You are on page 1of 9

SEISMIC RESPONSE OF LIQUEFACTION SITES

By J. W. Pease/ Associate Member, ASCE, and T. D. O'Rourke,z Member, ASCE

ABSTRACT: This paper focuses on the seismic response of liquefaction sites by evaluating the transient shear
strains and strong-motion characteristics at Treasure Island during the 1989 Lorna Prieta earthquake (Mw = 6.9).
These strain levels are used to model soil and buried pipeline deformations observed in the Marina District of
San Francisco during the same earthquake. Predicted and observed pipeline damage compare favorably, thereby
indicating that transient strains played a critical role in lifeline performance. A simple model for liquefaction
sites shows that shear modulus reduction caused by increased pore pressure can amplify either acceleration or
Downloaded from ascelibrary.org by Indian Inst Of Technology Library, Kharagpur on 09/16/16. Copyright ASCE. For personal use only; all rights reserved.

transient displacement, depending on soil properties and deposit thickness. Liquefaction of thick, loose sand
deposits can attenuate accelerations while amplifying long-period (>4 s) displacements. The model provides a
quantitative means of characterizing the phenomenon of ground oscillation that has been observed at liquefaction
sites.

INTRODUCTION period of a liquefaction site changes and influences damage to


structures.
As described by Youd (National Research Council 1985),
ground oscillation occurs when a surficial soil layer at a level SITE CONDITIONS
site oscillates over an underlying liquefied deposit, resulting
in multiple relative displacements, cracks, and compression Fig. 1 shows three areas of interest: the Marina District,
features between oscillating surficial blocks and adjacent firm Treasure Island, and Yerba Buena Island. Some of the most
ground. Case studies of the 1906 San Francisco (Pease and severe damage resulting from the Lorna Prieta earthquake was
O'Rourke 1993), 1977 San Juan (Youd and Keefer 1994), and associated with liquefaction in a 100 ha area of the Marina
1989 Lorna Prieta (Pease and O'Rourke 1995) earthquakes District of San Francisco (e.g., O'Rourke et al. 1991, 1992).
provide field evidence of ground-oscillation damage. As de- Strong ground motions, however, were not recorded in the
fined here, liquefaction is the strength and stiffness reduction Marina District during the main shock.
in cohesionless soils that results from pore-pressure accumu- Liquefaction also occurred at Treasure Island, a 160 ha ar-
lation during strong ground shaking. Dynamic characteristics tificial island between San Francisco and Oakland built from
will be changed by reductions in soil stiffness, and the result- 1936 to 1939 (Pease and O'Rourke 1995). Strong ground-sur-
ing site response may contribute to lifeline and structural dam- face motions were recorded at Treasure Island over liquefied
age in areas where only small permanent ground deformations fill deposits and at Yerba Buena Island on bedrock, 3.5 km to
occur. Until now, no model has been proposed to explain and the south.
properly quantify this phenomenon. Treasure Island and the central portion of the Marina Dis-
Recently, strong ground-motion records have been acquired trict were constructed on San Francisco Bay of loose, hydrau-
from vertical arrays of measurement devices located below and lic fill. The hydraulic fills at Treasure Island and the Marina
above a liquefield deposit. Records of this type have been District extend to depths between 10 and 16 m and 5 and 10
obtained and studied for the Wildlife Site during the 1987 m, respectively, and consist of very fine, loose clean sand to
Superstition Hills earthquake (Zeghal and Elgamal 1994), Port slightly silty sand (D so = 0.15-0.25 mm) interbedded with
of Kushiro, during the 1993 Kushiro-Oki earthquake (Iai et al. sandy silt and clay. Both the Marina District and Treasure
1995), and Port Island during the 1995 Kobe earthquake (El- Island are situated on roughly 80-90 m of soil, including soft
gamal et al. 1996). Such measurements provide critical field Holocene bay mud. Given the similar epicentral distance, po-
data on the characteristics and levels of transient shear strains tential for underlying ground amplification, deposit origin, and
experienced by soils subjected to liquefaction, and provide im- soil conditions of both sites, the measured Treasure Island site
portant supplemental information from which to develop a response is considered to be a reasonable analogy for the type
model for transient strains and displacements at liquefaction and magnitude of strong motion experienced in the Marina
sites. District.
The present paper focuses on the seismic response of liq- Fig. 2 shows the soil profile and shear-wave velocities as
uefaction sites by evaluating the transient shear strains and determined from multiple subsurface investigations near the
strong-motion characteristics at Treasure Island, San Francisco strong-motion instrument location at Treasure Island. Fill and
during the 1989 Lorna Prieta earthquake, and shows these to shoal deposits (D so = 0.15-0.25 mm), containing loose sands
be similar to response at other instrumented liquefaction sites. and silty sands, extend from the ground surface to a depth of
Pipeline damage in the Marina District is modeled based on
Treasure Island ground strains and is compared with observed
damage patterns. Reductions in soil stiffness due to liquefac-
tion are used in a generic model to show how the natural

'Engr., Dames and Moore, 221 Main St., San Francisco, CA 94105.
2Prof., School of Civ. and Envir. Engrg., Cornell Univ., Hollister Hall!
Ithaca, NY 14853.
Note. Discussion open until June I, 1997. Separate discussions should
be submitted for the individual papers in this symposium. To extend the
closing date one month, a written request must be filed with the ASCE
Manager of Journals. The manuscript for this paper was submitted for or-----<3km
review and possible publication on July 14, 1995. This paper is part of
the Journal of Geotechnical and Geoenvironmental Engineering, Vol.
123, No.1, January, 1997. ©ASCE, ISSN 1090-0241/97/0001-0037- FIG. 1. Map Showing Marina District, Treasure Island, and
0045/$4.00 + $.50 per page. Paper No. 11145. Verba Buena Island

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / JANUARY 1997/37

J. Geotech. Geoenviron. Eng., 1997, 123(1): 37-45


SbeIt Wav. Velocity. (mi.) Tip Reaia1IltCe, Qcl (MPa)
0 100 :zoo

-. -
300 400 500 0 4 I U ~
--0
'1IIIIdSltoll

20

"'--

.-..
];40
! 60
4 Middle fill

80
Downloaded from ascelibrary.org by Indian Inst Of Technology Library, Kharagpur on 09/16/16. Copyright ASCE. For personal use only; all rights reserved.

SbeIr Wave VeJoolliea NClIHiquofiabI•


. . . M........t(cleAlbaetoJ.• lm) clayfil1
.. ·····~(Qlbbo.\lJ..I992)
- - Meuured (IfJycIw et 01., 1991)
----M........t(Redpath, 1991)
- Input prolllo uaed Iilr .....,... _rl11
.- - - 1latlIIIIlled recIuetioD aIIer IIllIIboeI
oar1IIqIlMe .... SHAICB

FIG. 2. Subsurface Conditions and Liquefaction Susceptibility of Solis at Treasure Island Seismograph Site

approximately 11 m. From a depth of 11-85 m are Holocene motion and response spectra that compare favorably with those
bay mud, dense sands, and stiff Pleistocene clays. In Fig. 2(b), derived from actual records on soft soil sites. As concluded
cone penetration test (CPT) measurements reported by de Alba by Idriss (1991), SHAKE was suitable for modeling this lo-
et al. (1994) are plotted with depth relative to liquefaction cation in the Lorna Prieta earthquake because relatively low
threshold curves consistent with the soils encountered and de- acceleration did not result in high shear-strain levels and
termined by procedures recommended by Seed and de Alba highly nonlinear behavior in the Holocene bay mud and un-
(1986) and Mitchell and Tseng (1991). Normalized tip resis- derlying deposits. Hryciw et al. (1991) and Finn et al. (1993)
tance, qclt less than the threshold curves developed for M, = note, however, that discrepancies exist between SHAKE sim-
7.1 (Mw = 6.9), and a peak acceleration of 0.16 g, predict soil ulations and recorded motions at Treasure Island, in part due
layers that liquefied in the 1989 earthquake. Soil susceptible to the inability of SHAKE to account for development of liq-
to liquefaction is located in the middle fill from a depth of uefaction in the near-surface layers.
3.7-7.0 m, and soil is located in the bottom fill and shoal In the present work, the effects of site amplification from
deposits from a depth of approximately 9.3-11.0 m. Using the bedrock through deep soils were modeled to obtain a strong-
criteria proposed by Seed and de Alba (1986) and Mitchell motion record at 11 m depth, approximately at the depth of
and Tseng (1991), the net liquefiable thickness was determined the base of liquefiable deposits. The program SHAKE was
to be 4.3 and 4.7 m, respectively. The thickness of liquefied used to model the one-dimensional dynamic site response for
soil in 1989, therefore, is estimated as the average given by the full soil profile at Treasure Island by using the Verba
both methods, 4.5 m. Buena Island records as the basis for bedrock motion. Because
analytical motions were evaluated at the 11 m depth, nonlinear
ANALYTICAL APPROACH effects associated with liquefaction should be low, thereby re-
sulting in conditions well-suited for SHAKE analysis. By eval-
Differential horizontal motions across a liquefied soil layer uating simulated motions at 11 m depth, referred to in the
can be estimated with the aid of a vertical array of two or remainder of this paper as the "base record," relative to the
more instruments above and below the layer. For example, actual recorded surface motions, relative displacements across
Zeghal and Elgamal (1994) used measurements at the Wildlife the liquefiable deposits can be estimated.
Site for this purpose from two accelerometers located at the Corrected acceleration records provided by the California
surface and below the liquefiable layer at a 7.5 m depth. At Division of Mines and Geology (Brady and Shakal 1994) were
Treasure Island in 1989, only the surface motions were re- used. The low-frequency cutoff for these records was deter-
corded; the nearest base record representative of the underly- mined by a signal-to-noise algorithm. Because this cutoff was
ing deposits was on bedrock at Verba Buena Island roughly 0.100 and 0.167 Hz for the Treasure and Verba Buena Island
3.5 kID distant. Whereas Treasure and Verba Buena Islands are measurements, respectively, relatively long (==6-10 s) wave-
roughly 3 kID apart, they are over 95 kID from the earthquake forms are preserved in the records.
epicenter and are aligned on the same azimuthal path. There
are no reported seismic discontinuities in bedrock between the GROUND-MOTION RECORDS
two sites. Accordingly, strong-motion bedrock recordings at
Verba Buena Island should provide a valid baseline from Fig. 3 compares Treasure Island accelerations and displace-
which to simulate one-dimensional shear-wave propagation at ments, respectively, for the recorded surface motions and an-
Treasure Island. alytical motions at the surface and at an 11 m depth. The
Many studies have identified the influence of thick soil de- records are compared by aligning the initial strong-motion ar-
posits on site response during the Loma Prieta earthquake by rivals between 10 and 14 s. Alignment was performed by vi-
comparison of measured strong ground motions with analytical sual comparison of acceleration-time records and then further
results of the one-dimensional, equivalent, linear seismic aligned by evaluating maxima and minima in the stored record
model, SHAKE (Schnabel et al. 1972). Idriss (1991), Seed et by spreadsheet comparisons. Both the East-West (E-W) and
al. (1992), and Rollins et al. (1994), who included interpre- North-South (N-S) components were studied, but for brevity
tation of the Treasure Island and Verba Buena Island records, only the E-W motions are shown. Displacements in Fig. 3(c)
observed that SHAKE resulted in simulated records of strong- were derived from double integration of accelerations in Fig.
38/ JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / JANUARY 1997

J. Geotech. Geoenviron. Eng., 1997, 123(1): 37-45


ISO placements for Treasure Island as the result of processing er-
i
......
JOO
SO
rors arising from the step in acceleration at 15 s; this judgment
was based in part on the apparent dissimilarity of this record
and liquefaction response at the Wildlife Site in the 1987 Su-
.~ o~~~ perstition Hills earthquake. Data obtained at other liquefaction
Ii .50 T...... loImd ...moe reocrd B-W sites since the Loma Prieta earthquake suggest that such long-
l -100
-J SO
NUlDlriea1ly aimulJlled ...... noord B-W
-f-,""T""T..,.,..TT-,...,....,..,...."*.,...,..,...,....,..,.....,...........,....,....,...,......,..,...,...TT-.............,..,........,
period motion is real, and stress-strain analyses show that a
liquefaction response at Treasure Island is not inconsistent
with behavior at the Wildlife Site. In a more recent earthquake,
o 10 IS 20 25 30 35 40
long-period (4 s) displacements are shown clearly in the re-
Time (5)
corded surface motions at Port Island overlying fills that liq-
(a)
Downloaded from ascelibrary.org by Indian Inst Of Technology Library, Kharagpur on 09/16/16. Copyright ASCE. For personal use only; all rights reserved.

uefied during the 1995 Kobe earthquake (Elgamal et al. 1996).


ISO
SHEAR-WAVE VELOCITY
'"'100
1 so As previously explained, the liquefiable soils were located
in two layers resulting in a net liquefiable layer thickness of
g 0 ~~H'tW#lAAflI\. 4.5 m. Assuming that this net thickness of soil liquefied during
l -so Truo... lolmd m .... reocrd B-W
the Loma Prieta earthquake, it is possible to determine the
! -100 NIIIIl.ne.uy aimulJlled ..... reoonI B-W
reduction in shear-wave velocity across the liquefied material
by computing time lags between strong-motion peaks at the
-I 50 +r......."TT~ .............-rl"T"T......."TT........,..,..................,...........,..,..,..,...,...,....,,...., ground surface and at the 11 m depth. The 4.5 m net thickness
o 5 10 IS 20 2S 30 3S of liquefied soil is then divided by the sum of the time lag
Time (s) and the initial travel time. The initial shear-wave velocity prior
(b)
to liquefaction may be taken as 160 mis, as determined from
downhole seismic tests (de Alba et al. 1994). For a 4.5 m thick
\0
T...... lolond m .... reocrd B-W layer, the initial travel time is 0.03 s.
Numerically aimuloted ..... reoonI B-W Time lags were quantified by measuring the delays between
! S peaks in a spreadsheet. Delay of peaks after 15 s in Fig. 3(c)

is.
) 0

-5
was obtained primarily for long-period motions of approxi-
mately 6 s duration. To identify peaks of long-period motions
in the base record, perturbations due to higher-frequency
.!l
Q waves were smoothed using a 1 s running average. The peaks
-\0 in the smoothed base record were compared to the actual peaks
in the surface record to determine time delay. Degraded shear-
0 5 10 IS 20 25 30 35
Timc(s) wave velocity is computed as
(c) Va = h2 /(!J.to + !J.t'ag) (1)
where h2 =liquefied thickness, 4.5 m; !J.to = initial travel time
FIG. 3. Comparison of Measured and Numerically Simulated
Surface and Base Strong-Motion Records at Treasure Island: (a) at 160 mls; and the time delay, !J.t1llll , is obtained from com-
Measured and Simulated Surface Accelerations; (b) Measured parison of peaks in surface and base records previously de-
Surface and SimUlated Base Accelerations; (c) Measured Sur- scribed.
face and Simulated Base Displacements Fig. 4 presents a plot of the shear-wave velocity with time.
Given that negligible time delays occur before 14 s, the initial
shear-wave velocity is undegraded until that time. Shortly
3(b). Drift in displacement estimates resulting after integration
thereafter, increasing delays in the records result in decreasing
were eliminated by removing linear trends, and no additional
shear-wave velocities. By 16 s shear-wave velocity drops to
filtering (of long-period components) was performed.
less than 10 mls. Long-period waves from 17 s to the end of
There is a remarkably strong coherence in waveforms of
the record show increasing time lag with apparent shear-wave
recorded surface and analytical surface and base accelerations
velocities of 2-5 mls. As discussed by Pease and O'Rourke
until 14 s. At this time, a notable lack of coherency develops
(1995), alternate explanations do not account for time delays
between acceleration at the surface and the 11 m depth. The
of sufficient magnitude to explain the observed delays of
drop in surface accelerations and decoupling of the two ac-
celeration records is interpreted as the result of liquefaction 0.5-2.3 s.
and softening of the saturated loose fill and shoal deposits. STRESS-STRAIN RELATIONSHIPS
As with the acceleration records, there is strong concor-
dance in Fig. 3(c) of displacement peaks and waveforms to The following expressions represent shear stress 'T, and av-
about 14 s. After 14 s, the waveforms of long-period displace- erage shear-strain "y histories of the liquefied layers:
ments are similar but show an increasing delay in arrival of
peaks in the surface record.
ShoIr Wave Voloohy. mi.
After 16 s, a predominant period of approximately 6 s can
be observed in both records. The surface record shows a very
smooth, sinusoidal displacement profile, which suggests that
short-period waves are attenuated, but long-period waves are
transmitted and amplified through the liquefied deposit. Re-
sponse spectra also were evaluated for the time records (Pease
and O'Rourke 1995) that show two to threefold amplification s 10 15 20 30 35 40
of displacements at periods of about 6 s in the surface record Time(s)

compared with the record at the 11 m depth. FIG. 4. Shear-Wave Velocity History of Liquefied Deposit at
Hanks and Brady (1991) interpreted the long-period dis- Treasurel.'and

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / JANUARY 1997/39

J. Geotech. Geoenviron. Eng., 1997, 123(1): 37-45


Tit) = pza.(t) (2) siderable similarity in the stress-strain relationships and pro-
gressive reductions in the shear modulus.
'Y(t) = [x.(t) - xb(t)]lh 2 (3) Figs. 5(b) through 5(d) show distinct segments of the Treas-
where a" x" and Xb =top accelerations, top displacements. and ure Island record for 8-14, 14-22. and 22-32 s, respectively.
base displacements at 11 m depth determined from recorded Fig. 5(b) shows stress"'strain behavior before liquefaction.
and analytical strong motion. The overburden density is p; z Abruptly after 12 s, the character of the stress-strain curve
=depth of the middle of the liquefied layer; H2 =thickness of changes. as shown in Fig. 5(c). Initially. the stiffness drops as
the liquefied layer; and t = time. As explained previously, the indicated by the shallow slope of the curve. However, for peak:
thickness of the liquefiable layer. H2 • is estimated as 4.5 m. strains at 15. 16. and 18 s. increasing stresses reflect a mod-
Because the mass of the nonliquefied surface layer plays a erate stiffening. expressed in the present paper as a peak: re-
prominent role in generating shear stress across the liquefied sidual stiffness. For decreasing strains and for subsequent cy-
Downloaded from ascelibrary.org by Indian Inst Of Technology Library, Kharagpur on 09/16/16. Copyright ASCE. For personal use only; all rights reserved.

layers, (2) was used to estimate shear stress. The average shear cles of strain lower than the maximum previous strains. there
stress given by T z = 1I2pz[a,(t) + ab(t»). where ab = base ac- is a lower modulus referred to as low-strain-residual stiffness.
celeration; and z = 5.5 m. was also evaluated. Stress-strain Low stiffness characterizes the backbone of curves in Fig.
relationships developed from average shear stress were similar 5(c). The low stiffness results in continued large shear strains
to. although somewhat more erratic than. those developed greater than 0.01 (single amplitude) after accelerations have
from (2). dropped to very small levels, as illustrated in Fig. 5(d).
Fig. 5(a) shows the stress-strain plot for the strong motions Approximate shear moduli can be estimated by measuring
between 6 and 30 s in the Treasure Island record. For com- the tangents of the three representative stress-strain curves in
parison. Fig. 6 shows the stress-strain plot for the Wildlife Site Fig. 5. It should be recognized that these moduli are derived
by Zeghal and Elgamal (1994). Although some deviation from from approximate relative motions averaged across approxi-
the well-defined stress-strain paths for the Wildlife Site are mate thicknesses of liquefiable soil. They are not precise meas-
evident in the Treasure Island plots, there nevertheless is con- urements and should be treated as representative values suit-
able for simple models and approximate sensitivity studies.
(0) (b) The initial shear modulus Gi , estimated from Fig. 5(b), is be-
10 -r---------~ 10
tween 6 and 9 MPa. Peak: residual modulus Grp, at high strain
levels after 12 s. is roughly 400 kPa. The residual modulus at
~ , lower strain levels in Fig. 5(c) is roughly 80 kPa. Although
...

'"B.,
0

., ~ the record in Fig. 5(d) shows considerable scatter or noise,


there is a low but consistent trend that shows a shear modulus
of roughly 20 kPa.
J Similar to the plots for Treasure Island. the Wildlife record
{/) ·10 -10
'- in Fig. 6 shows three characteristic tangents: initial modulus,
·1' +-.........- -r-...--r-....--...-----.---J ·1' peak: modulus associated with large increases in shear strains.
-O.QI
-0.02 -0.01 0.00 0.01 0.02 -0.02 0.00 0.01 0.02
and a low strain modulus at a shallow slope. Tangent moduli
(0) (eI)
for the Treasure Island and wildlife sites are summarized in
10 -r---------~ 10
Table 1. Shear-wave velocity V. is related to shear modulus G
by
(4)
., Comparing stiffness values in Table I, the peak: residual
moduli show a reduction of 15-35 times. and the low-strain
-10 residual modulus shows a reduction of 75-500 times lower
than initial modulus. The table shows good agreement between
·1' +-""""'--r-...--r-....--...-----.- ·1' +-.........--r~...-.,..-~-r-~- shear-wave velocities determined by alternate methods. There
-0.02 ..a.OJ 0.00 0.01 0.02 -0.02 ..0.01 0.00 0.01 0.02
Shear Strain, y(dec) Shear Strain, y(d") is a moderate discrepancy between initial shear-wave veloci-
ties, which may result from difficulties in estimating moduli
FIG. 5. East·West Shear Stress·Straln History of Liquefied De- from steep stress-strain plots. A shear-wave velocity of 160
posit at Treasure Island: (a) Full Record; (b) 8-14 Sj (c) 14-22 s;
(d) 22-32 s mls measured for Treasure Island by de Alba et al. (1994) is
considered to be high for loose. silty sand. which may also
account for some of the discrepancy.
10 Site-exploration data for the Wildlife Site show in-situ den-
sities higher than those of the Treasure Island hydraulic fill.
'2 These differences in density seem to be reflected in the dif-
5
:3- TABLE 1. Summary of Soli-Stiffness Properties Evaluated
l:l
..
~
!il
0
from Strong-Motion Records at Liquefaction Sites
Treasure
.8
en Soli-stiffness properties Island Wildlife Site
en -5 (1 ) (2) (3)
Z
Initial shear-wave velocity (m/s) 160 (55 - 70)' 110 (64-90)
Initial shear modulus (kPa) 6,000-9,000 8,000-16,000
Peak residual shear-wave velocity (m/s) 10 (14) 20 (15-30)
-10 Superstition Hills 450-1,300
Peak residual modulus (kPa) 400
Low-strain residual shear-wave velocity
-1.5 -1 -0.5 0 0.5 1.5 (m/s) 2-5 (3-6) 8 (7 -10)
NS Shear strain (%) Low-strain residual modulus (!cPa) 20-80 100-180
'Measurements in parentheses refer to equivalent shear-wave velocity calculated
FIG. 6. North·South Shear Stress·Straln History of Liquefied from shear modulus as obtained from transient stress-strain plot (see Fig. 6).
Deposit at Wildlife Site (Zeghaland ElgamaI1994)

40 I JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING I JANUARY 1997

J. Geotech. Geoenviron. Eng., 1997, 123(1): 37-45


ferences in approximate shear moduli for liquefied soils at both
sites. Given that Treasure Island was constructed of loose hy-
draulic fill and that liquefaction at the Wildlife Site was ex-
perienced in a loose to moderately dense layer, the data from
both sites may provide a range of representative moduli for
liquefaction sites encountered in practice.

PIPELINE DAMAGE IN MARINA


Pipeline performance and subsurface conditions in the Ma-
rina have been studied and described by O'Rourke et a1.
Downloaded from ascelibrary.org by Indian Inst Of Technology Library, Kharagpur on 09/16/16. Copyright ASCE. For personal use only; all rights reserved.

(1992). The similarities in hydraulic fill and site conditions at


both the Marina and at Treasure Island imply that both sites
were subjected to comparable levels of transient shear strain.
By comparing the spatial variation of computed surface dis-
tortion in the Marina with the distribution of buried pipeline
damage, one can decide if transient shear strains are capable FIG. 8. Contours of Pipeline Repair Intensity In Marina
of causing the observed levels of lifeline damage. By linking (O'Rourke et al. 1992)
transient motion with observed performance, one can establish
a basis for predicting lifeline damage from realistic estimates occurred at the junction of the hydraulic fill, seawall, and 1857
of transient strain and knowledge of three-dimensional sub- shoreline, except for an isolated area on Rico Way, where un-
surface conditions. usual constraints occurred as a result of pipeline construction
Only a brief description of the Municipal Water Supply Sys- along the curved street.
tem (MWSS) pipelines in the Marina is provided here, and
reference should be made elsewhere (O'Rourke et al. 1992) TRANSIENT SHEAR-STRAIN EFFECT ON BURIED
for additional details. Within the area bounded by the 1857 PIPELINES
shoreline and the current shoreline, there are approximately
11.3 km of pipeline belonging to MWSS. The MWSS mains Fig. 9 is a schematic of simplified ground response at a level
are predominantly 100, 150, 200, and 300 mm in diameter, liquefaction site. A loose, potentially liquefiable deposit both
composed of pit-cast iron, and installed principally between overlies and is adjacent to a denser, nonliquefiable soil. The
1924 and 1925. All pipes were buried at nominal depths to thickness of the liquefiable soil decreases from the center of
top of pipe between 0.9 and 1.2 m. the deposit toward its boundary with the adjacent nonliquefi-
The locations of MWSS pipelines and repairs relative to the able soil. The figure shows a pipeline that is embedded in an
street system, 1906 waterfront, and 1857 shoreline are shown upper nonliquefiable layer and is located across the boundary
in Fig. 7. Repairs were made at points of sheared or disen- between underlying liquefiable and nonliquefiable soils. Foun-
gaged service connections with mains, flexural round cracks dations, pavements, and pipelines typically are constructed
in mains, and longitudinally split sections of mains. A total of above the ground-water level and within a nonliquefiable sur-
123 repairs were made to the MWSS mains and services in face layer.
the Marina District, more than three times the number in the Fig. 9(b) shows settlement caused by postliquefaction con-
entire MWSS elsewhere. A total of 69 repairs were made to solidation. If the volumetric strain is more or less uniform
mains, including those at or near gate valves; more than 80% throughout the liquefied deposit, settlement response of the
of these repairs were attributed to round cracks. surface layer is directly proportional to the thickness of the
To represent the distribution of MWSS damage, the Marina liquefied material. Settlement is zero at the boundary of sub-
District was divided into a grid of approximately 40 cells, and
the number of repairs within each cell was counted (O'Rourke Legend:
et al. 1991, 1992). The repairs were then normalized with re- - Direction of
spect to the reference length of 300 m to provide a consistent oscillatory motion
basis for evaluation. Contours of equal repair rates were drawn Liquefiable soil
and superimposed on the street system and previous shore-
lines, as shown in Fig. 8. The heaviest repair concentration o Non-liquefiable soil (a)

~~~64:t~~r"
Earthen
Mole of

~~1~;~~~~~~~~~~~]
Gas LllIht Co.
San Francisco
18991
/
(h) (e)

Lateral Ground

(2!!L--_~_7--!:!.":.~ (d)

FIG. 9. Modes of Burled Pipeline Response to Transient Sur-


face Deformation above Liquefied Deposit: (a) Static Condl-
FIG. 7. Pipeline Repairs In Municipal Water-Supply System In tlon8; (b) Settlement; (c) Transverse Movement; (d) Axial Defor-
Marlna{O'Rourkeetal.1992) mation

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / JANUARY 1997/41

J. Geotech. Geoenviron. Eng., 1997, 123(1): 37-45


merged fill. At the center of the deposit, settlement is equal to sponse to permanent ground deformations (Taki and O'Rourke
the maximum submerged thickness times the volumetric strain. 1984; O'Rourke and Harris 1983). These investigations have
Changes in slope and curvature at the ground surface, which shown that the threshold for acceptable tensile strain is be-
impose shear and bending deformation on buried conduits, are tween 500 and 600 ILE for the cast-iron pipe. Strains imposed
directly proportional to the change in thickness and second by ground deformation above this level are not advisable be-
derivative of thickness of the liquefied deposit, respectively. cause of increased risk with respect to tensile failure of the
Figs. 9(c) and 9(d) show three modes of deformation that pipe. The limiting tensile strain, Ea), was also evaluated in re-
can occur due to ground oscillation. If shear strain is parallel lation to the pullout capacity of cast-iron joints (Pease and
to the boundary of the liquefied deposit, then the surface layer O'Rourke 1995) and found to be in the range of 250 ILE.
will be subject to shear distortion in the horizontal plane. If Fig. 10 also shows the computed levels of lateral ground
liquefaction shear strain is perpendicular to the deposit bound- strain lOa superimposed on the map of the Marina. Areas with
ary, compression and extension features will result. Lateral dis- lateral ground strain between 200 and 500 JL€, and especially
Downloaded from ascelibrary.org by Indian Inst Of Technology Library, Kharagpur on 09/16/16. Copyright ASCE. For personal use only; all rights reserved.

placement in Fig. 9(d) occurs toward the liquefiable boundary areas with lateral ground strain greater than 500 ILE, have con-
such that the surface layer will experience passive failure, ex- siderable potential for damage. Maps of actual pipeline breaks
pressed as compression ridges or buckled pavements. Move- and contours of repairs are provided in Figs. 7 and 8, respec-
ment away from the opposite boundary results in extensional tively, which can be compared directly with the areas of high
failure of the surface deposit. Both extension cracks and ver- lateral ground strain in Fig. 10. To expedite comparison, areas
tical offsets may result from this movement. A pipeline em- bounded by repair contours of six repairs per 300 m are iden-
bedded in the surface layer will be subject to similar com- tified in Fig. 10 by means of stippled shading and show rea-
pressive and tensile strains. sonably good agreement with locations of damage potential
Although the actual motions of liquefied deposits are more based on ground strain.
complex than those depicted, the figure nevertheless illustrates Although pipeline damage correlates well with locations of
how pipeline damage can result from one- and two-dimen- maximum lateral ground strain, it should be recognized that
sional strains that are relatively simple to quantify. If it is the actual causes of pipeline damage are complex. Pipe failure
assumed that the pipeline deforms axially as the ground de- can result from large axial tensile strains, joint pullout, con-
forms (as would be appropriate for relatively thin-wall pipe centrated compression at buckled and heaved surfaces, exces-
anchored in the ground by multiple-service connections and sive bending at locations of pipe restraint and abrupt vertical,
tees), the maximum tensile strain due to axial movement t a is extensional, and lateral offsets through surface soils. Pease and
equal to the lateral ground strain as follows: O'Rourke (1995) report lateral and vertical offsets of 10-100
mm at several locations in the Marina, frequently near the
Ea = AdH/L (5)
margins of liquefied fill. Horizontal offsets and areas of con-
where AdH = differential lateral displacement in the upper, centrated relative settlement impose local distortion on buried
nonliquefiable layer over a horizontal distance, L, along the pipelines. These other sources of ground distortion appear to
pipeline. Maximum liquefaction strains in the Marina were be related indirectly to lateral ground strain. Hence, the choice
assumed to be approximately equal to 2%, consistent with E- of lateral ground strain serves not only as a direct, but indirect
W strains back-calculated for Treasure Island in Fig. 5. Shear index of ground deformation, and may be taken as an appro-
strains of 1.25% in the N-S direction were estimated from the priate measure of ground response for evaluating the damage
ratio of 1.6 between the E-W and N-S components of maxi- potential of buried lifelines.
mum ground motion. The strains were multiplied by sub- Lateral ground strains evaluated for the Marina are consis-
merged thickness to obtain estimates of maximum surface dis- tent with the failure of brittle pipelines, such as cast iron, as-
placements. Taking the first derivatives of these functions, bestos cement, and vitrified clay piping. These strain levels
zones of maximum lateral ground strain were developed in can also damage steel-distribution piping at areas of local
Fig. 10. Given that the pipelines have thin walls and are par- weakness and deterioration, but will not damage steel and pol-
tially anchored by closely spaced service connections, it is yethylene piping in good repair.
likely that the ground strains were approximately equal to axial
strains in the pipelines. SITE RESPONSE
Evaluation of cast-iron microstructure, casting practices, and
laboratory tests on cast-iron pipe specimens have been per- A closed-form solution for the natural period of two un-
formed as a part of a detailed study of cast-iron pipeline re- damped, elastic soil layers on a rigid base is derived in Ap-
pendix I. The undamped natural period, Tn, can be evaluated
by rewriting (10) as

27rh2)
tan ( - (27rhl)
- tan - - =VdYs2
-- (6)
Tn V'2 Tn Vs1 Vs1'Ysl
where V s2 , h 2, and 'Ys2 = shear-wave velocity, thickness, and
unit weight of an underlying liquefiable layer; and Vsh hi' and
'YSI are similarly defined for an overlying nonliquefiable layer.
It is well known that damping has a strong effect on amplifi-
cation but a much smaller influence on natural period. Hence,
the natural period for soil layers, even those with significant
damping, can be estimated for practical purposes by means of
(6).
Fig. 11 shows the relationship between liquefied layer thick-
ness, shear-wave velocity of liquefied soil, and the undamped
natural period as given by (6) for the approximate site ge-
FIG. 10. Map Showing Zones of Predicted Lateral Ground ometry and soil characteristics at Treasure Island. For liquefied
Strain, 4dH IL, In Relation to Areas of Largest Pipeline Repair soil with Vs2 = 10 mis, h 2 = 4.5 m, and 'YSI = 'Ys2, the natural
Rates period of the seismograph site is approximately 6 s, as shown
42/ JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / JANUARY 1997

J. Geotech. Geoenviron. Eng., 1997, 123(1): 37-45


15 age over the hydraulic fill appears to have been mitigated by
! N
0
. I
dissipation of seismic energy through the liquefied deposits.
.c In contrast, over the denser and stiffer land-tipped fill, severe

! 10
damage occurred to approximately 50% of structures of sim-
ilar construction. These locations correspond to regions of pre-
dicted amplification of 0.8-1.5 s period waves through the
~ liquefied deposits.
! S. K. Harris (personal communication, 1994) prepared a da-
~ tabase of all two- to four-story residential structures in the
'0 Marina in which the damage percentage of each building was
III
! assessed according to the rating system proposed by the Ap-
Downloaded from ascelibrary.org by Indian Inst Of Technology Library, Kharagpur on 09/16/16. Copyright ASCE. For personal use only; all rights reserved.

~ plied Technology Council (1985). Figs. 13 and 14 have been


t=
00 20 40 60 80 100
Shear Wave Velocity of Liquefiable Layer, VS2 (m/s)

FIG. 11. Variation of Natural Period for Liquefaction Site;


Thickness and Shear-Wave Velocity of Surface Layer are 4 m
and 160 m/s
! 10
§
in Fig. 11. This is in good agreement with the approximate 6
s period in the Treasure Island displacement records in Fig. !
OS
3(c).
To check that the simplified site model provides a reason-
able approximation of dynamic soil response, the damping ra-
tio ~ was back-calculated and compared with measured values.
j
~

The amplification factors determined from both response spec- o ~~


tra analysis (Pease and O'Rourke 1995) and a comparison of o W 40 ~ 80 100
recorded surface displacements relative to those analyzed at Shear Wave Velocity of Liquefiable Layer, Vsz (m/s)
the base of the liquefied layer are between 2 and 3. Given this FIG. 12. Site-Response Characteristics of Hydraulic and
amplification and a natural period of 6 s, damping ratios rep- land-Tipped Fill Sites In Marina
resentative of liquefied hydraulic fill at Treasure Island and the
Marina were back-calculated with (11) to obtain ~ = 0.2-0.3.
This range is consistent with damping ratios measured in the
laboratory for sands at comparable levels of shear strain (Seed
et al. 1986).

SEISMIC SITE RESPONSE IN MARINA DISTRICT


Fig. 12 shows the zones of liquefiable layer thickness, shear-
wave velocity of liquefied soil, and the natural period for the
hydraulic and land-tipped fills in the Marina. It is assumed
that the range of shear-wave velocity for hydraulic fill at
Treasure Island at 2-10 rn/s (see Table 1) is representative of
liquefied conditions of loose hydraulic fill in the Marina. In
contrast, the land-tipped fills at the Marina are comparable to
the medium dense and silty sands at the Wildlife Site. For the
wildlife site, Zeghal and Elgamal (1994) provide stress-strain
relationships for liquefied soil from which low-strain and peak FIG. 13. Contour Plot for Marina of Number of Buildings/Hec-
residual shear-wave velocities of 7 -10 rn/s and 15 -30 mis, tare with More Than 10% Damage
respectively, can be inferred. Given the range of liquefied layer
thicknesses in the Marina (2.5 m :s h 2 :s 7 m), zones identi-
fying probable ranges in natural period of the liquefied sites
have been superimposed on the analytical relationships in Fig.
12.
In areas of the Marina District overlying hydraulic fill, dam-
age is predicted to result primarily from amplification of long-
period displacements with periods of approximately 6-8 s.
The reduction in deposit stiffness results in attenuation of
short-period accelerations. Long-period displacements, accord-
ingly, tended to dominate site response in areas of liquefied
hydraulic fill. In contrast, for areas of liquefied land-tipped fill,
damage is predicted from either strong accelerations or large
transient displacements.
Harris and Egan (1992) perfonned a detailed study of se-
verity of damage to corner buildings in the Marina District
after the Loma Prieta earthquake. They calculated the natural
period of two- to four-story residential structures as typically FIG. 14. Contour Plot for Marina of Number of BulldlngalHec-
0.8-1.25 s. Harris and Egan (1992) noted that building dam- tare with More Than 60% Damage

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING I JANUARY 1997 I 43

J. Geotech. Geoenviron. Eng., 1997, 123(1): 37-45


developed from this database. They show intensity of building
damage in terms of number of buildings per hectare exceeding
10% (Fig. 13) and 60% (Fig. 14) damage. The average build-
ing density in the Marina is 20-30 buildings/hectare. These
figures show clearly the concentration of damage in the land-
tipped fill with the preponderance of damaged structures lo-
cated adjacent to the northwest boundary of the hydraulic fill.
There is a remarkable difference in the damage intensities for
the land-tipped and hydraulic fills, with significantly less
building damage in the hydraulic fill. A much different pattern
VS21 YS2
of damage applies for buried pipelines, as illustrated in Fig. 8,
Downloaded from ascelibrary.org by Indian Inst Of Technology Library, Kharagpur on 09/16/16. Copyright ASCE. For personal use only; all rights reserved.

which shows that the greatest intensities of pipeline repair co-


incide with the hydraulic fill.

CONCLUSIONS Firm base


Transient lateral shear strains in liquefied soils are a prime FIG. 15. Idealized Slngl..Oegree-of-Freedom Model of Llqu..
cause of horizontal displacements and seismic damage to bur- faction Site
ied lifeline systems. Analyses of instrumented sites at the Port
of Kushiro (Iai et al. 1995) Wildlife Site (Zeghal and Elgamal and downward with respect to the interface between layer 2
1994) and Kobe Port Island (Elgamal et al. 1996) show max- and the firm base. Adopting notation similar to Roesset's,
imum transient shear strains in liquefied soils of approximately whereby PI =WIV'h P2 =wIV12 , ql =V,I'Y,.lV'2'Y12' 'Y, =soil unit
1.4-1.6%. Similar analyses performed for Treasure Island in weight; w = 27fIT; and T = wave period. By applying the
this work disclose about 1.8% maximum shear strain. recursive formulas for wave amplitudes at the layer interfaces,
Using lateral shear-strain levels of 1-2%, which are con- one obtains
sistent with earthquake measurements, it is shown that tran-
sient horizontal ground deformation in the Marina was of suf- A3 + Ai = 1/2A t [(1 + ql)(e 1a + e- 1a ) + (I - ql)(el~ + e-I~)]
ficient magnitude to damage cast-iron water mains, and the (8)
areal magnitude to damage cast-iron water mains and areal
distribution of the most severe transient deformation coincides where a = Ptht + P2h2 and ~ = P2h2 - P1h l for undamped
with that of the most intense damage to the pipeline system. elastic layers. Using trigonometric identities and by combining
This finding is highly significant because it shows for the first (7) and (8), one obtains
time how transient horizontal deformation at liquefaction sites 1
affects the performance of buried utilities. It also shows how AFI(w) = cOS(P2h2)cos(Plhl) ..
- qt sm(P2h2)sm(P 1h l )
(9)
the three-dimensional characteristics of liquefiable soils con-
trol the pattern of transient lateral displacements near the At resonance AFI(w) approaches infinity and
ground surface.
The simplified site-response model in this work shows am-
(10)
plified accelerations in the liquefied land-tipped fill of the Ma-
rina at natural periods similar to those of the heavily damaged
corner buildings. In contrast, accelerations are reduced through from which the natural period, Tn, that satisfies (10) can be
deep, loose hydraulic fill at the same time that transient dis- solved.
placements are amplified in the long-period range. These long- If damping is considered in the liquefied layer, (9) can be
period (4-6 s) displacement waves are what has been identi- re-expressed as
fied in previous postearthquake site investigations as ground
oscillation. They are responsible for damage to the buried
pipelines, which are vulnerable to lateral ground strains ex- AFI(w) = cos(p!h2)cos(Pthl) _ qr sin(p!h2)sin(P h 1 t
) (II)
ceeding 200-500 !-LE. Such lifelines include cast iron, asbestos where p! = P2fl; qr
= qlfl; fl = (1 - i~)(l + 4~2)-t/4(1 +
cement, and vitrified clay piping, as well as steel piping that ~2)-t; and ~ = damping ratio. The denominator of (11) is a
has been weakened by corrosion or low-strength welds. complex number so that the modulus of AFt(w) must be found
to evaluate the amplification factor.
ACKNOWLEDGMENTS
This paper presents the results of research supported by the National APPENDIX II. REFERENCES
Center for Earthquake Engineering Research, Buffalo, N.Y., under Project
No. NCEER94-2304(N) and by the U.S. Geological Survey, Reston, Va., Applied Technology Council (ATe). (1985). "Earthquake damage eval-
under Grant No. 1434-93-G-2332. Laurie McCall and Kristin Stewart uation data for California." Rep. ATC-13, Redwood City, Calif.
prepared the manuscript, and Ali Avcisoy prepared the figures. Brady, G. A., and Shakal, A. F. (1994). "Strong-motion recordings." The
Lorrw Prieta, California Eanhquake of October 17. 1989-Strong
Ground Motion and Ground Failure. Profl. Paper 1551-A, R. E. Bor-
APPENDIX I. DERIVATIONS FOR TWO-SOIL-LAYER cherdt, ed., U.S. Geological Survey, U.S. Govt. Printing Ofc., Wash-
SYSTEMS ington, D.C., AI09-A119.
de Alba, P., Benoit, J., Pass, D. G., Carter, J. J., Youd, T. L., and Shakal,
The natural period, Tnt for two elastic soil layers overlying A. F. (1994). "Deep instrument array at Treasure Island naval station."
a firm base (see Fig. 15) can be derived from the formulations The Lorrw Prieta. California Eanhquake of October 17. 1989-Strong
given by Roesset (1977) for the amplification factor, AFI(w) Ground Motion and Ground Failure, Profl. Paper 1551-A, R. E. Bor-
cherdt, ed., U.S. Geological Survey, U.S. Govt. Printing Ofc., Wash-
AFI(w) = 2A/[A + Ail
3 (7) ington, D.C., AI55-AI68.
Elgamal, A.-W., Zeghal, M., and Parra, E. (1996). "Liquefaction of re-
where Al = displacement amplitude at the surface of layer 1 claimed island in Kobe, Japan." J. Geotech. Engrg., ASCE, 122(1),
of a vertically propagating shear wave; and A 3 and Ai = dis- 39-49.
placement amplitudes of the shear waves propagating upward Finn, W. D. L., Ventura, C. E., and Wu, G. (1993). "Analysis of ground

44 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / JANUARY 1997

J. Geotech. Geoenviron. Eng., 1997, 123(1): 37-45


motions at Treasure Island site during the 1989 Loma Prieta earth- U.S. Geological Survey Profl. Paper 1551-F, U.S. Government Printing
quake." Soil Dyn. and Earthquake Engrg., 12(7),383-390. Office, Washington, D.C., 155-179.
Geomatrix Consultants. (1990). "Compilation of data and information Pease, J. W., and O'Rourke, T. D. (1993). "Liquefaction hazards in the
for evaluation of interior area pedormance, naval station Treasure Is- Mission District and South of Market areas, San Francisco, Califor-
land, San Francisco, California." Proj. Rep. 1539.09, Vol. 4, San Fran- nia." Final Rep.• U.S.G.S. Grant 14-08-0001-G2128, School of Civ.
cisco, Calif. and Envir. Engrg., Cornell Univ., Ithaca, N.Y.
Gibbs, J. F., Fumal, T. E., Boore, D. M., and Joyner, W. B. (1992). Pease, J. W., and O'Rourke, T. D. (1995). "Liquefaction hazards in the
"Seismic velocities and geologic logs from borehole measurements at San Francisco Bay region: site investigation, modeling, and hazard as-
seven strong-motion stations that recorded the Loma Prieta earth- sessment at areas most seriously affected by the 1989 Loma Prieta
quake." Open File Rep. 92-287, U.S. Geological Survey, Menlo Park, earthquake." Final Rep.• U.S.G.S. Grant No. 1434-93-G-2332, School
Calif. of Civ. and Envir. Engrg., Cornell Univ., Ithaca, N.Y.
Hanks, T. C., and Brady, A. G. (1991). "The Loma Prieta earthquake Redpath Geophysics. (1991). "Seismic velocity logging in the San Fran-
ground motion and damage in Oakland, Treasure Island, and San Fran- cisco Bay area," Prepared for Rep., EPRI Agreement No. RP3014-06,
Downloaded from ascelibrary.org by Indian Inst Of Technology Library, Kharagpur on 09/16/16. Copyright ASCE. For personal use only; all rights reserved.

cisco, Bull. of the Seismological Soc. of Am., 81(5), 2019-2047. Electric Power Research Institute, Palo Alto, Calif.
Harris, S. K., and Egan, J. A. (1992). "Effects of ground conditions on Robertson, P. K., and Campanella, R. G. (1983). "Interpretation of cone
the damage to four-story corner apartment buildings." The Lomo penetration tests. Part I: sand," Can. Geotech. J., Ottawa, Canada,
Prieta. California Earthquake of October 17. 1989-Marina District. 20(4),718-733.
Profl. Paper 1551-F, T. D. O'Rourke, ed., U.S. Geological Survey, U.S. Roesset, J. M. (1977). "Soil amplification of earthquakes," Numer. Meth-
Govt. Printing Ofc., Washington, D.C., FI81-FI94. ods in Geotech. Engrg., McGraw-Hili Book Co., Inc., New York, N.Y.,
Hryciw, R. D., Rollins, K. M., Homolka, M., Shewbridge, S. E., and 639-682.
McHood, M. (1991). "Soil amplification at Treasure Island during the Rollins, K. M., Hyrciw, R. D., Homolka, M., and Shewbridge, S. E.
Loma Prieta earthquake." Proc.• 2nd Int. Con/. on Recent Adv. in Geo- (1994). "Ground response on Treasure Island." The Lomo Prieta, Cal-
tech. Earthquake Engrg. and Soil Dyn.• Paper LP20, Univ. of Missouri- ifornia Earthquake of October 17, 1989-Strong Ground Motion.
Rolla, Rolla, Mo., 1679-1695. Projl. Paper IS51-A, R. E. Borcherdt, ed., U.S. Geological Survey, U.S.
Govt. Printing Ofc., Washington, D.C., AI09-AI19.
Iai, S., Morita, T., Kameoka, T., Matsunaga, Y., and Abiko, K. (1995).
Schnabel, P. B., Lysmer, J., and Seed, H. B. (1972). "SHAKE; a com-
"Response of a dense sand deposit during the 1993 Kushiro-Oki earth-
puter program for earthquake response analysis of horizontally layered
quake." Soils and Found., 35(1), 115-131.
sites," Rep. No. EERC 72-12, Univ. of California, Berkeley, Calif.
Idriss, I. M. (1991). "Response of soft soil sites during earthquakes,"
Seed, H. B., and de Alba, P. (1986). "Use of SPT and CPT tests for the
Proc.. H. B. Seed Memorial Symp., Vol. 2, Bitech Publishers, Ltd., liquefaction resistance of soils." Geotech. Special Publ. No.6, ASCE,
Vancouver, B.C., Canada, 273-289. New York, N.Y., 281-303.
Mitchell, J. K., and Tseng, D.-J. (1991). "Assessment of liquefaction Seed, H. B., Wong, R. T., Idriss, I. M., and Tokimatsu, K. (1986). "Mod-
potential by cone penetration resistance." Proc., H. B. Seed Memorial uli and damping factors for dynamic analyses of cohesionless soils."
Symp., Vol. 2, Bitech Publishers, Ltd., Vancouver, B.C., Canada, J. Geotech. Engrg., ASCE, 112(11), 1016-1032.
335-350. • Seed, R. B., Dickenson, S. E., and Mok, C. M. (1992). "Recent lessons
National Research Council. (1985). Liquefaction of soils during earth- regarding seismic response analysis of soft and deep clay sites." Proc.,
quakes. Com. on Earthquake Engrg., National Academy Press, Wash- 4th U.S.-Japan Workshop on Earthquake Resistant Des. ofLifeline Fac.
ington, D.C. and Countermeasures for Soil Liquefaction. NCEER-92-0019, Nat. Ctr.
O'Rourke, T. D., and Harris, C. W. (1983). "Effects of adjacent exca- for Earthquake Engrg. Res., Buffalo, N.Y., 131-146.
vation on gas pipelines." Operating Adjacent Sect.• Proc., Am. Gas Taki, H., and O'Rourk.e, T. D. (1984). "Factors affecting the pedormance
Assn., Arlington, Va., 254-263. of cast iron pipe," Geotech. Engrg. Rep. 84·1, School of Civ. and
O'Rourke, T. D., Gowdy, T. E., Stewart, H. E., and Pease, J. W. (1991). Envir. Engrg., Cornell Univ., Ithaca, N.Y.
"Lifeline and geotechnical aspects of the 1989 Loma Prieta earth- Youd, T. L., and Keefer, D. K. (1994). "Liquefaction during the 1977
quake." Proc.• 2nd Int. Conf. on Recent Adv. in Geotech. Engrg. and San Juan Province, Argentina earthquake (Ms = 7.4)," Engrg. Geol.,
Soil Dyn., Univ. of Missouri-Rolla, RoUa, Mo., 1601-1612. 37(3), 211-233.
O'Rourke. T. D., Pease, J. W., and Stewart, H. E. (1992). "Lifeline per- Zeghal, M., and Elgamal, A.-W. (1994). "Analysis of site liquefaction
formance and ground deformation during the earthquake." The Lomo using earthquake records," J. Geotech. Engrg., ASCE, 120(6), 996-
Prieta, California, Earthquake of October 17. 1989-Marina District. 1017.

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAl ENGINEERING / JANUARY 1997/46

J. Geotech. Geoenviron. Eng., 1997, 123(1): 37-45

You might also like