LIQUEFACTION-INDUCED GROUND SURFACE DISRUPTION
T. Leslie Youd
Professor of Civil Engineering
Brigham Young University
Provo, Utah 84602
and
Christopher T. Garris
Senior Staff Engineer
Woodward-Clyde Consultants
Salt Lake City, Utah 84047
ABSTRACT
Although liquefaction is a major cause of earthquake damage, little harm occurs unless
liquetaction generates some form of ground surface disruption or ground failure. Thus. ability
to accurately predict potential for ground surface disruption is a major concern for
hnologists charged with safe siting of constructed works. Ishihara presented preliminary
al criteria for assessing. the potential for ground surface disruption at liquefaction sites
Those criteria are based on relationships between the thickness of liquefiable sediment beneath
a site and the corresponding thickness of overlying non-liquefiable soil. The purpose of this
study is to further evaluate and verify these criteria by testing them against thicknesses calculated
from a wide range of earthquake and site conditions. The newly developed data lead to the
following conclusions: (1) For sites not susceptible to ground oscillation or lateral spread. the
thickness bounds proposed by Ishihara appear to be valid. (2) For sites susceptible to ground
oscillation or lateral spread, the bounds suggested by Ishihara are not sufficient for predicting
ground surface disruption.
27INTRODUCTION
Although liquefaction is a major cause of earthquake damage, little harm occurs unless the
liquefied condition leads to some form of ground surface disturbance or ground failure. ‘Thus
ability to accurately predict potential for ground surface disruption is a major concern for
geotechnologists charged with safe siting of constructed works
Ishihara (1985) presented preliminary empirical criteria for assessing the potential for ground
surface disruption at liquefaction sites . Those criteria are based on relationships between
thicknesses of liquefiable layers beneath a site, and corresponding thicknesses of the overlying,
non-lquefiable soil (Figures 1 and 2). Ishihara developed these relationships using analyses of
field data from sites with or without surface liquefection effects. He evaluated field
reconnaissance and borehole data from three large earthquakes--1964 Niigata. Japan (M=7.5),
1983 Nihonkai Chubu, Japan (M=7.7), and 1976 Tangshan, China (M=7.8). Using borehole logs
from sites with or without surface effects of liquefaction, Ishihara calculated thicknesses of
liquefiable layers and overlying non-liquefiable layers utilizing criteria published in the Japanese
bridge code (Japan Highway Association, 1980). Ishihara plotted those data as shown in
Figure I. and from those data sketched the bounds reproduced in Figure 2. These thickness
relationships have been used by many engineers to assess the potential for liquefaction-induced
damage to constructed works
The purpose of this study is to further evaluate and verify Ishihara’s criteria by testing the
proposed bounds against thicknesses of layers calculated from borehole data taken from field
investigations following several additional earthquakes. These investigations cover a wider range
of earthquake magnitudes and site conditions than the those considered by [shihara.
LAYER THICKNESSES FROM SITES OF PAST LIQUEFACTION
To test the bounds published by Ishihara (1985), we calculated layer thicknesses from data on
308 borehole logs taken from sites with or without surface effects of liquefaction shaken by 15
different earthquake ranging in magnitude from 5.3 to 8.0. The sites and the earthquakes that
shook them are listed in Table 1. Data from these sites were compiled by Bartlett and Youd
(1992) and by Loertscher (1994), Some of the data had been previously collected by Seed and
others (1983), Liao (1986). and Ambraseys (1988). Because the "simplified procedure developed
by Seed and his colleagues (Seed and others, 1985) is more widely used in the United States than
other procedures to evaluate I1quefaction hazard, we utilized the simplified procedure to calculate
layer thicknesses. For most sites, the simplified procedure and the procedure given in the
Japanese bridge code yield comparable thicknesses,
For the evaluation, we divided surface effects or lack of effects into four categories. (1) no
observed surtace effects: (2) typical liquefaction effects, including sand boils and small ground
fissures, but without noticeable lateral ground displacement; (3) typical liquefaction effects plus
effects generated by ground oscillation, such as buckled pavements, curbs, broken pipelines. ete.
28but without a consistent pattern of permanent lateral displacements; and (4) surface effects
generated by laterals spreads, including a consistent pattern of permanent ground displacements
accompanied by ground fissures and usually by sand boils.
Layer thickness combinations from our analyses are plotted on Figures 3 to 10. Figure 3 shows
thicknesses for earthquakes with magnitudes ranging from 5.3 to 6.6 and estimated peak
accelerations, aq... ranging from 0.13 g to 0.22 g, Although there were only two sites with
typical liquefaction effects (sand boils and small fissures) in this data set, the plotted thicknesses
agree well with Ishihara’s bound for 0.2 g peak acceleration, Most of the data on this plot are
from the Marina District in San Francisco which was shaken by the 1957 Daly City (M = 5.3)
without generating observed surface effects of liquefaction. This same area was severely
disturbed by liquefaction effects during the much larger 1989 Loma Prieta event (M = 6.9).
Figure 4 shows data from several earthquakes in the magnitude 5.9 to 7.0 range with estimated
pax Fanging from 0.26 g to 0.35 g. The data from sites with surface disruption are correctly
predicted by Ishihara’s bound with the exception of one badly misclassified datum noted as Vail
site, and three points plotted noted as Marina District site where liquefaction-induced ground
oscillation occurred during the 1989 Loma Prieta earthquake. The borehole at the Vail site
penetrated 4.9 m of non-liquefiable sediment before penetrating only 0.9 m of liquefiable silty
sand. Although several sand boils erupted within about 30 m of the drill hole during the 1981
Westmorland, California earthquake, the drill site was near the eastern margin of the disturbed
area, Perhaps the liquefiable sediment at the borehole locality was anonymously thin, or perhaps
the conduits for the sand boils drifted laterally from nearby thicker liquefiable deposits. In any
event, information from field measurements and observations may be expected to contain a small
percentage of anomalous ot spurious data because of field inhomogeneities. The data from the
Marina District indicate that Ishihara’s bounds may not be valid for ground oscillation sites.
Figure 5 shows thickness data for earthquakes with magnitudes between 5.9 and 7.0, but with
estimated a,,, between 0.37 and 0.55 Data on this plot come primarily from sites of lateral
spreads generated by the 1971 San Fernando and 1979 Imperial Valley earthquakes (M = 6.4 and
6.6. respectively). Many of the data plotted on this figure are poorly predicted by Ishihara’s
bounds. Most of the points plotting to the right of the bound on this figure are from the Jensen
water filtration plant site in Sylmar, Califorma that was badly damaged by a lateral spread during
the 1971 San Fernando earthquake. The Jensen facility was constructed on non-liquefiable
compacted fill, as thick as 15 m, overlying a 1- to 5-m thick layer of liquefiable alluvial silt and
sand. Lateral displacements at the filtration plant ranged from 0.1 m to more than a meter,
accompanied by ground fissures and sporadic sand boil eruptions ("Rourke and others, 1992,
Youd 1973). These data indicate that lateral spread can generate disruption to the ground surface
for liquefiable layer thicknesses well below the bounds suggested by Ishihara
Figure 6 shows thickness data for earthquakes with magnitudes between 5.9 and 7.0, but with
estimated a,,,, between 0.56 g and 0.78 g. Only one datum on this plot, noted as J.P Castro site,
lies well below Ishihara’s bound. That datum 1s from a site of ground oscillation where the
thickness of the liquetiable layer was about 0.6 m and the thickness of the surface layer was
6.1 m, Surface effects of liquefaction at the site included sand boils, ground fissures and buckled
‘curbs in close proximity to open fissures (Youd and Keefer, in press). The latter features are
29clear indications of ground oscillations, which characteristically buckle rigid linear bodies, such
as curbs, due to compressional impacts, while pulling apart adjacent ground and overlying
structures during extensional phases of oscillation. The location of this single point below
Ishihara’s curve on Figure 6 along with the three points from the Marina District plotted on
Figure 5 indicate that ground oscillation, as well as lateral spread, can generate disruptions to the
ground surface in areas predicted to be free of surface disruption by Ishihara’s bounds.
Figures 7 and 8, from the 1964 Niigata and 1988 Nihonkai-Chubu earthquakes, respectively.
contain several points that lie beyond Ishihara’s bounds. All of these apparently misclassified
data are from areas affected by lateral spreads. These data further support the postulate that
lateral spreads can produce surface disruptions that are not correctly predicted by Ishihara’s
bounds
INTERPRETATIONS FROM RESULTS.
Although we compiled a rather large data set, there are some biases in these data. For example,
post-earthquake drilling investigations in Noshiro, Japan, were conducted primarily at sites where
significant lateral displacement had occurred. Thus, for that locality, few sites were drilled where
surface effects were not observed or where liquefaction effects were observed but lateral
displacement did not occur. For the smaller magnitude earthquakes (M<6), nearly all of the
borehole data are from sites where liquefaction was predicted, but surface effects were not
evident. Many of these sites are in the Marina District of San Francisco where surface effects
of liquefaction occurred during the 1989 Loma Prieta earthquake (M=6.9), but did not develop
during the smaller 1957 Daly City earthquake (M=5.3). Although minor surface effects,
primarily small fissures and sand boils, have been observed in the epicentral region of many small
earthquakes (5