You are on page 1of 12

In Situ Pore-Pressure Generation Behavior

of Liquefiable Sand
Wen-Jong Chang1; Ellen M. Rathje2; Kenneth H. Stokoe II3; and
Kenan Hazirbaba4

Abstract: To overcome current limitations in predicting in situ pore-pressure generation, a new field testing technique is used to measure
directly the coupled, local response between the induced shear strains and the generated excess pore pressure. The pore-pressure genera-
tion characteristics from two in situ liquefaction tests performed on field reconstituted specimens are presented, including the pore-
pressure generation patterns at various strain levels, the observed stages of pore-pressure generation, and pore-pressure generation curves.
Comparisons of the in situ pore-pressure generation curves with data in the literature and from laboratory strain-controlled, cyclic direct
simple shear tests support the in situ testing results. In addition, the effects of effective confining stress on threshold shear strain and pore-
pressure generation curves are discussed. Comparisons of the rate of pore-pressure generation among the in situ tests, laboratory strain-
controlled tests, and a model based on stress-controlled tests reveal that in situ pore pressures generated in reconstituted soil specimens
during dynamic loading develop more similarly to those from cyclic strain-controlled laboratory testing. This observation implies that the
evaluation of induced strains rather than induced shear stresses may be more appropriate for the simulation of pore-pressure generation.
DOI: 10.1061/共ASCE兲1090-0241共2007兲133:8共921兲
CE Database subject headings: Liquefaction; In situ tests; Pore water pressure; Shear tests.

Introduction Laboratory testing can only provide limited information re-


garding in situ pore-pressure generation, because of the compli-
Liquefaction remains a critical earthquake-induced hazard for cated processes that take place in situ during the pore-pressure
civil infrastructure. The key to fully understanding the effects of generation process. For example, the temporal and spatial varia-
liquefaction is to understand the generation and dissipation of tion of in situ pore-pressure generation can be affected by several
excess pore water pressure. Excess pore-pressure generation has factors, such as upward seepage 共Yoshimi and Kuwabara 1973兲,
been used to characterize the liquefaction potential of soil depos- formation of a water film 共Kokusho and Kojima 2002兲, stratified
its 共Dobry et al. 1982; Marcuson and Hynes 1990兲, estimate the soil effects 共Amini and Sama 1999兲, and alternation of ground
reduction of soil stiffness during earthquake loading 共Iwasaki motion. Therefore, in situ monitoring of pore-pressure generation,
et al. 1984; Marcuson et al. 1990兲, and design drainage systems redistribution, and dissipation during dynamic loading is highly
for liquefaction remediation 共Seed and Booker 1977; Pestana recommended and has been addressed in NRC 共1985兲.
et al. 1997兲. The generation of excess pore pressure also plays a At present, only a few studies have systematically evaluated
role in the nonlinear, dynamic response of granular soil deposits the in situ pore-pressure generation of soil. A summary of these
共Finn et al. 1977; Yang et al. 2003兲. Therefore, current semi- studies is listed in Table 1. In situ pore-pressure measurements
empirical liquefaction procedures and numerical modeling efforts have been made at even fewer sites around the world during
can be enhanced by providing new data and insights regarding the previous earthquakes 共Ishihara et al. 1981; 1989; Shen et al. 1991;
in situ pore-pressure generation and dissipation behavior of satu- Youd and Holzer 1994兲, and these data have provided invaluable
rated granular soils. information regarding in situ pore-pressure generation. However,
the unpredictable timing of earthquakes means that one may wait
1 many years between recorded events, and the unreliability of in-
Assistant Professor, Dept. of Civil Engineering, National Chi Nan
strumentation over long periods of time has lead some to question
Univ., Nantou 545, Taiwan. E-mail: wjchang@ncnu.edu.tw
2
Associate Professor, Dept. of Civil Engineering, Univ. of Texas, the credibility of pore-pressure data collected from in situ instru-
Austin, TX 78712. mentation 共Scott and Hushmand 1995兲. Recently, research has
3
Professor, Dept. of Civil Engineering, Univ. of Texas, Austin, TX focused on active experiments that use various dynamic loading
78712. sources such as downhole blasting 共Charlie et al. 1992; Gohl et al.
4
Senior Staff Engineer, GeoSyntec Consultants, 475 14th St., Suite 2001; Rollins et al. 2003兲. However, concerns have been raised
450, Oakland, CA 94612. regarding the applicability of blast loading to earthquake loading
Note. Discussion open until January 1, 2008. Separate discussions because of the rate of loading, the induced blast pressures, and
must be submitted for individual papers. To extend the closing date by difficulties in monitoring pore pressures after the blast pressure.
one month, a written request must be filed with the ASCE Managing
To improve current limitations in the field measurement of
Editor. The manuscript for this paper was submitted for review and pos-
sible publication on December 28, 2005; approved on July 28, 2006. This pore-pressure generation, a new in situ testing technology has
paper is part of the Journal of Geotechnical and Geoenvironmental been developed that measures the pore pressures and shear strains
Engineering, Vol. 133, No. 8, August 1, 2007. ©ASCE, ISSN 1090- induced by a near-surface, high-energy dynamic source 共Rathje
0241/2007/8-921–931/$25.00. et al. 2005兲. This paper presents results from a series of field

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / AUGUST 2007 / 921


Table 1. Summary of Previous In Situ Liquefaction Measurements
Site Soil type Collected data Source Reference
a b
Owi Island 共Japan兲 Silty fine sand SPT, CPT, pore pressure, Mid-Chiba earthquake, Ishihara et al. 共1981兲
surface acceleration amax = 0.1 g
Sunamachi 共Japan兲 Fine sand SPT, Vs,c pore pressure, Chiba-Toho-Oki earthquake, Ishihara et al. 共1989兲
downhole acceleration amax = 0.12 g
Lotung site 共Taiwan兲 Sand Vs, CPT, pore pressure, 18 earthquakes during 1985–1986 Shen et al. 共1991兲
downhole acceleration
Wildlife site 共United States兲 Silty sand SPT, CPT, Vs, pore pressure, Superstition Hills earthquake, Youd and Holzer 共1994兲
downhole acceleration amax = 0.21 g
South Platte River 共United States兲 Coarse sand SPT, CPT, Vs, pore pressure, Downhole explosions Charlie et al. 共1992兲
particle velocity, settlement
Boundary Bay 共Canada兲 Sandy silt CPT, Vs, pore pressure, Downhole explosions Gohl et al. 共2001兲
surface acceleration,
settlement
Treasure Island 共United States兲 Sand and silty sand SPT, CPT, pore pressure, Downhole explosions Rollins et al. 共2003兲
Vancouver 共Canada兲 settlement
a
SPT: Standard penetration test.
b
CPT: Cone penetration test.
c
Vs: Shear wave velocity measurement.

liquefaction tests using this new testing procedure, and these data number of loading cycles, and the earthquake loading in the soil,
are compared with existing models of pore-pressure generation. typically expressed in terms of the induced shear stresses or the
Laboratory tests are also performed to compare with the field induced shear strains.
results. The locally measured pore pressures are presented to- Using data from undrained, stress-controlled cyclic tests, Seed
gether with the ground response and the induced shear strains to et al. 共1975兲 defined a pore-pressure generation function repre-
provide more insight into the coupled soil response. Further, the senting the relationship between excess pore-pressure ratio 共ru兲
spatial variation of pore-pressure generation and dissipation and the cycle ratio 共N / Nl兲:
are presented to demonstrate the pore-pressure redistribution
characteristics.
ru =
ug 2
= sin−1
␴⬘0 ␲
N
冉冊
Nl
1/2␪
共1兲

Pore-Pressure Generation Models where ug⫽generated excess pore pressure, ␴0⬘⫽initial mean effec-
tive stress under triaxial conditions or the initial vertical effective
Several relatively simple models have been proposed to predict stress for simple shear conditions; N⫽number of loading cycles;
the development of excess pore pressure during cyclic loading. Nl⫽number of uniform stress cycles causing liquefaction; and
These models use a pore-pressure generation function 共e.g., Fig. ␪⫽empirical constant ranging from 0.5 to 0.9, with a mean value
1兲 to express the relationship between excess pore pressure, the of about 0.7 关Fig. 1共a兲兴. The term Nl in Eq. 共1兲 is determined from

Fig. 1. Pore-pressure generation models: 共a兲 from cyclic stress-controlled tests 共adapted from Seed et al. 1975兲; 共b兲 from strain-controlled cyclic
triaxial tests 共adapted from Dobry et al. 1982兲

922 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / AUGUST 2007


the estimated earthquake-induced shear stresses, and thus repre- arrangement of instrumentation are designed to create an axisym-
sents the earthquake loading in the soil. The pore-pressure metric loading system, which simplifies the three-dimensional vi-
generation function demonstrates that the rate of pore-pressure bration into a two-dimensional 共2D兲 case. Consequently, the shear
generation varies during the loading process. The rate of pore- strains calculated represent the shear strains induced in the verti-
pressure generation is relatively constant up to cyclic ratio of 0.8, cal plane.
but accelerates rapidly as the soil specimen approaches liquefac- To compute shear strain, the recorded particle velocity–time
tion 关Fig. 1共a兲兴. These pore-pressure generation functions have histories are converted into displacement–time histories by nu-
been used in engineering practice because of their simplicity. merical integration, with a baseline correction included to remove
Rather than relating pore-pressure generation to the number of the drift induced by noise. The two components 共horizontal and
stress cycles, Dobry et al. 共1982兲 used shear strain to quantify the vertical兲 of displacement at the four corner sensor points and an
state of disturbance to predict pore-pressure generation. Based on assumed linear variation of displacement between sensor points
undrained, strain-controlled, cyclic triaxial tests, Dobry et al. are used to calculate the shear strain-time history within the in-
共1982兲 developed a relationship between excess pore-pressure strumentation array 共Rathje et al. 2004兲.
ratio 共ru兲, cyclic shear strain amplitude, and number of loading To be able to capture in a single setup the pore-pressure gen-
cycles 关Fig. 1共b兲兴. A unique relationship was observed, although eration curve 共PPGC兲 of the soil, expressed as the excess pore-
different soils and different specimen preparation techniques were pressure ratio 共ru兲 versus mean shear strain amplitude for a
used. The unique relationship between excess pore-pressure ratio specific number of loading cycles, a multistage testing procedure
and shear strain amplitude reveals that shear modulus plays an is used. During the initial testing stage, a small loading level is
important role in pore-pressure buildup. Also, the relationship de- used that generates shear strains below the threshold shear strain
fines a threshold shear strain, the strain level above which excess and no accumulated excess pore pressure. The load amplitudes for
pore pressure begins to generate 共Dobry et al. 1982兲. Below the the subsequent stages are increased based on the induced shear
threshold strain, no excess pore pressure is generated even under strain level and/or generated excess pore pressure during the pre-
a large number of loading cycles. vious stage. Between stages, any excess pore pressure is allowed
Threshold shear strain is stress dependent and Dobry and to dissipate over a 30-min period. The loading levels are gradu-
Swiger 共1979兲 proposed a relationship between threshold shear ally increased until a moderate amount of pore pressure is ob-
strain 共␥t兲 and isotropic confining effective stress 共␴⬘兲: served 共ru ⬃ 10– 20% 兲. Because loads that generate significant
pore pressures 共ru ⬃ 20– 80% 兲 will result in permanent changes to
␥t = 1.33 ⫻ 10−5共␴⬘兲 共2兲 the soil structure and adversely affect subsequent loading stages,
where ␥t is in percent and ␴⬘ is in Pascal. Eq. 共2兲 predicts a these load levels are avoided. After the load level with ru ⬃ 20%,
threshold shear strain between 0.01 and 0.04% for normal stresses the maximum output of the vibroseis is applied and the largest
between 0.25 and 2 atm, respectively. Additionally, the model for shear strains and excess pore water pressures are generated. Ad-
modulus reduction curves of sand by Menq 共2003兲 can be used to ditional details regarding the testing procedure, instrumentation,
derive the confining pressure dependence for a threshold strain and data analysis can be found in Rathje et al. 共2005兲 and Chang
based on a given value of G / Gmax 共typically ⬃0.8兲. This model 共2002兲.
produces a relationship in which ␥t varies with 冑␴⬘. Convention-
ally, the threshold strain for pore-pressure generation is assumed
Field Testing Specimens
to be equal to about 0.01% for confining pressure close to 1 atm.
The in situ dynamic liquefaction test was used to dynamically
load two reconstituted test specimens 共T1, T2兲. In each test series,
Description of Field Testing Program several dynamic tests were performed with the vibroseis to estab-
lish the pore-pressure generation curve for the reconstituted test
The main objective of the developed in situ testing technique is to specimen. Each loading within each test series was named ac-
capture the coupled response of the induced shear strains and the cording to its chronological order and was expressed as an exten-
generated excess pore water pressures during dynamic loading. sion after the test series name 共e.g., T1-1, T1-2, etc.兲. Eight tests
Additionally, the subsequent pore-pressure dissipation and settle- were performed in Test Series T1 and each test was dynamically
ment after dynamic loading are also measured during testing. loaded for 20 cycles at a loading frequency of 20 Hz. The second
A vibroseis truck vibrating vertically on a circular foundation test series 共T2兲 had the same configuration as T1, but was de-
is used to generate stress waves 共shear, compression, and signed to study pore-pressure generation patterns and the varia-
Rayleigh-type surface waves兲 that propagate through an instru- tion of soil stiffness during a large number of loading cycles. Test
mented test area and cyclically shear the soil at a specified Series T2 consisted of four loading levels applied for 60 cycles at
frequency for a given number of cycles. The embedded instru- a frequency of 20 Hz.
mentation includes geophones, which measure particle velocity– The reconstituted test specimens were constructed in an exca-
short-time histories, and miniature pore-pressure transducers, vated trench 共approximately 1.2 m by 1.2 m by 1.2 m兲, and the
which measure excess pore-pressure–time histories 共Rathje et al. vibroseis truck was located 3.3 m from the trench. The typical
2005兲. The sensor signals 共particle velocities, excess pore pres- configuration of the reconstituted test specimens is shown in Fig.
sures兲 are recorded by dynamic data acquisition systems. Two 2. Because the natural ground water table at the site is about 1 m
geophones 共horizontal and vertical兲 and one pore-pressure trans- below the bottom of the trench, an impermeable liner was placed
ducer are integrated in a small 共3.8 cm by 8.9 cm兲 waterproof within the trench to keep water from seeping out of the test speci-
case and the integrated unit is called a liquefaction sensor. Lique- men. A PVC pipe was placed at the side of the trench furthest
faction sensors are deployed at several locations within the soil to from the vibroseis to provide casing for a crosshole source, such
form a square instrumentation array. The instrumentation array is that crosshole seismic testing could be used for specimen charac-
located in a vertical plane normal to the cylindrical surface of the terization. The liquefaction sensors were placed in a square array
foundation. The vertically vibrating circular foundation and the with specific orientations in order to implement the strain calcu-

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / AUGUST 2007 / 923


Table 3. Physical Properties of Reconstituted Test Specimens
T1 T2
Physical properties Initial Final Initial Final
a
Average relative density 共Dr, %兲 34.7 61.2 36.0 52.0b
Water content 共w, %兲 21.1 19.0 21.0 18.8
Total unit weight 共␥t, kN/ m3兲 20.4 20.8 20.3 20.7
Vsc at 0.22 m deep 共m/s兲 83 83 90 88
V pc at 0.22 m deep 共m/s兲 918 —d 878 —
Vs at 0.82 m deep 共m/s兲 109 98 107 97
V p at 0.82 m deep 共m/s兲 1500 — 1500 —
a
Corresponding to 4.3 cm settlement at top of specimen.
b
Corresponding to 2.4 cm settlement at top of specimen.
c
Vs, V p⫽shear and compression wave velocities, respectively.
d
Not measured.

Fig. 2. Schematic layout of in situ liquefaction test In Situ Soil Characterization


Small-strain seismic testing techniques were employed to verify
lation method. To avoid disturbance of the loose sand, liquefac- specimen saturation and to monitor the stiffness variation of the
tion sensors were installed during the specimen preparation test specimens before and after dynamic loading. The saturation
process. Also, the pore-pressure transducers in the liquefaction verification was conducted by measuring the compression-wave
sensors were submerged during the installation process to main- 共P-wave兲 velocity, while the shear stiffness at different depths
tain saturation. Settlement plates were installed at different depths was monitored by shear wave 共S-wave兲 velocity measurements.
to measure the settlement at various depths. The small-strain seismic tests were performed in accordance with
In preparing the specimens, the main objective was to build typical crosshole and downhole testing procedures, using a vari-
loose, uniform, saturated specimens at a known saturated density. ety of sources and the embedded array sensors as receivers.
The reconstituted specimens were constructed by wet pluviation. The S-wave velocities at depths of 0.22 and 0.82 m, measured
Moist soil at a known water content was weighed and mixed with before and after testing, and the P-wave velocities measured be-
water in a bin at the ground surface to remove air bubbles, and the fore testing are shown in Table 3. The results indicate that the
soil/water mixture was carefully placed in the test pit. Throughout shear wave velocity is greater between the deeper sensors due to
sample preparation, the water level was maintained about 30 cm the increased vertical effective stress. For a saturated soil, the
共1 ft兲 above the top of the backfilled sand to allow for wet plu- theoretical P-wave velocity is around 1,500 m / s 共5,000 ft/ s兲
viation, to minimize the air trapped in the specimen, and to keep 共Allen et al. 1980兲, and high frequency waves should be observed
the sensors submerged during the installation process. A 0.8-m- in the crosshole records after the first arrival of the P waves. The
thick layer of sand was placed on top of the test specimens to cross-hole records from Sensors 3 and 4 共Fig. 2兲 displayed these
provide some additional confinement and to impede drainage at characteristics; therefore, the soil between Sensors 3 and 4
the ground surface. 共0.82 m depth兲 is assumed to be saturated. The P-wave velocity
The soil used for the reconstituted specimens is a clean aggre- between Sensors 1 and 2 generally was much lower 共approxi-
gate sand 共0.43% fines, SP classification in Unified Soil Classifi- mately 900 m / s兲 and no high frequency waves were observed,
cation System兲 with soil properties tabulated in Table 2. The indicating less than full saturation. However, the soil between
physical properties of the reconstituted specimens are summa- Sensors 1 and 2 was likely close to full saturation because
rized in Table 3. The initial saturated density and void ratio for P-wave velocity drops off very quickly as saturation decreases,
each specimen were calculated from the measured weight of soil with the saturation level likely above 99.5% 共Allen et al. 1980;
backfilled in the trench and the measured volume of the trench, Ishihara et al. 2001兲. P-wave velocities were also measured after
assuming full saturation. The specimen preparation procedure shaking, using a push-in geophone as the receiver and an alumi-
produced consistent specimens at about 35% relative density. The num rod as the source; these measurements also indicated that the
postshaking densities were estimated from the vertical settlement area within the instrumented array is close to saturation with V P
measured at the top of the specimen assuming no lateral defor- greater than 900 m / s 共Chang 2002兲.
mation. These densities were confirmed with large-diameter tube
samples obtained after testing. Estimation of Effective Vertical Stress
The effective vertical stress inside the reconstituted specimen is
an important property for computing the excess pore-pressure
Table 2. Soil Properties of Aggregate Sand
ratio, as well as for evaluating the liquefaction potential of the
Soil properties Values soil. The effective stresses at the sensor locations can be com-
Specific gravity 2.68 puted using the soil unit weight, location of the water table, and
Maximum void ratio 0.64
thickness of the overburden layer. However, questions arose
regarding whether the in situ effective stresses matched the effec-
Minimum void ratio 0.43
tive stresses computed in this manner. Assuming that the average
USCS classification SP
total unit weight of the 0.8-m-thick unsaturated overburden sand
Fines content 共%兲 0.43
is 17.6 kN/ m3 and that the overburden can be modeled as an

924 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / AUGUST 2007


Fig. 3. Pore pressure generation from Test T1-1 共small strain level test兲: 共a,b兲 time histories; 共c兲 Fourier amplitude spectra of excess pore-pressure
ratios at Sensors 2 and 3; 共d兲 hydrodynamic pore pressure and vertical particle velocity–time histories at Sensor 3

infinite layer of soil, the effective vertical stresses at depths of Characteristics of In Situ Pore-Pressure Generation
0.22 and 0.82 m are 16.4 and 22.8 kPa, respectively. However,
the measured shear wave velocities 共80 and 110 m / s at depths of Excess pore-pressure generation under seismic conditions is in-
0.22 and 0.82 m, respectively兲 indicate approximate vertical ef- duced by stress wave propagation 共shear, compression, and sur-
fective stresses of about 2.2 and 7.0 kPa at these depths, based on face waves兲. Generally, shear waves 共S-waves兲 cause distortion
resonant column testing of reconstituted specimens of the aggre- 共shear兲 of the soil, compression waves 共P-waves兲 compress the
gate sand at a range in confining pressures. These stresses are soil, and surface waves both shear and compress the soil. Shear
unreasonably low and suggest that some of the weight of the stresses from S-waves induce shear strains and cause a rearrange-
overburden layer was not transferred to the specimen due to the ment of soil particles that leads to permanent, or residual, excess
fact that the overburden layer acted over a finite area 共Fig. 2兲. pore pressure. P-waves, on the other hand, increase the mean
To estimate accurately the vertical effective stresses at various stress in the soil, which leads to excess pore pressure, but this
depths, another test pit was constructed to measure the vertical excess pore pressure disappears after passage of the P-wave.
stress using an earth pressure cell placed at the bottom of the Thus, this excess pore pressure is elastic, or recoverable, and
trench. The vertical effective stress at the bottom of the test pit oscillates in phase with the passing P-waves. This component of
was computed by taking the measured pressure, which represents excess pore-pressure generation is called hydrodynamic excess
the total stress, and subtracting the water pressure at the same pore pressure, after Scott and Hushmand 共1995兲, whereas the
depth, as computed by the water level. Using this procedure, the shear-induced permanent pore pressure is called the residual ex-
measured vertical effective stress at the bottom of the test was cess pore pressure, after Dobry et al. 共1982兲. Conventionally, re-
about 65% of the theoretical value based on the unit weights, searchers are more interested in residual excess pore pressure
water table, and overburden pressure. Axisymmetric, linear- because it is directly related to liquefaction and soil stiffness
elastic finite element analyses of a trapezoidal overburden layer, reduction.
indicated that 68–72% of the theoretical overburden is transferred
to the top of the test pit. Thus, for the test results presented here,
Pore-Pressure Generation at Small Strain Levels
the vertical effective stress at depths 0.22 and 0.82 m are taken as
10.7 and 14.8 kPa 共65% of 16.4 and 22.8 kPa兲, respectively. At For shear strain levels below the threshold shear strain, only
the center of the array 共Sensor 5兲, the vertical stress is then esti- hydrodynamic excess pore pressure is generated and the time
mated to be 12.7 kPa 共65% of 19.6 kPa兲. domain characteristics such as phase, amplitude, and frequency
Implementing the estimated states of stress and the measured content are directly related to the seismic source and wave propa-
values of Vs, the stress-corrected shear wave velocities 共Vs1 gation. The excess pore-pressure ratio–time histories of Test T1-1
= Vs / 冑␴⬘v兲 within the instrumentation array are between 158 and are presented in Fig. 3 to show these characteristics. Results from
168 m / s. Based on the liquefaction-shear wave correlation by Sensors 2 and 3 共Fig. 2兲 are selected to represent different depths
Andrus and Stokoe 共2000兲, these values indicate a highly liquefi- and radial distances from the source.
able soil. To highlight the hydrodynamic and residual excess pore pres-

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / AUGUST 2007 / 925


sures, various digital filters are applied. First, a low-pass filter
with a cutoff frequency of 28 Hz is applied to the raw data to
remove background noise. A low-pass filter with a cutoff fre-
quency of 2 Hz is used to highlight the residual pore pressure,
whereas a band-pass filter, preserving the signals between 15 and
25 Hz, is used to highlight the hydrodynamic pore pressure. In
Fig. 3, ru–time histories 关Figs. 3共a and b兲兴 and the corresponding
Fourier amplitude spectra 关Fig. 3共c兲兴 from Sensors 2 and 3 are
shown for the full loading period. No residual excess pore pres-
sure is generated and the amplitude of hydrodynamic ru during
shaking is almost constant. The spectra clearly show that the hy-
drodynamic ru is focused around 20 Hz, the excitation frequency
of the vibroseis. The induced excess pore pressure and the mea-
sured vertical particle velocity at Sensor 3 are shown in Fig. 3共d兲
and an in-phase behavior is observed. The coincidence of the
dominant frequency of the hydrodynamic excess pore pressure
and the vibrating source frequency, and the in-phase behavior
between the induced hydrodynamic ru with the local ground mo-
tion indicate that the hydrodynamic pressure is a result of wave
passage from the source. Further, a smaller hydrodynamic ru at
depth 共Sensor 3兲 was observed because the average particle ve-
locity, and thus wave amplitude, was smaller. Finally, the shear
strain-time histories for the small-strain tests showed constant
amplitude during shaking.

Pore-Pressure Generation at Large Strain Levels


In contrast to small strain level tests, where the residual excess
pore pressures were zero, significant excess pore pressures were
observed at large strain levels. The pore-pressure ratio-time his-
tories at the center of the array 共i.e., Sensor 5兲 during Test T1-8
are plotted in Fig. 4. Fig. 4共a兲 shows the full record of excess pore
pressure, including the generation stage during shaking and the
postshaking dissipation. The maximum residual pore-pressure
ratio occurred at the end of the shaking and dissipation started Fig. 4. Pore-pressure generation from Test T1-8 共large strain level
immediately thereafter, in an exponential fashion. During dissipa- test兲: 共a兲 complete recorded ru–time history; 共b兲 ru–time history
tion, no hydrodynamic excess pore pressure is recorded. during shaking; and 共c兲 shear strain–time history
Fig. 4共b兲 displays only the pore-pressure generation during
dynamic loading. Again, a 28-Hz low-pass filter was applied to
remove background noise. In comparison with the small strain
and the amplitudes of the induced shear strains, the pore
level tests, a component of residual excess pore pressure is ob-
pressure–time history can be divided into four phases.
served in addition to the hydrodynamic pore pressure. The re-
In Phase 1, the rate of residual pore-pressure generation is
sidual excess pore pressure is separated from the recorded excess
large and the amplitude of the hydrodynamic pore pressure is
pore pressure by a 2-Hz low-pass filter. Also, the induced shear
almost constant. During this phase, the shear strain amplitude also
strain–time history at the same location is shown in Fig. 4共c兲. In
increases as the soil softens. When the residual ru reaches ap-
contrast to small strain level tests, the shear strain amplitude var-
proximately 50%, the pore-pressure response transfers to Phase 2,
ies as residual excess pore pressure is generated, an indication of
where the residual pore pressure continues to increase but the rate
the nonlinear process at work.
of residual pore-pressure generation decreases. The shear strain
amplitude decreases in this phase because the soil has softened so
Phases of Pore-Pressure Generation much that it cannot transfer large stresses/strains. Also in this
phase the amplitude of the hydrodynamic pore pressure continu-
Because the induced shear strain and the generated excess pore ously increases. After the soil reaches an ru of 100%, the ampli-
pressure are coupled, the level of shear strain should be consistent tudes of both the residual and hydrodynamic pore pressures
with the level of excess pore pressure. In addition, the reduction remain constant until the end of shaking 共Phase 3兲. In Phase 3, the
in stiffness due to the generation of excess pore pressure will shear strain amplitude is small and relatively constant, and no
affect stress wave propagation and alter the ground response. The further residual pore pressure is generated. In general, during
measured excess pore pressure and shear strain–time histories Phase 3 the shear strain amplitude is about 0.005%, which is close
should reflect this coupled behavior. The ru–time history from to the threshold strain for pore-pressure generation. In Test T1-8,
Sensor 5 of Test T2-4, which induced a ru value of 100%, is which was a large strain level test that lasted only 20 cycles 关Figs.
shown in Fig. 5 along with the shear strain–time history computed 4共b and c兲兴, only the first two phases were observed because the
for the same location. Based on the rate of residual pore-pressure vibration stopped before the specimen reached Phase 3. When
generation, the amplitudes of the hydrodynamic pore pressure, shaking ends, the excess pore pressure starts to dissipate 共Phase

926 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / AUGUST 2007


Fig. 5. 共a兲 Pore-pressure generation; 共b兲 induced shear strains from Test T2-4 共center of array兲

4兲. In the beginning of Phase 4, the excess pore-pressure dissi- in the field. The overburden soil is unsaturated, and therefore
pates quickly but the rate of dissipation slows as the pore pres- provides some impedance to drainage that slows the dissipation
sures are reduced. Although only the result from Sensor 5 is process.
shown, all of the other sensors recorded a similar pattern.
Pore-Pressure Generation Curves
Pore-Pressure Dissipation
One of the major goals of the in situ dynamic liquefaction test is
All of the liquefaction sensors recorded pore pressure for several to develop relationships between excess pore-pressure ratio 共ru兲,
minutes after the end of shaking, which allows for an assessment shear strain level 共␥兲, and number of loading cycles 共N兲 using the
of the pore-pressure dissipation process. To simplify the 2D test data collected from the embedded instrumentation. These rela-
specimen 共Fig. 2兲 into a one-dimensional 共1D兲 system, a 1D ver- tionships are PPGC. The recorded residual excess pore pressures
tical profile through the location of liquefaction Sensor 5 was are used to compute values of ru for different numbers of loading
chosen. The average pore pressure from Sensors 1 and 2 was used cycles. Because the shear strain amplitude varies during vibration,
to represent the excess pore pressure at a depth of 0.22 m, the mean shear strain amplitude over the number of cycles of
whereas the average pore pressure from Sensors 3 and 4 was used
to represent the excess pore pressure at a depth of 0.82 m. The
pore pressure from Sensor 5 represented the excess pore pressure
at a depth of 0.52 m. Because of the impervious liner placed
around the test pit, 1D vertical pore-pressure dissipation through
the drainage boundary at the top of the specimen is expected.
The excess pore-pressure profile at selected time steps after the
end of shaking 共EOS兲 for Test T2-4 is shown in Fig. 6. The peak
residual excess pore pressures at each depth occurred at the end of
shaking and the residual excess pore pressure at EOS increases
linearly with depth, which agrees with the liquefaction condition
of the instrumented area. The excess pore pressure dissipates over
time, with the quickest rate of dissipation occurring immediately
after the end of loading because the largest gradients are present.
By t = EOS+ 10 s, the excess pore-pressure profile is relatively
constant with depth. The recorded pore-pressure profiles during
dissipation show the effects of a non-free draining boundary at the Fig. 6. Residual excess pore-pressure profiles during dissipation in
top of the specimen due to the existence of the overburden layer Test T2-4

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / AUGUST 2007 / 927


Fig. 7. Pore pressure generation curves from Test Series T2

dynamic loading is plotted in the pore-pressure generation curves


to represent the average induced shear strain level.
Fig. 7 shows the excess pore-pressure ratio 共ru兲 versus shear
strain at the center of the instrumentation array for different num-
bers of loading cycles from Test Series T2, which included four
different dynamic loading levels. The first two loading stages
共T2-1 and T2-2兲 were very small, inducing mean shear strains less
than 1.5⫻ 10−3% and residual excess pore pressures less than 1%.
Although the residual pore pressures were less than 1%, the hy-
drodynamic pore pressures were between 1.5 and 2.5 % for these Fig. 8. 共a兲 Field measured pore-pressure generation curves and
tests. The loading levels were increased further in the third test curves from Dobry et al. 共1982兲; 共b兲 pore-pressure generation curves
stage 共T2-3兲, but the mean shear strain remained around 7 normalized by the threshold strain
⫻ 10−3% and the residual excess pore pressure was below 20%.
The hydrodynamic pore pressure in this stage increased to 9%
because the wave amplitudes were larger. In Test Stage T2-4, the ratory by Dobry et al. 共1982兲. However, the in situ test data reflect
largest loading was applied and it generated a mean shear strain that the residual pore-pressure ratio increases quickly after the
of 1.3⫻ 10−2% and a pore-pressure ratio of 100% after 35 cycles induced shear strain level exceeds the threshold shear strain,
of loading. The hydrodynamic pore-pressure ratio varied between which is consistent with the Dobry et al. 共1982兲 data.
12 and 20% for this stage. No settlement was measured in the test In view of the threshold shear strain, Eq. 共2兲 provides a rela-
specimen in the smaller loading levels 共T2-1 to T2-3兲, but 2.4 cm tionship between threshold strain and confining pressure. The
was measured at the top of specimen after the largest loading stress exponent in Eq. 共2兲 indicates that a reduction in confining
level 共T2-4兲. This settlement corresponds to 2.1% volumetric pressure from 100 kPa 共1 atm兲 to 8.5 kPa 共mean effective stress
strain. The pore-pressure generation curve in Fig. 7 indicates a at the center of array, assuming K0 = 0.5兲 can reduce the threshold
threshold strain for pore-pressure generation of between 3 strain to 20% of its value at 1 atm. The Menq 共2003兲 model
⫻ 10−3 and 5 ⫻ 10−3%, which is smaller than the typical value of predicts a reduction to 30% of the value at 1 atm. Although the
0.01%. field data only indicate a reduction to 30– 50% of the conven-
tional value of 0.01%, the trend agrees with those predicted by
Eq. 共2兲 and Menq 共2003兲.
Discussion To compare the excess pore-pressure ratios at strain levels ex-
ceeding the threshold shear strain, the shear strains in the PPGCs
from in situ testing and from Dobry et al. 共1982兲 are normalized
Comparison of Pore-Pressure Generation Curves
to their respective threshold shear strains. The strain-normalized
with Dobry et al. „1982…
PPGCs are shown in Fig. 8共b兲. The results reveal that even after
The compiled mean shear strains and residual excess pore- the effect of confining pressure on the field-evaluated threshold
pressure ratios for different loading cycles from Sensor 5 of Test strain is taken into account, larger excess pore-pressure ratios are
Series T1 and T2 are shown in Fig. 8共a兲, along with pore-pressure still observed in the field. This discrepancy could, again, be due to
generation curves from Dobry et al. 共1982兲. The Dobry et al. differences in confining pressure, or it could be the result of dif-
共1982兲 pore-pressure generation curves are based on strain- ferent shearing conditions.
controlled cyclic triaxial tests on reconstituted Monterrey sand The main differences in shearing conditions between the field
prepared at a relative density of 45% and performed at an effec- and laboratory tests are multi- versus unidirectional loading and
tive confining pressure of 95.6 kPa. simple shear-type versus triaxial loading. The shear strains gen-
One important difference between the field-measured pore- erated in the field are predominantly in the vertical plane due to
pressure generation curves from this study and those from the horizontal passage of Rayleigh-type surface waves, and to a lesser
Dobry et al. 共1982兲 study is the threshold strain at which pore extent passage of shear waves maove in a plane perpendicular to
pressures start to develop. The threshold shear strain from this the plane of wave propagation 共SV-waves兲. The shaking is not
study is smaller than from Dobry et al. 共1982兲, 0.5⫻ 10−2 versus truly multidirectional because no shear strain is induced in the
1.0⫻ 10−2%. As a result, the field measured pore-pressure ratios out-of-plane direction, and in fact most of the shear strain is in-
at specific strain levels are larger than those measured in the labo- duced by vertical particle motion 共Rathje et al. 2004兲. Thus, the

928 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / AUGUST 2007


Fig. 9. Comparisons of pore-pressure generation curves from in situ
tests and from laboratory strain-controlled, cyclic direct simple shear
tests
Fig. 10. Pore-pressure generation as a function of cycle ratio from in
situ testing, strain-controlled CDSS testing, and Seed et al. 共1975兲
model based on stress-controlled testing
multidirectional effect in the field is minimal. However, the field
tests induce shear strains in a manner that is analogous to simple
shear testing, rather than triaxial testing, in terms of the rotation
of principal planes. Finn et al. 共1971兲 reported that for stress- shows that the slopes of the pore-pressure generation curves at
controlled testing the cyclic resistance measured under simple strains greater than the threshold strain appear to increase with
shear conditions is generally smaller than under triaxial condi- decreasing effective stress, which is similar to the comparison
tions. Thus, the different shearing conditions may explain the dis- between the in situ data and the data from Dobry et al. 共1982兲.
crepancy in pore-pressure development at strains greater than the Thus, the laboratory results indicate that the differences in pore-
threshold strain. However, effective stress may also explain the pressure generation between this study and Dobry et al. 共1982兲
difference, as discussed in the next section. most likely can be attributed to differences in confining pressure.

Comparison of Pore-Pressure Generation Curves Comparison of Rate of Pore-Pressure Generation


with Laboratory Data and Seed et al. „1975… Model
A series of strain-controlled, undrained, cyclic direct simple shear To further investigate the pore-pressure generation process, the
共CDSS兲 tests were performed on the aggregate sand used in the rate of pore-pressure generation 共ru versus normalized number of
field testing 共Hazirbaba and Rathje 2004; Hazirbaba 2005兲 for loading cycles兲 from in situ measurements, laboratory strain-
comparison with the in situ measurements and to assess the effect controlled CDSS tests, and the model proposed by Seed et al.
of confining pressure on pore-pressure generation. The laboratory 共1975兲 关Eq. 共1兲, Fig. 1共a兲兴 are compared. The processed results
specimens were prepared by water sedimentation to replicate the from in situ Test T2-4, the laboratory test with a shear strain
field sample preparation procedure described previously. A total amplitude 0.3% and ␴⬘v = 100 kPa, and the Seed et al. 共1975兲
of 13 reconstituted specimens were tested. Among the 13 tests, 7 model are presented in Fig. 10.
tests were performed at a vertical effective stress of 100 kPa, The excess pore-pressure ratio from in situ tests shows signifi-
whereas 6 tests were performed at 25 kPa. Specimens were not cant differences from the Seed et al. 共1975兲 model. The Seed et al.
tested at 12.7 kPa because the testing apparatus could not apply 共1975兲 model indicates a relatively constant rate of pore-pressure
the small loads required. The relative densities of the reconsti- generation until about 70–80% of ru is reached. At this point, the
tuted specimens ranged from 32 to 44%, with a mean value 38%, rate of pore-pressure generation increases rapidly, with liquefac-
which is close to the field value. Each specimen was subjected to tion occurring shortly thereafter. The strain controlled CDSS
0.2-Hz sinusoidal, constant amplitude shear strain loading until a laboratory test indicates an initial rapid generation of pore pres-
maximum ru greater than 0.9 was reached or 30 cycles of loading sure, but the rate continuously declines until an asymptotic value
were applied. of ru is reached. The in situ test results reveal a rate of pore-
The laboratory results are compared with the in situ results pressure generation that is more similar to the laboratory results,
from Sensor 5 and shown in Fig. 9. The pore-pressure generation an initial rapid rate of generation that continuously decreases. The
curves corresponding to 10 and 20 cycles of straining from the discrepancy in pore-pressure generation is mainly from the load-
laboratory 共␴⬘v = 100 and 25 kPa, Dr = 32− 44%兲 and from the in ing condition.
situ tests 共␴⬘v = 12.7 kPa, Dr = 36%兲 are shown. The curves from The Seed et al. 共1975兲 model is based on stress-controlled tests
laboratory tests with ␴⬘v = 100 and 25 kPa show that ␥t is within where a constant stress amplitude is applied until liquefaction
0.7⫻ 10−2 to 0.9⫻ 10−2% and that ␥t appears to decrease slightly occurs. During these tests, the shear strain induced in each sub-
as confining pressure decreases. sequent cycle of loading increases because the increasing pore
The effect of effective stress on pore-pressure generation at pressure decreases the stiffness of the soil 共i.e., shear modulus,
shear strains larger than the threshold shear strain can be observed G兲. As the stiffness decreases, more strain is induced, which in-
from Fig. 9 as well. Because all tests were conducted on recon- duces more pore-pressure generation in each loading cycle. Con-
stituted specimens of the same material prepared by similar tech- sidering the shear modulus to be proportional to the square root of
niques 共wet pluviation兲, the variations of the three test series can the mean effective stress, the reduction in G as a function of ru
be predominantly attributed to variations in effective stress. Fig. 9 can be written as

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / AUGUST 2007 / 929


G/G0 = 共1 − ru兲0.5 共3兲 shear tests at different confining pressures. Comparisons of the
rate of pore-pressure generation from the in situ tests, the labora-
where G0⫽initial, maximum shear modulus and ru is in decimal. tory strain-controlled tests, and the Seed et al. 共1975兲 model based
Because the shear modulus is affected by the square root ru, the on stress-controlled tests revealed that the in situ pore pressure
reduction is not as significant for ru ⬍ 80% 共G ⬃ 0.45G0 at developed more like the strain-controlled tests than the stress-
ru = 80%兲. For ru ⬎ 80%, the shear modulus reduces dramatically, controlled tests, although the in situ tests are not truly responding
which leads to increased rates of shear strain and pore-pressure in either a stress-controlled or strain-controlled manner. Nonethe-
generation in stress-controlled testing. The relationship between less, this observation regarding the rate of pore-pressure genera-
ru and G explains the rate of pore-pressure generation from stress- tion implies that the evaluation of induced strains rather than
controlled testing, as shown in Figs. 1共a兲 and 10. induced shear stresses may be more appropriate for the simulation
The pore-pressure generation from the in situ tests more of pore-pressure generation.
closely resembles that from the laboratory, strain-controlled The developed in situ dynamic liquefaction test has many po-
CDSS tests than the stress-controlled tests. The results imply that tential applications in both engineering practice and academic
the reconstituted soil specimens tested in situ were responding research, such as improvements of the current empirical correla-
more closely to the induced shear strains than the induced tions used to evaluate liquefaction initiation, more accurate esti-
stresses. However, the shear strain amplitudes were continuously mates of expected deformation and strain potential in liquefiable
changing during the in situ tests, which is not captured by con- soils during seismic events, evaluation of the liquefaction resis-
stant amplitude, strain-controlled testing. Nonetheless, the rate of tance of gravelly soils, and assessment of the effectiveness of
excess pore-pressure generation was similar 共Fig. 10兲. The bound- various liquefaction remediation techniques. The test not only es-
ary conditions in the in situ liquefaction test are analogous to a tablishes a new alternative for the evaluation of liquefaction sus-
liquefiable soil deposit overlain by a nonliquefiable layer, which ceptibility, but it also captures the coupled response between
represents typical liquefiable natural soil deposits. This finding shear strain and excess pore-pressure generation. With large vi-
indicates that evaluation of induced shear strain rather than in- brator equipment now available to the entire earthquake engineer-
duced shear stress may be more appropriate for the simulation of ing community through the George E. Brown, Jr., Network for
pore-pressure generation. Earthquake Engineering Simulation 共NEES, http://nees.
utexas.edu兲, the developed testing technique has the potential to
make a great impact in earthquake engineering research.
Conclusions

Pore pressure generation measurements in a reconstituted granu-


lar soil using a newly developed in situ dynamic liquefaction test
Acknowledgments
were presented and compared with strain-controlled laboratory
Financial support was provided by the National Science Founda-
test results and data published in the literature. Because the newly
tion under CAREER Award No. CMS-9875430 and Grant No.
developed test acquires local measurements of the excess pore
CMS-9973717. This support is gratefully acknowledged. Any
pressure and coupled ground response, several observations re-
opinions, findings and conclusions or recommendations expressed
garding pore-pressure generation can be made.
in this material are those of the writers and do not necessarily
For the in situ liquefaction test, the induced excess pore pres-
reflect the views of the National Science Foundation.
sure consists of both hydrodynamic and residual components. The
oscillating hydrodynamic excess pore pressures are in phase with
the particle motion and are caused by increases in mean stress
during wave passage. The residual excess pore pressures reflect References
the accumulated plastic strains in the soil, and are responsible for
soil liquefaction and soil stiffness reduction. The measured excess Allen, N. F., Richart, F. E., and Woods, R. D. 共1980兲. “Fluid wave propa-
pore pressures and induced shear strains reveal four distinct gation in saturated and nearly saturated sand.” J. Geotech. Engrg.
stages of pore-pressure generation that represent different rates of Div., 106共3兲, 235–254.
Amini, F., and Sama, K. M. 共1999兲. “Behavior of stratified sand-silt-
pore-pressure generation, as well as pore-pressure dissipation.
gravel composites under seismic liquefaction conditions.” Soil Dyn.
During the pore-pressure dissipation process the unsaturated Earthquake Eng., 18共6兲, 445–455.
overburden layer impeded drainage of the test specimens. The in Andrus, R. D., and Stokoe, K. H., II. 共2000兲. “Liquefaction resistance of
situ measurements of excess pore pressure and shear strain were soils from shear-wave velocity.” J. Geotech. Geoenviron. Eng.,
used to develop pore-pressure generation curves, which indicated 126共11兲, 1015–1025.
a threshold shear strain of about 0.5⫻ 10−2%, and rapid genera- Charlie, W. A., Jacobs, P. J., and Doehring, D. O. 共1992兲. “Blast-induced
tion of pore pressure at shear strains between 0.5⫻ 10−2 and liquefaction of an alluvial sand deposit.” Geotech. Test. J., 15共1兲,
2 ⫻ 10−2%. 14–23.
The results from the in situ dynamic liquefaction test were Chang, W.-J. 共2002兲. “Development of an in situ dynamic liquefaction
compared with other studies of pore-pressure generation. Com- test.” Ph.D. dissertation, Univ. of Texas at Austin, Austin, Tex.
Dobry, R., Ladd, R. S., Yokel, F. Y., Chung, R. M., and Powell, D.
parisons with the pore-pressure generation curves developed by
共1982兲. “Prediction of pore water pressure buildup and liquefaction of
Dobry et al. 共1982兲 revealed that the results from this study dis- sands during earthquake by the cyclic strain method.” NBS building
played a smaller threshold strain and more rapid pore-pressure science series 138, National Bureau of Standards, Gaithersburg, Md.
generation at shear strains greater than the threshold strain. These Dobry, R., and Swiger, W. F. 共1979兲. “Threshold strain and cyclic behav-
differences were attributed predominantly to the small effective ior of cohesionless soils.” Proc., 3rd ASCE/EMDE Specialty Conf.,
confining pressures found in the reconstituted test specimen 共␴⬘v Austin, Tex, 521–525.
⬃ 12.7 kPa at the center of the instrumentation array兲. This result Finn, W. D. L., Lee, K. W., and Martin, G. R. 共1977兲. “An effective stress
was further confirmed by strain-controlled cyclic direct simple model for liquefaction.” J. Geotech. Engrg. Div., 103共6兲, 517–533.

930 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / AUGUST 2007


Finn, W. D. L., Pickering, D. J., and Bransby, P. L. 共1971兲. “Sand lique- Ph.D. dissertation, Univ. of Texas at Austin, Austin, Tex.
faction in triaxial and simple shear tests.” J. Soil Mech. and Found. National Research Council 共NRC兲. 共1985兲. Liquefaction of soils during
Div., 97共4兲, 639–659. earthquakes, National Academic Press, Washington, D.C.
Gohl, W. B., Howie, J. A., and Rea, C. E. 共2001兲. “Use of controlled Pestana, J. M., Hunt, C. E., and Goughnour, R. R. 共1997兲. “FEQDrain: A
detonation of explosives for liquefaction testing.” Proc., 4th Int. Conf. finite element computer program for the analysis of the earthquake
on Recent Advances in Geotechnical Earthquake Geotechnical Engi- generation and dissipation of pore water pressure in layered sand de-
neering and Soil Dynamics and Symposium in Honor of Professor W. posits with vertical drains.” Rep. No. UCB/EERC 97-15, Earthquake
D. Liam Finn, San Diego, Paper No. 9.13. Engineering Research Center, UC-Berkeley, Calif.
Hazirbaba, K. 共2005兲. “Pore pressure generation characteristics of sands Rathje, E., Chang, W.-J., and Stokoe, K. H. 共2004兲. “Evaluation of
and silty sands a strain approach.” Ph.D. dissertation, Univ. of Texas ground strain from in situ dynamic response measurements.” Proc.,
at Austin, Austin, Tex.
13th World Conf. on Earthquake Engineering, Vancouver, Canada.
Hazirbaba, K., and Rathje, E. M. 共2004兲. “A comparison between in situ
Rathje, E. M., Chang, W. J., and Stokoe, K. H., II 共2005兲. “Development
and laboratory measurements of pore water pressure generation.”
of an in situ dynamic liquefaction.” Geotech. Test. J., 28共1兲, 65–76.
Proc., 13th World Conf. on Earthquake Engineering, Vancouver,
Rollins, K. M., Anderson, J., McCain, A., and Goughnour, R. 共2003兲.
Canada.
Iwasaki, T., Arakawa, T., and Tokida, K. 共1984兲. “Simplified procedures “Vertical composite drains for mitigating liquefaction hazard.” Proc.,
for assessing soil liquefaction during earthquake.” Int. J. Soil Dyn. 13th Int. Offshore and Polar Engineering Conf., Honolulu, 498–505.
Earthquake Eng. 3共1兲, 49–58. Scott, R. F., and Hushmand, B. 共1995兲. “In situ calibration of dynamic
Ishihara, K., Muroi, T., and Towhata, I. 共1989兲. “In situ pore water pres- pore pressure transducers.” Final Rep. to the U.S. Geology Survey,
sure and ground motions during the 1987 Chiba-Toho-Oki Earth- Grant No. 1434-92-G-2169, California Institute of Technology, Pasa-
quake.” Soils Found., 29共4兲, 75–90. dena, Calif.
Ishihara, K., Shimizu, K., and Yamada, Y. 共1981兲. “Pore water pressure Seed, H. B., and Booker, J. R. 共1977兲. “Stabilization of potentially lique-
measured in sand deposits during an earthquake.” Soils Found., fiable sand deposits.” J. Geotech. Engrg. Div., 103共7兲, 757–768.
21共4兲, 85–100. Seed, H. B., Martin, P. P., and Lysmer, J. 共1975兲. “The generation and
Ishihara, K., Tsuchiya, H., Huang, Y., and Kamada, K. 共2001兲. “Recent dissipation of pore water pressures during soil liquefaction.” Rep. No.
studies on liquefaction resistance of sand-effect of saturation.” Proc., UCB/EERC 75-26, Earthquake Engineering Research Center, UC-
4th Int. Conf. on Recent Advances in Geotechnical Earthquake Geo- Berkeley, Calif.
technical Engineering and Soil Dynamics and Symposium in Honor of Shen, C. K., Wang, Z., and Li, X. S. 共1991兲. “Pore pressure response
Professor W. D. Liam Finn, San Diego, Keynote Lecture. during 1986 Lotung Earthquake.” Proc., 2nd Int. Conf. on Recent
Kokusho, T., and Kojima, T. 共2002兲, “Mechanism for postliquefaction Advances in Geotechnical Earthquake Engineering and Soil Dynam-
water film generation in layered sand.” J. Geotech. Geoenviron. Eng., ics, St. Louis, March 11–15.
128共2兲, 129–137. Yang, Z., Elgamal, A., and Parra, E. 共2003兲. “Computational model for
Marcuson, W. F., III, and Hynes, M. E. 共1990兲. “Stability of slopes and cyclic mobility and associated shear deformation.” J. Geotech.
embankments during earthquakes.” Proc., ASCE/Pennsylvania De- Geoenviron. Eng., 129共12兲, 1119–1127.
partment of Transportation Geotechnical Seminar, Hershey, Pa. Yoshimi, Y., and Kuwabara, F. 共1973兲. “Effects of subsurface liquefaction
Marcuson, W. F., III, Hynes, M. E., and Franklin, A. G. 共1990兲. “Evalu- on the strength of surface soil.” Soils Found., 13共2兲, 67–81.
ation and use of residual strength in seismic safety analysis of em- Youd, T. L., and Holzer, T. L. 共1994兲. “Piezometer performance at Wild-
bankment.” Earthquake Spectra, 16共3兲, 529–572. life liquefaction site, California.” J. Geotech. Engrg., 120共6兲, 975–
Menq, F. Y. 共2003兲. “Dynamic properties of sandy and gravelly soils.” 995.

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / AUGUST 2007 / 931

You might also like