Effects of Excess Pore Pressure Redistribution in

Liquefiable Layers
Sumeet K. Sinha, A.M.ASCE 1; Katerina Ziotopoulou, M.ASCE 2; and Bruce L. Kutter, M.ASCE 3
Downloaded from ascelibrary.org by Indian Institute of Technology Bombay on 07/21/25. Copyright ASCE. For personal use only; all rights reserved.

Abstract: Existing simplified procedures for evaluating soil liquefaction potential or for estimating excess pore pressures during earthquakes
are typically based on undrained cyclic tests performed on saturated soil samples under controlled loading and boundary conditions. Under
such conditions, the effect of excess pore pressure (ue ) dissipation and redistribution to neighboring soil layers cannot be accounted for.
Existing simplified procedures treat liquefiable layers as isolated soil layers without any boundary conditions even if dense and loose layers
are very thin, permeable, and adjacent to each other. However, redistribution is likely to increase and decrease ue in the neighboring dense and
loose layers respectively. Until now, no procedure short of fully coupled numerical analysis is available to estimate the importance of redis-
tribution. This paper presents an approximate analytical procedure for assessing the effects of ue redistribution in (1) soil layers that would
have liquefied if they were undrained, and (2) soil layers that would have not liquefied even if undrained. It is found that a layer that is initially
assumed liquefied under undrained conditions might not even liquefy accounting for the ue redistribution to neighboring layers. On the other
hand, a layer initially assumed to not liquefy can develop significant ue and can even liquefy due to pore pressure migration from the
neighboring layers. Thus, accounting for redistributed ue is important for liquefaction consequence assessment quantification, particularly
in systems that span the depth of these effects like deep foundations. Migration of u toward the tip of a pile can reduce its capacity, even if the
tip is embedded in a dense sand layer. On the other hand, if redistribution can result in the reduction of ue in initially assumed liquefied layers,
risks associated with liquefaction might be avoided. A criterion is also developed to evaluate the thicknesses of a layer below which redis-
tribution could prevent liquefaction even if the layer is deemed liquefied according to the existing liquefaction-triggering procedures. Finally,
the proposed procedure is illustrated by application to selected shaking events of centrifuge tests involving liquefaction of layered soil pro-
files. The predictions from the procedure matched the centrifuge test results reasonably. DOI: 10.1061/JGGEFK.GTENG-11857. © 2024
American Society of Civil Engineers.
Author keywords: Excess pore pressure redistribution; Liquefaction; Reconsolidation; Pore pressure dissipation.

Introduction (ue ) dissipation and redistribution due to migration that may be

realized to the neighboring layers in the field cannot be accounted
Over the last decades, there has been significant progress in the re- for. Undrained tests on saturated specimens (e.g., cyclic triaxial
finement of simplified procedures for the assessment of earthquake- compression or simple shear) have been used to determine the stress
induced soil liquefaction (e.g., Youd et al. 2001; Idriss and conditions under which the soil would generate ue equal to the ap-
Boulanger 2008). In addition, methods have been developed to es- plied confining stress (initial liquefaction) and undergo large strains
timate post-liquefaction reconsolidation settlement (e.g., Tokimatsu (cyclic mobility). For example, Ishihara and Yoshimine (1992) de-
and Seed 1984) and lateral spreading (Zhang et al. 2004), amongst veloped a methodology for estimating liquefaction-induced ground
numerous liquefaction effects. This progress has enabled the study settlement based on many undrained cyclic simple shear tests on
of case histories (Bray and Macedo 2017; Chiaradonna et al. 2015) uniformly saturated specimens.
for assessing the site performance under specified earthquake con- Natural soil deposits may often be layered such that ue migrates
ditions. The foundation for these procedures is based on undrained from one layer to another during and after shaking. Accounting for
cyclic shear tests on uniform soil specimens (e.g., Lee and Albaisa redistribution is essential for quantifying liquefaction consequences
1974; Nagase et al. 1988), where the effect of excess pore pressure (e.g., pile foundation capacity) and developing liquefaction mitiga-
1
tion approaches. Sinha et al. (2021a) showed how ue redistribution
Assistant Professor, Dept. of Civil Engineering, Indian Institute of
from a liquefied layer into a nonliquefied layer will increase ue and
Technology Delhi, Hauz Khas, New Delhi 110016, India (corresponding
author). ORCID: https://orcid.org/0000-0002-2011-4887. Email: sksinha@
decrease pile tip resistance, potentially causing large settlements.
civil.iitd.ac.in At the same time, the decreasing ue within the liquefied layers
2
Associate Professor, Dept. of Civil and Environmental Engineering, can quickly increase their liquefaction resistance and potentially
Univ. of California Davis, One Shields Ave., Davis, CA 95616. ORCID: benefit projects affected by thin and deep liquefiable soil layers
https://orcid.org/0000-0001-5494-497X. Email: kziotopoulou@ucdavis by eliminating the need for ground improvement or other liquefac-
.edu tion mitigation efforts.
3
Professor Emeritus, Dept. of Civil and Environmental Engineering, In this paper, the excess pore pressure ratio-based definition of
Univ. of California Davis, One Shields Ave., Davis, CA 95616. ORCID: liquefaction is adopted. We distinguish earthquake-induced un-
https://orcid.org/0000-0002-0628-1275. Email: [email protected]
drained excess pore pressure (uue ) from excess pore pressure due
Note. This manuscript was submitted on April 15, 2023; approved on
November 1, 2023; published online on January 30, 2024. Discussion per- to redistribution (ude ). The superscript u refers to the undrained
iod open until June 30, 2024; separate discussions must be submitted for earthquake loading, and d refers to the drainage or dissipation of
individual papers. This paper is part of the Journal of Geotechnical and ue from redistribution. Similarly, the excess pore pressure ratio ru is
Geoenvironmental Engineering, © ASCE, ISSN 1090-0241. defined as ruu and rdu for the undrained condition and accounting for

© ASCE 04024014-1 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2024, 150(4): 04024014

 redistribution, respectively. Accordingly, the liquefiable layers are relatively impermeable soil above and below. In these idealized ex-
categorised as NLu, Lu, NLd, and Ld layers as discussed here. The amples, redistribution is assumed to occur after the undrained load-
layers that would liquefy (i.e., ruu ¼ 1.0) assuming the undrained ing. As the Lu soil sediments and reconsolidates, water moves from
condition and subject to a given earthquake loading are referred the Lu layer to the NLu layer. The migration of water decreases ue
to as Lu layers. On the other hand, the soil layers that would in the Lu layer and increases ue in the NLu layer, achieving a peak
not liquefy (i.e., ruu < 1.0) under undrained earthquake loading value in the NLu layer. The peak excess pore pressure in the NLu
are referred to as NLu layers. This aims to make a clear distinction layer is presumed to occur when ue at the interface becomes con-
between the description of soil layers as NLu and Lu and is impor- tinuous and results in a redistributed excess pore pressure profile as
tant because potentially all soil layers can be liquefied or not de- shown in Fig. 1. Interestingly, although the time required for
pending on the intensity of shaking and the drainage of water from achieving the peak ue in the two layers depends on the permeability
the adjacent soil layers. For example, the water migration from re- of the Lu and NLu layers, its magnitude can be estimated independ-
Downloaded from ascelibrary.org by Indian Institute of Technology Bombay on 07/21/25. Copyright ASCE. For personal use only; all rights reserved.

distribution can cause liquefaction in NLu layers and, at the same ently. The ude in the NLu layer can be estimated by tracking the
time, also prevent liquefaction in the Lu layers. The portion of an water movement between the NLu and the reconsolidating Lu
NLu layer with rdu ¼ 1.0 due to redistribution is referred to as the layers.
Ld layer. Similarly, the portion of an Lu layer with rdu < 1 due to On the other hand, accounting for redistribution during shaking
redistribution is referred to as the NLd layer. Early on, Seed and (discussed later in the paper) resulting in partially drained condi-
Lee (1966) showed that even during a shaking event uue from tions, requires knowledge of compressibility and permeability in
the Lu layers would dissipate to the adjacent NLu layers, resulting the soil layers. The maximum ue in the Lu layer is likely to occur
in an increased ue in NLu layers and decreased ue in the Lu layers. during shaking but may be suppressed if the permeability of the
Yoshimi and Kuwabara (1973) studied ude development in an NLu layers is sufficient to drain the ue through the layer thicknesses on
layer from an overlying reconsolidating Lu layer. They found that the time scale of the duration of shaking. However, suppose the soil
as the Lu layer reconsolidated, ude in the NLu layer first increased, permeabilities are relatively high and the thickness of the Lu layer
attained a peak value, and then decreased. They also found that if is smaller than the NLu layer; in that case, dissipation caused by
the compressibility of the NLu layer was much less than the com- redistribution during shaking could potentially prevent liquefaction
pressibility of the Lu layer, a very large ude value could be devel- (Cubrinovski et al. 2019). In this paper, while the time required to
oped in the NLu layer. Seed et al. (1976) developed a numerical achieve redistributed excess pore pressure is not explicitly mod-
model to estimate ude in the soil layers, accounting for the ue gen- eled, a procedure has been developed using consolidation theory
eration from cyclic loading and dissipation from reconsolidation. to determine the time required for redistribution. Thus, determina-
While these studies have contributed significantly to understanding
tion of the thickness of the Lu layer that can be prevented from
ude development in the layers adjacent to Lu layers, their usage in
liquefying requires estimating the ue generation and dissipation
simplified procedures has been limited. Mele et al. (2021) devel-
rates within the layer and then integrating them over the entire du-
oped a simplified model to estimate ue in soil layers following
ration of shaking.
liquefaction triggering procedures from Idriss and Boulanger
In the following sections, first, an analytical framework is de-
(2008); however, they did not consider redistribution effects.
veloped to study the effects of redistribution (occurring following
This paper describes a procedure for approximating the effects
the undrained loading) on the idealized two-layered soil profiles as
of redistribution of increased excess pore pressures in the NLu layer
shown in Fig. 1. The results obtained are then used to describe the
and decreased excess pore pressures (hence increased liquefaction
redistribution effects on multilayered systems. Lastly, a procedure
resistance) of the Lu layer. It should be noted that redistribution can
accounting for redistribution during shaking (i.e., partially drained
occur both during and after shaking. First, an analytical framework
is developed where redistribution is considered to occur following condition) is developed to evaluate its effect on the increased lique-
undrained loading. The framework describes redistribution effects faction resistance of the Lu layer.
on two primary types of layered systems: the NLu layer below the
Lu layer and the NLu above the Lu layer. The developed framework
is then used to study the effects of redistribution and estimate peak
excess pore pressures in an NLu layer. Later, a procedure for
partially drained conditions (where redistribution occurs during
shaking) is developed for evaluating the conditions of increased
liquefaction resistance of the Lu layer. The procedure is then used
to define a criterion on the minimum thickness of the liquefiable
layer below which redistribution could prevent liquefaction in that
layer. Finally, the proposed procedure is applied to selected shak-
ings of centrifuge tests involving liquefaction of layered soil pro-
files, and the results are compared.

Analytical Framework–Concepts and Assumptions

Excess pore water pressure redistribution depends on soil proper-

ties (i.e., permeability, compressibility, cyclic resistance), soil state
Fig. 1. Illustration of possible redistributed excess pore pressure
(effective stress, initial excess pore pressure), hydraulic boundary
ðude Þ profiles (corresponding to the time when peak ue is developed
conditions, and ground motion characteristics (i.e., intensity, dura-
in the NLu layer) due to migration of excess pore pressures from the
tion, and magnitude). Initially, we consider the system of Fig. 1: an
reconsolidating Lu layer present (a) below; and (b) above the NLu
Lu layer that has developed ruu−Lu ¼ 1 adjacent to an NLu layer
layer.
with ruu−NLu < 1 at the end of earthquake shaking surrounded by

© ASCE 04024014-2 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2024, 150(4): 04024014

 Redistributed Excess Pore Pressures (u de ) in respectively. Solving Eq. (1) for the different soil profiles for
Two-Layered Systems the primary two types of layered systems (Fig. 1) can provide
estimates of rdu−Lu and rdu−NLu in the Lu and NLu layers, respec-
The redistribution of ue is first analyzed for two types of layered tively. While solving Eq. (1), the reconsolidating Lu layer has
systems, depending on whether the Lu layer is below [Fig. 1(a)] or ruu−Lu ¼ 1.0.
above [Fig. 1(b)] the NLu layer. The two-layered systems are as-
sumed to be surrounded by impermeable layers, neglecting drain-
age outside these two layers (conservative). Consequently, the NLu Layer above an Lu Layer
effect of soil ejecta is also not modeled. In both cases, the top
When an NLu layer overlies an Lu layer, the high compressibility
of the NLu layer is defined at a depth, z. The approximate possible
of the Lu layer and the comparatively lower initial effective stress
ude profiles (at times when it achieves its peak value in the NLu 0
ðσvo−NLu Þ of the NLu layer can lead to high rdu−NLu (i.e., high
Downloaded from ascelibrary.org by Indian Institute of Technology Bombay on 07/21/25. Copyright ASCE. For personal use only; all rights reserved.

layer) for the primary two-layered systems are shown by solid red
volume of water flowing into the NLu layer). For relatively thick
lines in Fig. 1. The approximate ude profiles were estimated based
Lu layers, giving out a lot of water, rdu−NLu values may even reach
on the numerical procedure defined by Yoshimi and Kuwabara
one [Fig. 1(a)]. Yoshimi and Kuwabara (1973) note that the NLu
(1973), where Darcy’s law is assumed to be valid with a constant
layer can liquefy due to the redistribution of excess pore pressures
permeability within each stratum. The time required to achieve
from a reconsolidating Lu layer (of relative compressibility 10
peak ude in the NLu layer is affected by the relative permeability
times or higher) depending upon the relative H, ruu , and Z values
of the soil layers and the hydraulic boundary conditions. Sinha et al.
of the soil layers. The possible ude profiles, when peak ue is
(2022b) conducted several centrifuge tests and observed the redis-
achieved in the NLu layer, are shown in Fig. 1(a). The redistrib-
tribution of excess pore pressures within the Lu and NLu layers.
ution results in the equalization of pore pressures starting from
The duration of redistribution can be very fast (within seconds)
the bottom of the Lu layer. Depending on the magnitude of excess
as it requires only a small amount of water to migrate within
pore pressure redistribution (which in turn depends on the relative
the Lu and NLu layers to achieve the redistributed excess pore pres-
compressibility, thickness, and effective stress of the NLu and Lu
sure. The prevention of water draining out of the two-layered sys-
tems ensures the conservation of the volume of water between the layers), the ude profile in the NLu layer would change from linearly
earthquake-induced and redistributed excess pore pressure. When varying to equalized excess pore pressure with depths as shown in
the NLu is above the Lu layer, redistribution can potentially liquefy profiles 1, 2, and 3, respectively, in Fig. 1(a).
the NLu layer [Fig. 1(a)]. When the NLu layer is below the Lu During reconsolidation, excess pore pressures begin dissipat-
layer, the pore pressures will increase but the NLu will never ing from the bottom of the Lu layer, as shown in the ude Profile 1 of
liquefy (i.e., ru ≠ 1.0) because the ude in it will not exceed the Fig. 1(a). Redistribution can cause complete liquefaction of an
uue in the overlying Lu layer [Fig. 1(b)]. The results from these overlying NLu layer if l < H Lu , where l is calculated from
Eq. (2)
two-layered systems act as a basis for ude in multilayered soil pro-
files (described later in the paper). The average redistributed excess
pore pressure ratio (rdu ) in the NLu and Lu layers can then be sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
σ0
obtained by assuming the volume conservation of water between l ¼ HLu 2H̄ mv vo−NLu ð1 − ruu−NLu Þ ð2Þ
the impermeable boundaries, which can be written as γ 0 H Lu

0
mv−Lu HLu σvo−Lu ðruu−Lu − rdu−Lu Þ
where mv ¼ mv−NLu =mv−Lu is the compressibility ratio; H̄ ¼
0
¼ mv−NLu HNLu σvo−NLu ðrdu−NLu − ruu−NLu Þ ð1Þ HNLu =HLu is the thickness ratio of the NLu layer with respect
to the Lu layer; and γ 0 is the average effective unit weight of
where H Lu and HNLu = thicknesses; mv−Lu and mv−NLu = average the soil layer.
0
compressibilities; σvo−Lu 0
and σvo−NLu = average initial effective As the thickness of the NLu layer increases, redistribution could
0
u u
stresses; ru−Lu and ru−NLu = average earthquake-induced excess result in ude profile 2 of Fig. 1(a), where ude−NLu equals σvo at its top
pore pressure ratios [described later in Eq. (20)]; and rud−Lu and d
and ue−Lu at its bottom (Yoshimi and Kuwabara 1973). For an even
rud−NLu = average redistributed excess pore pressure ratios for Lu larger thickness of the NLu layer, redistribution would result in the
and NLu layers [described later through Eqs. (3)–(7)] (Fig. 1). Sim- equalization of pore pressures in both the Lu and NLu layers, as
ilarly, the terms , uue−Lu and uue−NLu and ude−Lu and ude−NLu , represent shown in ude profile 3 of Fig. 1(a). The resulting rdu in the Lu
0
the average earthquake-induced ðuue ¼ ruu σvo Þ and redistributed and NLu layers is obtained by solving Eq. (1) for the assumed ude
d 0
ðue ¼ ru σvo Þ excess pore pressures in the Lu and NLu layers,
d
Profiles 2 and 3 and is given as

8
>
> l
>
> 1; <1
>
>
>
>
H Lu
>
>
> 0 0
< ruu−NLu H̄ mv þ σvo−Lu þσvo−NLu −γ 0 H NLu =2
σ0 γ 0 HNLu =2
rdu−NLu ¼ vo−NLu
; 1− 0 ≤ rdu−NLu ≤ 1.0 ð3Þ
>
> H̄ mv þ2 σvo−NLu
>
>
>
> σ0
>
> ruu−NLu H̄ mv þ σ 0vo−Lu γ 0 H NLu =2
>
> ; rdu−NLu ≤ 1 −
>
vo−NLu
: 1 þ H̄ mv 0
σvo−NLu

© ASCE 04024014-3 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2024, 150(4): 04024014

 8
>
> γ 0 l2 l
>
> 1− ; ≤1
>
> 0
2H Lu σvo−Lu
>
>
HLu
>
>
> 0
< ð2ru−NLu −1Þσvo−NLu
u þγ 0 H nliq =2
H̄ mv þ2 0
0
σvo−Lu σvo−NLu − γ 0 HNLu =2 γ 0 HLu =2
rdu−Lu ¼ ; 0 ≤ r d
≤ 1 − 0
ð4Þ
>
> H̄ mv þ2 σvo−Lu u−Lu
σvo−Lu
>
>
>
> 0
> ru−NLu H̄ mv σvo−NLu
> u
þ1 0
σvo−NLu − γ 0 HNLu =2
>
>
0
σvo−Lu
>
: ; rdu−Lu ≤ 0
1 þ H̄ mv σvo−Lu
Downloaded from ascelibrary.org by Indian Institute of Technology Bombay on 07/21/25. Copyright ASCE. For personal use only; all rights reserved.

It can be observed from Eq. (3) that for a higher compressibility Following redistribution, the ude−Lu in the thickness l of the Lu
ratio (mv ), rdu−NLu decreases. Fig. 2 shows the rdu for the NLu and layer equalizes and attains a valued equal to the initial uue−Lu at
Lu layers as a function of thickness and compressibility ratio a distance l from the bottom of the Lu layer. The equalized redis-
(H̄ mv ) with mv ¼ 1=50, ruu−NLu of 0, 0.5, and 0.9, and a unit thick- tributed excess pore pressures for the case when ðl < H Lu Þ is
ness of the Nlu layer ðH NLu ¼ 1 mÞ at a depth of Z ¼ 10 m. As shown as ude profile 1 in Fig. 1(b). For very thick NLu layers,
expected, rdu values in both layers (Lu and NLu) decrease as H̄ or the entire thickness of the Lu layer ðl ¼ HLu Þ contributes to
mv increases. However, for very large values of H̄ mv , rdu−NLu redistribution and correspondingly results in ude profile 2 as
asymptotically approaches ruu−NLu , whereas rdu−Lu asymptotically shown in Fig. 1(b). The thickness of the Lu layer contributing
approaches ruu−NLu =ð1 þ HNLu =ZÞ (Fig. 2). The rdu in both Lu to redistribution can be obtained by solving Eq. (1) for ude profile
and NLu layers decreases with depth (Z); however, since NLu 1 of Fig. 1(b) as
is above the Lu Layer, rdu−NLu is always greater than rdu−Lu (Fig. 2).    
1 l 2 l 0
σvo−NLu H̄
As redistribution increased ue in the NLu layer (rdu−NLu > ruu−NLu ), þ mv H̄ − mv H̄ ð1 − ru−NLu Þ 0
u
− ¼0
it decreased ue in the Lu layer ðrdu−Lu < ruu−Lu Þ. Assuming redistrib- 2 HLu HLu γ HLu 2
ution occurs during shaking, such a decrease of ue in the Lu layer ð5Þ
will significantly increase its liquefaction resistance, especially for
the deep thin Lu layers. A discussion of the role of redistribution in where a solution of the thickness l ≤ HLu indicates that only a
increasing the liquefaction resistance of Lu layers and the factors it small thickness (l) of the Lu layer participates in redistribution
depends on is presented later. and results in the ude profile 1 of Fig. 1(b). Any other solution would
indicate the participation of the full Lu thickness (l ¼ HLu ) result-
NLu Layer below an Lu Layer ing in ude profile 2 of Fig. 1(b). The resulting rdu in the NLu and
Lu layers is given as
When an NLu layer is under an Lu layer, the movement of water
from the Lu layer results in the equalization of excess pore pres-
sures while forming a water film at the impermeable boundary
above the Lu layer (Sinha et al. 2022a) [Fig. 1(b)]. With the
assumption that reconsolidation of the Lu layer results in water
movement only toward the NLu layer with no formation of the
water film (above the Lu layer), Eq. (1) can be solved to obtain
the rdu values in the Lu and NLu layers. Please note, in case of
the formation of a water film layer, the redistributed excess pore
pressure in NLu would be smaller due to the drainage of some
water to the water film. As a result, the assumption of no water
film formation results in conservative ue estimates in the NLu layer.
Fig. 1(b) shows the approximate possible ude profiles (1 and 2)
when peak ue is achieved in the NLu layer. The redistribution re-
sults in the equalization of pore pressures in the NLu layer. In the
Lu layer, equalization of excess pore pressure occurs from the
bottom of the layer which slowly progresses through the entire
Lu Layer with the increase in magnitude of redistribution as illus-
trated in profiles 1 and 2 of Fig. 1(b).
During reconsolidation, ue begins dissipating from the bottom
of the Lu layer of a thickness (l) as shown in the ude profile 1 in
Fig. 1(b). The maximum possible value of ude−NLu is equal to the
effective stress at the bottom of the Lu layer (i.e., ude−NLu ¼ Fig. 2. Redistributed excess pore pressure ratio ðrdu Þ in the layered sys-
0
σvo−Lu þ γ 0 HLu =2 for l → 0). If uue−NLu is larger than the effec- tem with an NLu layer above a Lu layer as a function of thickness and
tive stress at the bottom of the Lu layer (i.e., uue−NLu > compressibility ratio (H̄ mv ) for earthquake-induced excess pore pres-
0 sure ratio (ruu−NLu ) of 0, 0.5, and 0.9, with a compressibility ratio ðmv Þ
σvo−Lu þ γ 0 HLu =2Þ, no redistribution can occur toward the NLu
of 1/50, and a unit thickness of the NLu layer ðHNLu ¼ 1 mÞ at a depth
layer. The thickness (l) of the Lu layer participating in the redis-
of Z ¼ 10.
tribution increases with the thickness of the NLu layer [Fig. 1(b)].

© ASCE 04024014-4 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2024, 150(4): 04024014

 8
>
> γ 0 HNLu =2
>
> ruu−NLu ; ruu−NLu ≥ 1 −
>
> 0
σvo−NLu
>
>
>
>
< γ 0 ðl þ HNLu γ 0 H =2
2 Þ l
rdu−NLu ¼ 1 − 0 ; < 1; ruu−NLu ≤ 1 − 0 NLu ð6Þ
>
> σvo−NLu HLu σvo−NLu
>
>
>
>
0
σvo−Lu
>
> σ 0 þ ruu−NLu H̄ mv 0
σvo−Lu − γ 0 HLu =2 u γ 0 H =2
>
> vo−NLu
; rdu−NLu ≤ ; ru−NLu ≤ 1 − 0 NLu
: ð1 þ H̄ mv Þ 0
σvo−NLu σvo−NLu
8
γ 0 H NLu
>
>
>
> 1.0; ruu−NLu ≥1− 2
>
> 0
σvo−NLu
Downloaded from ascelibrary.org by Indian Institute of Technology Bombay on 07/21/25. Copyright ASCE. For personal use only; all rights reserved.

>
>
>
>
< γ 0 l2 l NLu γ 0H
rdu−Lu ¼ 1− 0 ; < 1; ruu−NLu ≤ 1 − 0 2 ð7Þ
>
> 2HLu σvo−Lu HLu σvo−NLu
>
>
>
> 0
σvo−NLu
>
> 1 þ ru−NLu H̄ mv σvo−Lu
u
σ0 −γ
0H γ 0 H NLu
> 0 Lu
>
: ; rdu−Lu ≤ vo−Lu0 2
; ruu−NLu ≤1− 2
0
ð1 þ H̄ mv Þ σvo−Lu σvo−NLu

It can be observed from Eqs. (6) and (7) that rdu−NLu increases It should be noted that redistribution always exists between the
with mv decreasing and ruu−NLu increasing. Fig. 3 shows rdu in the Lu Lu and NLu layers, resulting in the decrease and increase of ue in
and NLu layers as a function of thickness and compressibility ratio the Lu and NLu layers, respectively. The magnitude of redistribu-
(H̄ mv ) for mv ¼ 1=50, ruu−Nlu of 0, 0.5, and 0.9 and HNLu ¼ 1 m at tion effects depends on the relative earthquake-induced excess pore
a depth of Z ¼ 10 m. As expected, as H̄ or mv increases, rdu de- pressures ratio (ruu−Lu ), the thickness (H), and the compressibility
(mv ) of the soil layers. Increasing the shaking intensity would result
creases and asymptotically approaches to rdu−NLu ¼ ruu−NLu and
in a lower factor of safety against liquefaction (FSliq ) and thus
rdu−Lu ¼ ruu−NLu ð1 þ 0.5HNLu =ZÞ in the NLu and Lu layers, respec-
larger development of earthquake-induced excess pore pressures
tively. Again, rdu is higher in the Lu layer since the Lu is above the (ruu ) in the Lu and NLu layers. The final redistributed excess pore
NLu layer. For this layered profile, redistribution also resulted in
pressures (rdu ) would also depend on the relative thickness and com-
decreased ue in the Lu layer ðrdu−Lu < ruu−Lu Þ and increased ue pressibility (H̄ mv ) of the soil layers. As shown in Figs. 2 and 3,
in the NLu layer ðrdu−NLu > ruu−NLu Þ. The resulting increase in lique- for higher values of thickness and compressibility ratio (H̄ mv ),
faction resistance of the Lu layer due to redistribution is the redistributed excess pore pressures would be similar to the
discussed later. earthquake-induced pore pressures.

Redistributed Excess Pore Pressure in Multi-

Layered Systems

Redistributed excess pore pressures in a multilayered system can

be far more complex than in simple two-layered systems. While
the primary contributing factors are the same as before (soil pro-
perties and state, loading conditions), the complexity of drainage
conditions gets accentuated proportionally to the number of layers
(Bekir Afacan 2020). Furthermore, the timing of liquefaction can
vary across multiple Lu layers within such a system thus compli-
cating the redistribution, especially when drainage is considered
during shaking.
To apply the developed analytical framework for a multilayered
system, it is assumed that it can be decomposed into many primary
layered systems (of the NLu layer above or below the Lu
layer–Fig. 4) and there exists no migration of excess pore pressure
between the primary layered systems. In multilayered layered
systems, an increase of ue in the NLu layer can occur from the
Lu layer above and below it. Similarly, the dissipation of ue from
the Lu layer can also occur in either direction: to the NLu layer
Fig. 3. Redistributed excess pore pressure ratio ðrdu Þ in the layered sys- above and below it. As described earlier, the maximum possible
tem with an NLu layer below a Lu layer as a function of thickness and rdu in an NLu layer below an Lu layer equals the effective stress
compressibility ratio (H̄ mv ) for earthquake-induced excess pore pres- at the bottom of the Lu layer. If the NLu layer is above the Lu layer,
sure ratio (ruu−NLu ) of 0, 0.5, and 0.9, with a compressibility ratio ðmv Þ
redistribution can cause liquefaction in the NLu layer (rdu−NLu ¼
of 1/50, and a unit thickness of the NLu layer ðHNLu ¼ 1 mÞ at a depth
1.0). Thus, a reasonably simple way to split the multilayered
of Z ¼ 10 m.
system and prevent double counting the redistribution effect in both

© ASCE 04024014-5 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2024, 150(4): 04024014

Downloaded from ascelibrary.org by Indian Institute of Technology Bombay on 07/21/25. Copyright ASCE. For personal use only; all rights reserved.

Fig. 4. Illustration of the multi-layered soil systems decomposition into smaller units of the two primary layered soil systems of the NLu layer above/
below a Lu layer to estimate redistributed excess pore pressure in the NLu layers.

(upward and downward) directions is to (wherever possible) de- both NLu layers. The ude−Lu in the sandwiched Lu layer can be con-
compose the multilayered system into multiple units of the primary servatively taken equal to the minimum of the ude−Lu calculated
layered system of an NLu layer above an Lu layer (Fig. 4). from the two primary systems. Since the effect of only one NLu
If sublayers cannot be decomposed into a set of NLu layers layer (instead of both NLu layers) is considered on the redistrib-
above Lu layers, then two types of three-layer systems are intro- ution of excess pore pressures from the Lu layer, it results in a
duced to account for redistribution: an NLu layer sandwiched be- conservative estimate of ude−Lu . For example, in Fig. 4, the Lu layer
tween two Lu layers and an Lu layer sandwiched between two NLu (# 4) sandwiched between two NLu layers (# 3 and # 5) is decom-
layers. For example, in the multilayered system presented in Fig. 4, posed into the two systems: an NLu layer above an Lu layer and an
the presence of the clay layer results in two subsystems: an Lu layer NLu layer below an Lu layer. To find ude−NLu#3 and ude−NLu#5 in the
(# 4) sandwiched between the NLu layers (# 3 and 5) and an NLu NLu layers and ude−Lu#4 in the Lu layer, we assume that ue−NLu#3
d−Lu#4
layer (# 8) sandwiched between the Lu layers (# 7 and 9). Similarly, d−Lu#4 d
and ue−NLu#5 represent the ue in NLu layers (# 3 and # 5) due to
the no drainage condition beneath layer 14 results in a subsystem
redistribution from the adjacent Lu layer (# 4). Similarly, ud−NLu#3
e−Lu#4
with an Lu Layer (# 13) sandwiched between two NLu layers
and ud−NLu#5 d
e−Lu#4 represent the ue in the Lu layer (#4) due to redistrib-
(# 12 and 14). The following subsections describe the estimation
ution from the adjacent NLu layers (# 3 and # 5). Then, ude in the Lu
of rdu for these two additional types of subsystems.
and NLu layers is taken as
ude−NLu#3 ¼ ue−NLu#3
d−Lu#4
ð8Þ
Lu Layer Sandwiched between NLu Layers
For an Lu layer sandwiched between two NLu layers, ude−NLu can ude−NLu#5 ¼ ue−NLu#5
d−Lu#4
ð9Þ
be conservatively estimated by assuming the contribution of the
full Lu layer in developing ude in both the NLu layers above and ude−Lu#4 ¼ minðue−Lu#4 ; ue−Lu#4 Þ
d−NLu#3 d−NLu#5
ð10Þ
below it. The ude in the NLu layers can be individually calculated
accounting for redistribution from the middle Lu layer assuming
NLu Layer Sandwiched between Lu Layers
there is no migration of the excess pore pressures to the other
NLu layer. Equivalently, one could further decompose the subsys- For an NLu layer sandwiched between two Lu layers, ude−NLu can
tem into two systems: an NLu layer above the Lu layer and an NLu be conservatively estimated by taking the contributions from both
layer below the Lu layer, and then individually calculate ude−NLu in Lu layers. First, ude−NLu is estimated from the Lu layer above it.

© ASCE 04024014-6 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2024, 150(4): 04024014

 Then this is assumed as uue−NLu to obtain the final ude−NLu from the Fig. 1, the assumption of linear scaling would still be conservative
Lu layer below it. For example, in Fig. 4, the ude in the NLu layer as those profiles would result in higher rdu−Lu ðtÞ. The rate of dis-
(# 8) sandwiched between two Lu layers (# 7 and # 9) is obtained sipation from excess pore pressure redistribution (ṙdu−Lu ðtÞ) can
by decomposing the profile into two systems: an NLu layer below then be estimated as
an Lu layer and an NLu layer above an Lu layer. First, the redis-
tributed excess pore pressure ud−Lu#7 ṙdu−Lu ðtÞ ¼ rpd
u−Lu ðtÞṙu−Lu
d
ð12Þ
e−NLu#8 in NLu layer # 8 is calcu-
lated from the Lu layer # 7 above it. Then, the obtained ude is taken
as an earthquake-induced pore pressure (i.e., uue−NLu#8 ¼ ud−Lu#7 e−NLu#8 ) ð1 − rdu−Lu Þ
to estimate the final redistributed excess pore pressure in the NLu ṙdu−Lu ¼ ð13Þ
td
layer (ude−NLu#8 ¼ ue−NLu#8
d−Lu#9
) as a result of the Lu layer # 9 below it.
The ude in the Lu layers is taken as ude−Lu#7 ¼ ud−NLu#8 and where td = time required for redistribution, i.e., the duration re-
Downloaded from ascelibrary.org by Indian Institute of Technology Bombay on 07/21/25. Copyright ASCE. For personal use only; all rights reserved.

e−Lu#7
ude−Lu#9 ¼ ud−NLu#8 . Later in this paper, the developed procedure quired for the earthquake-induced excess pore pressures (ruu−Lu )
e−Lu#9
is applied to centrifuge tests of multilayered soil systems to illus- to achieve rdu−Lu in the reconsolidating Lu layer. As expected, the
trate its utility and to provide a preliminary partial validation of the time required for redistribution (td ) is larger for smaller rdu−Lu
approach. values. Later in the paper, a procedure is defined to estimate td .

Rate of Excess Pore Pressure Generation from

Increased Liquefaction Resistance by Shaking (ṙ uu−Lu )
Redistribution
The average rate of excess pore pressure ratio generation, ṙuu−Lu ðtÞ,
This section evaluates the potential of redistribution for preventing in the Lu layer can be estimated from the rate of undrained loading
liquefaction in an Lu layer. The analytical framework described in (Seed et al. 1976) at an arbitrarily defined stress level (τ ) as
the above sections showed that for a large thickness ratio (H̄), re-  
distribution can significantly reduce ue in an Lu layer. During shak- Nu 1
ṙuu−Lu ðtÞ ¼ ṙuu−Lu ¼ ð14Þ
ing, a partially drained condition can exist, where the undrained N Lu tu
earthquake loading generates ue while redistribution to neighboring
layers decreases it. Herein, the rate of excess pore pressure dissi- where N Lu = number of shear stress (τ ) cycles required to cause
pation due to redistribution is the rate at which undrained excess liquefaction; N u = equivalent number of shear stress (τ ) cycles;
pore pressures are converted to redistributed excess pore pressures. and tu is the time for ue generation. As a simplification, tu is taken
If the dissipation rate from redistribution is fast enough, it can pre- equal to the duration of undrained (or earthquake) loading. The
vent liquefaction in the Lu layer. However, if redistribution occurs empirical procedure by Idriss and Boulanger (2008) also follows
too slowly, liquefaction may not be prevented. Thus, determining the same philosophy of rate of excess pore pressure generation
the rate of excess pore pressure generation (from undrained load- as defined through Eq. (14) by Seed et al. (1976). The cycle ratio
ing) relative to the rate of dissipation (from redistribution) is re- ðN u =N Lu Þ can be substituted in terms of FSliq−Lu as defined as the
quired. The net effect of generation and dissipation processes is factor of safety against liquefaction and the parameter b defined as
referred to as the partially drained (pd) excess pore pressure ratio the slope of cyclic resistance ratio (CRR) (Idriss and Boulanger
2008) as
ðrpd
u−Lu Þ in the Lu layer.
1
ṙuu−Lu ¼ ðFSliq−Lu Þ−1=b ð15Þ
Rate of Excess Pore Pressure Dissipation from tu
Redistribution (ṙ du−Lu )
Previously, rdu−Lu was calculated at the end of shaking when the Lu Partially Drained Excess Pore Pressure in
layer was reconsolidating from its liquefied state (i.e., ruu−Lu ¼ 1) Lu Layer ṙ pd
u−Lu
[Eqs. (4) and (7)]. Assuming that rpd
u−Lu ðtÞ is known, the simplified The average rate of partially drained excess pore pressure in the
and conservative way of estimating the rdu−Lu ðtÞ at any time (t) Lu layer, ṙpd u−Lu ðtÞ, is given by the difference in the generation
during shaking is the linear scaling of rdu−Lu with rpdu−Lu ðtÞ (ṙuu−Lu ðtÞ) and dissipation (ṙdu−Lu ðtÞ) rates, resulting in a first-order
ordinary differential equation
rdu−Lu ðtÞ ¼ rpd
u−Lu ðtÞru−Lu
d
ð11Þ
ṙpd u pd
u−Lu ðtÞ ¼ ṙu−Lu − ru−Lu ðtÞṙu−Lu
d
ð16Þ
The assumption of linear scaling of the rdu−Lu with rpd u−Lu ðtÞ
would be generally valid for the ude profiles [such as # 3 in This differential equation can be solved with the initial boun-
Fig. 1(a) and # 2 in Fig. 1(b)] where equalization of ue occurs dary condition of rpd pd
u−Lu ¼ 0 to obtain ru−Lu ðtÞ as
between the Lu and NLu layers. The ude profiles #3 and #2 in
ṙuu−Lu
rpd
u−Lu ðtÞ ¼ ð1 − e−ṙu−Lu t Þ; rpd
d
Figs. 1(a and b), respectively, represent the cases when the uue is u−Lu ðtÞ < 1.0 ð17Þ
not enough to liquefy the Lu layer. As a result, during the majority ṙdu−Lu
of the shaking period, while ue is developing in the Lu Layer
From Eq. (17), the partially drained excess pore pressure ratio in
(i.e., while ruu−Lu < 1), redistribution would result in equalized
excess pore pressures in the Lu and NLu layers as shown in ude the Lu layer (rpd
u−Lu ) at the end of earthquake loading (i.e., t ¼ tu ) is
profiles # 3 and # 2 of Figs. 1(a and b), respectively. Furthermore, given as
for evaluating the case of increased liquefaction resistance of thin 1
rpd
u−Lu ¼ ðFSliq−Lu Þ
−1=b
ð1 − e−ð1−ru−Lu Þt̄ Þ;
d
Lu layers adjacent to thick NLu layers, rdu−Lu would likely result ð1 − ru−Lu Þt̄
d
from fully equalized ude profiles in the Lu and NLu layers. Consid-
ering that the rdu−Lu is the result from other ude profiles shown in rpd
u−Lu < 1 ð18Þ

© ASCE 04024014-7 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2024, 150(4): 04024014

Downloaded from ascelibrary.org by Indian Institute of Technology Bombay on 07/21/25. Copyright ASCE. For personal use only; all rights reserved.

Fig. 5. Partially drained excess pore pressure ratio in the Lu layer Fig. 6. Partially drained excess pore pressure ratio in the Lu layer
½rpd −1=b  in the layered system with an NLu layer above ½rpd −1=b  in the layered system with an NLu layer below
u−Lu =ðFSliq−Lu Þ u−Lu =ðFSliq−Lu Þ
a Lu layer as a function of thickness and compressibility ratio (H̄ mv ) a Lu layer as a function of thickness and compressibility ratio (H̄ mv )
for time ratio (t̄) of 0.2, 1, and 5, earthquake-induced excess pore pres- for time ratio (t̄) of 0.2, 1, and 5, earthquake-induced excess pore pres-
sure ratio (ruu−NLu ) of 0 and 0.9, and a unit thickness of the NLu layer sure ratio (ruu−NLu ) of 0 and 0.9, and a unit thickness of the NLu layer
ðHNLu ¼ 1 mÞ at a depth of Z ¼ 10 m. ðHNLu ¼ 1 mÞ at a depth of Z ¼ 10 m.

where t̄ ¼ tu =td is the time ratio of the ue generation and dissipa-

tion (from redistribution). It should be noted that the dissipation
time (i.e., td ) required for redistribution of excess pore pressures
within the Lu and NLu layers can be very fast (even completed
within the shaking period) and has been observed in the centrifuge
tests by Sinha et al. (2022b). This is because (1) the permeability of
the soil can increase up to five times for the liquefied soils (Ueng
et al. 2017) and (2) only a small amount of water is required to
migrate within the Lu and NLu layers to achieve the redistributed
excess pore pressure. In addition to the permeability, td is also af-
fected by the thickness, effective stresses, and compressibility of
the soil layers, and the hydraulic boundary conditions as shown
later in the paper while developing a procedure for its approximate
estimation. In Eq. (18), ðFSliq−Lu Þ−1=b accounts for the extent of
liquefaction in the Lu layer if it were undrained. It can be observed
that for soils with a larger b parameter, redistribution results in
smaller rpd pd
u−Lu. The ru−Lu as the function mv H̄ for t̄ of 0.2, 1,
and 5, ru−NLu ¼ 0 and 0.9, and a HNLu ¼ 1 m (at Z ¼ 10 m)
u

for the primary two-layered systems are shown in Figs. 5 and 6,

respectively. As expected, rpd u−Lu decreases with increases in Fig. 7. The minimum thickness and compressibility ratio (H̄ mv ) in the
thickness and compressibility ratio (H̄ mv ) and time ratio (t̄). A layered system with an NLu layer above a Lu layer as a function of the
smaller ruu−NLu leads to smaller rpd u−Lu. For very large values of
time ratio (t̄) for which liquefaction can be prevented in the Lu layer
−1=b (i.e., rpd
u−Lu ≤ 0.9) having earthquake-induced excess pore pressure ratio
H̄ mv , ru−Lu asymptotically approaches rpd
pd
u−Lu ¼ ðFSliq−Lu Þ
−t̄ð1−rdu−Lu Þ
(ruu−NLu ) of 0, 0.5, and 0.9, and a unit thickness of the NLu layer
ð1=t̄ð1 − ru−Lu ÞÞð1 − e
d
Þ. ðHNLu ¼ 1 mÞ at a depth of Z ¼ 10 m with an undrained loading re-
presented by ð1=FSliq−Lu Þ1=b of 1,10, and 100.
The Criterion for Liquefaction Prevention in Lu Layer
Liquefaction of the Lu layer can be considered prevented if rpd u−Lu
falls below a critical value [Eq. (18)]. For example, the critical value Lu layer for a unit thickness of the NLu layer (HNL ¼ 1 m) at
can be assumed to be 0.9. The criterion on the thickness and com- a depth of Z ¼ 10 m for the primary two-layered systems are
pressibility ratio (H̄ mv ) and the time ratio (t̄) to prevent liquefac- shown in Figs. 7 and 8, respectively. A larger t̄ means faster ue
tion in the Lu layer for a given undrained loading represented by dissipation from redistribution and thus would prevent liquefaction
ðFSliq−Lu Þ−1=b can be obtained by solving Eq. (18) for rpd u−Lu ¼ 0.9. in a larger thickness of the Lu layer. As a result, the minimum
The minimum thickness and compressibility ratio (H̄ mv ) as a func- H̄ that can prevent liquefaction in Lu is smaller for a larger time
tion of the time ratio (t̄) required to prevent liquefaction in the ratio (t̄). The minimum H̄ is also smaller for layers with larger

© ASCE 04024014-8 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2024, 150(4): 04024014

 (3) estimation of redistributed excess pore pressures (ude ), and
check for liquefaction prevention in the Lu layer, and (4) estimation
of partially drained excess pore pressure (upd
e−Lu ). The steps are de-
scribed in what follows.

Determination of Lu and NLu Layers

A soil liquefaction hazard analysis can identify soil layers as Lu or
NLu. Among many methods [e.g., Youd et al. (2001), Idriss and
Boulanger (2008), Robertson (2015), and Cetin et al. (2018)],
the procedure by Idriss and Boulanger (2008) is widely used for
Downloaded from ascelibrary.org by Indian Institute of Technology Bombay on 07/21/25. Copyright ASCE. For personal use only; all rights reserved.

performing a liquefaction hazard analysis for a design earthquake

loading, typically quantified via an expected magnitude (M w ) and
peak ground acceleration (PGA). These are used to estimate the
imposed demand in terms of cyclic stress ratio (CSR) in the various
soil layers. The cyclic strength of soil is determined via liquefaction
triggering correlations between penetration resistance (standard
penetration test N 160cs or cone penetration test qc1Ncs ) and CRR.
The factor of safety against liquefaction ðFSliq ¼ CRR=CSRÞ is
Fig. 8. The minimum thickness and compressibility ratio (H̄ mv ) in the calculated to categorize the layers as “liquefied undrained” (Lu)
layered system with an NLu layer below a Lu layer as a function of the (with FSliq ≤ 1) and “non-liquefied undrained” (NLu) (with
time ratio (t̄) for which liquefaction can be prevented in the Lu layer FSliq > 1). It should be noted that the identification of layers,
(i.e., rpd
u−Lu ≤ 0.9) having earthquake-induced excess pore pressure ratio whether liquefied or non-liquefied, is always in reference to the
(ruu−NLu ) of 0, 0.5, and 0.9, and a unit thickness of the NLu layer design earthquake loading. A layer initially identified as NLu
ðHNLu ¼ 1 mÞ at a depth of Z ¼ 10 m with an undrained loading for a given shaking may be an Lu layer under a stronger shaking
represented by ð1=FSliq−Lu Þ1=b of 1, 10, and 100. event.

relative compressibility (mv ), smaller factor of safety against lique- Estimation of Earthquake-Induced Excess Pore
faction (FSliq−Lu ), or a larger b parameter (i.e., overall smaller Pressures (u ue )
ðFSliq−Lu Þ−1=b ) and smaller ruu−NLu. A large value of minimum Earthquake-induced excess pore pressure (uue ) in the soil layers is
H̄ means redistribution cannot prevent liquefaction in the Lu layer. estimated using the simplified equations by Mele et al. (2021). The
For example, in Fig. 7, for ruu−NLu ¼ 0, the minimum H̄ approaches method uses FSliq computed in Step (1) to estimate uue as
infinity for ðFSliq−Lu Þ−1=b > 10 and t̄ ≤ 10. The figure also shows 0
uue ¼ ruu σvo ð19Þ
that reducing the time required for redistribution (tr ) such as by
installing earthquake drains, would result in a larger t̄, resulting 8
>  1 
in liquefaction prevention in the Lu layer. For the primary layered < 2.0 arcsin FS−2bβ FSliq > 1
system with the NLu layer below the Lu layer, since redistribution ru ¼
u
π liq
ð20Þ
does not occur when uue−NLu is greater than the effective stress at the >
: 1.0 FSliq ≤ 1
bottom of the Lu layer, for such cases, liquefaction cannot be pre-
vented. For example, in Fig. 8, for ruu−NLu ¼ 0.9, the minimum where b and β are the parameters defined by Mele et al. (2021) in
H̄ mv for preventing liquefaction is in the order of 105 . terms of qc1Ncs and N 160cs , respecdtively. The equation for Mele
From the minimum thickness ratio (H̄) estimated from Figs. 7 et al. (2021) was slightly modified (following the original equation
and 8 [or from Eq. (18)], the maximum thickness of the Lu layer by Booker et al. 1976) to include the liquefaction triggering con-
that can be prevented from liquefaction can be computed as dition of ruu ¼ 1.0 for FSliq ≤ 1 (instead of ruu ¼ 0.9 for FSliq ¼ 1).
H Lu ¼ HNLu =H̄. For example, in Fig. 8, for ðFSliq−Lu Þ−1=b ¼ 1
and ruu−NLu ¼ 0, and t̄ ¼ 1, the thickness and compressibility ratio
Estimation of Redistributed Excess Pore
(H̄ mv ) is about 0.3. Assuming a compressibility ratio of mv ¼
Pressures (u de )
1=20, the maximum thickness of the Lu layer that can be prevented
from liquefaction is about 167 mm (i.e., equal to 16.6% of The redistributed excess pore pressures (ude ) in the Lu and NLu
HNLu ¼ 1 m). Knowing a Lu layer cannot liquefy because of layers depend on the thickness ratio (H̄ ¼ HNLu =HLu ), compress-
redistribution (as opposed to liquefiable under undrained loading ibility ratio ðmv ¼ mv−NLu =mv−Lu Þ, depth to the top of the NLu
with FSliq-Lu < 1.0 as predicted by the simplified liquefaction- layer (Z), average effective unit weight (γ 0 ), average initial mean
0 ), and the earthquake-induced pore pressure ra-
effective stress (σvo
triggering procedures) can prove to be extremely valuable in
u
tio (ru ) computed in Step (2). The compressibility of the soil layers
reducing the risk of liquefaction-related problems and their
remediation costs. is nonlinearly dependent on the excess pore pressure. When a soil
layer liquefies, the compressibility of the layer significantly in-
creases depending upon the extent of liquefaction and initial rela-
Procedure for Estimating Redistribution Effects tive density (Seed et al. 1976). The increase in the compressibility
is due to two mechanisms: (1) the nonlinearity of the unloading-
The procedure for estimating redistribution effects in the soil layers reloading compression curve, and (2) volume change that occurs
involves four steps: (1) determination of Lu and NLu layers, due to sedimentation while the soil is liquefied (i.e., while ru ≈ 1)
(2) estimation of earthquake-induced excess pore pressures (uue ), (Scott 1986). As a simplification, the compressibility ratio (mv ) of

© ASCE 04024014-9 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2024, 150(4): 04024014

Downloaded from ascelibrary.org by Indian Institute of Technology Bombay on 07/21/25. Copyright ASCE. For personal use only; all rights reserved.

Fig. 9. Estimation of compressibility (mv ) of liquefiable soils using (a) the compressibility ratio (mv =mvo ) relation as a function of earthquake-
induced excess pore pressure ratio (ruu ) from Seed et al. (1976); and (b) compressibility (mvo ) of normally consolidated sand and silts at mean effective
0
stress (σvo ) with ruu ¼ 0 from Janbu (1985).

the soil layers for the analytical framework is assumed to be con- generation (tu ) is taken as the duration of earthquake loading. The
stant and is taken for the state of ruu in the Lu layer and NLu layers, redistribution time (td ) can be estimated from the dimensionless
i.e., the time when redistribution starts. The compressibility (mv ) of time factor (T d ¼ ðcv =H2 Þtd ) associated with the degree of consoli-
the Lu and NLu layers can be estimated using the relation by Seed dation ðU ¼ 1 − rdu−Lu Þ. Taylor (1948) describes the relationship
et al. (1976), which approximates the lab test results from Lee and between estimating T d and U for a single layer as
Albaisa (1974). The relationship models mv as a function of relative
density (DR ) and ruu as follows: 8
< π U2 ;
>
U < 0.6
mv
¼
expðyÞ
ð21Þ Td ¼ 4 ð25Þ
mvo 1 þ y þ y2 =2 >
: −0.9332log10 ð1 − UÞ − 0.0851; U ≥ 0.6

y ¼ 5ð1.5 − DR Þðruu Þz ð22Þ

For a two-layered system, the rate of dissipation depends on
z ¼ 3ð4Þ−DR ð23Þ the permeability, compressibility, and thickness of both layers.
The ratio (cv =H 2 ) can be conservatively taken to be equal to the
where mvo = compressibility of the normally consolidated sand at lesser of Lu or NLu (cv−Lu =H 2Lu or cv−NLu =H 2NLu ). The time ratio
0
mean effective stress (σvo ); and ruu ¼ 0 calculated using the empiri- ðt̄ ¼ tu =td Þ and estimated rdu−Lu (in Step (3)) is then used in Eq. (18)
cal correlations from Janbu (1985) as
to estimate rpd u−Lu . The maximum thickness of the Lu layer that can
1 be prevented from liquefaction (for the assumed liquefaction pre-
mvo ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð24Þ vention criteria of rpd
m Patm σvo 0 u−Lu < 0.9) can be obtained by solving Eq. (18)
iteratively with the computed time ratio ðt̄Þ and factor of safety
against liquefaction (FSLu ).
where m = modulus parameter depending upon the porosity (n)
of the sand layer (from Fig. 9); and Patm = atmospheric pressure
equal to 101.3 kPa. Finally, redistributed excess pore pressures
0 ) and their distribution (as shown in Fig. 1) in the
(ude ¼ rdu σvo Application in Centrifuge Model Tests
Lu and NLu layers are estimated using Eqs. (2)–(7) with the
0
Sinha et al. (2022b) conducted several centrifuge models tests with
obtained soil layer parameters (H̄; mv ; σvo ; ruu ; Z; γ 0 ). loose sand above the dense sand layer and observed redistributed
excess pore pressures in the soil layers at the end of shaking.
Estimation of Partially Drained Excess Pore Pressure While the simplified liquefaction triggering procedures (Idriss and
in the Lu Layer (u pd Boulanger 2008) predicted liquefaction in the loose sand layer and
e−Lu )
very small excess pore pressures in the dense sand layer; however,
Estimation of partially drained excess pore pressure (upd
e−Lu ) in because of redistribution during shaking, liquefaction was pre-
the Lu layer requires estimating the time for excess pore pres- vented in the bottom part of the loose sand layer and significant
sure generation (tu ) and redistribution (td ). The time for ue excess pore pressures developed in the dense sand layer.

© ASCE 04024014-10 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2024, 150(4): 04024014

Downloaded from ascelibrary.org by Indian Institute of Technology Bombay on 07/21/25. Copyright ASCE. For personal use only; all rights reserved.

Fig. 10. Redistribution effects for shaking event EQM3 of centrifuge model test SKS02: (a) measured normalized overburden corrected cone tip
resistance (qc1Ncs ); (b) cyclic stress ratio (CSR) and factor of safety against liquefaction (FSliq ); and (c) comparison of estimated earthquake-induced
(uue ) and redistributed (ude ) excess pore pressures with measured excess pore pressures. (Data from Sinha et al. 2021a.)

Description of Centrifuge Model Tests pressure (ue ) generated in the soil layers during the shaking event
are shown in Figs. 10(a and c) and 11(a and c), respectively.
Two shaking events, EQM3 and EQM4 , from two centrifuge model
tests, SKS02 (Sinha et al. 2021a) and SKS03 (Sinha et al. 2021b),
were chosen to study redistribution effects and as a check on the Estimating Redistribution Effects in Lu and
procedures discussed so far. These centrifuge model tests were con- NLu Layers
ducted on the 9-m radius centrifuge facility at the Center for Geo-
technical Modeling at the University of California, Davis. The tests Results on the cyclic stress ratio (CSR), the factor of safety against
were performed at a centrifugal acceleration of 40 g. All the units liquefaction (FSLu ), and earthquake-induced excess pore pressure
reported for the centrifuge test are in the prototype scale following (uue ) for the selected shaking events from SKS02 and SKS03 are
centrifuge scaling laws by Garnier et al. (2007). The models con- shown in Figs. 10(b and c) and 11(b and c), respectively. The fig-
sisted of 21 m of soil with an undrained boundary condition at ures also show the categorization of soil layers as Lu and NLu
the bottom of the layers (i.e., at 21 m) because of the impermeable layers depending on whether FSliq ≤ 1 or FSliq > 1, respectively.
base of the model container. In prototype scale, SKS02 consisted of For EQM3 in SKS02, the Lu layer consisted of the loose sand layer
a 9-m-thick liquefiable loose sand layer (DR ≈ 43%, n ≈ 0.41) between the depths of 5 to 14 m (i.e., HLu ¼ 9 m) and the NLu
sandwiched between a 4-m-thick layer of over-consolidated clay layer consisted of the dense sand layer below it up to 21 m
with an undrained shear strength su ≈ 20 kPa on the top and a (i.e., H NLu ¼ 7 m). The average initial mean effective stresses at
dense sand layer (DR ≈ 85%, n ≈ 0.36) on the bottom (Fig. 10). the mid depth in the Lu and NLu layers were 79.6 kPa and
Above the clay layer, the SKS02 model had a 1 m of Monterey sand 157.3 kPa, respectively. These layers developed earthquake-
layer. The soil profile of SKS03 consisted of 1 m of Monterey sand, induced pore pressures of uue−Lu ¼ 79.6 kPa and uue−NLu ¼
2 m of clay crust (su ≈ 28–35 kPa), 4.7 m of the loose liquefiable 15.03 kPa, resulting in excess pore pressure ratios of 1.0 and
sand layer (DR ≈ 40%, n ≈ 0.41), 1.3 m of a clayey silt layer 0.096, respectively. The layers resulted in a compressibility ratio
(20% clay and 80% silt), 4 m of the medium dense sand layer of mv ¼ 1=20, which was estimated using Eqs. (21) and (24) with
(DR ≈ 60%, n ≈ 0.39), and a dense sand layer (DR ≈ 83%, compressibility (mvo ) computed by taking the mean value of the
n ≈ 0.36) (Fig. 11). The effective unit weight (γ 0 ) of all of the sand modulus parameter (m) for the sand layers (Fig. 9). Such a relative
layers was very close to 10 kN=m3 . The permeabilities (k) of the compressibility ratio of mv ¼ 1=20 for the soil layers is reasonable
loose sand, the medium dense, and the dense sand layer were considering the loose liquefied Lu layer will certainly be much
0.026 m=s, 0.022 cm=s, and 0.022 cm=s, respectively. The models more compressible than the dense NLu layer (see Fig. 9). For
were shaken with scaled Santa Cruz earthquake motions of Mw ¼ EQM4 in the SKS03 test, the Lu layer consisted of about 1.8 m
6.9 from the Loma Prieta 1989 earthquake. The duration of shaking of the medium dense sand below the relatively impermeable silt
of the earthquake motion was about 30 s. The peak ground accel- layer (i.e., H Lu ¼ 1.8 m) and the NLu layer consisted of 2.2 m of
erations (PGAs) used for the liquefaction potential assessment for medium dense sand and dense sand layer below it (i.e., H NLu ¼
the shaking events EQM3 and EQM4 were 0.23 g and 0.16 g, 10.2 m with Z ¼ 10.8 m). Other values computed for the Lu
respectively (Sinha et al. 2023). The measured normalized overbur- and NLu layers were 89.54 kPa and 151.8 kPa for their initial
den corrected cone tip resistance (qc1Ncs ) and peak excess pore mean effective stresses, respectively, 1.0 and 0.24 for their excess

© ASCE 04024014-11 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2024, 150(4): 04024014

Downloaded from ascelibrary.org by Indian Institute of Technology Bombay on 07/21/25. Copyright ASCE. For personal use only; all rights reserved.

Fig. 11. Redistribution effects for shaking event EQM4 of centrifuge model test SKS03: (a) measured normalized overburden corrected cone tip
resistance (qc1Ncs ); (b) cyclic stress ratio (CSR) and factor of safety against liquefaction (FSliq ); and (c) comparison of estimated earthquake-induced
(uue ) and redistributed (ude ) excess pore pressures with measured excess pore pressures. (Data from Sinha et al. 2021b.)

pore pressure ratios, respectively, uue−Lu ¼ 89.54 kPa, uue−NLu ¼ seen from the figures that the uue (estimated without considering
36.04 kPa, and mv ¼ 1=12.5. redistribution) significantly underestimates ue in the NLu layers.
In both centrifuge models, the NLu layer was below the Lu layer In the SKS02 test, uue−NLu ¼ 15.03 kPa compared to the measured
with an impermeable clay or silt layer above the Lu layer and an ue ≈ 90 kPa [Fig. 10(c)]. Similarly, in the SKS03 test, uue−NLu ¼
impermeable boundary condition below the NLu layer. As 36.05 kPa compared to the measured ue ≈ 76 kPa [Fig. 11(c)].
a result, Eqs. (6) and (7) were used to estimate ude in the NLu On the other hand, excess pore pressure in the NLu layer considering
and Lu layers. Figs. 10(c) and 11(c) show the estimated ude profiles. redistribution (ude−NLu ¼ 98 kPa and ude−NLu ¼ 75 kPa) matched
Estimated ude−NLu values for EQM3 in SKS02 and EQM4 in SKS03 quite well with both centrifuge test (SKS02 and SKS03) results.
were 98 kPa and 75 kPa, respectively. While redistribution increased ue in the NLu layers, it prevented
It should be noted that redistribution of excess pore pressures, as liquefaction in the Lu layers. The predicted thickness of the Lu
well as the complex overall dynamic response due to liquefaction in layer where redistribution prevented liquefaction is consistent with
the adjacent layers, could be the reasons for smaller excess pore the observations of the centrifuge test. For EQM3 of SKS02, the
pressures in some thicknesses of the loose and medium dense developed approximate procedure predicted liquefaction preven-
Lu sand layer. However, the large excess pore pressure developed tion in the 1.9 m thickness of the loose sand layer from its bottom
in the underlying dense sand layer (Figs. 10 and 11) clearly dem- compared to the 3 m thickness as observed in the centrifuge test
onstrates the significance of redistribution. Similar observations [Fig. 10(c)]. The difference in the predicted thickness of the Lu
have been made by Cubrinovski et al. (2019) in the system-level layer may be due to the conservativeness in the proposed approxi-
response of interbedded liquefiable deposits. mate procedure. For EQM4 of SKS03, the procedure predicted
The thickness of the Lu layer for liquefaction prevention liquefaction prevention in the entire 1.8 m of the Lu layer, like
(rpd the observations from the centrifuge test [Fig. 11(c)]. It must be
u < 0.9) was found by solving Eq. (18). The ue generation time
(tu ) was taken as 30 s, equal to the duration of earthquake shaking. noted that more data is required to fully validate the developed
The sand used in the centrifuge model tests was Ottawa F-65 with procedures.
parameter “ b” equal to 0.15 (Bastidas 2016). For EQM3 of SKS02,
the thickness of the Lu layer where liquefaction would be prevented Summary and Conclusions
(for calculated t̄ ¼ 42.3) was predicted to be 1.9 m from the bottom
of the loose sand layer. For EQM4 of SKS03, the procedure pre- This paper described a new procedure to account for the redistrib-
dicted liquefaction prevention (for calculated t̄ ¼ 10.7) in the full ution of excess pore pressures that will either increase ue in layers
thickness (1.8 m) of the Lu layer. determined to be non-liquefied (NLu) (under undrained conditions)
or decrease ue in layers determined to be liquefied (Lu) (under un-
drained conditions). The development of the procedure involved
Comparison of Results with Centrifuge Test
studying redistributed excess pore pressures (ude ) in two types of
Excess pore pressures in the NLu layer considering redistribution simple layered systems: an NLu below an Lu layer and an NLu
matched quite well with the centrifuge test. The comparison of es- layer above an Lu layer, which formed the basis for estimating
timated ude with the measured ue for EQM3 in SKS02 and EQM4 in redistributed excess pore pressures in multilayered systems. While
SKS03 are shown in Figs. 10(c) and 11(c), respectively. It can be redistribution increased ue in the NLu layer, it also decreased ue in

© ASCE 04024014-12 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2024, 150(4): 04024014

 the Lu layers. Estimating accurate ue values in the NLu layer is policies. All the data used in this study are made available through
critical for estimating the end-bearing resistance of deep founda- DesignSafe project PRJ-2828 (Sinha et al. 2021a, b).
tions. Development of a large ue at a pile’s tip bearing (an NLu)
layer can significantly decrease the tip capacity and cause large set-
tlements. At the same time, the decrease of ue can be beneficial for Acknowledgments
increasing the liquefaction resistance of the Lu layer. Simplified
steps and equations were provided to estimate ude in soil layers. This work was performed with funding from the California Depart-
The paper also described a criterion for the maximum thickness ment of Transportation under Agreement 65A0688. The centrifuge
of the Lu layer that can be prevented from liquefaction due to tests were made possible by the staff and facilities of the Center for
redistribution effects. Preventing liquefaction in a deep thin Lu Geotechnical Modeling at UC Davis. The centrifuge facility at UC
layer because of pore pressure migration to the adjacent NLu layers Davis is part of the NSF Natural Hazards Research Infrastructure
Downloaded from ascelibrary.org by Indian Institute of Technology Bombay on 07/21/25. Copyright ASCE. For personal use only; all rights reserved.

might prove extremely valuable in reducing the risk of liquefaction- (NHERI) program under Award CMMI 2037883.
related failures and the cost associated with remediation. Applying
the procedure to centrifuge tests showed that the developed analyti-
cal procedure reasonably predicted ue in the soil layers, much Notation
better than the existing simplified procedures that ignore redistrib- The following symbols are used in this paper:
ution effects. Finally, a Microsoft Excel spreadsheet has been
b = slope of cyclic resistance curve;
developed (see Supplemental Materials) following the proposed
cv = coefficient of consolidation;
analytical procedure for estimating redistribution effects in liquefi-
DR = relative density of sand;
able layers.
Several simplifying assumptions were used to present a com- d = superscript to denote dissipation from redistribution;
plete procedure with many of these assumptions likely be more FSLu = factor of safety against liquefaction;
conservative than necessary. For example, the redistribution of ex- H = thickness;
cess pore pressures in the NLu layer assumed no water drainage H̄ = thickness ratio H̄ ¼ HNLu =H Lu ;
outside the Lu and NLu layers until redistribution was achieved. Ld = subscript to denote a layer that liquefies due to
This condition would be applicable for the case when the surround- redistribution;
ing soil layers are relatively impermeable (such as clay, silt, and Lu = subscript to denote a layer that would liquefy for a given
sand silt mixtures). For the case of partially or fully drained hy- earthquake loading;
draulic boundary conditions, the presented approximate procedure mv = compressibility;
would result in conservative estimates of redistributed excess pore mv = compressibility ratio mv ¼ mv−NLu =mv−Lu ;
pressures. Extending the two-layer systems to multilayer systems NLd = subscript to denote a layer that will not liquefy due to
conservatively assumed that the Lu layer fully contributed to excess redistribution;
pore pressures above and below the layer. In the analytical study of NLu = subscript to denote a layer that would not liquefy for a
increased liquefaction resistance in the Lu layer from redistribution, given earthquake loading;
the assumption of a constant compressibility ratio might be overly n = porosity;
conservative since it increases non-linearly with excess pore pres- pd = superscript to denote partially drained condition during
sure development. On the other hand, factors such as the duration earthquake loading;
of shaking and the extent of liquefaction are not well captured in the qc1Ncs = overburden corrected cone tip resistance;
present analytical study, which might result in the underestimation ru = excess pore pressure ratio;
of excess pore pressure estimates in the Lu and NLu layers. For
ṙu = time rate of excess pore pressure ratio;
long-duration shakings and with prominent liquefaction in the
su = undrained shear strength;
Lu layer, continuous redistribution can occur throughout the
T r = time factor of redistribution;
shaking resulting in very high excess pore pressures in soil layers.
Future refinements of the procedure may be able to estimate redis- t = time during a shaking event;
tribution effects better while avoiding some excessive conserva- td = time required for ue dissipation from redistribution;
tism. For sites where the previously listed factors may play an tu = duration of ue generation during an undrained
important role, an advanced 1-D or 2-D site response analysis loading;
with a fully coupled analysis of excess pore pressure generation/ t̄ = time ratio t̄ ¼ tg =tr ;
dissipation models might also be performed to estimate realistic U = degree of consolidation;
excess pore pressures in soil layers. u = superscript to denote undrained earthquake loading;
In summary, it is recommended that redistribution effects in ue = excess pore pressure;
liquefiable layers are evaluated in practice, especially for the design Z = depth to the top of the NLu layer;
0
of deep foundations. The proposed procedures are a first attempt σvo = initial effective stress;
toward the evaluation of redistribution effects. Until the proposed γ 0 = effective unit weight of soil; and
simplified procedures are better validated, they may be regarded as γ w = unit weight of water.
a screening tool to help engineers decide whether more sophisti-
cated analyses are required for this purpose.
Supplemental Materials

Data Availability Statement The Microsoft Excel spreadsheet that has been created using the
developed analytical procedure for estimating redistribution effects
Some or all data, models, or code generated or used during the in the liquefiable layers can be found online in the ASCE Library
study are available in a repository online per funder data retention (www.ascelibrary.org).

© ASCE 04024014-13 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2024, 150(4): 04024014

 References Robertson, P. K. 2015. “Comparing CPT and Vs liquefaction triggering
methods.” J. Geotech. Geoenviron. Eng. 141 (9): 04015037. https://doi
Bastidas, A. M. P. 2016. “Ottawa F-65 sand characterization.” Ph.D. .org/10.1061/(ASCE)GT.1943-5606.0001338.
dissertation, Dept. of Civil and Environmental Engineering, Univ. of Scott, R. F. 1986. “Solidification and consolidation of a liquefied sand col-
California Davis. umn.” Soils Found. 26 (4): 23–31. https://doi.org/10.3208/sandf1972
Bekir Afacan, K. 2020. “Estimation of excess pore pressure generation .26.4_23.
and nonlinear site response of liquefied areas.” In Geotechnical Seed, H. B., and K. L. Lee. 1966. “Liquefaction of saturated sands during
engineering—Advances in soil mechanics and foundation engineering. cyclic loading.” J. Soil Mech. Found. Div. 92 (6): 105–134. https://doi
London: IntechOpen. https://doi.org/10.5772/intechopen.88682. .org/10.1061/JSFEAQ.0000913.
Booker, J. R., M. S. Rahman, and H. B. Seed. 1976. GADFLEA: A com- Seed, H. B., P. P. Martin, and J. Lysmer. 1976. “Pore-water pressure
puter program for the analysis of pore pressure generation and dissi- changes during soil liquefaction.” J. Geotech. Eng. Div. 102 (4):
pation during cyclic or earthquake loading. Berkeley, CA: Earthquake 323–346. https://doi.org/10.1061/AJGEB6.0000258.
Sinha, S. K., K. Ziotopoulou, and B. L. Kutter. 2021a. “SKS02: Centrifuge
Downloaded from ascelibrary.org by Indian Institute of Technology Bombay on 07/21/25. Copyright ASCE. For personal use only; all rights reserved.

Engineering Research Center.

Bray, J. D., and J. Macedo. 2017. “Simplified procedure for estimating test of liquefaction-induced downdrag in uniform liquefiable deposit.”
liquefaction-induced building settlement.” Soil Dyn. Earthquake Eng. In Centrifuge testing of liquefaction-induced downdrag on axially
102 (Feb): 215–231. https://doi.org/10.1016/j.soildyn.2017.08.026. loaded piles. Miami: DesignSafe-CI. https://doi.org/10.17603/ds2
Cetin, K. O., R. B. Seed, R. E. Kayen, R. E. S. Moss, H. T. Bilge, M. Ilgac, -d25m-gg48.
and K. Chowdhury. 2018. “SPT-based probabilistic and deterministic as- Sinha, S. K., K. Ziotopoulou, and B. L. Kutter. 2021b. “SKS03: Centrifuge
sessment of seismic soil liquefaction triggering hazard.” Soil Dyn. Earth- test of liquefaction-induced downdrag in interbedded soil deposits.” In
quake Eng. 115 (Jan): 698–709. https://doi.org/10.1016/j.soildyn.2018 Centrifuge testing of liquefaction-induced downdrag on axially loaded
.09.012. piles. Miami: DesignSafe-CI. https://doi.org/10.17603/ds2-wjgx-tb78.
Chiaradonna, A., G. Tropeano, A. d’Onofrio, F. Silvestri, and D. Park. Sinha, S. K., K. Ziotopoulou, and B. L. Kutter. 2022a. “Centrifuge model
tests of liquefaction-induced downdrag on piles in uniform liquefiable
2015. “Application of a simplified model for the prediction of pore pres-
deposits.” J. Geotech. Geoenviron. Eng. 139 (9): 04022048. https://doi
sure build-up in sandy soils subjected to seismic loading.” In Proc., 6th
.org/10.1061/(ASCE)GT.1943-5606.0002817.
Int. Conf. on Earthquake Geotechnical Engineering. Christchurch, NZ:
Sinha, S. K., K. Ziotopoulou, and B. L. Kutter. 2022b. “Numerical model-
International Society for Soil Mechanics and Geotechnical Engineering.
ing of liquefaction-induced downdrag: Validation against centrifuge
Cubrinovski, M., A. Rhodes, N. Ntritsos, and S. Van Ballegooy. 2019.
model tests.” J. Geotech. Geoenviron. Eng. 148 (12): 04022111.
“System response of liquefiable deposits.” Soil Dyn. Earthquake Eng.
https://doi.org/10.1061/(asce)gt.1943-5606.0002930.
124 (Jun): 212–229. https://doi.org/10.1016/j.soildyn.2018.05.013.
Sinha, S. K., K. Ziotopoulou, and B. L. Kutter. 2023. “Displacement-based
Garnier, J., C. Gaudin, S. M. Springman, P. J. Culligan, D. Goodings, D.
design of axially loaded piles for seismic loading and liquefaction-
Konig, B. Kutter, R. Phillips, M. F. Randolph, and L. Thorel. 2007.
induced downdrag.” J. Geotech. Geoenviron. Eng. 149 (9): 1–15.
“Catalogue of scaling laws and similitude questions in geotechnical https://doi.org/10.1061/JGGEFK.GTENG-11178.
centrifuge modeling.” Int. J. Phys. Modell. Geotech. 7 (3): 1–23. https:// Taylor, D. W. 1948. Fundamentals of soil mechanics. New York: Wiley.
doi.org/10.1680/ijpmg.2007.070301. Tokimatsu, K., and H. B. Seed. 1984. Simplified procedures for the evalu-
Idriss, I. M., and R. W. Boulanger. 2008. Soil liquefaction during earth- ation of settlements in clean sands. NSF/CEE-84027. Berkeley, CA:
quakes. Oakland, CA: Earthquake Engineering Research Institute. Univ. of California Berkeley.
Ishihara, K., and M. Yoshimine. 1992. “Evaluation of settlements in sand Ueng, T. S., Z. F. Wang, M. C. Chu, and L. Ge. 2017. “Laboratory tests for
deposits following liquefaction during earthquakes.” Soils Found. permeability of sand during liquefaction.” Soil Dyn. Earthquake Eng.
32 (1): 178–188. https://doi.org/10.3208/sandf1972.32.173. 100: 249–256. https://doi.org/10.1016/j.soildyn.2017.05.037.
Janbu, N. 1985. “Soil models in offshore engineering.” Géotechnique Yoshimi, Y., and F. Kuwabara. 1973. “Effect of subsurface liquefaction on
35 (3): 241–281. https://doi.org/10.1680/geot.1985.35.3.241. the strength of surface soil.” Soils Found. 13 (2): 67–81. https://doi.org
Lee, K. L., and A. Albaisa. 1974. “Earthquake induced settlements in /10.3208/sandf1972.13.2_67.
saturated sands.” J. Geotech. Eng. Div. 100 (4): 387–406. https://doi Youd, T. L., et al. 2001. “Liquefaction resistance of soils: Summary report
.org/10.1061/AJGEB6.0000034. from the 1996 NCEER and 1998 NCEER=NSF workshops on evalu-
Mele, L., A. Chiaradonna, S. Lirer, and A. Flora. 2021. “A robust empirical ation of liquefaction resistance of soils.” J. Geotech. Geoenviron. Eng.
model to estimate earthquake-induced excess pore water pressure in 127 (10): 817–833. https://doi.org/10.1061/(ASCE)1090-0241(2001)
saturated and non-saturated soils.” Bull. Earthquake Eng. 19 (Feb): 127:10(817).
3865–3893. https://doi.org/10.1007/s10518-020-00970-5. Zhang, G., P. K. Robertson, and R. W. I. Brachman. 2004. “Estimating
Nagase, H., K. Ishiharai, and K. Ishihara. 1988. “Liquefaction-induced liquefaction-induced lateral displacements using the standard penetra-
compaction and settlement of sand during earthquakes.” Soils Found. tion test or cone penetration test.” J. Geotech. Geoenviron. Eng. 130 (8):
28 (1): 65–76. https://doi.org/10.3208/sandf1972.28.65. 861–871. https://doi.org/10.1061/(ASCE)1090-0241(2004)130:8(861).

© ASCE 04024014-14 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2024, 150(4): 04024014