Vo1．4．No．1 EARTHQUAKE ENGINEERING AND ENGINEERING VIBRATION June， 2005

Article ID：1 67 l一3664(2005)0 1．0047．1 1

Physical modeling and visualization of soil liquefaction

under high confining stress

Lenart GonzfilezH，Tarek Abdoun”，Mourad Zeghal ，Vivian Kallou and Michael K．Sharp

1．Dept．ofCivil and Environmental Engineering，Rensselaer Polytechnic Inst． PI)，Troy,NY,USA

2．MueserRutledge ConsultingEngineers，225 W 34 Street,New York,NY,USA

3．USArmy Engineer Research andDevelopment Center Vicksburg,MS,USA

Abstract：The mechanisms ofseismically—induced 1iquefaction ofgranular soils under high confining stresses are stil1 not

fully understood．Evaluation ofthese mechanisms is generally based on extrapolation ofobserved behavior at shallow depths．

Three centrifuge mode1 tests were conducted at RPI’s experimental facility to investigate the effects ofconfining stresses on

the dynamic response of a deep horizonta1 deposit of saturated sand．Liquefaction was observed at high confining stresses in

each of the tests．A system identification procedure was used to estimate the associated shear strain and stress time histories．

These histories revealed a response marked by shear strength degradation and dilative patterns．The recorded accelerations

and pore pressures were employed to generate visual animations ofthe models．These visualizations revealed a liquefaction

front traveling downward and 1eading to 1arge shear strains and isolation of upper soil1ayers．

Keywords：centrifuge modeling；high confining stress；liquefaction；system identification；visualization

1 IntrOductiOn

Liquefaction is a major factor affecting the earthquake resistance of civil systems．Remedial countermeasures．

costing hundreds of millions of dollars of public funds

annually,are usually undertaken to address concerns

over the liquefaction possibility of soils in seismically

active areas．Historically,numerous foundations，dams

and other systems were constructed with little to no

consideration ofearthquake excitations and must now be

reexamined in light of potential damage from this threat．

An open question in this regard is the effect of depth

and confining stresses on liquefaction potentia1．There

have been no field observations of liquefaction at depths

greater than about 30 m fYoud et a1． 2001)．However,

this lack of evidence is not a proof that high confining

stresses beneath a foundation or an embankm ent dam

would hinder the development of significant excess pore

pressures and associated stiffness degradation．

Centrifuge tests have been used by researchers

to study the development of soil liquefaction in both

level and inclined models，including homogeneous

COrrespOndence tO： Lena~ Gonzfilez， Dept． of Civi1 and

Environmenta1 Engineering， Rensselaer Polytechnic Inst．

(RPI1，Troy，N Y_USA

Tel：(51 8)276 8143：Fax：(518)276 4833 E—mail：gonzal rp1．edu

tResearch Associate；{Assistant Professor；~Associate Professor；

*Project Engineer；#Technica1 Director

Received 2005—03—2 1：Accepted 2005—04—20

and multiple layers deposits． However。 most of

these studies have simulated shallow layers of 6 m

to l 5m．Currently．there is little inform ation about

the development of liquefaction at larger depths and

confining stresses。either from the field or laboratory

fincluding 1．g and centrifuge test)． Centrifuge tests were conducted at Rensselaer

Polytechnic Institute(RPI1 fGonzfilez et a1． 20021 to investigate the behavior of saturated soils under

high confining stresses．More specifically,three level

ground models，simulating deep deposits with either

a homogeneous profile or a dense layer overlaying a

medium dense stratum．were tested in a large laminar

box．In two of the models．a surcharge steel plate was

placed on top of the deposit to increase confinement．

Simple efficient tools of system identification and

visualization were used to analyze the seismic response

of the tested soils under high confining stresses．

2 Testing program

This paper presents results and analyses of two

centrifuge tests fFig．1 1 corresponding to models 1 and 3

ofthe experimental study mentioned above(Gonzfilez et

a1．．2002)．Table 1 lists the main features and parameters

of both models(in prototype units)．Whereas，both models simulate deep level deposits with a maximum

effective confining pressure at the bottom of 380 kPa；

there are significant differences in their configurations．

Model 1(Fig．1(a))simulates a 38 m unifonTl deposit of

a saturated medium．dense sand at a D一

= 55％ ．M odel 3

维普资讯 http://www.cqvip.com

48 EARTH0UAKE ENGINEERING AND ENGINEERING VIBRATION VO1．4

(Fig．1(b))idealizes a two layers deposit；a 1 6 m stratum of dense sand at a D

一

= 75％ over an 8 m medium—dense

layer at a D一

= 55％ ．A steel plate was placed on top

of Model 3 to apply a surcharge of 1 40 kPa(prototype

units)．The response of the deposits to dynamic base excitations were monitored using dense vertical

arrays of accelerometers(measuring the lateral motion、

and pore pressure transducers which were installed

at different elevations within the models (Fig．1、．

Table l Testing parameters of centrifuge models

,Note． 75fora16m thicktoplayer．55foran 8m thickbottom layer

3 System identification and visualization

A simple system identification procedure was

used to investigate the response of the tested deposits．

The lateral response of these deposits was idealized

using a one—dimensional shear beam and the involved

A

T 。 ，

Nevada sand ’Ac6 ’Ac7 r可 D 55％ 。Ac5 =

·P5 -Ac4 ·P6

iAc3 ·P4

-At2 ．P3

·P1 -Ac1 ·P2

Ac

nput m otion—·一 —— —---50 cycIes．24g．1 80Hz

fa、M ode1 1

t-VDTl LVDT2 Ac7 ● ■ 一

Ne ■A a ． ’ = co 1

D ：75％ ~P7 mAc5 ·P8 ·

~P5 mAc4 ·P6

·P4 -Ac3

Ncvada sand IAc2 ~P3 E D 一55％ ·P1 _Ac1 ·P2 l

Input motion · ———- 50 cycIes，1 6g．1 20Hz

·Pore pressure transdtlcer - Accelerometer

● LVDT Surcharge=1．75kPa

fb、M ode1 3

z=0m ：=1 3111 ==6．3m

：一l3 1m

z=24．81n

=一30．8111 ==37m ：一38m

=--0m z-0 61n ==3 6m

=一7 4111

==13．4m

==19 3m

==22 8m z=23 7m

Fig．1 Setups and instrument locations of centrifuge m odels

shear stress and shear strain histories were evaluated

directly from the recorded accelerations at a number

of locations within the instrumented soil layers(Zeghal P，a1．．1995、．At the location of the accelerometers and

halfway between accelerometers fFig．2)these stresses

are respectively given by：

r ， (，)=r (，)+p 三堡二 ±堕 ；TO=0, =1，2，3，

in which the subscript

accelerometer at level

refers to the location of the

，， the subscript i-1／2 refers

to the location halfway between the rf一1 th and the

accelerometer(at level(二H+二，)／2)，／／~=ii(z，，，)is acceleration at level and time instant，．P is average

，

mass density between levels ___and and A7 is

spacing interval between accelerometers(Fig．2、．The corresponding shear strains are given by：

y，(f)= t+ l+

= l 2 3，

A7i ]； at the accelerometer locations，and by

Free surface

△Z+1

口 Accelerometer

—

l

+1

Fig．2 Schematic of a downhole arra) of accelerometers

used t0 evaluate shear stresses and strains

维普资讯 http://www.cqvip.com

N()．1 Lenan Gonzfilez el a1．：Physical modeling and visualization of soil liquefaction under high confining stress 49

(f) ； ，2’3．⋯ (4)

halfway between accelerometers． In Eqs．(3)and(4)，

Hi~H(Z，，，)is absolute displacement(evaluated through double time—integration ofthe corresponding acceleration

history／／=(z，，，)．This simple technique was shown to be an effective tool to evaluate the seismic shear stress．strain

histories of actual sites(Zeghal et a1．．1 995；Zeghal and

Elgamal，1 994)，as well as a number ofcentrifuge models

ofsoil systems(Zeghal et a1． 1 998；Elgamal et a1．．1 996)．

The shear beam assumption was also employed

along with linear interpolations to evaluate the

displacement and pore pressure fields ofM odels 1 and 3．

These fields were then used to develop visual animations

of the dynamic response of these models．The visual

animations may be accessed on RPI’s Geotechnical

Centrifuge Center web page at http：www．nees．rp i．edu

under Research and under Visualizations．

4 Experimental results and analysis

4．1 M odel 1

A sketch ofM odel1．the associated instrumentation．

and employed laminar box is presented in Fig．1 fa1．The

model height was approximately 0．32 m．simulating

under a l 20 g level a 3 8 m prototype soll deposit．

The soll consisted of Nevada sand placed at a relative

density of about 55％．This soll was fully saturated

with a de．ionized．de．aired water．metulose solution

having 60 times the viscosity of water．Thus．M odel l

simulates a field deposit with a permeability twice that

ofNevada sand rabout l 0～m／s)．The model was excited by 50 cycles of sinusoidalinput acceleration parallel to

the base of the laminar box．The input had a uniform

amplitude of about 24 g and a frequency of l 80 Hz．

For the l 20 g-centrifuge acceleration of the test．this

corresponds in prototype units to a peak acceleration of

0．2 g and a frequency of 1．5 Hz(Fig．31． 4．1．1．Acceleration and excess pore pressure records

The input motion and recorded accelerations at

difIferent depths in the deposit are shown in Fig．3．

Near the ground surface，the acceleration amplitudes

decreased significantly after about 2 cycles of shaking

due to liquefaction．This phenomenon of very low

accelerations propagated to deeper locations in the

deposit as a liquefaction front(Sharp and Dobry 20021 moved downward during shaking．The arrows in Fig．3

indicate the time instants when the liquefaction front

reached the accelerometer depths．These instants were

estimated based on the pore pressure measurements

(Fig．41．The decrease in acceleration amplitudes，after the passage of the liquefaction front is related to a

significant reduction in shear stiffness and strength as

compared with the original pre．excitation conditions．

Note however that the recorded acceleration at 3 7 m

depth(Fig．3、exhibited no decrease in amplitude after

liquefaction occurred(as indicated by a pore pressure

ratio about equal to 1．0 in Fig． 4、 because of the

effects of the nearby imperm eable rigid base． Figure 3

also indicated that the residual acceleration after

liquefaction had an amplitude that increased with depth

within the deposit．

The excess pore pressure time histories for Model l

(Fig．4)revealed that liquefaction extended along the whole soll deposit after about 24 cycles of shaking．The

number of cycles necessary to reach liquefaction at a

specific location increased with depth．The recorded

pore pressures indicated clearly that liquefaction started

near the free surface and progressed downward．in

agreement with the trend exhibited by the acceleration

time histories．The pore pressure isochrones(Fig．51 confirm ed the downward pattern of liquefaction

development during shaking．Figure 5 shows that the

maximum excess pore pressure did not occur at the

bottom of the model for the first l 3 s of shaking．but

rather at some intermediate depth (ranging from 25

m to 30 m、．This pattern contrasts with that observed

in centrifuge tests of shallow models fe．g．Abdoun．

1997)． The slow rate of pore pressure buildup at greater

depths generated hydraulic gradients pointing

downward．The excess pore pressure at a 25 m depth

was about l 50 kPa at time，= 4s．while this excess

was about ll 5 kPa at levels deeper than 30 m fFig．51．

A similar response was also observed at，= 2 s．6 s

and 8 s．These negative hydraulic gradients must have

induced an increase in excess pore pressure near the

bottom ofthe mode1．

4．1．2．System identification analysis

The recorded accelerations were employed to

estimate the deposit shear stress and strain histories(Eqs． 1．4、．Figures 6 and 7 present these histories at locations

halfway between accelerometers(the time instants of the arrival of the liquefaction front are indicated by arrows

on these figures)．The evaluated shear strains near the ground surface decreased substantially after about 2

cycles of shaking due to liquefaction．A significant

decrease in strain amplitudes was observed at deeper

location at subsequent time instants，in consistency with

the liquefaction front moving downward(as described

above)． Figure 6 shows a zone of high cyclic shear strains

in the vicinity of the liquefaction front．In other words．

1arge shear strain amplitudes started developing as the

liquefaction front approached a given elevation．As soon

as the liquefaction front passed this elevation．the soll

at that location became isolated and accelerations．shear

stresses and shear strains reduced significantly．This

phenomenon i s in agreement with the observ ations made

by Sharp and Dobry (2002)．The zone of high shear strains may be explained by the continuous softening

and decrease in soil stiffness as the liquefaction front

approached a certain location．which culminates with

liquefaction and a substantial decrease in effective

维普资讯 http://www.cqvip.com

50 EARTHQUAKE ENGINEERING AND ENGINEERING VIBRATION 、，o1．4

0．4

0．2

0．0

— 0．2

— 0．4

0．4

0．2

0．0

— 0．2

— 0．4

0．4

0．2

0．0

— 0．2

— 0．4

0．4

0．2

0．0

— 0．2

— 0．4

0．4

0．2

0．0

— 0．2

— 0．4

0．4

0．2

0．0

— 0．2

— 0 4

0．4

0．2

0．0

— 0．2

— 0．4

0 20 30 40

Time(s)

Fig．3 Acceleration time histories，Model 1(the arrows

indicate the time instants of the arrival of the lique—

factionfront)

stresses and shear strength．

The shear stress—strain histories at different depths of

the deposit are shown in Fig．8．A significant reduction

400

300

200

100

0 20 30 40

Time(s)

Fig．4 Excess pore pressure buildup time histories，M odel 1

0

g 、

专

0 20

30

0 100 200 300 400

Excess pore pressure(kPa)

Fig．5 Isochrones of excess pore pressure buildup during

shaking，M odel 1

in soil shear stiffness and strength occurred at 9．7 1TI

depth after about three cycles of base excitation．At 34

m depth，the soil exhibited a significant stiffness during

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 D 0伽 姗 0彻 姗 姗 0彻 姗 0

一 )I一。．I ∞∞。．I 。．10 ∞∞。u)(

一 0 BJ0Iauu《

维普资讯 http://www.cqvip.com

No．1 Lenan Gonza1ez Pt a1．：Physical modeling and Visualizati。n。f s。il liquefacti。n under high c。nfining stress 51

1．0

0．5

0．0

— 0．5

． 1．0

1．0

0．5

0．0

— 0．5

— 1．0

1．0

0．5

0．0

— 0．5

— 1．0

1．0

0．5

0．0

— 0．5

— 1．0

1．0

0．5

0．0

— 0．5

— 1．0

0 10 20 30 40

Time(s)

Fig．6 Shear strain time histories，Model 1(the arrows

indicate the time instants of the arrival of the lique-

factionfront)

the first ten seconds of shaking(Fig．81，or about 1 5

cycles of base excitation．The stress-strain response

pattern changed considerably once the pore pressure

buildup reached values approaching the initial venical

effective stresses at this depth．The response was marked

bv stiffness degradation with small stresses and large

strains(as large as 0．5％、along with some evidence of a dilative behavior associated with the observed decrease

in excess pore pressure during each load cycle(Fig．4)． 4．1．3．Visualization analysis

The field of excess pore pressure was evaluated

using simple linear interpolations between

measurements provided by the vertical array of the

transducers P2．P4 and P6 of Fig．1 fa1：in view of the

fact that the lateral variations in pressure were minor

f0r any given level(as shown in Fig．41．Figure 9

shows the pore pressure ratios(Fu ratio of excess pore pressure buildup and the corresponding initial venica1

：

j 0 10 20 30 40

Time(s)

Fig．7 Shear stress time histories，Model 1 (the arrows

indicate the time instants of the arrival of the lique—

factionfront)

effective stress)and lateral displacements of Model l at selected time instants of a visual animation of the

whole time history．The displacements were amplified

3 0 times for visual clarity．Figure 9 also includes the

input motion，where the dots correspond to the selected

time instants．The visual animation confirmed clearly

that the liquefaction front(areas of dark blue1 started near the free surface and propagated downward as

the shaking progressed．The upper third of the model

liquefied and became isolated after 7 S of cyclic base

excitation．The whole soil deposit reached liquefaction

after l 8 S(Fig．9(e11_The visualization analysis also

showed：(1 1 a noticeable zone of large shear strains fi_e．，

relatively large displacement gradients)traveling with

the liquefaction front(Figs．9(b)and 9(c))，and(2)the process of the upper layers becoming isolated while the

liquefaction front moved downward．

∞ 0 ∞ ∞ 如 0 ∞∞ 如 0 如 ∞∞ 0 ∞ ∞∞ 如 0 如 ∞

1 - 1 1 _ 1 1 — 1 1 — 1 1 — 1

一≈ )I一∞∞0-I=l∞-I≈oL{∽

(_ 一【I一≈-I=l∞-I≈oL{∽

维普资讯 http://www.cqvip.com

52 EARTHQUAKE ENGINEERING AND ENGINEERING VIBRATION V61．4

— 1．2 —0．8 —0．4 0 0．4 0．8 】．2

Shear strain f％1

6 12 18 24 30 36

Time(S)

Fig．8 Shear stress-strain histories．M odel 1

4．2 M odel 3

A sketch of M odel 3 and the associated

instrumentation iS shown in Fig．1 fb1．The model height

was about 0．30 m．simulating under an 80 g level a 24

m prototype soil deposit．The soil consisted of 0．1 m

of Nevada sand having a D．≈55％．topped with 0．2 m

ofthe same Nevada sand at a D ≈ 75％ ．A steel plate

was placed on top of the deposit before starting the

saturation process in order to increase the prototype

vertical effective stress at all depths by 1 40 kPa The

steel plate had a number ofholes to allow some drainage

of the pore fluid of the deposit．

The two layers of soil were fully saturated with a

de．ionized．de．aired water．metulose solution having 40

times the viscosity ofwater．Similarly to Model 1．Model

3 simulates a field deposit wi【h a permeability twice

善 0．2

看 0 一 0．2

< 0 5 l0 15 20 25 30 35

Time(S)

3 5S

(b)，=7s

(C)，1 10s

(d)，1 14s

(e)，1 1 8s

} 1 0．4

} 1 0．2

【1j 0,0

Fig．9 Pore pressure ratios and estimated lateral displace-

ments for selected time instants．M odel 1

that of Nevada sand(about 1 0。m／s)．The model was excited by 50 cycles of a sinusoidal input acceleration

如 0 Ⅷ 如 0 Ⅷ 如 0 Ⅷ 如 0 Ⅷ 如 0 Ⅷ

∞∞0 ∽磊0L{∽

维普资讯 http://www.cqvip.com

N。1 Lenan G。nzMez et a1．：Physica1 mode1ing and Visua1izati。n。f soi1 liquefacti。n under hi

gh c。n ning stress

0 20 30 40

Time(S)

Fig．1 l Excess pore pressure buildup time histories，M odel 3

0

5

1 5

20

Fig．1 0 Acceleration time histories，Model 3(the arrows 0

indicate the time instants of the arrival of the lique—

faction front)

1O0 200 300 400 500

Excess pore pressure(kPa)

Fig．1 2 lsochrones of excess pore pressure buildup during

shaking，M odel 3

姗 0蜘 Ⅻ 0枷 Ⅻ 0伽 湖 瑚 0枷 姗 0

一高 )I一0_IT1∽∞0_I【=l 0_10 CI∞∽00×

．

互 ． 4 2 0 2 4 4 2 0 2 4 4 2 0 2 4 4 2)2 4 4； 2 4 }； i l； i 叫 舢州 舢删 叭 眦 ̈ 叭

一 —c0I】 言 00<

维普资讯 http://www.cqvip.com

54 EARTHQUAKE ENGINEERING AND ENGINEERING VIBRATION VO1．4

1．0

0．5

0．0

— 0．5

— 1．0

1．0

0．5

0．0

— 0．5

— 1．0

1．0

0．5

0．0

— 0．5

— 1．0

1．0

0．5

0．0

— 0．5

— 1．0

1．0

0．5

0．0

— 0 5

一 1．0

Fig．13 Picture of M odel 3 after the test

l ’1

0 10 20 30 40

Time(S)

Fig．14 Shear strain time histories，Model 3(the arrows

indicate the time instants of the arrival of the lique—

factionfront)

which was parallel to the base of the deposit and had

an amplitude ofabout 16 g and a frequency of 120 Hz．

For the 80 g—centrifuge acceleration of the test． this

corresponds in prototype units to a peak acceleration of

0．2 g and a frequency of 1．5 Hz(Fig．10)．

4．2．1．Acceleration and excess pore pressure records

l he input motion and recorded accelerations at

different depths in the deposit are shown in Fig． 1 0，

along with the acceleration of the surcharge plate． The

corresponding excess pore pressure time histories are

presented in Fig．1 1，while the pore pressure isochrones

are shown in Fig．1 2．Figure 1 3 presents a picture of

Model 3 just after the test，showing the steel plate and sand boils at the location of drainage holes

．

The recorded accelerations and excess pore pressures

exhibited response patterns that are significantly different

from those of M odel 1． These records suggested that

liquefaction of the deposit occurred first within the

100

50

0

— 50

— 100

100

50

0

— 50

— 100

100

50

0

— 50

— 100

100

50

0

— 50

— 100

100

50

0

— 50

． 100

0 10 20 30 40

Time(S)

Fig．15 Shear stress time histories，Model 3 (the arrows

indicate the time instants of the arrival of the

liquefaction front)

一 )I一∞∞o．I ∞．I o LI∽

一 一c 暑∽急0 LI∽

维普资讯 http://www.cqvip.com

No．1 Lenan Gonzfilez et a1．：Physical modeling and visualization of soil liquefaction under high confining stress 55

medium dense layer at a depth ofabout 19．3 m after 21

S of shaking．The acceleration time histories in the strata

above this elevation，as well as on the surcharge showed

a gradual decrease in acceleration amplitudes．as those

strata became progressively isolated．The corresponding

pore pressures indicated that the top dense soi1 liquefied

only in the vicinity and below the 1 3．4 m depth．The

recorded acceleration at this depth (1 3．4 m1 showed

a reduction in amplitudes only towards the end of

shaking，in agreement with the measured excess pore

pressures·

The recorded high pore fluid pressures that

developed first within the medium dense sand layer

provided an explanation of the observed response

paRern．These high pressures induced an upward pore

fluid flow towards the ground surface and increased the

pore pressures at the bottom of the dense sand． This

100

50

0

— 50

— 100

100

50

0

— 50

。 100

— 100

50

墨 。

。 50

∽ 。 100

100

50

0

— 50

。 100

100

50

0

— 50

。 100 1．2 。o．8 。o．4 o o．4 o．8 1．2

Shear strain f％1

6 l2 18 24 30 36

Time(s)

Fig．16 Shear stress-strain histories，M odel 3

increase in pore pressure led eventually to the delayed

liquefaction of the dense layer at about 1 3．4 m towards

the end of shaking．A similar phenomenon of delayed

liquefaction of a shallow denser sand layer occurred

after the end of shaking of the 1 964 Niigata earthquake

in Japan(Seed and Idriss．1 967)． 4．2．2．System identification analysis

9 。

c-

0．2

董 0 0．2

吕

Dense

sand

Loose

sand

Dense

sand

Loose

sand

Dense

sand

Loose

sand

～ 一 >

Input motion

0 10 15 20 25 30 35

Time(S)

fa1，=6s

Dense

sand

Loose

sand

Dense

sand

fb1，= 12S

fc1，=18s

～ 一 >

Loose

sand

(d)，=23s

(e1， 30s

1．0

0．8

0．6

0．4

0．2

0．0

Fig．1 7 Pore pressure ratios and estimated lateral displace。

ments for selected time instants，M odel 3

—===㈠===㈠===U

维普资讯 http://www.cqvip.com

56 EARTHQUAKE ENGINEERING AND ENGINEERING VIBRAT10N VOI．4

The recorded accelerations of the deposit and the

stee1 plate were employed to estimate the soi1 shear

stress and strain histories．as shown in Figs．1 4 and 1 5

fin which the times of arriva1 of the liquefaction front

are indicated by arrows)．The 1argest cyclic strains were observed at depths of 1 6．4 m and 2 1 m．The strain peaks

at these two 1ocations occurred when the soi1 liquefied．

The same phenomenon was discussed above for M ode1

1 in relation to the passage of the liquefaction front．

At the shallow depths of 2．1 m and 5．5 m within the

dense layer．the cyclic shear strain amplitudes(Fig．1 41

remained constant at the beginning of the shaking，and

then started to decrease gradually unti1 vanishing once

the medium dense 1ayer started experiencing large pore

pressures at about l 8 s．

The shear stress—strain histories exhibited

contrasting response patterns．as illustrated by Fig．1 6．

The evaluated shear stress—strain history at 2．1 m and

5．5 m depths fwithin the dense layer1 indicated only

a minor soi1 stiffhess fslope of the stress—strain loops)

reduction while the strain amplitudes decreased due to

the isolation effect mentioned above．This is consistent

with the observation that at these depths the dense layer

did not liquefy．At a depth of 2 1 m．an initia1 high shear

stiffness was followed by cycles of large shear strains

and substantia1 stresses after the onset of liquefaction．

4．2．3．Visualization analysis

The field of excess pore pressure ratio( ，)was

evaluated using simple 1inear interpolations between

measurements of the array of the transducers P 1．P4．

P5 and P7 of Fig．1 fb、．This field is shown for selected

time instants in Fig．1 7．along with the mode1 1atera1

displacements and the associated input motion(where

the dots correspond to the selected time instants)． Similarly to Mode1 1．the displacements were amplified

30 times for visua1 clarity．

Excess pore pressures started to develop in the

medium dense 1ayer early in the shaking．Atier 1 8 s，

the pore pressure ratios r，

reached values between 0．7

and 0．9(Fig．1 7(d))．Thereafter，the denser upper layers became isolated．and large shear strains developed in the

medium dense layer(Figs．1 7(d)and 1 7(e))．By the end ofthe shaking the dissipation ofpore pressures frO1TI the

medium dense layer 1ed to the liquefaction ofthe bottom

ofthe dense 1ayer fFig．1 7(e))．Note again that the denser

layer did not liquefy fully even though significant pore

water pressures were generated(Fig．1 7(e))．

5 Conclusions

This paper presented the experimental results，data

interpretation and analyses oftwo centrifuge model tests

which simulated the effect of high confining stresses on

the development of seismically—induced liquefaction．

The major conclusions are as follows： · Consistent results were obtained from both

centrifuge mode1 tests and provided detailed information

on the development ofsoil liquefaction at high confining

stresses．

· Full liquefaction of medium dense sand layers

was reached at high confining stresses and depths larger

than 30 m ．

· The use of a viscous pore fluid was vita1 to

properly simulate the development and dissipation of

the excess pore pressures．

· The use of a stee1 plate surcharge increased the

vertica1 stresses within the tested mode1．but had two

maj or shortcomings：(1)a large excess pore pressure

gradient was generated near the surface；and f2、the

total shear stresses of the deposit did not increase

significantly due to the 1ow friction between the plate

and the soi1．

· The employed system identification procedure

to investigate the recorded dynamic response was

advantageous in analysis of the liquefaction process

and the associated stiffness and strength degradation．

This procedure allowed the observation of phenomena

1ike shear strength degradation．stress—strain dilative

patterns，and a zone of large shear strains traveling

downward with the liquefaction front．

Acknowledgments

This research was supported by the National Science

Foundation，Grant No．CMS一984754 (Dr．C．Astill

program manager)，and the US Army Engineer Research and Development Center．This support is gratefully

acknowledged．

References

Abdoun T (1 997)，“Modeling of Seismically Induced Lateral Spreading ofMulti—layered Soil and Its Effect on

Pile Foundations，”Ph．D．thesis，Rensselaer Polytechnic

Institute，Troy，N USA．

Elgamal AW,Zeghal M，Taboada V and Dobry R

(1 996)，“Analysis of Site Liquefaction and Lateral Spreading Using Centrifuge Testing Records，”Soils and

Foundations，36(2)：111—121．

Gonzfilez LL，Abdoun T and Sharp M (2002)，“Modeling of Seismically Induced Liquefaction Under High

Confining Stress International Journal( Physical

Mode ling in Geotechnics，2(3)：1—15．

Seed HB and Idriss IM (1 967)，“Analysis of Soil

Liquefaction：Niigata Earthquake，”Journal of the Soil Mechanics and Foundations Division，ASCE，93：

rSM3)：83—108．

Sharp M and Dobry R(2002)，“Sliding Block Analysis of Lateral Spreading Based on Centrifuge Results，’’

International Journal 0f Physical M odeling and

Geotechnics．2：1 3—32．

Youd TL Idriss IM ，Andrus

G， Christian JT， Dobry R，

LF，Hynes M E，Ishihara K，

RD，Arango I，Castro

Finn W D L．

Koester JP,

Harder Jr．

Liao SSC．

维普资讯 http://www.cqvip.com

No l Lenan GonzMez et a1．：Physical modeling and Visualization of soil liquefaction under high confining stress 57

M arcuson W F，M artin GR，M itchell JK． M oriwaki Y_

Power M S，Robertson PK，Seed RB and Stokoe KH

(2001)，“Liquefaction Resistance of Soils：Summary Report from the 1 996 NCEER and l 998 NCEER／NSF

Workshop on Evaluation of Liquefaction Resistance of

Soils，’’Journal of Geotechnical and Geoenvironmental Engineering，ASCE 127(10、：8l7—833．

Zeghal M and Elgamal AW (1 994)，“Analysis of Site

Liquefaction Using Earthquake Records，’’Journal oJ

Geotechnical Engineering，ASCE， 120(6)：996—1 0 1 7

Zeghal M，Elgamal AW,Zeng X and Arulmoli K(1 998)， “Identification of Liquefaction Response Mechanisms

in Sand—Silt Dynamic Centrifuge Tests，’’Soil Dynamics

and Earthquake Engineering，18(1)：7 l一85．

Zeghal M ， Elgamal AW, Tang HT and Stepp JC

(1 995)，“Lotung Downhole Array I1：Evaluation of

Site Nonlinear Properties，’’Journal of Geotechnical Engineering，ASCE，121：363—378．

维普资讯 http://www.cqvip.com