You are on page 1of 18

Two Case Histories Demonstrating the Effect of Past

Earthquakes on Liquefaction Resistance of Silty Sand


W. El-Sekelly, M.ASCE 1; R. Dobry, M.ASCE 2; T. Abdoun, M.ASCE 3; and J. H. Steidl 4

Abstract: The paper compares two liquefaction case histories in California: (1) the response of the Wildlife site in the Imperial Valley to the
2010 El-Mayor Cucapah earthquake (M w ¼ 7.2, amax ¼ 0.15 g); and (2) the response of the Treasure Island Fire Station (F.S.) site in the San
Downloaded from ascelibrary.org by University Of Florida on 10/31/17. Copyright ASCE. For personal use only; all rights reserved.

Francisco Bay area to the 1989 Loma Prieta earthquake (M w ¼ 6.9, amax ¼ 0.16 g). Both histories involve silty sand critical layers with
nonplastic fines contents, FC ¼ 24–27%, similar normalized shear wave velocities, V s1 ¼ 145–155 m=s, low cone penetration test (CPT)
cone penetration resistances, and groundwater tables at essentially the same depth. The corresponding data points plot almost on top of each
other on the shear wave velocity field liquefaction charts, which predict liquefaction at both sites. While Treasure Island F.S. did liquefy
during the shaking, Wildlife did not and was far from liquefaction as indicated by piezometers at the site. This paper constitutes an attempt
to understand the reason for these very different pore pressure responses through a detailed analysis of similarities and differences between
the two histories. It is concluded that preshaking by previous earthquakes is the most probable explanation of the higher liquefaction re-
sistance exhibited by the Wildlife site and other sites in the Imperial Valley. While the Wildlife critical layer was subjected to about 60–70
earthquakes capable of generating significant excess pore pressures between its estimated 1907 deposition and the 2010 earthquake,
the Treasure Island F.S. layer was subjected to only about two earthquakes capable of doing so between deposition in the 1930’s and
the 1989 earthquake. This difference is due to the very high seismic activity in the last 100-plus years in the Imperial Valley compared
with a seismically quiet San Francisco Bay Area after the 1906 earthquake. The significance of the prior seismic history is corroborated
by recent results from centrifuge and large-scale experiments. These results as well as the methodology developed in the paper may be
helpful when analyzing the observed high liquefaction resistance of sandy sites located in other seismic regions. DOI: 10.1061/
(ASCE)GT.1943-5606.0001654. © 2017 American Society of Civil Engineers.

Introduction reduction coefficient, which is unity at the ground surface and


decreases depth. The soil resistance to liquefaction in the field
Soil liquefaction occurs repeatedly during earthquakes, is very is defined by its cyclic resistance ratio (CRR), with liquefaction
damaging to soil and soil-structure systems, and is also a complex predicted if CSR > CRR.
phenomenon. As a result of this complexity, liquefaction evalua- Deterministic and probabilistic liquefaction charts have been
tions in most practical applications are conducted using liquefac- developed relating CRR to several field tests: standard penetration
tion charts calibrated by field case histories and based on the test (SPT), cone penetration test (CPT), and Shear Wave Velocity
Simplified Procedure proposed by Seed and Idriss (1971). A (V s ). All charts have been empirically calibrated by field case
number of updates have been implemented based on this procedure histories of liquefaction and no liquefaction. Fig. 1 shows the
(Youd et al. 2001; Idriss and Boulanger 2008). The earthquake V s -based charts proposed by Andrus and Stokoe (2000) and Kayen
shaking demand is quantified by the cyclic stress ratio (CSR) et al. (2013) in deterministic applications, for an earthquake having
   moment magnitude, M w ¼ 7.5. (The Andrus-Stokoe curve in Fig. 1
a σv0
CSR ¼ 0.65 max 0 rd ð1Þ was originally developed for clean sands, but Kayen et al. (2013)
g σv0 and Dobry et al. (2015a) later concluded that the two curves in
Fig. 1 can be used for both clean and silty sands.) The normalized
where amax = peak horizontal ground surface acceleration; g =
0 shear wave velocity in Fig. 1 is defined as
acceleration of gravity; σv0 = effective vertical overburden pres-
sure; σv0 = total vertical overburden pressure; and rd = shear stress  0.25
Pa
V s1 ¼ V s 0 ð2Þ
1
Lecturer, Mansoura Univ., Mansoura, Dakahlia Governorate 35516, σv0
Egypt (corresponding author). E-mail: waleed.elsekelly@gmail.com
2 0 is in kPa, and P ¼ 101.33 kPa is the
where V s1 , V s are in m=s, σv0
Institute Professor, Dept. of Civil and Environmental Engineering, a
Rensselaer Polytechnic Institute, 110 8th St., JEC 4049, Troy, atmospheric pressure, approximated as Pa ¼ 100 kPa by Andrus
NY 12180. and Stokoe (2000). All field case histories in this paper were ex-
3
Iovino Chair Professor, Dept. of Civil and Environmental Engineering, tracted from the Andrus and Stokoe database, so they are used here
Rensselaer Polytechnic Institute, 110 8th St., JEC 4049, Troy, their version of Eq. (2), V s1 ¼ V s ð100=σv0 0 Þ0.25 .
NY 12180. Several authors have pointed out the limitations of these SPT,
4
Research Seismologist, Earth Research Institute, UC Santa Barbara,
CPT, and V s curves, including: (1) the deterministic charts provide
6710 Ellison Hall, Santa Barbara, CA 93106-1100.
Note. This manuscript was submitted on September 14, 2015; approved a lower bound for the occurrence of liquefaction being conservative
on September 26, 2016; published online on February 7, 2017. Discussion by definition; and (2) neither deterministic nor probabilistic charts
period open until July 7, 2017; separate discussions must be submitted for account for geologic setting and loading history of the sand
individual papers. This paper is part of the Journal of Geotechnical and layer (Youd et al. 2001, 2003; Pyke 2003; Dobry and Abdoun
Geoenvironmental Engineering, © ASCE, ISSN 1090-0241. 2015a). Also, each of the three techniques has its own problems.

© ASCE 04017009-1 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2017, 143(6): 04017009


The data points compiled by Dobry et al. (2015a) and included
in Fig. 1, correspond to all field case histories of liquefaction and
no liquefaction in the original Andrus and Stokoe (2000) database
(Andrus et al. 2003), which are clean and silty sands having non-
plastic fines content, FC ≤34%. That is, Fig. 1 does not include the
case histories in the Andrus and Stokoe database that are gravels,
gravelly sands, clayey sands, silts, and silty sands of FC > 34%.
In their paper, Dobry et al. (2015a) were able to reduce the con-
servatism of the original Andrus and Stokoe curve of Fig. 1, which
includes a number of false positives (points above the curve that did
not liquefy). This was accomplished by identifying two separate
groups of case histories in Fig. 1: recent uncompacted fills and nat-
urally deposited sands. It was shown that the location of the original
Andrus and Stokoe curve was controlled by the uncompacted fills,
Downloaded from ascelibrary.org by University Of Florida on 10/31/17. Copyright ASCE. For personal use only; all rights reserved.

with essentially all false positives corresponding to geologically


young, natural silty sands in the Imperial Valley. Finally, Dobry
et al. (2015a) proposed the two different charts for uncompacted
fills and natural sands in the Imperial Valley of Fig. 2. The two
new curves are parabolas having equations of the form, CRR ¼
AðV s1 =100Þ2 , where A ¼ 0.033 for the recent uncompacted clean
or silty sand fills [Fig. 2(a)], and A ¼ 0.065 for the natural silty
sands in the Imperial Valley [Fig. 2(b)]. That is, the Imperial Valley
soils have a resistance to liquefaction for a given V s1 which is about
Fig. 1. Case histories of clean and silty sands of FC ≤ 34% (modified twice that of the uncompacted fills. The parabolic shapes of the
from Dobry et al. 2015a) curves in Fig. 2 correspond to lines of constant cyclic shear strain,
γ cl . For M w ¼ 7.5, γ cl ≈ 0.03% in the uncompacted fills and
γ cl ≈ 0.1–0.2% in the Imperial Valley natural sands. Dobry et al.
(2015a) speculated that this increase in liquefaction resistance in
For example, the SPT and CPT measure a smaller penetration in the Imperial Valley may be due to the intense seismic activity tak-
silty sand than in clean sand for a given liquefaction resistance, ing place in the Valley in the last century. Previous small-sample
requiring a fines correction before entering the chart (Idriss and laboratory studies had shown that low-level cyclic shear prestrain-
Boulanger 2008). In the case of the V s , the shear wave velocity ing of either dry or saturated sand can increase significantly the
fails to capture the improved liquefaction resistance due to preshak- liquefaction resistance, and more recent centrifuge and large-scale
ing, prompting the need for regional liquefaction charts as shaking tests have confirmed this beneficial effect of preshaking.
discussed in the next paragraph for the Imperial Valley of California Additional details on these relevant studies are provided later in
(Dobry et al. 2015a). this paper and in Appendix S1.

Fig. 2. Constant cyclic shear strain liquefaction charts proposed by Dobry et al. (2015a) for (a) recent uncompacted clean and silty sand fills;
(b) natural silty sands in Imperial Valley of California (modified from Dobry et al. 2015a)

© ASCE 04017009-2 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2017, 143(6): 04017009


Scope of Paper and they generated similar peak surface horizontal accelerations as
recorded by accelerometers at the two sites (0.15–0.16 g). The
The paper focuses on two case histories singled out in Figs. 1 and 2, groundwater level depth was essentially the same (1.4–1.5 m).
taken from these two different groups of sites. They are (1) the Finally, the soils in the critical layers are also very similar (silty
response of the Treasure Island Fire Station (F.S.) site in the sands with nonplastic fines and FC ¼ 24–27% and measured
San Francisco Bay Area to the 1989 Loma Prieta earthquake V s1 ¼ 145–155 m=s). For the tip (CPT) cone penetration resistan-
[uncompacted fill, Fig. 2(a)]; and (2) the response of the Wildlife ces, qc was also similarly low in both critical layers (2–10 MPa and
site in the Imperial Valley of California to the 2010 El-Mayor 1–7 MPa for Wildlife and Treasure Island, respectively). Table 1
Cucapah earthquake [natural soil, Fig. 2(b)]. These two case his- also includes values of representative qc1N for the liquefiable criti-
tories were chosen because several of their main parameters are cal layers at both sites, as reported by Boulanger and Idriss (2014),
quite similar, as shown in Figs. 3–6 and summarized in Table 1. where qc1N = value of qc1N for the layer after normalizing by both
The two sites are in California, the earthquakes had similar mag- effective overburden pressure and atmospheric pressure. The two
nitudes (6.9–7.2), they occurred at similar distances (93–98 km), numbers are qc1N ¼ 71 for Wildlife and 53 for Treasure Island,
suggesting that the CPT point penetration resistance at Wildlife
Downloaded from ascelibrary.org by University Of Florida on 10/31/17. Copyright ASCE. For personal use only; all rights reserved.

is about 34% higher than at Treasure Island. However, given the


variability of the two CPT profiles within each layer shown by
[Fig. 5(a)], clearly the determination of a single value of qc1N
for the layer involves considerable uncertainty. Further examination
of the possibility that the CPT point resistance may be a better
indicator than V s of the effect of preshaking on liquefaction resis-
tance is included later in the “Discussion” section.
As a result of their similarities, the two case histories plot
essentially on top of each other on the liquefaction graph of Fig. 1,
where the two sites are predicted to liquefy by the Andrus and
Stokoe and Kayen et al. charts. However, the responses of the
two sites to these earthquake events were very different. While
the Treasure Island F.S. site did liquefy in 1989, the Wildlife site
did not liquefy in 2010, and in fact the 2010 event generated low
excess pore water pressures in the critical layer that were far from
liquefaction, as recorded by piezometers at the site [Fig. 3(b)]. This
is despite two differences that may seem to point to a higher like-
lihood of liquefaction at Wildlife: (1) the probability of liquefaction
using the method developed by Kayen et al. (2013) is higher for
Wildlife in 2010 than for Treasure Island in 1989 (52% versus
32%, see Table 1); and (2) the critical liquefiable layer at Wildlife
is thicker [4.3 m versus 2.5 m, see Table 1 and Fig. 2(a)].
This paper is an attempt toward understanding the very different
behavior of both sites in response to broadly similar events. The
two sites as well as the 1989 and 2010 earthquakes are documented
and compared with discussion of both similarities and differences.
This includes looking at the deposition, geologic age, seismic pre-
shaking, and liquefaction history of each of the sites before 1989
and 2010. Similar to the conclusion by Dobry et al. (2015a), it is
found that the much more intense seismic preshaking history in the
Imperial Valley is the most probable explanation of the two differ-
ent liquefaction responses. This hypothesis is validated in detail
using several tools, including: (1) identification and counting of
relevant earthquakes that affected the two sites between times of
deposition and earthquake; and (2) comparison with the response
of saturated sand deposits to multiple shakings measured in centri-
fuge and large-scale tests. The insensitivity of V s to preshaking,
already noticed in the field by Dobry et al. (2015a), is confirmed
by these centrifuge and large-scale tests, which also provide
additional information on the increase in liquefaction resistance
due to preshaking for clean and silty sands and different seismic
Fig. 3. (a) Recorded ground accelerations at Wildlife site during
environments.
the 2010 El Mayor-Cucapah earthquake in NS and EW directions;
(b) histories of excess pore pressure ratios recorded at different depths
in critical layer at Wildlife site during 2010 El Mayor-Cucapah earth-
quake; (c) recorded ground acceleration histories at Treasure Island F.S. Overview of Liquefaction Responses in 1989 and
site during 1989 Loma Prieta earthquake in NS and EW directions; 2010
(d) recorded ground accelerations at nearby Yerba Buena Island rock
This section summarizes the acceleration and liquefaction re-
outcrop during 1989 Loma Prieta earthquake in NS and EW directions;
sponses of the Wildlife site and the Treasure Island F.S. site to the
(data from Wildlife Liquefaction Array 2016; CESMD 2016)
2010 El-Mayor Cucapah and 1989 Loma Prieta earthquakes,

© ASCE 04017009-3 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2017, 143(6): 04017009


Downloaded from ascelibrary.org by University Of Florida on 10/31/17. Copyright ASCE. For personal use only; all rights reserved.

Fig. 4. Locations of: (a) Wildlife site and (b) Treasure Island site, along with active faults and epicenters of major earthquakes in their vicinity
(modified from Power et al. 1998; Rockwell and Klinger 2013)

respectively. As shown by Figs. 3(a and c), surface accelerometers indicating an amplification by the soil of more than two times.
recorded the horizontal acceleration time histories in both cases. This amplification in the peak acceleration is consistent with
In addition, piezometers installed since 2005 at the Wildlife site similar amplifications observed in soft sites throughout the San
(Youd et al. 2007; Steidl and Seale 2010) recorded the buildup Francisco Bay Area (Idriss 1990).
and dissipation of excess pore pressures at several depths during Fig. 3(c) indicates that there was a sudden change in the
the 2010 event [Fig. 3(b)]. characteristics of the recorded accelerogram about 5 s after the be-
ginning of shaking, with much smaller accelerations and longer
periods afterward. This is typical of earthquake records at sites that
Response of the Wildlife Site to the 2010 Earthquake
have liquefied, starting with Niigata, Japan, in the 1964 earthquake
On April 4, 2010, the Wildlife site was subjected to the M w ¼ 7.2 (NRC 1985). Therefore, the Treasure Island F.S. record of Fig. 3(c)
El Mayor Cucapah earthquake. The recorded surface ground hori- has been interpreted as providing strong evidence of onset of
zontal acceleration time histories are shown in [Fig. 3(a)]. The peak liquefaction of the site (Idriss 1990; de Alba et al. 1994; Power
accelerations were 0.15 g (NS) and 0.12 g (EW). There were no et al. 1998). As stated by de Alba et al. (1994): “ : : : the surface-
surface manifestations of liquefaction at the site. This nonliquefac- acceleration record obtained at this site, in contrast to those at other
tion behavior was confirmed by the piezometers installed in the soft-soil sites, shows a sudden drop about 15 s into the record and
critical silty sand layer, which recorded a maximum pore pressure practically no response after 16 s (Idriss 1990). This behavior is
ratio, ðru Þmax ¼ 19% [Fig. 3(b)]. almost certainly due to liquefaction of the underlying sand.” This
accelerogram in Fig. 3(c) is the only evidence of liquefaction at the
site as there were no sand boils or other surface manifestations,
Response of the Treasure Island F.S. Site to the 1989
most probably due to the presence of a very thick nonliquefiable
Earthquake
layer (4.5 m) on top of a liquefiable hydraulic fill sand layer of
On October 17, 1989, the Treasure Island F.S site was subjected to significant smaller thickness (2.5 m), see Ishihara (1985) and
the M w ¼ 6.9 Loma Prieta earthquake. The recorded horizontal Liu and Mitchell (2006). On the basis of the strong evidence pro-
surface acceleration histories are shown in Fig. 3(c). The peak ac- vided by Fig. 3(c), a consensus has developed that the site did in-
celerations were 0.16 g (NS) and 0.10 g (EW). The maximum of deed liquefy in 1989, and this case has been incorporated since as a
0.16 g may be compared to the maximum acceleration of 0.067 g liquefaction data point in the calibrated SPT, CPT, and V s charts
recorded at the nearby Yerba Buena Island rock outcrop [Fig. 3(d)], (Andrus et al. 2003; Cetin et al. 2004; Boulanger and Idriss 2014).

© ASCE 04017009-4 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2017, 143(6): 04017009


Downloaded from ascelibrary.org by University Of Florida on 10/31/17. Copyright ASCE. For personal use only; all rights reserved.

Fig. 5. (a) Stratigraphy and CPT results of Wildlife and Treasure Island F.S. sites (data from Bennett et al. 1984; de Alba et al. 1994; Andrus et al.
2003); (b) average grain size distributions of soil in critical layers at both sites (data from Vucetic 1986; Power et al. 1998)

Wildlife Site 2010 Case History The April 4, 2010, El Mayor-Cucapah earthquake is also vari-
ously known as 2010 Baja California earthquake, 2010 Easter
This section discusses in detail the Wildlife case history in terms of: earthquake, and 2010 Sierra El Mayor earthquake. It had an
characteristics of the 2010 El-Mayor Cucapah earthquake and its M w ¼ 7.2, and the epicenter was located at 32.286°N, 115.295°W,
location relative to the site; site conditions, stratigraphy and dep- approximately 47 km SSE of Mexicali, Baja California, Mexico,
osition history; threshold acceleration of the critical liquefiable and 180 km SE of San Diego, California (Liao and Meneses
sand layer at the site; measured surface ground accelerations 2013). The hypocentral depth was 10 km. The earthquake occurred
and excess pore pressures in the critical layer during the 2005–
along the principal plate boundary between the North American
2014 instrumented period; estimated seismic preshaking history
and Pacific plates (Gonzalez-Ortega et al. 2014). The earthquake
of the site since deposition of the critical liquefiable layer; and,
was felt in a large area encompassing northern Baja California
finally, liquefaction history of the site.
in Mexico and Southern California, Arizona, and Nevada in the
United States, with the largest damage occurring in Mexico but
2010 Earthquake Location and Characteristics with damage also occurring up to the Salton Sea to the north,
The Wildlife site is located in the Imperial Valley of Southern see Fig. 4(a) (Liao and Meneses 2013). This was the largest earth-
California, 160 km east of San Diego and close to the Mexican quake in Baja California after the 1892 Laguna Salada earthquake,
border, as seen in Fig. 4(a), which also shows the major faults and with the 2010 earthquake being well recorded in Southern
earthquake epicenters in the area. California (Liao and Meneses 2013).

© ASCE 04017009-5 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2017, 143(6): 04017009


0
the top thirty meters of the site, V S30 (Dobry et al. 2000). The
Wildlife site (Bierschwale and Stokoe, 1984)
Wildlife site (Cox, 2006)
V S30 of the Wildlife site was measured to be 177 m=s (Steidl et al.
1 Treasure Island F.S. site (Fuhriman, 1993) 2014), with the site classified as Site Class E, as per ASCE 7
(ASCE 2010).

2
Threshold Acceleration of Critical Layer at Site
As shown by Dobry et al. (1981) and Dobry and Abdoun (2015a),
3 the accumulation of excess pore pressure in a sand layer in the field
starts only when the soil ground surface amax exceeds the threshold
Depth [m]

acceleration, of the layer, at . The value of at is controlled by the


4 threshold cyclic shear strain, γ tv , of the soil in the layer. The value
of the threshold shear strain in normally consolidated sands is no-
tably constant, γ tv ≈ 0.01–0.02%, with this range consistently
Downloaded from ascelibrary.org by University Of Florida on 10/31/17. Copyright ASCE. For personal use only; all rights reserved.

5
measured by a number of researchers in the lab and in the field
over a 40-year period (Dobry et al. 1982; Dobry and Abdoun
2015a). Specifically, γ tv is largely independent of the number of
6
cycles, sand type including nonplastic fines content, relative den-
sity, and sand fabric (deposition method). Also, γ tv is about con-
7
stant and independent of effective confining pressure for the range
of pressures of interest in most liquefaction evaluations (between
0 30 60 90 120 150 180 about 0.2 and 2 atm). Cox (2006) measured the threshold strain of
Shear wave velocity, V [m/s] the liquefiable critical layer at the Wildlife site using a truck-
s
mounted shaker and sensors embedded in the soil. The results
Fig. 6. Crosshole shear wave velocity measurement of critical layers are plotted in Fig. 7, and they confirm that γ tv ¼ 0.01–0.02%
at Wildlife and Treasure Island F.S. sites (data from Bierschwale and for the critical layer at the site. Similarly, Steidl et al. (2014) mea-
Stokoe 1984; Fuhriman 1993; Cox 2006) sured the onset of excess pore pressure generation in the liquefiable
layer at the Wildlife site through analysis of earthquake data from
the permanent accelerometers and pressure transducers at the site,
Site Conditions and Deposition History and found this to be at γ c ¼ γ tv ≈ 10−2 % (Steidl et al. 2014).
Since the threshold acceleration, at , is controlled by the value of
The Wildlife site is composed of a floodplain sediment that lies on
the threshold shear strain, γ tv , which is independent of the number
top of denser sedimentary deposits. Holzer and Youd (2007) sug-
of cycles, at is also independent of shaking duration and hence of
gest that this floodplain sediment may have been deposited by
the magnitude of the earthquake. This is unlike the value of amax
floodwater from the Colorado River that flowed down the Alamo
needed to trigger liquefaction, which depends on earthquake mag-
River in 1905–1907. Specifically, the sediment was deposited by
nitude (Ishihara 1981; Dobry et al. 1981; Dobry and Abdoun
currents from the flooding river and the bedload that was left be-
2015a; Abdoun et al. 2013). Dobry et al. (1981) developed from
hind once the currents waned (T. L. Holzer, personal communica-
the following approximate expression to calculate at for a given
tion, 2016). The floodplain sediment consists of 2.5–3 m of a
critical sand layer
nonliquefiable silty clay to clay silt layer, followed by the 3.5 to
 
4 m loose silty sand critical liquefiable layer. The 6–7 m of sedi- at γ tv ðG=Gmax Þtv 2
ment lies on top of a thick layer of silty clay to clay [Fig. 5(a)]. ¼ Vs ð3Þ
g gzrd
The groundwater level is shallow, fluctuating at a depth of about
1–1.5 m. Fig. 5(a) also shows the CPT tip resistance (qc ) and fric- where ðG=Gmax Þtv = modulus reduction factor corresponding
tion ratio (Rf ) reported by Bennett et al. (1984), with qc ≈ to a cyclic shear strain, γ c , equal to the threshold strain, γ c ¼ γ tv ;
2–10 MPa in the critical liquefiable silty sand layer. The character- z = depth of the midpoint of the critical layer. Modulus reduction
istics of the soil at the site have been amply investigated and doc- curves, ðG=Gmax Þc versus γ c have been proposed for sands by Seed
umented (Bennett et al. 1984; Youd and Wieczoreck 1984; Vucetic and Idriss (1970) and Darendeli (2001), among others.
1986; Dobry et al. 1992). Based on radiocarbon information, the Use of Eq. (3) in conjunction with the range, γ tv ¼ 0.01–0.02%,
floodplain sediment liquefiable layer is less than 230 years old. yields a range of threshold accelerations, at ¼ 0.029–0.05 g for the
The grain size distribution of the silty sand in the critical liquefiable critical liquefiable silty sand layer at the Wildlife site. This range
layer is shown in Fig. 5(b). has been included in Table 1. Steidl et al. (2014) examined at
The shear wave velocity of the liquefiable layer of the Wildlife through analysis of earthquake data from the permanent accelerom-
site was measured using the crosshole method by Bierschwale and eters and pressure transducers at the site and found essentially the
Stokoe (1984) and more recently in 2005 by Cox (2006), as shown same threshold acceleration range of 0.02–0.05 g.
in Fig. 6. The figure indicates that the shear wave velocity profile,
reflecting the small-strain shear stiffness of the deposit, remained
Measurement of Surface Accelerations and Pore
practically unchanged between 1984 and 2005. This small change
Pressures in 2005–2014
in shear wave velocity over a period covering two decades is gen-
erally consistent with other field and laboratory evidence, as exam- The Wildlife site was instrumented with accelerometers and
ined later in the “Discussion” section. pore pressure transducers twice in the last 30 years or so. The first
ASCE 7–10 Standard Ch. 20 classifies most sites for seismic time was in 1982 by the USGS and second times was in 2004 and
design into five site classes, A through E, where A is the stiffest 2005 by the Network for Earthquake Engineering Simulation
site (hard rock) and E is the least stiff (soft soil). The classification (NEES) and the University of California at Santa Barbara (UCSB)
is performed using mainly the average shear wave velocity of (Youd et al. 2007; Steidl and Seale 2010). In 2004 a newsite

© ASCE 04017009-6 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2017, 143(6): 04017009


Table 1. Comparison between Liquefaction Responses of Wildlife and Treasure Island Sites (Data from Dobry et al. 1981; Bennett et al. 1984; Idriss 1990; de
Alba et al. 1994; Power et al. 1998; Andrus and Stokoe 2000; Holzer and Youd 2007; Steidl and Seale 2010; Kayen et al. 2013; Boulanger and Idriss 2014;
Steidl et al. 2014; Dobry et al. 2015a)
Site and earthquake Wildlife site (Imperial Valley, Treasure island F.S. Site (San Francisco
features Feature Southern California) Bay Area, Northern California)
Site features Deposit type Natural site Artificial fill
Deposition process Alluvial/fluvial Hydraulic fill
Year of deposition 1905–07 (1907 assumed) 1936–37 (1937 assumed)
Thickness of critical 4.3 2.5
liquefiable layer (m)
Thickness of nonliquefiable 2.5 4.5
surface layer (m)
Groundwater level depth (m) 1.5 1.4
0 (kPa)
σv0 53.9 60.9
Downloaded from ascelibrary.org by University Of Florida on 10/31/17. Copyright ASCE. For personal use only; all rights reserved.

Nonplastic fines content (%) 27 24


V s1 (m=s) 146–148 145–155
qc (MPa) 2–10 1–7
qc1N 71 53
V s30 (m=s) and Site Class 177, Site Class E 155, Site Class E
K DR 1.09/1.01 1.05
Threshold acceleration (g) 0.029–0.05 0.028–0.05
Total estimated number of 61 4
earthquakes in 1907–2010
Earthquake Earthquake name and year 2010 El Mayor-Cucupah 1989 Loma Prieta
features Mw 7.2 6.9a
Depth (km) 10 19
Epicentral distance from Site (km) 93 98
Peak ground surface 0.15 (NS), 0.12 (EW), 0.16 (NS), 0.1 (EW),
acceleration, amax (g) and 0.135 (Av.) and 0.13(Av.)
(NS, EW, and average,
respectively)
ðCSRÞ7.5 0.13 0.11
Probability of liquefaction by 52% 32%
Kayen et al. (2013) method
Years since deposition 103 52
Years since last liquefaction event 23 53 (never liquefied before)
Site liquefaction — Did not liquefy [ðru Þmax ¼ 0.19 at end Liquefied during shaking (from
response of shaking from piezometric recording] recorded ground surface accelerogram)
M w ¼ 6.9 is the official moment magnitude reported by the U.S. Geological Survey (USGS) for the 1989 Loma Prieta earthquake; a slightly different value,
a

M w ¼ 7.0, has been used in a number of liquefaction databases and correlations published by several authors over the years, including Dobry et al. (2015a); the
USGS value of Mw ¼ 6.9 is used in this paper.

was instrumented about 80 m downriver from the 1982 USGS site.


In 2005 the 1982 USGS site was reinstrumented and is referred to
as the Old Array site. During the last 10-year instrumented period
(2005–2014), the Wildlife site was subjected to many earthquakes,
most of them local events of magnitude less than 5.5. Using data
from the newly instrumented Old Array site, 13 of these earth-
quakes generated a significant maximum pore pressure ratio,
ðru Þmax ≥ 0.02 in the critical layer, with these 13 earthquakes also
having a recorded peak ground surface horizontal acceleration at
the site 0.09 g≤ amax ≤ 0.3 g (Table 2). Fig. 8(a) plots these thir-
teen pairs of recorded ðru Þmax versus amax . The figure also includes
two data points for the two earthquakes in the 1980’s that are
known to have liquefied the site (Table S1 in Appendix S1), which
have been assigned the value ðru Þmax ¼ 1.0 in Fig. 8(a). There
seems to be different curves of ðru Þmax versus amax depending
on the magnitude of the earthquake, related to the fact that greater
magnitudes are associated with longer shaking durations. The most
important result from the plot is that a minimum value of amax ≈
0.09–0.1 g is needed at the site to generate an excess pore pressure
Fig. 7. Field determination of threshold shear strain, γ tv , in critical silty
ratio, ðru Þmax ≥ 0.02 in the critical layer, irrespective of earthquake
sand layer at Wildlife site, using surface vibration induced by truck-
magnitude.
mounted shaker and measurements with embedded sensors (modified
Under the assumption that the 2005–2014 experience reflected
from Cox 2006)
in Fig. 8(a) is representative of the 100-plus years since deposition

© ASCE 04017009-7 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2017, 143(6): 04017009


Table 2. Peak Accelerations (amax ) and Maximum Pore Pressure Ratios ðru Þmax Recorded in 2005–2014 by the Reinstrumented NEES at UCSB Array (Old
Array), Wildlife Site, during Events Having Moment Magnitude, Mw ≥ 4.0 and amax ≥ 0.09–0.1 g (Wildlife Liquefaction Array 2016)
Date and time (UTC = coordinated universal time) Magnitude (Mw ) Depth (km) Epicentral distance (km) amaxðgÞ ðru Þmax
08-27-2012 04:41:37 (UTC) 4.9 4.1 8.58 0.3 0.45
08-26-2012 23:33:25 (UTC) 4.61 7.4 7.74 0.3 0.26
08-26-2012 21:15:29 (UTC) 4.25 4.7 6.75 0.11 0.03
08-26-2012 20:57:58 (UTC) 5.5 9 8.36 0.15 0.10
08-26-2012 19:40:13 (UTC) 4.33 9.7 13.78 0.12 0.04
08-26-2012 19:33:01 (UTC) 4.89 4.3 9.67 0.13 0.06
2012-08-26-2012 19:31:23 (UTC) 5.3 12.3 8.86 0.19 0.15
08-26-2012 19:20:05 (UTC) 4.59 4.7 9.09 0.1 0.04
04-04-2010 22:40:42 (UTC) 7.2 10 93 0.14 0.19
(El Mayor–Cucupah earthquake)
09-02-2005 01:27:20 (UTC) 5.11 9.8 12.12 0.13 0.05
Downloaded from ascelibrary.org by University Of Florida on 10/31/17. Copyright ASCE. For personal use only; all rights reserved.

09-01-2005 13:50:20 (UTC) 4.48 0 13.1 0.09 0.02


08-31-2005 22:50:24 (UTC) 4.45 2 11.25 0.1 0.03
08-31-2005 22:47:46 (UTC) 4.59 4 12.38 0.1 0.02
Note: Two arrays of sensors were installed by UCSB at the Wildlife site. The new site in 2004, and the Old Array site in 2005, at the same location as the one
installed by the USGS in the 1980s.

of the critical layer at Wildlife, it seems reasonable to take amax ¼ calculated in the previous section. Inspection of the piezometric re-
0.09–0.10 g as the lower bound of earthquakes generating signifi- sponse at the site in 2005–2014 reveals that, indeed, for a number of
cant pore pressures that affect the critical layer’s resistance to lique- earthquakes having 0.05 g ≤ amax < 0.09 g, small excess pore pres-
faction to future earthquakes. In this line of reasoning, the authors sures were measured, with ðru Þmax < 0.02 in all cases (Steidl et al.
have defined significant pore pressure to be ðru Þmax ≥ 0.02. 2014). Therefore, while amax ¼ at ¼ 0.029–0.05 g is the accelera-
This lower bound, amax ≈ 0.10 g, is two to three times greater tion value needed to initiate pore pressure buildup, the larger value,
than the calculated threshold acceleration, amax ¼ 0.029–0.05 g, amax ≈ 0.10 g, is required to generate at least ðru Þmax ≈ 0.02, as-
sumed by the authors to be the minimum level needed to have an
effect on the liquefaction resistance of the critical layer.
Therefore, it will be assumed in the next section that any time
an earthquake took place with amax ≥ 0.10 g at the site, the resis-
tance to liquefaction of the critical layer to future earthquake events
was modified. In most of these earthquakes that did induce pore
pressures without liquefying the site [most of the data points in
Fig. 8(a)], this resistance modification consisted of strengthening
the soil (preshaking), but in those events that liquefied the layer (the
two data points in Fig. 8(a) with ðru Þmax ¼ 1.0), the modification
may have consisted of decreasing the liquefaction resistance of the
soil (Heidari and Andrus 2012). The reason for this according to
Yasuda and Tohno (1988), is that liquefaction of a site by a strong
earthquake generates a strong upward flow of excess water from the
lower part of the deposit that disturbs and weakens the upper part,
allowing this upper part to be liquefied again by a subsequent
earthquake.

Seismic Preshaking History of Site


The seismic preshaking history of the site is defined here by the
total number of earthquakes that affected the critical layer at the
site since its deposition and generated a maximum ground surface
acceleration about equal or greater than the lower bound of 0.10 g
defined in the previous section. In this context, the word preshaking
is defined here rather loosely, as it also includes those few earth-
quakes where the pore pressure buildup was large enough to liquefy
the layer. This can be done because, both in the field for sites in
the Imperial Valley including Wildlife [Figs. 1 and 2(b)], as well
Fig. 8. (a) Plot of ðru Þmax versus amax for events of magnitude ≥ 4.0 re-
as in centrifuge and large-scale shaking tests in the laboratory
corded at Wildlife site in 2005–2014 (Wildlife liquefaction array 2016),
(El-Sekelly 2014), the combination of the more numerous strength-
augmented by liquefaction case histories at site ðru Þmax ¼ 1.0 during
ening events that did not cause liquefaction (true preshaking with
1981 Westmorland and 1987 Superstition Hill earthquakes (Andrus
increase in liquefaction resistance), in conjunction with the few
et al. 2003); the number besides each data point is moment magnitude
earthquakes that did cause liquefaction and hence decreased the
of earthquake, M w ; (b) comparison between number of earthquake
liquefaction resistance, the final result has been a significant net
events of magnitude ≥ 4.0 with predicted amax ≥ 0.1 g in both Treasure
increase in the current liquefaction resistance of the sand after
Island F.S. and Wildlife sites in 1907–2010 (data from Field et al. 2013)
several dozen earthquakes. This conclusion is at the core of the

© ASCE 04017009-8 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2017, 143(6): 04017009


Table 3. Estimated Number of Earthquakes Causing amax ≥ 0.1 g at the Wildlife and the Treasure Island F.S. Sites during the Period 1907–2010 (Data from
Field et al. 2013)
Number of earthquakes in 1907–2010
Magnitude Maximum distance Wildlife site, Treasure Island Fire Station site,
range M w for amax ≥ 0.1 g (km) Imperial Valley San Francisco Bay Area
4–4.5 15 14a 2a
4.5–5 20 25 0
5–6 20 13 1 (1957 Daly City earthquake)
6–7 60 8 0b
7–7.5 120 1 1b (1989 Loma Prieta earthquake)
Total — 61 4
a
The numbers represent only 36% of the events with magnitudes in the range 4.0–4.5 within a distance of 15 km (14 out of 38 for the Wildlife site and 2 out of 4
for the Treasure Island F.S. site); based on the recorded acceleration during the 2005–2014 instrumented period, only four out of 11 (36%) of events with
magnitudes in the range 4.0–4.5 within a distance of 15 km generated amax ≥ 0.1 g at the Wildlife site.
Downloaded from ascelibrary.org by University Of Florida on 10/31/17. Copyright ASCE. For personal use only; all rights reserved.

b
While the 1989 Loma Prieta earthquake had an M w ¼ 6.9, it was decided to count it in the Mw ¼ 7.0–7.5 range; this earthquake had a surface wave
magnitude of 7.1, and while it occurred at a distance of 98 km from the Treasure Island F.S. site, it had a recorded amax ¼ 0.16 g, significantly higher
than the cutoff, amax ¼ 0.10 g. It must be added that the authors went back to the seismic catalog to find out what other earthquakes occurred after
1907 with M w > 6.0 and a distance less 120 km; they found two more events; an M w ¼ 6.4 at 84 km from the site and an M w ¼ 6.2 at 83 km from
the site (Morgan Hills earthquake listed in Table 4 with recorded amax ¼ 0.03 g).

evaluation of the Wildlife case history presented here and is elabo- affected the Wildlife site in 1907–2010, that is, since deposition
rated further in the rest of the paper, including a summary in the of the layer. The numbers of earthquakes broken down by magni-
“Discussion” section. The “Discussion” section also examines again tude are also plotted in Fig. 8(b). As shown by Table 3 and Fig. 8(b),
the insensitivity of the shear wave velocity to the preshaking process, most of them were low magnitude events at the local faults nearby,
already mentioned, as well as alternative field measurements that including especially the Brawley fault zone [Fig. 4(a)], which
may correlate better with the increased liquefaction resistance. periodically generates swarms (sequences of earthquakes in a
The seismic preshaking history of the Wildlife site before the short time period). Swarms were reported in this fault zone in
2010 earthquake was estimated by extracting the corresponding 1975, 1976, 1977, 2005, and 2012 (Johnson and Hadley 1976;
numbers of earthquakes for different relevant combinations of mag- Johnson and Hill 1982; Hauksson et al. 2013); and about half of
nitudes and distances that are estimated to have generated amax ≥ the records in 2005–2014 listed in Table 2 were associated with the
0.10 g at the site, from the available combined historical and in- 2012 swarm.
strumental UCERF3 Catalog during the period 1907–2004 (Felzer It is important to note that the Imperial Valley has a deep basin
and Cao 2008; Hutton et al. 2010; Field et al. 2013). As mentioned geologic structure and is characterized for having many earthquake
before, amax after 2005 was measured rather than predicted, as the records of long duration (one minute or more), due to the later
site was instrumented that year with accelerometers; and these re- arrival of long period surface waves. Surface waves can be expected
corded events were added to the count. The Catalog includes only to reflect off the basin edge boundaries, with a large part of the
information about each earthquake’s magnitude and distance, with energy trapped in the basin (Olsen et al. 1995). This is clear for
no indication of the peak ground acceleration level. However, the 2010 El Mayor-Cucupah earthquake acceleration time histories
guided by the data from the more recent history of earthquake re- in Fig. 3(a), with Fig. 3(b) showing corresponding long period fluc-
cordings in 2005–2014 for which actual values of M w , distance, tuations of the pore water pressures after approximately 26–27 s,
and amax are available at the site, the following events from the caused by the surface wave arrivals after the main (body wave)
Catalog were counted and included in Table 3, as they were as- shaking had ended. Furthermore, the piezometric records in
sumed to have generated amax ≥ 0.10 g at the site: Fig. 3(b) also reveal that by t ¼ 26–27 s, the excess pore pressures
• Earthquake events with M w ¼ 4.0–4.5 within a distance of 15
km from the site. (As shown in Table 3, only 36% of the events
from the Catalog were considered in the count, because the re- Table 4. Peak Accelerations (amax ) of All Earthquakes Recorded at
corded accelerations during the 2005–2014 instrumented period Treasure Island in 1974–2015 by the Old Fire Station Accelerometer
showed that only four out of 11 (36%) of the events with mag- (Station 58117, Active in 1974–2004) and by the New Geotechnical
nitudes in the range 4.0–4.5 within a distance of 15 km gener- Array (Station 58642, Active in 1992–Present)
ated amax ≥ 0.1 g at Wildlife in 2005–2014); Date and time PDT–Pacific Epicentral
• All earthquake events with M w ¼ 4.5–6.0 within a distance daylight time; PST–Pacific Magnitude Depth distance
of 20 km from the site; standard time (M w ) (km) (km) amaxðgÞ
• All earthquake events with M w ¼ 6–7 within a distance of
04-24-1984 14:15:19 (PDT) 6.2 9 83.2 0.03
60 km from the site; and (Morgan Hills earthquake)
• All earthquake events with magnitudes in the range 7–7.5 within 10-17-1989 17:04:00 (PDT) 7.0 18 97.6 0.160
a distance of 120 km from the site. (Loma Prieta earthquake)
While these magnitude-distance combinations for amax ≥ 01-15-1993 06:29:34 (PST) 5.3 5.0 120.3 0.014
0.10 g at Wildlife were obtained from the experience of recordings 08-02-2006 20:08:12 (PDT) 4.4 9.1 62.6 0.014
at the site in 2005–2014, they are also roughly consistent with gen- 12-20-2006 19:12:28 (PST) 3.7 9.0 12.7 0.02
erally accepted attenuation relations for the western United States, 10-30-2007 20:04:54 (PDT) 5.4 9.2 68.5 0.012
as discussed in Appendix S1 (Petersen et al. 2008). 01-07-2011 06:10:16 (PST) 4.1 7.1 86.5 0.005
Table 3 shows that about 61 earthquakes having an estimated 08-24-2014 03:20:44 (PDT) 6.0 11.3 43.6 0.017
(Napa earthquake)
peak ground acceleration at the site, amax ≥ 0.1 g, may have

© ASCE 04017009-9 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2017, 143(6): 04017009


were already dissipating in most of the critical layer, with the large justified assumptions rather than actual measurements. Also, some
fluctuations in ru after 26–27 s being elastic fluctuations rather than of the discussions are based on concepts already defined in the
true pore pressure buildup. That is, most probably these oscillations previous section.
in the pore pressures were caused by fluctuations generated by the
surface waves in the total vertical pressure, σv , without change in
the effective pressure, σv0 . In fact, many small to moderate earth- 1989 Earthquake Location and Characteristics
quakes recorded at the site exhibit this dynamic oscillation in the Treasure Island is located in the San Francisco Bay of Northern
pore pressure as the total vertical pressure changes with the passage California. The island is about 80 km north to the closest part
of both body and surface waves, without any excess pore pressure of the Loma Prieta earthquake rupture zone, as shown in Fig. 4(b),
generation. In this context, the recorded value of ðru Þmax ¼ 0.19 which also shows the major faults and earthquake epicenters in the
that occurred after 40 s in Fig. 3(b) and is included in Tables 1 area (Rockwell and Klinger 2013).
and 2, may be a slight overestimation of the true pore pressure The Treasure Island F.S site was subjected on October 17, 1989,
buildup relevant to possible liquefaction of the critical layer in to the Loma Prieta earthquake having a moment magnitude,
the El Mayor-Cucupah earthquake. This 2010 earthquake, reflected M w ¼ 6.9. The epicenter of the earthquake was located at
Downloaded from ascelibrary.org by University Of Florida on 10/31/17. Copyright ASCE. For personal use only; all rights reserved.

in the records of Fig. 3, occurred outside the Valley at a distance 37.04°N 121.88°W, near the summit of Loma Prieta Mountain, with
from the site of 93 km [Table 1 and Fig. 4(a)]. Significant surface a hypocentral depth of about 18 km, approximately 16 km north-
waves have also been observed in closer, lower magnitude earth- east of Santa Cruz and about 30 km south of San Jose, on a section
quakes associated with local faults. A notable example is the 1987, of the San Andreas Fault. An important aspect was the bilateral
Mw ¼ 6.5 Superstition Hills earthquake that did liquefy the site,
character of the fault rupture, which developed along both the
which occurred at a distance of 30.5 km [Fig. 4(a)]. The 1987 ac-
northwest and southwest directions of the fault, with a rupture du-
celeration records also had a long duration due to surface wave
ration of about 7–10 s, resulting in shorter durations of ground
arrivals that caused large cyclic shear strains in the softened critical
shaking (typically less than 15 s) than typically expected for an
silty layer, and the surface waves may have contributed to the lique-
earthquake of M w ¼ 6.9 (EERI 1990; Rockwell and Klinger 2013).
faction process (Porcella et al. 1987; Brady et al. 1989; Dobry et al.
The damage was heavy in Santa Cruz County, with the heavy
1992; Holzer and Youd 2007).
damage extending well to the north into the San Francisco Bay
Area, where Treasure Island is located, both on the San Francisco
Liquefaction History of Site Peninsula and across the bay in Oakland (EERI 1990).
For some of the estimated total of 61 earthquakes that produced As discussed later, the 1989 Loma Prieta earthquake was the
significant accelerations and pore pressures in the critical layer largest earthquake felt up to that point at the Treasure Island F.S.
at Wildlife in 1907–2010, liquefaction did occur. Youd et al. site since the soil was deposited in the 1930s. This is an artificial
(2007) reported that six earthquakes in the previous 75–80 years island built in the 1930s, about 30 years after the major 1906 San
generated observed liquefaction effects at or within 10 km of Francisco earthquake, which had relieved stress on the San Andreas
Wildlife. Fig. 1 includes data points from five earthquakes of mag- fault in the region and resulted in a relatively quiet seismic period
nitudes ranging from 5.9 to 7.2 that shook the Wildlife site in the throughout the rest of the twentieth century (ATC 2005).
31-year period between 1979 and 2010, with these earthquakes
listed in Table S1 of Appendix S1. Two of these five events are
known to have liquefied the silty sand layer: the 1981 Westmorland Site Conditions and Deposition History
earthquake and the 1987 Superstition Hills earthquake, with the Treasure Island is an artificial island created by hydraulic filling of
liquefaction in the 1987 Superstition Hills earthquake confirmed sand over soft mud sedimentary deposit in 1936–1937. The specific
both by surface manifestations and recordings from the piezometers method of deposition was from a pipe into open water, so the criti-
installed at the time (Holzer et al. 1989; Andrus et al. 2003; Zeghal cal liquefiable layer of interest was probably formed by settling
and Elgamal 1994; Holzer and Youd 2007; Dobry et al. 2015a). No (T. L. Holzer, personal communication, 2016). The island was
liquefaction has been reported since 1987. Two other earthquakes formed from 21 cubic yards of dredged sand under the direction
in Table S1 of Appendix S1 are known not to have liquefied the of the U.S. Army Corps of Engineers. The dredged sand was ob-
site: the 1987 Elmore Ranch and the 2010 El Mayor-Cucupah tained from different borrow sources in the San Francisco Bay
earthquakes, with this nonliquefaction status confirmed in both (Power et al. 1998). The soil profile of the Treasure Island F.S. site
cases by lack of surface manifestations and piezometric recordings was summarized by Pass (1994) and Faris and de Alba (2000). It
(Zeghal and Elgamal 1994). It is not clear if the site liquefied or not consists of [Fig. 5(a)]: (1) from 0 to about 4.6 m, fine sand with
in the 1979 Imperial Valley earthquake, also listed in Table S1, as rock and concrete fragments; (2) from about 4.6 to 12 m, fine gray
no observer visited it at the time (Youd et al. 2007; Youd 2013). silty sand, which was deposited by hydraulic filling with a clayey
The corresponding question mark for this case history has been zone in the middle located approximately from about 7 to 9 m;
added in Table S1. (3) from about 12 to 14 m, a transitional clayey sand zone; and
(4) from about 14 to 30 m, the Young Bay Mud native normally
Treasure Island Fire Station Site 1989 Case History consolidated clay. Fig. 5(a) also shows the CPT tip resistance (qc ,
MPa) and friction ratio (Rf , %) reported by de Alba et al. (1994),
This section discusses in detail the Treasure Island F.S. site case with qc ≈ 1–7 MPa in the critical liquefiable silty sand layer.
history utilizing the same viewpoints used in the previous section The critical layer of the deposit was defined by Andrus et al.
for the Wildlife site case history. Specifically, the titles of the sub- (2003) to be between 4.5 and 7 m, which is confirmed by the
sections are the same in both sections. The main difference between low cone tip resistance of this layer. The average grain size distri-
the discussion here as compared with the similar discussion in the bution curve of the silty sand in the critical layer is shown in
previous section about the Wildlife case history, is associated with Fig. 5(b). The crosshole shear wave velocity profile in the critical
the absence of pore pressure measurements in the Treasure Island layer at the Treasure Island F.S. site was measured by Fuhriman
F.S. site. Therefore, some of the results shown here are based on (1993) and is reproduced in Fig. 6.

© ASCE 04017009-10 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2017, 143(6): 04017009


The Center for Engineering Strong Motion Data (CESMD), eight events recorded during these 41 years, with only the 1989
classifies the Treasure Island F.S. site as a Site Class E site Loma Prieta earthquake resulting in an amax greater than the
withV S30 ¼ 155 m=s(http://strongmotioncenter.org/cgibin/CESMD practical threshold acceleration of 0.10 g. Perhaps even more
/stationhtml.pl?stationID=CE58117&network=CGS). interesting, except for the Loma Prieta earthquake, all recorded
earthquakes in the table have amax ≤ at ¼ 0.029–0.05, the range
previously calculated for the actual threshold acceleration for the
Threshold Acceleration of Critical Layer at Site critical layer at the site. Therefore, it is reasonable to conclude
The threshold acceleration (at ) of the liquefiable layer of the that—with the exception of Loma Prieta in 1989—none of the
Treasure Island F.S. site was obtained using Eq. (3) and a threshold earthquakes recorded by the two instruments in 1974–2015 and
strain, γ tv ¼ 0.01–0.02%; that is, using the same procedure utilized listed in Table 4, generated any excess pore pressure at the critical
in the previous section to estimate at for the Wildlife site. For layer in the Treasure Island F.S. site, and none had any effect on the
Treasure Island, the equation yields a range of at ¼ 0.028–0.05 g, liquefaction resistance of the layer.
essentially identical to the range, at ¼ 0.029–0.05 g obtained for
Wildlife. This range for Treasure Island has been included in
Downloaded from ascelibrary.org by University Of Florida on 10/31/17. Copyright ASCE. For personal use only; all rights reserved.

Table 1. Seismic Preshaking History of Site


The same as for Wildlife, these at ¼ 0.028–0.05 g correspond The same as for Wildlife, the seismic preshaking history of the
to the level of ground surface acceleration that initiates the pore Treasure Island F.S. site is defined as the total number of earth-
pressure buildup. The instrumental measurements in 2005–2014 quakes with amax ≥ 0.10 g that affected the critical layer between
in Wildlife, discussed in the previous section and summarized in deposition (taken to be in the year 1937), and the Loma Prieta
Fig. 8(a), showed that the minimum value of amax associated with earthquake in 1989.
a significant pore pressure ratio, ðru Þmax ≥ 0.02, sufficient to start The instrumental information just discussed and listed in Table 4
affecting the liquefaction resistance of the layer, was significantly covers the 15 years before the Loma Prieta earthquake (1974–
greater than at ¼ 0.029–0.05 g. Specifically, it was determined that 1989), which is about one third of the 52-year period between
at the Wildlife site, where the range for at is essentially identical to deposition of the critical layer in 1937, and the shaking and lique-
that of the Treasure Island F.S. site, the minimum acceleration value faction of the site in 1989. Still, Table 4 is strongly suggestive of the
needed to generate such significant pore pressure buildup was scarcity of relevant earthquakes affecting the critical layer in those
amax ≈ 0.10 g. Due to the lack of pore pressure measurements 52 years, confirming how quiet the seismic environment of the San
and the scarcity of significant earthquake shakings affecting the Francisco Bay Area has been in the last century or so. In fact, with
Treasure Island F.S. site, it is not possible to determine this mini- the exception of the 1989 Loma Prieta earthquake itself, and per-
mum amax empirically the way it was done at the Wildlife site. On haps also the 2014, M w ¼ 6.0 Napa earthquake, there is no other
the other hand, as shown in the next two headings, exact determi- event in Table 4 within any of the magnitude-distance combinations
nation of this minimum value of amax needed for significant pore listed in the previous section for the Wildlife site needed to generate
pressure buildup is less critical for the counting of relevant earth- at the site an amax ≥ 0.10 g. This is also confirmed by the fact
quakes at Treasure Island than for the Wildlife site, due to the scar- that—except again for Loma Prieta—all amax values listed in
city of preshaking events for any level of acceleration at Treasure Table 4 are much smaller than 0.10 g.
Island. The next step is to estimate the total number of earthquakes with
Given the similarity of the values of at at both sites, as well as amax ≥ 0.1 g that the Treasure Island F.S. site may have experi-
the broad similarity in the rest of the site conditions and soil char- enced in the 52 years between deposition and the Loma Prieta
acteristics summarized in Table 1 and Figs. 5 and 6, it seems rea- earthquake. In order to facilitate a direct comparison between
sonable to select for Treasure Island in first approximation, the the level of seismic activity at the Treasure Island site, with that
same minimum value of amax ¼ 0.10 g previously found from at the Wildlife site, the earthquakes at Treasure Island between
the data in Fig. 8(a) for Wildlife. Therefore, it will be assumed 1907–2010 (103 years) are counted first and then adjusted for
in the next subsections that any time an earthquake took place with the shorter period 1937–1989 (52 years).
amax ≥ 0.10 g at the Treasure Island site, the resistance to liquefac- The same as is done for Wildlife, the main tool to count earth-
tion of the critical layer was modified (increased or decreased). quakes in Treasure Island is the available combined historic and
That is, only earthquakes with amax ≥ 0.10 g at the site will be instrumental UCERF3 Catalog (Field et al. 2013). However, before
counted to evaluate the preshaking effect. entering the Catalog for the Treasure Island F.S. site, it is necessary
to define first the relevant combinations of magnitudes and distan-
ces estimated to have induced amax ≥ 0.1 g at the site.
Measurement of Surface Accelerations in 1974–2015
It was decided to use for Treasure Island exactly the same mag-
The site was instrumented with a surface accelerometer in 1974 by nitude-distance combinations previously utilized for Wildlife and
the California Geological Surveys’ Strong Motion Instrumentation listed in Table 3, on the basis of the following considerations:
Program, with the instrument remaining active until 2004. A sep- 1. The selected criterion for shaking is the same at both sites
arate array of surface and buried accelerometers—labeled Geotech- (amax ≥ 0.1 g).
nical Array—was installed in 1992 about 10 m from the fire station 2. Both sites are constituted by soft soils and are classified as Site
instrument; this Geotechnical Array was still active at the time of Class E, with V s30 ¼ 177 and 155 m=s for the Wildlife and
submitting this paper (2015). The site was also instrumented with Treasure Island F.S. sites, respectively. This means that both
pore pressure transducers; however, no pore pressure data has been sites are expected to have similar amplification characteristics
made available and the status of the instrumentation is unknown. and are treated similarly by ASCE 7–10 for seismic design
Given the low levels of ground motion at the site, it is expected that purposes.
no significant excess pore pressure buildup has occurred. Table 4 3. The same western United States attenuation relations at low
lists all earthquakes recorded by either the old or the new surface periods are used by the USGS in the Imperial Valley and
accelerometer covering a total period of 41 years (1974–2015), as San Francisco Bay Area for generation of the corresponding
reported by the CESMD website (CESMD 2016). The table lists U.S. National Seismic Hazard Maps (Petersen et al. 2008).

© ASCE 04017009-11 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2017, 143(6): 04017009


It is useful to elaborate further on Point 3. The tectonic environ- lateral spreading, ground cracks, landslides, settlements, and so
ments of the Imperial Valley and Treasure Island are similar, with forth). The focus of the compilation is on the 1906 San Francisco
predominantly strike-slip faulting related to the San Andreas fault earthquake as well as earthquakes afterward, specifically between
system. In the Imperial Valley, the proximity to the Gulf of 1906 and 1970, for which more observations are available than be-
California extensional spreading system, and the step over between fore 1906. No cases of ground failure of any kind are identified by
the San Andreas and Imperial Fault systems does produce more Youd and Hoose for Treasure Island. This is strong evidence that no
normal faulting sources relative to the San Francisco Bay Area. site in Treasure Island experienced liquefaction between construc-
However, this difference should be negligible in terms of ground tion of the island in 1937 and 1970. This, combined with the mea-
motions for similar regional geologies and local site conditions, as sured low accelerations at the Treasure Island F.S. site after 1974
normal faulting sources and strike-slip faulting sources produce listed in Table 4, shows that no liquefaction took place at the site
similar motions compared to reverse faulting sources (Bommer before the 1989 Loma Prieta earthquake. In this respect, the earth-
et al. 2003). As noted before, the local site conditions at the quake listed in Table 3 with M w ¼ 5–6 and d < 20 km is especially
Treasure Island F.S. and Wildlife sites are similar, with both clas- interesting. It corresponds to the 1957 Daly City earthquake, with
sified as Site Class E. However, as discussed in the previous sec- M w ¼ 5.3 and epicenter located at the west end of the city of
Downloaded from ascelibrary.org by University Of Florida on 10/31/17. Copyright ASCE. For personal use only; all rights reserved.

tion, the Imperial Valley has a very deep basin structure that tends San Francisco, about 15 km from Treasure Island (Bonilla
to generate surface waves that arrive late in the shaking records and 1960). The earthquake caused extensive liquefaction and land slid-
may increase both the duration and the long period response at the ing of hydraulic fills in Lake Merced, very close to the epicenter
Wildlife site. This is very different from the Treasure Island F.S. (Youd and Hoose 1978). However, while it probably generated
site, where Franciscan bedrock is located at about a 90-m depth. some excess pore pressures in the critical layer at the Treasure
On the other hand, this difference should not affect the short period Island F.S. site, it certainly did not liquefy the site.
response of the site, with the peak ground acceleration, amax , not Therefore, the Treasure Island F.S. site has experienced lique-
being affected except for a very large magnitude event happening at faction only once—in 1989—with a couple of other earthquakes
a great distance, an event not relevant to this study. This reasoning possibly generating excess pore pressures short of liquefaction after
is confirmed by the fact that for the Wildlife site, as per Table 3, 1937, both before and after 1989.
earthquakes in the magnitude range, M w ¼ 7–7.5 at distances,
d < 120 km, are expected to produce a value of amax ≥ 0.10 g.
The Loma Prieta earthquake, with Mw ¼ 6.9, is close to that earth- Comparison between the Two Case Histories
quake magnitude range and occurred at d ¼ 98 < 120 km from
Treasure Island; this earthquake had a recorded amax ¼ 0.16 g > As mentioned previously, the Wildlife and the Treasure Island F.S.
0.10 g, consistent with the Wildlife magnitude-distance relation- case histories plot very close to each other on the V s -based lique-
ship of Table 3. faction charts (Fig. 1). As shown in Fig. 6, the shear wave velocity
Therefore, the same magnitude-distance combinations utilized profiles of both sites also plot on top of each other, indicating al-
for the Wildlife site and listed in Table 3, were used to count earth- most identical small-strain shear stiffness of the two critical layers.
quakes from the UCERF3 Catalog, expected to have had amax ≥ In addition, the two silty sands have similar grain size distributions,
0.10 g at the Treasure Island F.S. site. The results of this counting the two peak accelerations were 0.15–0.16 g, and the magnitudes
are listed in Table 3 and plotted in Fig. 8(b). As seen in Table 3, a of both earthquakes were also similar (6.9–7.2); see Table 1. As a
total of only four earthquakes are estimated to have caused amax ≥ result, the magnitude corrected CSR of the liquefiable layers of
0.10 g at the Treasure Island F.S. site between the years 1907– both sites during the 1989 and 2010 earthquakes—presented by
2010. This is a very small number for the 103-year period, which Andrus et al. (2003) and Dobry et al. (2015a) using the Simplified
is reduced even further when one considers the shorter 52-year Procedure—are found to be 0.11 and 0.13, respectively. These sim-
period between construction of the island in 1937 and just before ilar CSR and shear wave velocities are the reason why the two data
the 1989 Loma Prieta earthquake. That is, after considering that points are so close on the chart of Fig. 1. These and other similar-
one of the four events in Table 3 is the Loma Prieta event itself; ities presented in Table 1 support the validity of the one-to-one
the unavoidable conclusion from Table 3 is that between deposition comparison between both case histories. The main question ad-
and the 1989 earthquake, the critical layer was subjected to about dressed in this paper is why, despite these similarities, the pore
two earthquakes producing amax ≥ 0.10 g and generating signifi- pressure responses of the sites to the 1989 and 2010 earthquakes
cant excess pore pressures in the critical layer. This is in turn were so different. The Treasure Island F.S. site liquefied while at
confirmed by the instrumental records in Table 4, which indicate Wildlife a small pore pressure was generated that was far from
that no record of this type occurred in the 15 years between 1974 liquefaction.
and just before the Loma Prieta event in 1989. This very small On the basis of field and small-sample laboratory investigations
number of earthquakes in 1937–1989 (≈2) potentially affecting presented in the literature and discussed in Appendix S1, the two
the liquefaction resistance of the critical layer at the Treasure Island factors that could potentially explain the higher field liquefaction
F.S. site, confirms the fact that the San Francisco Bay Area was resistance in Wildlife than in Treasure Island at essentially equal
seismically very quiet in the twentieth century after the stress V s1 , are (1) the longer geologic age of the Wildlife critical layer
relief caused by the large 1906 San Francisco earthquake (ATC at the time of the 2010 earthquake (103 years since deposition),
2005). compared with the age of the Treasure Island F.S. site at the time
of the 1989 earthquake (52 years since deposition); and (2) the
much more intense seismic preshaking that has taken place in
Liquefaction History of Site
the Imperial Valley compared with the San Francisco Bay Area dur-
Youd and Hoose (1978) performed a comprehensive compilation of ing the twentieth century and the beginning of the 21st century.
historic ground failures in Northern California triggered by earth- These two factors are explored in the next section.
quakes, including detailed tables and maps. The study included A question could arise about the possible influence of the differ-
earthquakes between 1800 and 1970 and all phenomena typically ent fabrics the two layers had immediately after deposition on their
associated with surface manifestations of liquefaction (sand boils, different liquefaction responses in 1989 and 2010. As mentioned

© ASCE 04017009-12 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2017, 143(6): 04017009


earlier, the most probable deposition methods of the layers were not In seismically quiet areas and for long geologic times, effect
identical: At Wildlife the layer was bedload left behind once the (1) should tend to predominate, while in seismically active areas
river currents waned, while at Treasure Island the layer was settled effects (1) and (2) act together and both must be considered.
in water from deposition from a pipe into open water. However, as Appendix S1 includes a more detailed discussion on both preshak-
shown by Abdoun et al. (2013) and Dobry and Abdoun (2015b), ing and geologic age and their relevance to the two case histories
the influence of fabric on liquefaction resistance for an uncom- under consideration here. The main conclusions from the discus-
pacted deposit immediately after deposition is mainly a shear stiff- sion in Appendix S1 are summarized in the rest of this section.
ness effect, which is accounted for by the value of shear wave The critical layer in Wildlife is classified as geologically young
velocity. The wave velocities at the two sites are currently in the at the time of the 2010 earthquake (about 103 years old; 23 years
same narrow range (V s1 ¼ 146–155 m=s, see Table 1). Based on since last liquefaction in 1987). At the time of the 1989 Loma Prieta
the evidence presented herein, this range is probably still earthquake the critical layer of the Treasure Island F.S. site was
representative of the value V s1 had at both sites immediately after about 52 years old, which makes it also geologically young. Based
deposition, because (1) both deposits are less than or about a cen- on their field work and work by others, Hayati and Andrus (2009)
tury old, which for a nonplastic silty sand means that the increase in proposed that the increase in liquefaction resistance due to geologic
Downloaded from ascelibrary.org by University Of Florida on 10/31/17. Copyright ASCE. For personal use only; all rights reserved.

V s1 due to geologic age is limited at most to 5–10% (Anderson age can be taken into account by multiplying the CRR obtained
and Stokoe 1978; Kokusho et al. 1982); and (2) centrifuge and from a chart such as Fig. 1, by a factor K DR. That is, the CRR ¼
large-scale experiments discussed later in this paper indicate that ðCRRÞyoung associated with the curves in such liquefaction charts is
V s1 is not significantly affected by a preshaking sequence such assumed to correspond to a geologically young deposit having a
as that affecting the Wildlife site before the 2010 earthquake representative age, t ¼ 23 years. For deposits older than 23 years,
(El-Sekelly et al. 2016b). Based on this, the difference in fabrics ðCRRÞold ¼ ðCRRÞyoung × K DR , where K DR ¼ 0.13 logðtÞ þ 0.83,
between the two layers probably did not result in a significant dif- and t (years) is either the time since the layer last liquefied, or if
ference in their shear wave velocities after deposition, so it is rea- never liquefied, the time since deposition. More details about K DR
sonable to infer that their liquefaction resistances at that time were can be found in Appendix S1. The authors calculated this factor,
also similar. and obtained K DR ¼ 1.01–1.09 for the Wildlife layer (using respec-
tively, 23 years since last liquefaction in 1987, or 103 years since
deposition in 1907); and K DR ¼ 1.05 for the Treasure Island F.S.
Possible Effects of Geologic Age and Preshaking layer (52 years since deposition). These values of K DR are all very
low and similar, which rules out geologic age as an explanation of
After a sediment is deposited and buried by other sediments there the significant difference in pore pressure responses at the two sites
are a number of processes that may occur within the deposit and during the 2010 and 1989 earthquakes.
change its engineering properties over geologic time under static On the other hand, Fig. 8(b) compares the numbers of preshak-
load, typically stiffening and strengthening it, which can finally re- ing earthquake events affecting the Wildlife and Treasure Island
sult in lithifying the soil into rock. These processes are collectively F.S. sites between 1907 and 2010 (values extracted from Table 3),
known as diagenesis (e.g., Press and Siever 1978; Finn 1981). In for several magnitude ranges as well as for the total number of
areas that are seismically active during the same geologic time, earthquakes. The difference between the two sites is striking.
earthquakes occurring from time to time also have an effect, either Whether the comparison is performed by earthquake magnitudes
additionally stiffening and strengthening the soil, or, if they pro- or by looking at the total number of events, the number of earth-
duce a great enough disturbance by, for example, massively lique- quakes preshaking the Wildlife site has been greater by more than
fying and deforming the ground, by weakening the soil and making one order of magnitude, when compared with the Treasure Island
it perhaps less stiff. Therefore, in principle, the two factors that F.S. site (61 versus two earthquakes). In other words, while the
could help explain the higher field liquefaction resistance at equal critical layer at the Wildife site was preshaken (including some
V s1 shown by the Wildlife site compared to the Treasure Island F.S. liquefaction events) several dozen times by earthquakes of amax ≥
site are as follows: 0.10 g before the El Mayor-Cucupah earthquake in 2010, the criti-
1. The interlocking (Schmertmann 1991; Arango and Migues cal layer at the Treasure Island F.S. site was preshaken only about
1996) and/or cementation (Seed 1979; Mitchell and Solymar twice since deposition. This is consistent, again, with the general
1984) that strengthens the contacts between neighboring sand observation that the Imperial Valley has been subjected to very in-
grains (with the cementation caused by small amounts of miner- tense seismic activity in the last century, while the San Francisco
al precipitation), that accumulate with time (labeled here as geo- Bay Area was very quiet seismically between the 1906 and 1989
logic age effect for simplicity); and/or earthquakes.
2. The preshaking due to repeated earthquakes. Several laboratory Therefore, it seems that after looking at the situation from a
studies were conducted in the 1970s involving cyclic loading number of viewpoints, the seismic preshaking history emerges
undrained tests of saturated sand samples that had been sub- as the most significant difference between the two sites. The results
jected to a previous sequence of events of cyclic undrained from this side-by-side comparison suggest that seismic preshaking
loading (cyclic prestraining or preshaking), while allowing dis- history is the most significant difference between the two sites, and
sipation of the excess pore pressures between events. Low-level is therefore the most probable reason for the increased liquefaction
cyclic prestraining was found to increase liquefaction resistance resistance of the Wildlife site compared to the Treasure Island F.S.
rather rapidly (Youd 1977; Seed et al. 1977; Seed 1979), while site as evidenced by their different responses to the 2010 and 1989
large liquefying cyclic prestraining was found to decrease lique- earthquakes.
faction resistance (Finn et al. 1970). Although the low-level pre-
straining did not change the relative density of the soil much, it
probably induced a stronger fabric with added particle interlock- Centrifuge and Large-Scale Experimental Results
ing. Additional discussion of possible mechanisms associated
with preshaking that are relevant to liquefaction resistance is This section provides experimental evidence on the effect of pre-
presented in the “Discussion” section. shaking on the liquefaction resistance of sand deposits. The effect

© ASCE 04017009-13 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2017, 143(6): 04017009


of preshaking on the liquefaction resistance of four silty sand and to test different shaking sequences and combinations. As listed in
clean sand deposits, was studied experimentally by El-Sekelly Appendix S1, only Events A, B, and C were applied to the deposits
(2014). The work included three centrifuge tests performed at Re- in Experiments 1, 2, and 4, with Events A, B, C, and one strong
nsselaer Polytechnic Institute (RPI) (Experiments 1–3; El-Sekelly Event D applied in Experiment 3. The excess pore pressures were
2014), and one large- (full) scale, 1 g test performed at the Univer- allowed to dissipate fully between shakings. While generally the
sity at Buffalo (UB) (Experiment 4; Thevanayagam et al. 2009; sequences of shakings used in the experiments did not correspond
El-Sekelly 2014). The research involved simulating from several to any specific site, the sequence used in Experiment 1 to shake the
decades to several centuries of earthquake events applied to a silty sand deposit was made to correspond crudely to the shaking
6-m uniform soil deposit, including shakings that liquefied the de- history that affected the critical silty sand layer of the Wildlife site
posit and others that did not liquefy it. The program used different before the 2010 earthquake (El-Sekelly et al. 2016b). In addition to
shaking sequences in the different experiments, in order to explore Events A, B, C, and D, weak nondestructive shakings, non-destructive
the complex relation in the field between preshaking seismic events (ND), were also applied for System Identification purposes.
that increase the liquefaction resistance of young deposits, and Fig. 9(a) shows the data points for centrifuge and large-scale
liquefying events that may decrease it totally or partially. Experi- Experiments 1–4 on clean and silty sands corresponding to the first
Downloaded from ascelibrary.org by University Of Florida on 10/31/17. Copyright ASCE. For personal use only; all rights reserved.

ments 1 and 2 were performed on silty sand deposited by dry three or four shakings, plotted on the V s -based field liquefaction
pluviation, while Experiments 3 and 4 were performed on a clean chart for recent uncompacted fills proposed by Dobry et al.
sand deposited by dry pluviation and hydraulic filling, respectively. (2015a) and previously shown in Fig. 2(a). For Experiments 2–4,
More details about the experimental program conducted by El- the four data points in Fig. 9(a) correspond to the first four shakings
Sekelly (2014) can be found in Appendix S1. The results of Experi- (Events ND, A, B, and C). For Experiment 1, there are only three
ment 1 are also reported by El-Sekelly et al. (2016a, b). data points in Fig. 9(a), as no Event C was applied at the beginning or
The shaking sequences in the four experiments were different, end of the test. At this early stage of the experiments, the centrifuge
However, in all four tests the soil deposits were subjected to the and large-scale deposits tested are comparable to recent uncom-
same three or four types of events applied at the base of the model; pacted artificial fills in the field which have not been subjected to
Events A, B, and C/D. In all four tests, Events A were defined as five significant preshaking. As expected, the curve in Fig. 9(a) generally
sinusoidal cycles of a peak base acceleration, apb ≈ 0.035–0.05 g; separates well liquefaction from no liquefaction events, thus validat-
Events B as 15 sinusoidal cycles of apb ≈ 0.04–0.05 g in prototype ing both the Dobry et al. (2015a) chart and Experiments 1–4.
units; and Events C/D as 15 sinusoidal cycles of apb ≈ 0.1=0.25 g, Fig. 9(b) shows the data points corresponding to the same three
respectively, all in prototype units. The prototype frequency in all or four shakings of Experiments 1–4 on clean and silty sands
cases was 2 Hz. The 15-cycle duration of Events B and C/D cor- plotted on the same V s -based field liquefaction chart, but now cor-
responds approximately to an earthquake of moment magnitude, responding to the end of the corresponding experiment. Fig. 9(b)
M w ≈ 7.5; while the five-cycle duration of Events A corresponds also includes the higher liquefaction curve proposed by Dobry et al.
to M w ≈ 6 (Idriss and Boulanger 2008). The total number of shak- (2015a) for the Imperial Valley [dashed curve in Fig. 9(b)].
ing events in the experiments ranged from 37 to 91 shakings in order The solid and dashed curves in Fig. 9(b) are the same previously

Fig. 9. (a) Locations of first three or four shakings (recently deposited uncompacted fills before preshaking) from Experiments 1–4 on clean and silty
sand deposits on Dobry et al. (2015a) V s -based liquefaction chart; (b) locations of last three or four shakings (heavily preshaken fills) from Experi-
ments 1–4, Dobry et al. (2015a) V s -based liquefaction chart; full data points represent liquefaction, open data points represent no-liquefaction (mod-
ified from El-Sekelly 2014; Dobry et al. 2015d); all data points in the figure correspond to events A, B, and C as well as weak nondestructive
ND shakings (stronger event D was applied to deposit only once in middle of Experiment 3)

© ASCE 04017009-14 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2017, 143(6): 04017009


calibrated by field case histories in Fig. 2(b). At this late stage of the The sequence of 66 shakings used in centrifuge Experiment 1 sum-
experiments and after being subjected to 37–91 earthquake events, marized in Fig. 9, which simulated crudely the seismic history of
some of which liquefied the deposit, the centrifuge and large-scale the Wildlife site, increased the relative density (Dr) of the silty sand
models are comparable to heavily preshaken deposits in the field. from 38 to 50% (El-Sekelly et al. 2016b). The third mechanism
Fig. 9(b) shows several points that did not liquefy despite being would be an increase in lateral effective pressures and a correspond-
well above the solid line (false positives), or even above the dashed ing increase in the lateral pressure factor K 0 that presumably took
line. In fact, the excess pore pressures measured in these shakings place during the preshaking, both in the field at Wildlife and in the
were all small and far from liquefaction (El-Sekelly 2014). centrifuge and large-scale Experiments 1–4 summarized in Fig. 9.
The authors did not measure the values of K 0 in these experiments,
but laboratory testing by Youd and Craven (1975) showed that the
Discussion increase in K 0 produced by shaking can be quite significant, and in
turn a higher K 0 is associated with a higher liquefaction resistance
The paper presented strong evidence pointing to the prior seismic (Seed 1979; Finn 1981). One common feature of these three mech-
preshaking history as the most probable explanation of the very anisms created by preshaking is that they all induce stronger par-
Downloaded from ascelibrary.org by University Of Florida on 10/31/17. Copyright ASCE. For personal use only; all rights reserved.

different pore pressure responses of the Wildlife site and the ticle interlocking in the sand.
Treasure Island F.S. site to the 2010 and 1989 earthquakes. This Both field and laboratory evidences suggest that shear wave
difference is consistent with several reported observations after velocity may not reflect the increased strength of soil deposit from
earthquakes which suggest that—other things being equal—natural seismic preshaking. Cox et al. (2013) reported the following about
sands appear to liquefy less than nearby artificial fills in highly Urayasu City in the 2011 Great East Japan earthquake: “Median V s
seismic regions in California and Japan (Pyke 2003; Ishihara et al. profiles showed no significant difference in small-strain shear
2011; Cox et al. 2013; Dobry et al. 2015a). Ishihara et al. (2011) modulus between soils of different age or between areas of different
reported the following for Urayasu City on Tokyo Bay after the liquefaction severity.” This is also clear from the results of the cen-
2011 Great East Japan earthquake: “It is of interest to notice that trifuge and large-scale experiments, which showed only a small in-
although the SPT N-value of 5–15 for the alluvial sand is only crease in the shear wave velocity of the deposit after a number of
slightly larger as compared to N ¼ 5–10 for the reclaimed sand, shaking events ranging between about 40 and 90 [El-Sekelly 2014;
the new sand was more vulnerable to liquefaction, as evidenced Fig. 9(b)]. These relatively unchanged values of V s1 measured in
by the clear manifestation of liquefaction in the reclaimed area the experiments are also consistent with the V s measured by cross-
in contrast to no liquefaction in the old area in the north.” These hole tests in the field at the Wildlife layer in 1984 and 2005
and other examples of increased liquefaction resistance of natural (Bieschwale and Stokoe 1984; Cox 2006), before and after the
sands in both highly seismic and less seismic areas have been 1987 liquefaction of the site, as shown by Fig. 6. The insensitivity
variously attributed to geologic age and/or seismic preshaking of V s to preshaking is consistent with other evidence from the lit-
by previous earthquakes (e.g., Arango et al. 2000; Heidari and erature. Specifically, two of the three mechanisms discussed earlier
Andrus 2012; Hayati and Andrus 2008, 2009; Dobry et al. 2015a). that would explain the effect of preshaking in increasing liquefac-
In the specific case of the Imperial Valley of California, the in- tion resistance are densification and increase in K 0 . It is useful to
creased liquefaction resistance of the geologically very young silty quantify the possible variation of V s at a given depth if the relative
sand deposits there, was almost surely caused by intense preshak- density increases from 38 to 50% (as measured in Experiment 1 and
ing rather than geologic age, as discussed throughout this paper. reported by El-Sekelly et al. 2016b), and K 0 experiences a simul-
The effect of geologic age is small and similar for these two taneous assumed increase from 0.5 to 0.7. Using the procedure
representative sites in the Imperial Valley and San Francisco Bay described by Seed and Idriss (1970), the predicted increase in
Area, so geologic age cannot possibly explain their very different V s is only 14%.
liquefaction resistances at the time of the 2010 and 1989 earth- This inability of shear wave velocity to capture the increase in
quakes. On the other hand, 60–70 earthquakes have induced sig- liquefaction resistance due to preshaking was a main reason behind
nificant excess pore water pressures on the critical layer at the the Dobry et al. (2015a) recommendation of a higher liquefaction
Wildlife site in the Imperial Valley, between its 1907 deposition curve for natural silty sand sites in the Imperial Valley of Southern
and the 2010 El Mayor-Cucupah earthquake, explaining its high California discussed in a previous section and shown in Figs. 2(b)
liquefaction resistance in 2010. Contrast this with about two earth- and 9(b). Based on the experimental results summarized in
quakes that may have induced significant excess pore pressures in Figs. 9(a and b) and especially Experiment 1, it is possible to quan-
the critical layer at the Treasure Island F.S. site, between its 1937 tify in a very preliminary way the number of shakings at the
deposition and the 1989 Loma Prieta earthquake, explaining why Wildlife site that were needed to increase the liquefaction resistance
this layer still behaved as a recent uncompacted fill with low lique- of the site to the level defined by the dashed curve in Fig. 9(b).
faction resistance in 1989. Of special interest is the data shown in Figs. 9(a and b) correspond-
The authors speculate that there are three main mechanisms ing to the triangular data point at V s1 ≈ 100 m=s and CSR in the
inducing this observed increase in liquefaction resistance of the range 0.15–0.20, associated to an Event B similar to the El Mayor-
Wildlife silty sand layer over the last century due to the 60–70 shak- Cucupah earthquake that shook the site in 2010 and did not induce
ings experienced by the layer. The first mechanism would be the liquefaction. As shown by Figs. 9(a and b), supplemented by the
collapse of the loose unstable array of particles present in the sand information in El-Sekelly et al. (2016b), the first shaking of this
immediately after deposition, prompted by the first few earthquakes Event B, S1, did induce liquefaction, while the last relevant Event
felt by the layer. As proposed by Youd (1977), this collapse pro- B for the site, S56, did not. Events B stopped liquefying the cen-
duces a stronger fabric more resistant to liquefaction without much trifuge model deposit somewhere between shakings S12 and S45,
change in relative density. The second and third mechanisms would suggesting that several dozen earthquakes happening during much
have acted simultaneously and more slowly over the 100-plus-years of the twentieth century were needed at Wildlife before the dashed
of the history of the layer. The second mechanism would be line in Fig. 9(b) became applicable to the site.
densification of the layer by the shakings, with a higher relative Given this insensitivity of shear wave velocity to the beneficial
density known to increase liquefaction resistance (Seed 1979). effect of preshaking in increasing liquefaction resistance, it is useful

© ASCE 04017009-15 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2017, 143(6): 04017009


to explore alternative methods to quantify this effect in the absence relevant preshaking earthquake events since deposition, by
of a regional V s liquefaction chart such as the dashed line of using site recordings if available in conjunction with available
Fig. 9(b). The two other popular methods for evaluating liquefac- regional earthquake catalogs and regional attenuation relations.
tion potential use penetration resistance (SPT and CPT). The au-
thors conducted CPT measurements at different times during the
preshaking process in large-scale Experiment 4, which caused a big Acknowledgments
increase in liquefaction resistance as shown by Fig. 9 (El-Sekelly
2014). The relative density also increased very significantly, going The authors want to acknowledge the California Department of
from Dr ¼ 40–80% at the end of Experiment 4. While the value of Fish and Wildlife that provides access to the monitoring site at
V s1 increased only by 15%, the CPT point resistance increased by the Wildlife Liquefaction Array. The NEES@UCSB field site
almost an order of magnitude (from qc1N ¼ 17 to 130; see facility received support from the George E. Brown, Jr. Network
El-Sekelly 2014). This is due to the much greater sensitivity of for Earthquake Engineering Simulation program through CMS-
the penetration resistance compared with V s to a higher relative 0217421 and CMMI-0927178. The California Geological Surveys’
density of the soil, as well as to the presumably increased lateral Strong Motion Instrumentation Program (CSMIP) provided the
Downloaded from ascelibrary.org by University Of Florida on 10/31/17. Copyright ASCE. For personal use only; all rights reserved.

stresses, and is consistent with other evidence in the literature ground motion observations for the Treasure Island site, and the
(Baldi et al. 1981; Salgado et al. 1997, 1998; Dobry 2010). While U.S. Geological Survey National Strong Motion Instrumentation
these results for Experiment 4 were for clean rather than silty sand project (NSMP) provided the 1987 ground motion and pore pres-
and also the sequence of shakings was quite different from that ex- sure observations from the Wildlife Site. The authors also want to
perienced by the Wildlife site, these results suggest that the CPT thank Professor R. Andrus for useful discussions about the effects
point resistance may have increased significantly at the Wildlife of geologic age, preshaking and liquefaction, Dr. T. Holzer for his
silty sand layer during its more-than-100-year history of shakings. help in clarifying the types of soil deposition at the Wildlife and
Additional research is needed on the effect of preshaking on CPT Treasure Island sites, and Professor S. Thevanayagam for his con-
resistance to clarify if the dramatic results of Experiment 4 on clean tribution to the experimental results included in Fig. 9.
sand are also valid for other clean and silty sands as well as for a
variety of shaking sequences.
Supplemental Data

Appendix S1, containing additional information related to USGS


Conclusions attenuation relations for the western United States, possible effects
Some main conclusions from this paper are of geologic age and preshaking, and Wildlife site liquefaction and
• Seismic preshaking may have a significant effect in increasing nonliquefaction case histories, is available online in the ASCE
the liquefaction resistance of sands. This is shown by the de- Library (http://www.ascelibrary.org).
tailed analysis of the two case histories in the paper, the rela-
tively high liquefaction resistance of other natural sites in the
Imperial Valley as well as some natural sands in other seismic References
areas, and the results of centrifuge and large-scale experiments.
Abdoun, T., et al. (2013). “Centrifuge and large scale modeling of seismic
• The shear wave velocity of liquefiable sands seems to change
pore pressures in sands: A cyclic strain interpretation.” J. Geotech.
little even after extensive preshaking, so this parameter does not Geoenviron. Eng., 139(8), 1215–1234.
account for the increase in liquefaction resistance due to pre- Anderson, D. G., and Stokoe, K. H. (1978). “Shear modulus: A time-
shaking in the corresponding V s -based charts. dependent soil property.” ASTM STP 654:66-90, ASTM, Baltimore.
• Existing V s -based liquefaction charts favorably predict the Andrus, R. D., and Stokoe, K. H. (2000). “Liquefaction resistance of soils
liquefaction and no liquefaction of uncompacted clean and silty from shear-wave velocity.” J. Geotech. Geoenviron. Eng., 126(11),
recent fills, but may be too conservative for heavily preshaken 1015–1025.
deposits in highly seismic areas. New regional charts are needed Andrus, R. D., Stokoe, K. H., Chung, R. M., and Juang, C. H. (2003).
to reduce this conservatism. “Guidelines for evaluating liquefaction resistance using shear wave
• Most of the conclusions in the paper about influence of preshak- velocity measurement and simplified procedures.” U.S. Dept. of
ing in increasing liquefaction resistance are based on the Commerce, Materials and Construction Research Division,
Gaithersburg, MD.
Wildlife site and other silty sand sites in the Imperial Valley,
Arango, I., Lewis, M. R., and Kramer, C. (2000). “Updated liquefaction
supplemented by four centrifuge and large-scale tests. The silty potential analysis eliminates foundation retrofitting of two critical struc-
sand layers in the Valley are all geologically young and have tures.” Soil Dyn. Earthquake Eng., 20(1-4), 17–25.
been subjected to the specific seismic environment of the Valley. Arango, I., and Migues, R. E. (1996). “Investigation of the seismic lique-
Additional field case histories and side-by-side comparisons as faction of old sand deposits.” Rep. on Research, Bechtel Corp., San
well as more centrifuge and large-scale tests are needed cover- Francisco.
ing a wider range of sands, layerings, geologic ages, and seismic ASCE. (2010). “Minimum design loads for buildings and other structures.”
environments in order to reach more general conclusions about ASCE 7, Reston, VA.
the effect of preshaking on liquefaction resistance. ATC. (2005). “San Francisco’s earthquake risk.” Applied Technology
• Tools used in this paper that may be useful to evaluate the degree Council, Redwood City, CA.
of preshaking of a given critical layer and site in a seismic region Baldi, G., Bellotti, R., Ghionna, V., Jamiolkowski, M., and Pasqualini, E.
(1981). “Cone resistance in dry normally and overconsolidated sands.”
include: calculation of the threshold surface acceleration to in-
Proc., GED Session on Cone Penetration Testing and Experience,
itiate pore pressure buildup in the layer; determination from the ASCE National Convention, St. Louis.
recorded data of a practical threshold acceleration associated Bennett, M. J., McLaughlin, P. V., Sarmiento, J. S., and Youd, T. L. (1984).
with the generation of a significant pressure buildup in the layer, “Geotechnical investigations of liquefaction sites, Imperial Valley,
if the site has been instrumented with accelerometers and piezo- California.” Open-File Rep. 84-252, U.S. Geological Survey, Menlo
meters for an extended period; and counting of the number of Park, CA.

© ASCE 04017009-16 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2017, 143(6): 04017009


Bierschwale, J. G., and Stokoe, K. H. (1984). “Analytical evaluation of Convention, In Situ Tests to Evaluate Liquefaction Susceptibility,
liquefaction potential of sands subjected to the 1981 Westmorland ASCE, New York.
earthquake.” Geotechnical Engineering Rep. GR-84-15, Civil Engineer- Dobry, R., Thevanayagam, S., El-Sekelly, W., and Abdoun, T. (2015d).
ing Dept., Univ. of Texas, Austin, TX. “Large scale testing of effect of earthquake preshaking history on
Bommer, J. J., Douglas, J., and Strasser, F. O. (2003). “Style-of-faulting in the liquefaction resistance of a hydraulic filled clean sand.” in press.
ground-motion prediction equations.” Bull. Earthquake Eng., 1(2), EERI (Earthquake Engineering Research Institute). (1990). “Loma Prieta
171–203. earthquake reconnaissance report.” Oakland, CA.
Bonilla, M. G. (1960). “Landslides in the San Francisco south quadrangle.” El-Sekelly, W. (2014). “The effect of seismic pre-shaking history on the
California Open-File Rep. 60-15, U.S. Geological Survey, Menlo Park, liquefaction resistance of granular soil deposits.” Ph.D. dissertation,
CA. Rensselaer Polytechnic Institute, Troy, NY.
Boulanger, R. W., and Idriss, I. M. (2014). “CPT and SPT based liquefac- El-Sekelly, W., Abdoun, T., and Dobry, R. (2016a). “Liquefaction resis-
tion triggering procedures.” Rep. No. UCD/CGM-14/01, Center for tance of a silty sand deposit subjected to preshaking followed by
Geotechnical Modeling, Dept. of Civil and Environmental Engineering, extensive liquefaction.” J. Geotech. Geoenviron. Eng., 10.1061
Univ. of California, Davis, CA. /(ASCE)GT.1943-5606.0001444, 04015101.
Brady, A. G., Mark, P. N., Seekins, L. C., and Switzer, L. C. (1989). El-Sekelly, W., Dobry, R., Abdoun, T., and Steidl, J. H. (2016b). “Centri-
Downloaded from ascelibrary.org by University Of Florida on 10/31/17. Copyright ASCE. For personal use only; all rights reserved.

“Processed strong-motion records from the Imperial Wildlife liquefac- fuge modeling of the effect of preshaking on the liquefaction resistance
tion array, Imperial County, California, recorded during the Superstition of silty sand deposits.” J. Geotech. Geoenviron. Eng., 10.1061/(ASCE)
Hills earthquakes, November 24, 1987.” Open File Rep. 97-87, U.S. GT.1943-5606.0001430, 04016012.
Geological Survey, Menlo Park, CA. Faris, J. and de Alba, P. (2000). “National geotechnical experimentation site
CESMD. (2016). “Center for engineering strong motion data.” 〈http:// at Treasure Island, California.” National geotechnical experimentation
strongmotioncenter.org/〉 (Jan. 16, 2016). sites, ASCE, Reston, VA.
Cetin, K. O., et al. (2004). “Standard penetration test-based probabilistic Felzer, K. R., and Cao, T. (2008). “WGCEP historical California
and deterministic assessment of seismic soil liquefaction potential.” earthquake catalog, appendix H in 2007 working group on California
J. Geotech. Geoenviron. Eng., 10.1061/(ASCE)1090-0241(2004) earthquake probabilities, the uniform California earthquake rupture
130:12(1314), 1314–1340. forecast, version 2 (UCERF2).” U.S. Geological Survey Open-File
Cox, B. R. (2006). “Development of direct test method for dynamically Rep., 2007-1437-H, USGS, Menlo Park, CA.
assessing the liquefaction resistance of soils in situ.” Ph.D. thesis, Univ. Field, E. H., et al. (2013). “Uniform California earthquake rupture forecast,
of Texas, Austin, TX. version 3 (UCERF3)—The time-independent model.” U.S. Geological
Cox, B. R., et al. (2013). “Liquefaction at strong motion stations in the Survey Open-File Rep. 2013-1165, USGS, Menlo Park, CA.
2011 great east Japan earthquake with an emphasis in Urayasu City.” Finn, W. D. L. (1981). “Liquefaction potential: Developments since 1976.”
Earthquake Spectra, 29(S1), S55–S8. Proc., Int. Conf. on Recent Advances in Geotechnical Earthquake
Darendeli, M. B. (2001). “Development of a new family of normalized Engineering and Soil Dynamics, Missouri Univ. of Science and
modulus reduction and material damping curves.” Ph.D. thesis, Univ. Technology, Rolla, MO.
of Texas, Austin, TX. Finn, W. D. L., Bransby, P. L., and Pickering, D. J. (1970). “Effect of strain
de Alba, P., Benoit, J., Pass, D. G., Carter, J. J., Youd, T. L., and Shakal, A. history on liquefaction of sand.” J. Soil Mech. Found. D., 96(SM6),
F. (1994). “Deep instrumentation array at the Treasure Island naval 1917–1934.
station.” U.S. Geological Survey Professional Paper 1551-A, R. D. Fuhriman, M. D. (1993). “Crosshole seismic tests at two northern
Borcherdt, ed., U.S. Goverment Printing Office, Washington, DC. California sites affected by the 1989 Loma Prieta earthquake,”
Dobry, R., et al. (2000). “New site coefficients and site classification system M.S. thesis, Univ. of Texas, Austin, TX.
used in recent building code provisions.” Earthquake Spectra, 16(1), Gonzalez-Ortega, A., et al. (2014). “El Mayor-Cucapah (Mw 7.2) earth-
41–67. quake: Early near-field postseismic deformation from InSAR and
Dobry, R. (2010). “Comparison between clean sand liquefaction charts GPS observations.” J. Geophys. Res., 119(2), 1482–1497.
based on penetration resistance and shear wave velocity.” Proc., 5th Hauksson, E. J., et al. (2013). “Report on the August 2012 Brawley earth-
Int. Conf. on Recent Advances in Geotechnical Earthquake Engineering quake swarm in Imperial Valley, southern California.” Seismol. Res.
and Soil Dynamics and Symp. in Honor of Prof. I. M. Idriss, Missouri Lett., 84(2), 177–189.
Univ. of Science and Technology, Rolla, MO. Hayati, H., and Andrus, R. D. (2008). “Liquefaction potential map
Dobry, R., and Abdoun, T. (2015a). “An investigation into why liquefaction of Charleston, South Carolina based on the 1886 earthquake.” J. Geo-
charts work: A necessary step toward integrating the states of art and tech. Geoenviron. Eng., 10.1061/(ASCE)1090-0241(2008)134:6(815),
practice.” J. Soil Dyn. Earthquake Eng., 68, 40–56. 815–828.
Dobry, R., and Abdoun, T. (2015b). “Cyclic shear strain needed for lique- Hayati, H., and Andrus, R. D. (2009). “Updated liquefaction resistance cor-
faction triggering and assessment of overburden pressure factor K σ.” rection factors for aged sands.” J. Geotech. Geoenviron. Eng., 10.1061
J. Geotech. Geoenviron. Eng., 10.1061/(ASCE)GT.1943-5606 /(ASCE)GT.1943-5606.0000118, 1683–1692.
.0001342, 04015047. Heidari, T., and Andrus, R. D. (2012). “Liquefaction potential assessment
Dobry, R., Abdoun, T., Stokoe, K., II, Moss, R., Hatton, M., and El of pleistocene beach sands near Charleston, South Carolina.” J. Geo-
Ganainy, H. (2015c). “Liquefaction potential of recent fills versus natu- tech. Geoenviron. Eng., 10.1061/(ASCE)GT.1943-5606.0000686,
ral sands located in high-seismicity regions using shear-wave velocity.” 1196–1208.
J. Geotech. Geoenviron. Eng., 10.1061/(ASCE)GT.1943-5606 Holzer, T. L., and Youd, T. L. (2007). “Liquefaction, ground oscillation,
.0001239, 04014112. and soil deformation at the Wildlife array, California.” Bull. Seismol.
Dobry, R., Baziar, M. H., O’Rourke, T. D., Roth, B. L., and Youd, T. L. Soc. Am., 97(3), 961–976.
(1992). “Liquefaction and ground failure in the Imperial Valley during Holzer, T. L., Youd, T. L., and Hanks, T. C. (1989). “Dynamics of lique-
the 1979, 1981 and 1987 earthquakes.” Case studies of liquefaction and faction during the 1987 Superstition Hills, California, earthquake.”
lifeline performance during past earthquakes, T. D. O’Rourke and Science, 244(4900), 56–59.
M. Hamada, eds., National Center for Earthquake Engineering Hutton, K., Woessner, J., and Hauksson, E. (2010). “Earthquake monitor-
Research, Buffalo, NY. ing in southern California for seventy years (1932–2008).” Bull.
Dobry, R., Ladd, R. S., Yokel, F. Y., Chung, R. M., and Powell, D. (1982). Seismol. Soc. Am., 100(2), 423–446.
“Prediction of pore water pressure buildup and liquefaction of sands Idriss, I. M. (1990). “Response of soft soil sites during earthquakes.” Proc.
during earthquakes by the cyclic strain method.” NBS Build. Sci. H. Bolton Seed Memorial Symp., Univ. of California, Berkeley, CA.
Ser. 138, National Bureau of Standards, Gaithersburg, MD. Idriss, I. M., and Boulanger, R. W. (2008). “Soil liquefaction during
Dobry, R., Stokoe, K. H., Ladd, R. S., and Youd, T. L. (1981). “Liquefac- earthquakes.” Monograph MNO-12, Earthquake Engineering Research
tion susceptibility from s-wave velocity.” Proc., ASCE National Institute, Oakland, CA.

© ASCE 04017009-17 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2017, 143(6): 04017009


Ishihara, K. (1981). “Pore water pressure rises during earthquakes.” Proc., Salgado, R., Mitchell, J. K., and Jamiolkowski, M. (1998). “Calibration
Int. Conf. on Recent Advances of Geotechnical Earthquake Engineering chamber size effect on the penetration resistance of sand.” J. Geotech.
and Soil Dynamics, Missouri Univ. of Science and Technology, Rolla, Geoenviron. Eng., 10.1061/(ASCE)1090-0241(1998)124:9(878),
MO. 878–888.
Ishihara, K. (1985). “Stability of natural deposits during earthquakes.” Schmertmann, J. H. (1991). “The mechanical aging of soils.” J. Geotech.
Proc., 11th Int. Conf. on Soil Mechanics and Foundation Engineering, Eng., 10.1061/(ASCE)0733-9410(1991)117:9(1288), 1288–1330.
A.A. Balkema, Rotterdam, Netherlands. Seed, H. B. (1979). “Soil liquefaction and cyclic mobility evaluation for
Ishihara, K., Araki, K., and Bradley, B. A. (2011). “Characteristics of level ground during earthquakes.” J. Geotech. Eng. Div., 105(GT2),
liquefaction-induced damage in the 2011 great east Japan earthquake.” 201–255.
Proc., Int. Conf. on Geotechnics for Sustainable Development–Geotec Seed, H. B., and Idriss, I. M. (1970). “Soil moduli and damping factors
(Keynote Lecture), Fecon Corporation, Hanoi, Vietnam. for dynamic response analyses.” Rep. No. EERC 70-10, Earthquake
Johnson, C. E., and Hadley, D. M. (1976). “Tectonic implications of the Engineering Research Center, Univ. of California, Berkeley, CA.
Brawley earthquake swarm, Imperial Valley, California, January 1975.” Seed, H. B., and Idriss, I. M. (1971). “Simplified procedure for evaluating
Bull. Seismol. Soc. Am., 66(4), 1133–1144. soil liquefaction potential.” J. Soil Mech. Found. Div., 97(SM9),
Johnson, C. E., and Hill, D. P. (1982). “Seismicity of the Imperial Valley.” 1249–1273.
Downloaded from ascelibrary.org by University Of Florida on 10/31/17. Copyright ASCE. For personal use only; all rights reserved.

Geological Survey Professional Paper 1254, U.S. Geological Survey, Seed, H. B., Mori, K., and Chan, C. K. (1977). “Influence of seismic
U.S. Government Printing Office, Washington, DC. history on liquefaction of sands.” J. Geotech. Eng. Div., 103(GT4),
Kayen, R. R., et al. (2013). “Shear wave velocity-based probabilistic 257–270.
and deterministic assessment of seismic soil liquefaction potential.” Steidl, J. H., Civilini, F., and Seale, S. (2014). “What have we learned
J. Geotech. Geoenviron. Eng., 10.1061/(ASCE)GT.1943-5606 after a decade of experiments and monitoring at the NEES@UCSB
.0000743, 407–419. permanently instrumented field sites?” Proc., 10th National Conf. on
Kokusho, T., Yoshida, Y., and Esashi, Y. (1982). “Dynamic properties of Earthquake Engineering, Earthquake Engineering Research Institute,
soft clay for wide strain range.” Soils Found., 22(4), 1–18. Oakland, CA.
Liao, Y., and Meneses, J. (2013). “Engineering characteristics of ground Steidl, J. H., and Seale, S. (2010). “Observations and analysis of ground
motion records from the 2010 Mw 7.2 El Mayor-Cucupah earthquake motion and pore pressure at the NEES instrumented geotechnical field
in Mexico.” Earthquake Spectra, 29(1), 177–205. sites.” Proc., 5th Int. Conf. on Recent Advances in Geotechnical Earth-
Liu, N., and Mitchell, J. K. (2006). “Effects of nonplastic fines on shear quake Engineering and Soil Dynamics, San Diego.
wave velocity based liquefaction procedures.” J. Geotech. Geoenviron. Thevanayagam, S., et al. (2009). “Laminar box system for 1-g physical
modeling of liquefaction and lateral spreading.” Geotech. Test. J.,
Eng., 10.1061/(ASCE)1090-0241(2006)132:8(1091), 1091–1097.
32(5), 438–449.
Mitchell, J. K., and Solymar, Z. V. (1984). “Time-dependent strength gain
Vucetic, M. (1986). “Pore pressure buildup and liquefaction at level
in freshly deposited or densified sand.” J. Geotech. Eng., 10.1061
sand sites during earthquakes.” Ph.D. thesis, Rensselaer Polytechnic
/(ASCE)0733-9410(1984)110:11(1559), 1559–1576.
Institute, Troy, NY.
NRC (National Research Council). (1985). Liquefaction of soils
Wildlife liquefaction array. (2016). “NEES at UCSB.” 〈http://www.nees
during earthquakes, R. V. Whitman, ed., National Academy Press,
.ucsb.edu/facilities/wla〉 (Jan. 16, 2016).
Washington, DC.
Yasuda, S., and Tohno, I. (1988). “Sites of reliquefaction caused by the
Olsen, K. B., Archuleta, R. J., and Matarese, J. R. (1995). “Three- 1983 Nihonkia-Chubu earthquake.” Soils Found., 28(2), 61–72.
dimensional simulation of a magnitude 7.75 earthquake on the San
Youd, T. L. (1977). “Packing changes and liquefaction susceptibility.”
Andreas fault in southern California.” Science, 270(5242), 1628–1632. J. Geotech. Eng. Div., 103(GT8), 918–922.
Pass, D. (1994). “Soil Characterization of the deep accelerometer site at Youd, T. L., et al. (2001). “Liquefaction resistance of soils: Summary report
Treasure Island, San Francisco, California.” M.S. thesis, Univ. of from the 1996 NCEER and 1998 NCEER/NSF workshops on evalu-
New Hampshire, Durham, NH. ation of liquefaction resistance of soils.” J. Geotech. Geoenviron.
Petersen, M. D., et al. (2008). “Documentation for the 2008 update of Eng., 10.1061/(ASCE)1090-0241(2001)127:10(817), 817–833.
the United States national seismic hazard maps.” Open-File Rep. Youd, T. L., et al. (2003). “Closure of ‘Liquefaction resistance of soils:
2008-1128, U.S. Geological Survey, Reston, VA. Summary report from the 1996 NCEER and 1998 NCEER/NSF work-
Porcella, R. L., Etheredge, E., Maley, R., and Switzer, J. (1987). “Strong- shops on evaluation of liquefaction resistance of soils’ by Youd, T. L.
motion data from the Superstition Hills earthquakes 0154 and 1315 et al. (2001).” J. Geotech. Geoenviron. Eng., 10.1061/(ASCE)1090
(GMT), Nov. 24, 1987.” Open-File Rep. 87-672, U.S. Geological -0241(2003)129:3(284), 284–286.
Survey, Menlo Park, CA. Youd, T. L. (2013). “Discussion of ‘Examination and reevaluation of
Power, M. S., Egan, J. A., Shewdridge, S. E., deBecker, J., and Faris, J. R. SPT-based liquefaction triggering case histories’ by Boulanger, R. W.,
(1998). “Analysis of liquefaction-induced damage on Treasure Island.” Wilson, D. W., and Idriss, I. M. (2012).” J. Geotech. Geoenviron. Eng.,
Professional Paper 1551-B, U.S. Geological Survey, Menlo Park, CA. 10.1061/(ASCE)GT.1943-5606.0000876, 1998–2000.
Press, F., and Siever, R. (1978). Earth, 2nd Ed., W.H. Freeman and Co., Youd, T. L., and Craven, T. N. (1975). “Lateral stress in sands during cyclic
San Francisco. loading.” J. Geotech. Eng. Div., 101(GT2), 217–221.
Pyke, R. (2003). “Discussion of ‘Liquefaction resistance of soils: Summary Youd, T. L., and Hoose, S. N. (1978). “Historic ground failures in Northern
report from the 1996 NCEER and 1998 NCEER/NSF workshops on California associated with earthquakes.” Geological Survey Profes-
evaluation of liquefaction resistance of soils’ by Youd, T. L. et al. sional Paper 993, U.S. Government Printing Office, Washington, DC.
(2001).” J. Geotech. Geoenviron. Eng., 10.1061/(ASCE)1090-0241 Youd, T. L., Steidl, J. H., and Steller, R. A. (2007). “Instrumentation of the
(2003)129:3(283), 283–284. Wildlife liquefaction array.” Proc., Int. Conf. on Earthquake Geotech-
Rockwell, T., and Klinger, Y. (2013). “Surface rupture and slip distribution nical Engineering, Thessaloniki, Greece.
of the 1940 Imperial Valley earthquake, Imperial Fault, Southern Youd, T. L., and Wieczorek, G. F. (1984). “Liquefaction during the 1981
California: Implications for rupture segmentation and dynamics.” Bull. and previous earthquakes near Westmorland, California.” Open-File
Seismol. Soc. Am., 103(2A), 629–640. Rep. 84-680, U.S. Geological Survey, Menlo Park, CA.
Salgado, R., Boulanger, R. W., and Mitchell, J. K. (1997). “Lateral stress Zeghal, M., and Elgamal, A. W. (1994). “Analysis of site liquefaction using
effects on CPT liquefaction resistance correlations.” J. Geotech. Geo- earthquake records.” J. Geotech. Eng., 10.1061/(ASCE)0733-9410
environ. Eng., 10.1061/(ASCE)1090-0241(1997)123:8(726), 726–735. (1994)120:6(996), 996–1017.

© ASCE 04017009-18 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2017, 143(6): 04017009

You might also like