress calcu-
was then evaluated to define base shear and moment histories lated using Eq. 共2兲 by normalizing by the preearthquake vertical
applied to the foundation. effective stress

1604 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / NOVEMBER 2008

J. Geotech. Geoenviron. Eng. 2008.134:1595-1608.

Downloaded from ascelibrary.org by Florida International University on 09/06/13. Copyright ASCE. For personal use only; all rights reserved.

Fig. 15. 共a兲 The range of CSR from soil–foundation–structure interaction under the footings of a six-story building on the west side of Site A;
共b兲 the range of CSR from SFSI under the mat foundation of a six-story building on the west side of Site A; and 共c兲 CSR from SFSI below the
middle edge of the spread footing for the single-story buildings on the east side of Site A

共␶cum兲max and 6, respectively兲. C2D accounts for the additional cycles asso-
CSR = 0.65 共3兲 ciated with multidirectional shaking relative to single-direction
␴⬘vc
shaking and is taken as 0.96 共Boulanger and Idriss 2007兲.
Profiles of CSR calculated in the above-presented manner are CRR profiles for foundation soils below the one and six-story
shown in Fig. 16 for soils below the single-story building and buildings are shown in Fig. 16. The presence of the building
soils below the footing and mat for the six-story building. For the affects the cyclic resistance 共CRR兲 relative to free-field conditions
footing and mat, CSR profiles at three locations are shown. Be- by increasing the preearthquake ␴⬘vc for a given depth. This re-
neath the center of the foundation 共point A兲, CSR is decreased duces OCR for a given ␴⬘p, which, in turn, reduces shear strength
relative to the free field. Beneath the corner 共point C兲 and edge
ratio 共su / ␴⬘vc兲. Per Eq. 共4兲, this causes CRR to decrease. The pres-
共point B兲, CSR is increased relative to the free-field near the
ence of the building also increases the static shear stress ratio,
surface 共upper 1 – 4 m兲, and is generally decreased at larger depth.
The presence of the building affects the seismic demand 共CSR兲 by ␣共␣ = ␶static / ␴⬘vc兲, which decreases 共␶cyc / su兲N,␣. Per Eq. 共4兲, this
increasing ␴⬘vc for a given depth, which reduces CSR, and increas- also causes CRR to decrease.
ing the cyclic shear stress acting on horizontal/vertical planes
共␶hv兲 due to base shear and rocking, which can increase CSR. Summary of Findings

Evaluation of Seismic Resistance A comparison of the CSR and CRR profiles indicates that the
foundation soils underlying the footings for the six-story struc-
The cyclic resistance ratio 共CRR兲 represents the ratio of cyclic tures would be expected to undergo cyclic failure 共as CRR
shear resistance to initial vertical effective stress. Boulanger and ⬍ CSR兲 from 1.5 to 3 m depth 共0 – 1.5 m below the foundation兲
Idriss 共2007兲 showed that CRR can be expressed as and from 8 to 11 m depth 共6.5– 10 m below the foundation兲. This

CRR = 冉 冊 冉 冊
␶cyc
␴⬘vc
= C2D
␶cyc
su
su
N,␣ ⬘
␴ vc
共4兲
result is generally consistent with the large observed settlement of
these footings 共and intermediate floor slab warping兲. The founda-
tion soils underlying the mat foundation generally have CRR
where 共␶cyc / su兲N,␣ represents the ratio of degraded-to-monotonic ⬎ CSR in the upper 6 m. Local stress concentrations near the
undrained strengths evaluated for an appropriate value of N 共num- edges of the mat associated with mat rigidity could worsen the
ber of cycles兲 and ␣ 共ratio of static shear stress on a horizontal situation relative to what is shown in Fig. 15. Hence, we cannot
plane to ␴⬘vc兲. The ratio 共␶cyc / su兲N,␣ and the shear strength ratio preclude the occurrence of cyclic softening at shallow depths be-
共su / ␴⬘vc兲 are evaluated from laboratory test results 共i.e., Figs. 8 neath the mat, although clearly the situation is less critical than

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / NOVEMBER 2008 / 1605

J. Geotech. Geoenviron. Eng. 2008.134:1595-1608.

Downloaded from ascelibrary.org by Florida International University on 09/06/13. Copyright ASCE. For personal use only; all rights reserved.

Fig. 16. CRR and CSR profiles for free-field and beneath-foundation cases on the east and west sides of Wufeng Site A

beneath the footings. This is generally consistent with the rela- pending on location兲. Considering the effects of anisotropy, the
tively good performance of the mat foundations on the west side averaged undrained shear strength 共su-ave兲 on a failure surface can
of the site. be taken as 80% of the shear strength from triaxial compression
In the free field on the west side, CRR is greater than CSR tests su 共Ladd, 1991兲, which is about 61 kPa. Using Vesic’s 共1973兲
from 1.5 to 8 m 共0 – 6.5 m below the foundation兲 depth. This re- bearing capacity theory, the ultimate bearing capacity for the foot-
sult is consistent with the lack of free-field ground failure. For the ing is 387 kPa, which provides a preearthquake factor of safety
CSR and CRR profiles under the single-story structures on the 共FSpreeqk兲 of 2.7. Following the earthquake, all conditions remain
east side, our analysis shows CRR⬎ CSR over the full depth of the same except that the undrained shear strength is reduced by a
the clay 共0 to approximately 5.0 m兲 both beneath the one-story factor ranging from 1.5 to 2.2; the postearthquake factor of safety
structures and in the free field. This is consistent with field obser- 共FSposteqk兲 is then estimated to be 1.2–1.8. These analyses indicate
vations of no ground failure in this area. that no significant bearing capacity problems would be expected
at the site either prior to or following the earthquake. This is
consistent with reports of local residents, which indicate no per-
Bearing Capacity Evaluation ceptible continuation of foundation settlement following the
earthquake. Similar analyses were carried out for the bearing ca-
In this section, we evaluate the bearing capacity of the foundation pacity under mat foundations. Our analysis indicates a pree-
soils beneath the six-story building on the west side of Site A to arthquake factor of safety 共FSpreeqk兲 of 4.1. If we conservatively
investigate whether bearing capacity theory, when used with care- assume that cyclic softening occurred, the postearthquake factor
fully chosen undrained strength parameters, can successfully pre- of safety 共FSposteqk兲 would be 1.9–2.7.
dict the observed performance. There are three conditions to be During the earthquake, the bearing capacity factor Nc is modi-
analyzed. Two involve static bearing capacity, one preearthquake fied to Nce to account for inertial forces in the soil mass 共some-
and one postearthquake with shear strengths reduced for the re- times referred to as a kinematical interaction effect on the bearing
molding effects of cyclic degradation. The third condition in- capacity兲. Paolucci and Pecker et al. 共1997兲 and Mylonakis et al.
volves bearing capacity during strong earthquake shaking. 共2002兲 have found that this effect is small for bearing capacity of
The long-term, static loading on the square footing is esti- undrained clays, and for an effective acceleration of approxi-
mated to be 140 kPa. Based on the ICU triaxial compression test mately 0.4 g 共taken as 2 / 3 of the peak acceleration兲, Nce / Nc is
results in Fig. 9, the averaged undrained shear strength 共su兲 of the approximately 0.88. The analysis of bearing capacity must also
upper 4 m clay is approximately 76 kPa 共range 34– 89 kPa, de- account for the “inclination” of the load acting on the footing.

1606 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / NOVEMBER 2008

J. Geotech. Geoenviron. Eng. 2008.134:1595-1608.

 Using a peak base shear of 3,550 kN, giving a V / mg ratio of failures of relatively tall buildings occurred in the low-plasticity
0.19, Fci = 0.94 using the load inclination factors of Vesic 共1973兲. clays at the site. The liquefaction susceptibility criteria of Bray
The vertical stress acting during the earthquake is increased near and Sancio 共2006兲 and Boulanger and Idriss 共2006兲 both indicate
the ends of the footing due to rocking from 140 to 231 kPa. If that about 50–60% of the low-plasticity clays 共CL, CL-ML兲
共suave兲 is taken as the preearthquake long-term static value but should be considered liquefiable unless shown otherwise by site-
increased by 33% to account for the faster earthquake loading rate specific material testing.
共Sheahan et al. 1996兲, the ultimate bearing capacity becomes Analyses using SPT- and CPT-based liquefaction correlations
426 kPa 关=387 kPaNce / NcFci共su−ave兲eqk / 共su−ave兲static兴, which pro- showed the soil has sufficiently low penetration resistance that
vides a during-earthquake factor of safety 共FSeqk兲 of 1.8. How- liquefaction would be expected, and its effects would be expected
ever, recognizing the cyclic softening of the clayey foundation to manifest at the surface. As this was not observed other than
soils predicted by the analyses described in the previous section, a beneath relatively tall buildings, we find that analysis of the
reduced shear strength corresponding roughly with the postcyclic site in a liquefaction framework is unable to explain the field
Downloaded from ascelibrary.org by Florida International University on 09/06/13. Copyright ASCE. For personal use only; all rights reserved.

undrained strength should be used. The effects of loading rate on performance.

the degraded shear strength are not well understood. Nonetheless, More detailed in situ and laboratory testing 共vane shear; con-
to be consistent with the ␶cyc / su ratios evaluated from laboratory solidation; ICU triaxial compression; ICU and ACU cyclic tri-
testing 共which already include rate effects, as ␶cyc is measured in axial compression兲 of the low-plasticity clays showed that they
a fast test兲 we remove the rate effect and reduce su-ave by factors exhibited stress-history-dependent normalized monotonic and cy-
ranging from 1.5 to 2.2. These reduced strengths provide a clic strengths and would be better evaluated using procedures
during-earthquake FSeqk = 0.6– 0.9 for footings on the degraded appropriate for clays.
clays. This suggests that the observed footing failures were in fact The potential for cyclic softening of the clays was evaluated
bearing failures that resulted from both earthquake-induced for three locations: 共1兲 beneath footing and mat foundations of a
strength reductions from pore pressure generation and large tran- six-story building 共where significant ground failure was observed
sient moment demands on the foundation. As for the bearing ca- beneath footings and minor to negligible ground failure occurred
pacity under the mat foundation, our analysis indicates FSeqk beneath mats兲; 共2兲 beneath the foundation of a single-story build-
= 3.1 with preearthquake shear strengths increased by 33% and ing 共no observed failure兲; and 共3兲 in the free-field 共no observed
FSeqk = 1.1– 1.6 with reduced shear strength to accommodate pos- failure兲. Cyclic softening potential is evaluated by comparing the
sible cyclic softening. Neither case is suggestive of a bearing CSR 共representing demand兲 and CRR 共representing resistance兲. If
failure. However, as noted previously, stress concentrations asso- CRR⬎ CSR at a given depth, cyclic softening is not expected,
ciated with slab rigidity could locally increase vertical stresses whereas cyclic softening is expected if CRR⬍ CSR. Beneath the
beneath the slab from gravity loads 关a factor of about 2.5 relative footing foundations for the six-story building, CSR⬎ CRR over
to the average stress is recommended in ATC-40 for the outer 1 / 6 two depth intervals 共0–1.5 and 6.5– 10 m below the foundation兲,
of mats; ATC 共1996兲兴. This could bring the mat close to local indicating a potential for cyclic softening. Bearing capacity analy-
bearing failures at the edge, although the situation is clearly much ses using reduced strengths to account for cyclic softening indi-
less critical than for footings. cate during-earthquake factors of safety against bearing capacity
The preceding calculations were all based on an average failure for these foundations that fall below one, which is consis-
monotonic undrained shear strength of 76 kPa in the upper 4 m of tent with the field performance. For the mat foundations for the
clay. The actual average strength 共as inferred from CPT data兲 six-story buildings, the potential for cyclic softening and bearing
varies with location across the site from 34 to 89 kPa, which is capacity failure is lower. For the single-story buildings and freef-
approximately 50–120% of the overall average strength of 76 kPa ield, the factors of safety against cyclic softening were greater
that was used in the preceding bearing capacity calculations. The than one, consistent with the lack of observed ground deformation
factor of safety against bearing failure of a square footing at a or foundation settlement. We find that analysis of the site in a
location where the average undrained shear strength was locally cyclic softening 共claylike behavior兲 framework is able to explain
only 34 kPa would then be only 1.8 for preearthquake static sta- the field performance.
bility and 0.8 during the earthquake even without degradation of Factors of safety for the six-story buildings on the west side
soil strengths. A mat foundation would load a larger area such that are generally similar for the cyclic softening and bearing capacity
a local minimum strength value would not likely be applicable. analyses. However, it is desirable to perform both types of analy-
Nonetheless, a reduction of average shear strength by a factor of ses. In fact, the cyclic softening results provide insight into the
0.8 to approximately 60 kPa is all that is required to lower during- undrained shear strengths that should be used for bearing capacity
earthquake factors of safety 共with degradation兲 to 0.9–1.3. These analysis 共i.e., use of peak strength or a reduced strength兲.
results illustrate that a range of foundation behaviors would be This case history demonstrates the potential benefits in prac-
expected due to spatial variations of shear strengths, but that the tice of performing more detailed in situ and laboratory testing in
overall pattern would include bearing failures for square footings low-plasticity fine-grained soil deposits. Reliance on the approxi-
and more localized 共less dramatic兲 bearing failures around the mate guidance provided by index test-based liquefaction suscep-
edges of mats. Thus, the analysis results in combination with the tibility criteria would have led to an overly conservative
spatial variations of undrained shear strengths are consistent with assessment of the potential for ground failure and foundation
observations of smaller building settlements for portions of build- settlement.
ings supported by mat foundations and observations of bearing
failures of shallow footings.
Acknowledgments
Summary and Conclusions
This project was sponsored by the Pacific Earthquake Engineer-
We document the seismic performance of free-field areas and ing Research Center’s Program of Applied Earthquake Engineer-
buildings in a region of Wufeng, Taiwan where bearing capacity ing Research of Lifeline Systems supported by the State Energy

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / NOVEMBER 2008 / 1607

J. Geotech. Geoenviron. Eng. 2008.134:1595-1608.

 Resources Conservation and Development Commission and the properties for foundation design.” Rep. No. EPRI EL-6800, Electric
Pacific Gas and Electric Company. This work made use of Earth- Power Research Institute, Palo Alto, Calif.
quake Engineering Research Centers Shared Facilities supported Kwok, A. O., et al. 共2007兲. “Use of exact solutions of wave propagation
problems to guide implementation of nonlinear seismic ground re-
by the National Science Foundation under Award No. EEC-
sponse analysis procedures.” J. Geotech. Geoenviron. Eng., 133共11兲,
9701568. In addition, the support of the California Dept. of Trans-
1385–1398.
portation’s PEARL program is acknowledged. Professor Ertugrul Ladd, C. C. 共1991兲. “Stability evaluation during staged construction.” J.
Taciroglu and professor John Wallace of UCLA are thanked for Geotech. Engrg., 117共4兲, 540–615.
their assistance with the structural response simulations. Professor Liu, A. H., Stewart, J. P., Abrahamson, N. A., and Moriwaki, Y. 共2001兲.
Scott Brandenberg and his graduate student Joseph Coe per- “Equivalent number of uniform stress cycles for soil liquefaction
formed the dynamic finite element analyses of a two-dimensional analysis.” J. Geotech. Geoenviron. Eng., 127共12兲, 1017–1026.
foundation–soil system and their work is appreciated. Mitchell, J. K., and Soga, K. 共2005兲. Fundamentals of soil behavior, 3rd
Ed., John Wiley & Sons, Hoboken, N.J.
Downloaded from ascelibrary.org by Florida International University on 09/06/13. Copyright ASCE. For personal use only; all rights reserved.

Moss, R. E. S., Seed, R. B., Kayen, R. E., Stewart, J. P., Der Kiureghian,
A., and Cetin, K. O. 共2006兲. “CPT-based probabilistic and determin-
References istic assessment of in situ seismic soil liquefaction potential.” J. Geo-
tech. Geoenviron. Eng., 132共8兲, 1032–1051.
Applied Technology Council 共ATC兲. 共1996兲. “Seismic evaluation and ret- Mylonakis, G., Gazetas, G., Nikolaou, S., and Chauncey, A. 共2002兲. “De-
rofit of concrete buildings.” Rep. No. 96-01, Seismic Safety Commis- velopment of analysis and design procedures for spread footings.”
sion, Sacramento, Calif. Technical Rep. No. MCEER-02-003, Dept. of Civil Engineering, City
Boulanger, R. W., and Idriss, I. M. 共2006兲. “Liquefaction susceptibility Univ. of New York, New York and Dept. of Civil, Structural and
criteria for silts and clays.”J. Geotech. Geoenviron. Eng., 132共11兲, Environmental Engineering, Univ. at Buffalo, State Univ. of New
1413–1426. York, Buffalo, N.Y.
Boulanger, R. W., and Idriss, I. M. 共2007兲. “Evaluation of cyclic soften- Paolucci, A., and Pecker, A. 共1997兲. “Soil inertial effects on the bearing
ing in silts and clays.” J. Geotech. Geoenviron. Eng., 133共6兲, 641– capacity of rectangular foundations on cohesive soils.” Eng. Struct.,
652. 19共8兲, 637–643.
Bray, J. D., and Sancio, R. B. 共2006兲. “Assessment of the liquefaction Poulos, H. G., and Davis, E. H. 共1974兲. Elastic solutions for soil and rock
susceptibility of fine-grained soils.” J. Geotech. Geoenviron. Eng., mechanics, John Wiley & Sons, Syndey, Australia.
132共9兲, 1165–1177. Seed, H. B., and Idriss, I. M. 共1982兲. Ground motions and soil liquefac-
Cetin, K. O., et al. 共2004兲. “Standard penetration test-based probabilistic tion during earthquakes, Earthquake Engineering Research Institute,
and deterministic assessment of seismic soil liquefaction potential.” J. Berkeley, Calif.
Geotech. Geoenviron. Eng., 130共12兲, 1314–1340. Sheahan, T. C., Ladd, C. C., and Germaine, J. T. 共1996兲. “Rate-dependent
Chu, D. B. 共2006兲. “Case studies of soil liquefaction of sands and cyclic undrained shear behavior of saturated clay.” J. Geotech. Engrg.,
softening of clays induced by the 1999 Taiwan Chi-Chi earthquake.” 122共2兲, 99–108.
Ph.D. dissertation, Dept. of Civil and Environmental Engineering, Silva, W. J. 共1986兲. “Soil response to earthquake ground motion.” Rep.
Univ. of California, Los Angeles. No. RP2556-07, Electrical Power Research Institute, Palo Alto, Calif.
Chu, D. B., et al. 共2004兲. “Documentation of soil conditions at liquefac- Skempton, A. W. 共1954兲. “The pore-pressure coefficients A and B.” Geo-
tion and non-liquefaction sites from 1999 Chi-Chi 共Taiwan兲 earth- technique, 4, 143–147.
quake.” Soil Dyn. Earthquake Eng., 24共9–10兲, 647–657. Stewart, J. P., coordinator. 共2001兲. “Chapter 4: Soil liquefaction. Chi-Chi,
Chu, D. B., Stewart, J. P., Boulanger, R. W., and Lin, P. S. 共2007兲. “Cy- Taiwan earthquake of September, 21, 1999 reconnaissance report, J.
clic softening of low-plasticity clay and its effect on seismic founda- Uzarski, and C. Arnold, eds.” Earthquake Spectra, 17 共Supplement
tion performance.” Proc., 4th Int. Conf. on Earthquake Geotechnical A兲, 37–60.
Engineering, Springer, The Netherlands. Vesic, A. S. 共1973兲. “Analysis of ultimate loads of shallow foundations.”
Computers and Structures, Inc. 共2000兲. “SAP2000, static and dynamic J. Soil Mech. and Found. Div., 99共1兲, 45–73.
finite element analysis of structures.” Version 7.40, Berkeley, Calif. Youd, T. L. et al. 共2001兲. “Liquefaction resistance of soils: Summary
Ishihara, K. 共1985兲. “Stability of natural deposits during earthquakes.” report from the 1996 NCEER and 1998 NCEER/NSF workshops on
Proc., 11th Int. Conf. Soil Mechanics and Foundation Engineering, evaluation of liquefaction resistance of soils.” J. Geotech. Geoenvi-
Vol. 1, San Francisco, Calif., 321–376. ron. Eng., 127共10兲, 817–833.
Kim, S., and Stewart, J. P. 共2003兲. “Kinematic soil-structure interaction Youd, T. L., and Garris, C. T. 共1995兲. “Liquefaction-induced ground sur-
from strong motion recordings.” J. Geotech. Geoenviron. Eng., face disruption.” J. Geotech. Engrg., 121共11兲, 805–809.
129共4兲, 323–335. Zergoun, M., and Vaid, Y. P. 共1994兲. “Effective stress response of clay to
Kulhawy, F. H., and Mayne, P. W. 共1990兲. “Manual on estimating soil undrained cyclic loading.” Can. Geotech. J., 31共5兲, 714–727.

1608 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / NOVEMBER 2008

J. Geotech. Geoenviron. Eng. 2008.134:1595-1608.