Académique Documents
Professionnel Documents
Culture Documents
www.elsevier.com/locate/soildyn
Abstract
Conventionally, evaluation of liquefaction potential of loose saturated cohesionless deposits as specified in Japanese design codes employs
peak ground acceleration (PGA). However, recent large-scale earthquakes in Japan revealed that liquefaction at some sites did not occur even
though large PGAs were recorded at or near these sites. As an alternative approach, an evaluation procedure based on peak ground motion
parameters, i.e. incorporating both PGA and the peak ground velocity (PGV), is proposed. By performing parametric studies using one-
dimensional seismic response analysis and formulating regression models, seismic-induced shear stresses within the deposit are expressed in
terms of peak ground motion parameters at the surface, and these are used to calculate the factor of safety against liquefaction. Application to
case histories in Japan indicates that the proposed two-parameter equation can adequately account for the occurrence and non-occurrence of
liquefaction at various sites as compared to the conventional PGA-based approach. Moreover, analyses of several strong motion records at
various sites show that liquefaction may occur when PGAR150 gal and PGVR20 kine, indicating that these values can serve as thresholds
in assessing the possible occurrence of liquefaction.
q 2005 Elsevier Ltd. All rights reserved.
Fig. 1. Guidelines for evaluating liquefaction potential for highway bridges (after Ref. [1]).
design codes in Japan [2–4] make use of similar empirical the gravitational acceleration, is different from the observed
formulas to estimate the cyclic strength of the target soil. one. Moreover, since the onset of liquefaction depends on
As for estimates of shear stress induced during seismic the number of loading cycles, the design codes assume that
loading, most design codes in Japan use an equation similar this level of acceleration is repeated in 15–20 cycles. In
to that initially proposed by Seed and Idriss [5]. In this contrast, the observed acceleration simply reveals the
equation, the cyclic shear stress ratio developed at a maximum value and does not show anything about the
particular depth beneath a level ground surface is expressed number of cycles.
in terms of the design seismic coefficient as shown on the With the advent of highly sensitive seismometers for use
right side of Fig. 1. The seismic coefficient is a function of in seismic monitoring networks in Japan, strong motion
the design peak ground acceleration (PGA). records showing short duration impulse of high frequency
In some practical applications though, the extent of (called acceleration spikes) are not uncommon. When used
liquefaction is assessed by using the observed peak ground with conventional liquefaction evaluation procedure, these
acceleration instead of the design acceleration. However, large acceleration peaks would provide unusually high
the design acceleration, being typically about 15–20% of intensities, and therefore, would overestimate the shear
R.P. Orense / Soil Dynamics and Earthquake Engineering 25 (2005) 225–240 227
stresses induced in soil deposits. For example, PGAs as high the occurrence or non-occurrence of liquefaction fairly
as 1000 gal (1 galZ1 cm/s2) have been observed during well, as compared to the more popular PGA-based approach
recent earthquakes in Japan. When these high values of specified in many Japanese codes.
observed peak ground acceleration are substituted into the Note that the discussion presented herein concentrates on
empirical design formula, the detection of liquefaction is the estimation of seismic-induced shear stresses. Although
often incorrectly judged. estimation of cyclic shear strength is important for liquefac-
The empirical methods of estimating seismic-induced tion potential evaluation, further research on this is
shear stresses based on PGAs, which have been incorpor- warranted and therefore is outside the scope of the present
ated in various Japanese codes, have been extensively research. Thus, the applicability of the proposed method may
validated using data mostly from the 1964 Niigata Earth- be confined to the Japanese practice only, which strongly
quake and the 1983 Nihonkai-Chubu Earthquake, both of relies on empirical equations to estimate cyclic strengths.
which were far-field earthquakes. However, since the
methods do not consider the frequency characteristics of
strong motion records, they do not work well on recent
large-scale earthquakes with near-field sources, which are 2. Conventional method of estimating maximum shear
characterized by high-frequency components. stress during earthquakes
Because of these shortcomings, alternative methods of
estimating liquefaction potential based on other ground In the original Seed-Idriss [5] simplified procedure, a soil
motion indices have been introduced. For example, column is considered as a rigid body. As the seismic loading
Towhata et al. [6] employed the spectrum intensity (SI), is excited at the base of the soil column, the shear wave
defined as the average value of the velocity response propagates to the ground surface. The shear stress generated
spectrum with 20% damping over a range of natural period in the soil column can be calculated by the following
from 0.1 to 2.5 s, as index for liquefaction detection. Kayen equation
and Mitchell [7] used the Arias Intensity, an energy-based amax
ðtmax Þr Z sv (2)
measure of earthquake shaking intensity, to assess liquefac- g
tion potential.
where (tmax)r is the maximum shear stress for rigid body, sv
Another commonly used ground motion index to indicate
is the total overburden pressure, amax is the peak horizontal
the severity of earthquake shaking is the peak ground
acceleration on the ground surface, and g is the acceleration
velocity (PGV). Studies made by Midorikawa and Waka-
due to gravity.
matsu [8] in sites of past liquefaction occurrence showed
In reality, however, soil behaves as a deformable body
that a PGV greater than 15 kine (1 kineZ1 cm/s) suggests
instead of as a rigid one. Hence, the rigid body shear stress is
liquefaction in the subsoil, while there is no clear correlation
reduced with a correction factor to give the deformable body
between liquefaction and maximum acceleration. More-
shear stress (tmax)d. The correction factor is called the stress
over, the close correspondence of the Japan Meteorological
reduction factor, rd, and measures the attenuation of peak
Agency (JMA) intensities and peak ground velocities [9]
shear stress with depth due to the non-elastic behavior of
shows that peak ground velocities are consistent with the
soil, i.e.
algorithm used by JMA in defining seismic intensities.
Another consideration in the choice of peak ground velocity ðtmax Þd
rd Z (3)
is that the kinetic energy available for inducing damage is ðtmax Þr
well-characterized by PGV. It has been mentioned that the
breakdown of soil structure that results in liquefaction is It was found that the parameter rd reduces from a value of
fundamentally more dependent on input energy than on a unity at the surface to typically between 0.3 and 0.7 at a soil
single level of acceleration [10]. depth of about 30 m (w100 ft), as shown in Fig. 2. Thus, in
As an alternative to the conventional approach of the simplified procedure, the maximum shear stress is
estimating seismic-induced shear stresses based on a single computed as
ground motion parameter, a method which combines both a
tmax Z rd sv max (4)
PGA and PGV is proposed in this paper. With this two- g
parameter approach, the proposed method can consider both Several empirical expressions for the reduction factor, rd,
near-source ground motions (often dominated by short- have been proposed by various researchers [11]. The
duration pulses and where PGV appears to be a robust equation widely used in Japan is based on studies made
measure of intensity for strong shaking) and far-source by Iwasaki et al. [12]. By performing parametric site
ground motions (where PGA is a good indicator, as proven response analyses on alluvial deposits, they showed that the
by the conventional procedure). These PGA and PGV-based stress reduction factor can be expressed in terms of depth, z,
shear stresses are then used to estimate the liquefaction as:
potential of soil deposits. Application to various sites in
Japan indicates that the proposed method evaluates rd Z 1:0 K 0:015z (5)
228 R.P. Orense / Soil Dynamics and Earthquake Engineering 25 (2005) 225–240
Depth (m)
10 10
Iwasaki et
al. (1978)
12 12
14 14
16 16
Iwasaki et
al. (1978)
18 18
20 20
Model 2 (2E=300 gal) Model 4 (2E=300 gal)
Fig. 3. Computed reduction factor corresponding to different T0 (in sec) for: (a) Model 2; (b) Model 4 profiles.
acceleration, PGA, mass density, r, shear wave velocity, VS, shear wave velocity VSO400 m/s, as usually employed in
natural circular frequency, u and depth, z, i.e. Japanese practice. The dynamic characteristics of each layer
tmax Z f ðPGA; r; VS ; u; zÞ (9) in the soil profile (i.e. strain-dependency of modulus and
damping) were obtained from laboratory tests or, when not
Next, the following equations are introduced: uZamax/ available, were estimated from relations recommended by
vmax for harmonic loading, as given in Eq. (7); overburden various researchers [15,16].
pressure, svZrz; and fundamental period of the ground, Five strong motion records, with predominant periods
TgZ4H/VS, where H is the thickness of the deposit. ranging from T0Z0.35 to 1.95 s were used as input motion
Moreover, some researchers [13] also contend that the at the bedrock. These include (1) Todai Seiken CO40NS
maximum shear stress is affected by the magnitude of motion observed during the 1987 Chibaken-toho oki
earthquake, M. Other factors, such as shear wave velocity Earthquake (T0Z0.35 s); (2) JMA Kobe NS motion
profile of the site and dynamic soil properties, may have recorded during the 1995 Hyogoken-Nambu Earthquake
profound effect on this value. Nevertheless, for practical (T0Z0.73 s); (3) MYGH01 EW motion monitored by Kik-
purposes, Eq. (9) can be re-written as Net during the 2001 Iwateken-Nambu Earthquake (T0Z
tmax Z f ðPGA; sv ; Tg ; PGA=PGV; M; zÞ (10) 1.11 s); (4) KNK GL-100 m NS motion obtained during the
1995 Hyogoken Nambu Earthquake (T0Z1.52 s); and (5)
Eq. (10) can serve as basis in defining the appropriate modified Itajima Bridge motion registered during the 1968
equation for the maximum shear stress induced by earth-
Hyuganada Earthquake (T0Z1.95 s). The time histories of
quake shaking within a deposit.
these acceleration records are shown in Fig. 7 while the
corresponding Fourier spectra are illustrated in Fig. 8. Both
low-pass and high-pass filters were used in the analysis, and
4. Estimation of maximum shear stress only the components of motion between 0.1 and 20 Hz were
considered.
The procedure proposed herein makes use of Eq. (10), In performing seismic response analyses, five intensities
which considers only essential parameters readily obtain- of bedrock outcrop input motion were used, i.e. 2EZ100,
able at the site before and during earthquake shaking. For 150, 200, 250 and 300 gal. Moreover, five levels of
this purpose, several series of one-dimensional seismic magnitudes were considered, i.e. MZ5.0, 5.5, 6.5, 7.5,
response analyses were performed using the computer and 8.0. These earthquake magnitudes were incorporated in
program SHAKE [14]. In the analyses, the ground models the seismic response analysis through the ratio of the
shown in Fig. 6 are employed. These five profiles, which equivalent uniform shear strain and the maximum shear
represent actual sites in the Kanto region (Japan), have strain. In SHAKE implementation, the shear strain ratio is
natural periods ranging from TgZ0.32 to 1.80 s. The shear estimated as (MK1)/10, as suggested in the AIJ Code [2].
wave velocities indicated in the figure were obtained by PS Thus, considering five model grounds, five acceleration
logging. Engineering bedrock is taken as the layer with records, five intensities of input motion, and five levels of
230 R.P. Orense / Soil Dynamics and Earthquake Engineering 25 (2005) 225–240
35 35
GL-6.5m GL-6.5m
30 30
τmax (kPa)
τmax (kPa)
25 25
20 20
Based on Eqs. (4) and (5) Based on Eqs. (4) and (5)
15 Model 2 15 Model 2
(PGA=200gal) (PGA=200gal)
10 10
0 1 2 3 4 5 6 0 1 2 3 4 5 6
Period (sec) Period (sec)
50 50
GL-10.5m GL-10.5m
40 40
τmax (kPa)
τmax (kPa)
20 20
Model 2 Model 2
(PGA=200gal) (PGA=200gal)
10 10
0 1 2 3 4 5 6 0 1 2 3 4 5 6
Period (sec) Period (sec)
70 70
GL-14.5m Model 2 GL-14.5m Model2
60 (PGA=200gal) 60 (PGA=200gal)
τmax (kPa)
τmax (kPa)
50 50
40 40
Based on Eqs. (4) and (5)
Based on Eqs. (4) and (5)
30 30
20 20
0 1 2 3 4 5 6 0 1 2 3 4 5 6
Period (sec) Period (sec)
Fig. 4. Relations between tmax and T0 at different depths for PGAZ200 gal Fig. 5. Relations between tmax and T0 at different depths for PGAZ200 gal
(Model 2). (Model 4).
earthquake magnitude, a total of 625 cases were analyzed. given by the following equation
For each case, the maximum shear stresses developed at
PGA
selected points within the profiles were considered. Shear tmax Z rd* sv (11)
g
stresses that develop only on the upper 20 m of each profile
were analyzed because most Japanese design codes limit the where g is the acceleration due to gravity (Z980 gal) and
evaluation of liquefaction potential to within 20 m from the the modified stress reduction coefficient, rd* is given by:
ground surface. In addition, peak ground acceleration and
PGA
peak ground velocity for each case were calculated. The *
lnðrd Þ Z 0:01034 C 0:00321 z (12)
velocity was computed by integrating the acceleration-time PGV
history with components of periods longer than 10 s with a correlation coefficient rZ0.90. In the above
removed in order to avoid baseline error. equations, PGA is in gal (cm/s2), PGV is in kine (cm/s), z
After performing parametric studies, multiple regression in m, while sv and tmax are in kPa. Note that Eq. (11) is
analyses were conducted to determine the best expression similar in form to that of Eq. (4). Consequently, rd* is also
for the maximum shear stress, tmax, at any point within the dimensionless, similar to the reduction factor in the
profile. In addition to those specified in Eq. (10), the conventional method (Eq. (5)).
predominant period of input motion, T0, was included as an Regression analyses showed that earthquake magnitude
independent parameter in the regression model. Based on does not have much effect on the maximum shear stress.
regression analysis, the best correlation achieved is that This is because seismic-induced shear stress obtained by
R.P. Orense / Soil Dynamics and Earthquake Engineering 25 (2005) 225–240 231
SHAKE is not highly sensitive to the effective strain used in which would indicate that either or both T0 and Tg should
the analysis. This is the reason why a strain ratio between appear in the regression model.
0.55 and 0.65 is usually adequate in most analyses, with the For comparison purposes, shear stresses estimated using
higher value appropriate for giving more uniform strain peak ground motions (Eqs. (11) and (12)) and those based
histories. on conventional method (Eqs. (4) and (5)) are plotted with
It is worthy to mention that Eq. (12) was obtained for a respect to the shear stresses computed by the program
wide range of predominant periods of strong motion, T0, and SHAKE. The plots are shown in Fig. 9. Note that for tmax
ground profiles, Tg. However, no pattern was observed estimated using the proposed method (Fig. 9a), almost all
160 160
τmax (SHAKE)
τmax (SHAKE)
120 120
80 80
40 40
N=12125 N=12125
R2 =0.932 R2 = 0.706
0 0
0 40 80 120 160 200 0 40 80 120 160 200
τmax (Estimated) τmax (Estimated)
Fig. 9. Comparison between computed (SHAKE) and estimated shear stresses based on: (a) proposed method; (b) conventional method.
R.P. Orense / Soil Dynamics and Earthquake Engineering 25 (2005) 225–240 233
the data points lie within G30% of the computed values, Shear Stress (kPa)
whereas based on conventional approach (Fig. 9b), the 0 20 40 60 80 100
estimated values obtained are much larger than the 0
computed ones, with more data scattering and thus leading PGV:
to lesser value of regression coefficient. Such overestima-
tion in shear stress when using the conventional method is 15 kine
4
consistent with that observed in Figs. 4 and 5.
20 kine
To illustrate the concept of the modified stress reduction
factor, vertical distributions of rd* with respect to various 30 kine
ratios of PGA/PGV are shown in Fig. 10. It can be seen that 8
rd* takes a value of 1.0 at the ground surface and decreases
Depth (m)
60 kine
with depth, with the attenuation rate decreasing as the PGA/
PGV value decreases. Apparently, the reduction factor Conventional
given by the conventional method (Eq. (5)) is the upper 12
bound.
The effect of PGV on the distribution of tmax is next
considered. For this purpose, a uniform level ground with
mass density rZ2.0 g/cm3 is employed, with the PGA 16
assumed to be constant at 300 gal. The distribution of tmax
with depth is shown in Fig. 11 for various values of PGV,
ranging from 15 to 60 kine. Compared to the conventional 20
method, it can be seen that although PGA is constant, the
stress distribution changes with PGV, with lower shear Fig. 11. Example calculation showing the effect of PGV on the vertical
stresses associated with smaller values of PGV. Again, distribution of maximum shear stress.
=10
8
Depth (m)
τmax(Estimated)
τmax(Estimated)
(PGA=200gal) (PGA=200gal)
τmax(SHAKE)
τmax(SHAKE)
1.5 1.5
1.0 1.0
τmax(Estimated)
(PGA=200gal) (PGA=200gal)
τmax(SHAKE)
τmax(SHAKE)
1.5 1.5
1.0 1.0
τmax(Estimated)
(PGA=200gal) (PGA=200gal)
τmax(SHAKE)
τmax(SHAKE)
1.5 1.5
1.0 1.0
Fig. 12. Relations between ratio of computed (SHAKE) and estimated stresses and T0 at different depths for PGAZ200 gal: (a) Model 2; (b) Model 4.
5.1. Application to case histories by block dots) and non-liquefaction (white dots) for both
methods are near FLZ1.0, indicating the applicability of the
To confirm its validity, an evaluation of the liquefaction proposed method. It is also worth mentioning that the
potential is performed on 44 sites during 12 earthquakes number of non-liquefied points with FL!1.0 appear to be
which shook Japan on or before 2000 and where the less for the proposed method as compared to that using the
occurrence or non-occurrence of soil liquefaction had been conventional procedure.
confirmed. The details of each boring data, as well as the
observed peak ground accelerations and peak ground 5.2. Application to recent earthquakes in Japan
velocities monitored at seismic monitoring stations nearest
the site, are given in Table 1. Moreover, the depth of ground On May 26, 2003, an earthquake of magnitude MZ7.0
water table and estimated liquefaction depths based on occurred off the coast of Miyagi Prefecture, with focal
published records are given in Table 1. depth of about 71 km. This earthquake was officially
Both conventional and proposed methods were employed called the 2003 Miyagiken-oki Earthquake. Two months
in calculating FL at each site. For both methods, the cyclic later, on July 26, a series of powerful earthquakes
strength ratio, R, of the deposit was estimated using the occurred inland in the northern portion of Miyagi
procedure incorporated in the earthquake-resistant design Prefecture, with the main shaking registering a magnitude
specifications for highway bridges [1] and summarized in MZ6.2. The main shaking of this earthquake, officially
Fig. 1. Note that in this procedure, the important parameters called the 2003 Miyagiken-Hokubo Earthquake, was
in calculating R are the SPT N-value, mean grain size (D50), preceded by foreshocks, the largest of which occurred
fines content (Fc) and effective overburden pressure, sv0 . about 7 h earlier with a magnitude MZ5.5. It was also
Thus, in calculating FL for each method, the cyclic shear followed by several aftershocks, with the largest having a
strength ratios, R, are similar, while the earthquake-induced magnitude MZ5.3 about 10 h after the main shock. The
shear stress ratios, L, are different. foreshock–main shock–aftershock events all had shallow
The calculated values of FL based on conventional foci, about 12 km from the surface. The epicenters of the
method and proposed method are shown in Fig. 13. It can be main shocks of these two seismic events are shown in
seen that the boundary separating liquefaction (represented Fig. 14.
R.P. Orense / Soil Dynamics and Earthquake Engineering 25 (2005) 225–240 235
Table 1
Data used in case history analyses for liquefaction evaluation
During the 5/26 Miyagiken-oki earthquake, JMA seismic Although these two earthquakes are characterized by
intensity of Lower 6 was recorded over wide zones in Iwate different earthquake mechanisms, unusually large surface
and Miyagi Prefectures. Similarly, several areas in accelerations were recorded at seismic stations near the
Miyagi Prefecture underwent severe shaking during 7/26 epicenter of both earthquakes. During the 5/26 earthquake,
Miyagiken-Hokubu Earthquake, with maximum intensity of horizontal accelerations greater than 1000 gal were
Upper 6 recorded in regions adjacent to the epicenter. recorded at JMA Oofunato station (1105 gal), K-Net Oshika
236 R.P. Orense / Soil Dynamics and Earthquake Engineering 25 (2005) 225–240
4
4
8
Depth (m)
Depth (m)
12
12
16
16
20 20
Fig. 13. Calculation of FL based on (a) conventional method; (b) proposed method.
station (1112 gal) and K-Net Kamaishi station (1038 gal). were not observed at the six sites, it was assumed that
On the other hand, horizontal accelerations exceeding liquefaction did not occur here.
800 gal were registered at JMA Kashimadaimachi station Fig. 15 shows the results of liquefaction potential
(1605 gal) and JMA Yamoto station (850 gal) during the analysis using the conventional method based on PGA
7/26 tremor [17,18]. only and using the proposed approach based on two peak
Following the two earthquakes, reconnaissance investi- ground motion indices. It can be seen that while the
gations were performed on six sites where oil tanks and oil conventional method shows numerous points corresponding
refineries are located. These include one site each in to FL values !1.0 even when liquefaction was not observed
Kamaishi and Oofunato, both in Iwate Prefecture, and one at these sites, the proposed method has less number of
site in Sendai, one site in Ishinomaki and two sites in
Kesennuma, all in Miyagi Prefecture. The locations of these
oil tanks and oil refineries are also shown in Fig. 14.
Considering the large levels of acceleration recorded
during these two events, it was initially thought that the oil
tanks would undergo extensive damage as a result of soil
liquefaction of the foundation ground. However, subsequent
ocular inspections at these sites revealed the absence of sand
boils, ground cracks and other evidences that soil liquefac-
tion indeed took place. Needless to say, the oil tanks and
refineries at the six sites underwent minor, if not zero,
damage after these two earthquakes, notwithstanding the
large levels of accelerations recorded [19].
In order to explain the absence of liquefaction at these
sites and to verify the applicability of the proposed
method in assessing liquefaction potential based on two
peak ground motion indices, analyses were made on
representative soil profiles at these 6 sites. The sites
contain loose sandy deposits with SPT N-values generally
less than 20. The strong motion records employed were
those obtained by K-Net stations [18] nearest the
appropriate site. The locations of the K-Net stations are
shown in Fig. 14, while the details of the boring logs, Fig. 14. Map showing the epicenters of 2003 Miyagiken (Japan)
peak ground accelerations, and peak ground velocities are earthquakes, including the locations of oil tanks and refineries and seismic
listed in Table 2. Note that since signs of liquefaction monitoring stations.
R.P. Orense / Soil Dynamics and Earthquake Engineering 25 (2005) 225–240 237
Table 2
Data used for liquefaction evaluation during the 2003 Miyagiken earthquakes
‘incorrectly judged’ points. Moreover, the lower limit of FL the seismic monitoring stations, where these records were
is nearer to 1.0 for the proposed method, indicating that obtained was confirmed based on evidence of liquefaction,
the method agrees fairly well with the observation of non- such as sand boils or surface cracks, as reported in the
occurrence of soil liquefaction. These computations high- literature. For comparison purposes, strong motions
light the shortcoming of the conventional method as far as recorded at the five K-Net stations during the 5/26 and 7/
detecting liquefaction occurrence under high levels of PGAs 26 Miyagiken earthquakes, all of which did not undergo
is concerned. liquefaction, are also included.
In the table, the peak ground displacement (PGD) is
incorporated. The displacement–time history is computed
6. Threshold values for liquefaction by double integration of the acceleration records, and the
PGD is taken as half the difference between the maximum
The above calculations show that the combination of positive and negative displacements. The PGD is a more
PGA and PGV is a good indicator of occurrence or non- direct manifestation of severity of shaking at a site since it
occurrence of liquefaction during earthquakes. In order to represents the induced shear strains within the layers, and
determine the threshold values that could trigger the onset of therefore a good indicator of damage potential.
liquefaction, actual earthquake motions recorded at various Based on this table, a plot showing the relation between
sites in Japan were analyzed, as shown in Table 3. The PGA and PGV is shown in Fig. 16(a). It can be seen that a
occurrence or non-occurrence of liquefaction at the sites of boundary separating the occurrence and non-occurrence of
Liquefaction Liquefaction
(a) Conventional Method No Liquefaction (b) Proposed Method
No Liquefaction
FL FL
0.0 0.5 1.0 1.5 2.0 2.5 0.0 0.5 1.0 1.5 2.0 2.5
0
0
4 4
Depth (m)
8 8
Depth (m)
12 12
16 16
20 20
Fig. 15. Calculation of FL at oil tank sites during the 2003 Miyagiken earthquakes based on (a) conventional method; (b) proposed method.
238 R.P. Orense / Soil Dynamics and Earthquake Engineering 25 (2005) 225–240
Table 3
Earthquake records employed for determining threshold values for liquefaction occurrence
Earthquake Site name Direction PGA (gal) PGV (kine) PGD (cm) Remarks
1964 Niigata Eq. Kawagishi NS 155 53.5 33.0 Liquefied sites
EW 159 58.5 25.5
1987 Superstition Hills Eq. Wildlife NS 201 30.5 11.4
EW 179 22.8 9.1
1983 Nihonkai-Chub Eq. Akita Port NS 190 31.3 13.5
EW 205 30.1 12.5
1968 Tokachi-oki Eq. Aomori Port NS 208 40.1 18.1
EW 180 31.8 14.7
1995 Hyogoken Nambu Eq. Port Island NS 341 91.1 34.4
EW 284 51.0 23.3
1995 Hyogoken Nambu Eq. Kobe Port NS 230 36.4 11.2
EW 525 109.0 38.9
2000 Tottori Eq. Sakai Minato NS 299 35.1 10.1
EW 748 81.3 25.4
2003 Tokachi-oki Eq. Kushiro NS 311 43.4 12.0
EW 407 39.7 11.9
2003 Tokachi-oki Eq. Chokubetsu NS 739 67.5 28.4
EW 785 107.6 53.7
1968 Tokachi-oki Eq. Hachinohe NS 233 34.8 8.9
EW 181 38.1 9.7
1978 Miyagiken-oki Eq. Kaihoku LG 190 9.9 2.6 Non- liquefied
TR 270 20.9 6.2 sites
1980 Chiba-Ibaraki Eq. Owi Island NS 95 7.4 1.2
EW 64 7.0 1.8
1987 Nihonkai-Chubu Eq. Sunamachi NS 84 12.4 2.5
EW 122 10.5 2.1
1995 Hyogoken Nambu Eq. Kobe Univ NS 270 55.2 13.5
EW 305 39.4 6.5
2003 Miyagiken-oki Eq. Ishinomaki EW 237 14.5 2.5 Observed at K-
NS 276 24.4 5.4 Net stations
2003 Miyagiken-oki Eq. Kamaishi EW 1038 36.7 11.2
NS 594 18.2 2.2
2003 Miyagiken-oki Eq. Kesennuma EW 359 16.5 3.7
NS 391 18.5 1.5
2003 Miyagiken-oki Eq. Oofunato EW 367 21.1 4.3
NS 273 10.3 1.9
2003 Miyagiken-oki Eq. Sendai EW 179 12.5 2.3
NS 145 11.1 1.8
2003 Miyagiken-Hokubu Eq. Ishinomaki EW 196 26.8 5.5 Observed at K-
NS 250 22.4 6.8 Net stations
2003 Miyagiken-Hokubu Eq. Kamaishi EW 74 3.9 0.6
NS 66 2.8 0.4
2003 Miyagiken-Hokubu Eq. Kesennuma EW 76 3.2 0.7
NS 89 3.6 0.6
2003 Miyagiken-Hokubu Eq. Oofunato EW 36 1.6 0.4
NS 20 1.2 0.6
2003 Miyagiken-Hokubu Eq. Sendai EW 185 12.2 2.7
NS 176 13.4 2.6
liquefaction is very clear. Although more strong motions Another interesting observation is that all the black dots in
records are necessary to support the observation, it appears the figure plot in the region where the PGA/PGV%10,
that based on the collected information alone, the threshold indicative of another possible threshold for liquefaction
value for liquefaction can be taken as PGA R150 gal and occurrence.
PGV R20 kine. The plot in Fig. 16(b) showing the relation Consider the data points corresponding to the 5/26 and
between PGA/PGV ratio and PGD indicates that relatively 7/26 earthquakes. Although some stations show high values of
large ground displacements (PGD R10 cm) can be PGA, liquefaction did not occur at these sites, possibly
observed at liquefied sites. Note that the Kobe University because the corresponding PGVs are less than 20 kine. These
site, which showed high values of PGV and PGD, is a rock low values of PGVs induced very small amount of shear
site, and therefore liquefaction potential is practically zero. deformation, indicating very small shear strains within
R.P. Orense / Soil Dynamics and Earthquake Engineering 25 (2005) 225–240 239
PGA (gal)
PGD (cm)
100 30
Liquefied site 20
10 Non-liquefied site
5/26 Earthquake 10
7/26 Earthquake
1 0
1 10 100 1000 0 20 40 60
PGV (kine) PGA/PGV (1/sec)
Fig. 16. Relations between (a) PGA and PGV; (b) PGD and PGA/PGV using data from various earthquakes in Japan.
the deposit. Therefore, these results clearly show that PGA 7. Concluding remarks
alone is not a good indicator of liquefaction potential of soil
deposits. An assessment of the liquefaction potential at soil sites
The aforementioned pattern cannot be explained by the based on peak ground motion parameters observed at the
conventional method based on PGA only. On the other surface during earthquakes is proposed. By performing
hand, the proposed method, which incorporates the effect parametric studies using one-dimensional seismic response
not only of PGA but also of PGV can provide reasonable
(a)
explanation for the above observation. To illustrate,
10000 0
simplified calculations were performed to estimate the 10
V=
vertical distribution of maximum shear stresses in a uniform /PG 0
GA =1
level ground with mass density rZ2.0 g/cm3. Five cases 1000 P
=1
were considered, with values of PGA and PGV indicated in 3 2
PGA (gal)
threshold condition (i.e. PGA O150 gal), but if the PGV Case 4: PGA=50, PGV=5
is low, attenuation of shear stress with depth may take place,
preventing the occurrence of liquefaction. 12
Case 5: PGA=50, PGV=50
Although the cases presented above considered very
simplified soil profile, they nevertheless show that the
proposed method, which incorporates both PGA and PGV, 16
can adequately explain the occurrence and/or non-
occurrence of liquefaction during actual earthquakes.
Hence, the method is a viable alternative for estimating 20
liquefaction potential of sandy deposits, vis-à-vis the
Fig. 17. Example calculations using the proposed method: (a) values of
PGA-based approach incorporated in current Japanese PGA and PGV for each case; (b) vertical distribution of maximum shear
design codes. stress for each case.
240 R.P. Orense / Soil Dynamics and Earthquake Engineering 25 (2005) 225–240