Principal Investigators and Institutions
Ricardo Dobry, Rensselaer Polytechnic Institute
Geoffrey R. Martin, University of Southern California
Objective
The primary objective of this task is to develop design guidelines for
large pile group foundations susceptible to liquefactioninduced lateral
displacement. This will be done through centrifuge model tests,
supplemented by analytical studies, intended to simulate lateral spread
loading of closely spaced pile groups. The Japanese Road Association (JRA)
has published a limit equilibrium approach to lateral loading, based on
limited Kobe earthquake case history data. This approach will be
reevaluated based on a series of 12 centrifuge tests, and reevaluation
of the Japanese case history data and JRA procedures.
This twoyear centrifugebased study has three specific objectives: (1)
to establish a database of centrifuge results directly useful to calibrate
the parameters needed for py and limit equilibrium analysis of pile
groups exposed to both liquefied and nonliquefied soil pressures, thus
supplementing the corresponding information already available from
previous RPI centrifuge tests for single piles; (2) to extract guidelines
for limit equilibrium and py analyses of pile groups from these
centrifuge tests; and (3) to reevaluate the JRA limit equilibrium
procedure for pile groups, on the basis of (1) and (2), plus reevaluation
of the original 1995 Kobe case histories and analyses used to develop this
procedure.
Approach
In Research Year 3, the work will consist mainly of conducting the
first six centrifuge model experiments, as shown in the following table,
and initiating corresponding interpretations and analytical modeling. The
remaining centrifuge tests will be conducted in Research Year 4.
Test Number 
No. of soil layers 
No. of piles 
Pile group configuration 
Pile cap embedded in soil 
Spacing (d = pile diameter) 
Laminar box size 
Research Year 3 Tests 
1
2
3
4
5
6 
2
2
2
2
2
3 
3
4
4
9
9
4 
3x1
2x2
2x2
3x3
3x3
2x2 
Yes
Yes
Yes
Yes
Yes
Yes 
3d
3d
3d
3d
3d*
3d 
small
small
large
large
large
large 
Research Year 4 Tests 
7
8 
3
3 
9
9 
3x3
3x3 
Yes
Yes 
3d
3d* 
large
large 
Tests 9 through 12  parameters to be determined 
* The liquefiable sand will be densified within and around the pile
group to simulate the effect of pile driving
The Year 3 tests constitute an expansion of similar centrifuge models
tested in the last few years, where an instrumented pile foundation model
is subjected to liquefaction and lateral spreading caused by inflight
shaking at the base of a "small" (46x25cm in plan dimensions)
laminar box container. These previous tests involved mostly single
endbearing piles in either two or threelayer soil profiles, with the
bending response of the pile in twolayer profiles controlled by the
lateral pressure of the liquefied layer, and the pile response in
threelayer profiles controlled mainly by the lateral pressure of the
shallow nonliquefied layer. The main thrust of these tests is the use of a
"large" (70x35cm) laminar box that can test 2x2, 3x3 and 4x4
pile groups. The basic soil, pile, and shaking parameters will be kept the
same as in previous RPI centrifuge tests. That is, the 6mthick prototype
liquefiable layer will be formed by fine Nevada sand saturated with water
and having a relative density of about 40%, with total soil profile depths
of 8m and 10m, endbearing pile models of these same lengths, prototype
pile diameter d = 0.6m; with the pilesoil models tested at a centrifuge
acceleration of 50g, and subjected in flight to a prototype peak base
acceleration of 0.3g with a shaking duration causing a permanent free
field lateral ground displacement of about 0.8m.
Test 1 corresponds to a row of three piles joined by an embedded pile
cap, with shaking perpendicular to the row, to study the response,
equivalent liquefied soil pressure, and py curves of the simplest
possible closely spaced pile system. Tests 2 and 3 correspond to a 2x2
square pile group in the small and large laminar boxes, respectively.
Tests 3 through 8 are all square 2x2 or 3x3 pile groups in the large
laminar box and covering the 2layer and 3layer soil profiles. In Tests 5
and 8, the 3x3 pile groups are repeated but with the liquefiable sand
layer densified within the groups and around it to simulate the effect of
pile driving. The pile cap will be embedded in the liquefiable layer for
the case of the 2layer profile, and it will be embedded in the shallow
nonliquefiable layer for the case of the 3layer profile.
In addition, contact will be established with the Japanese researchers
who studied the case history data of the Kobe earthquake and developed the
JRA lateral loading design procedure for pile groups, in preparation for a
reanalysis of these data and procedures.
In Research Year 4, the work will consist of will consist of:
 Additional interpretations and analyses of the centrifuge results in
order to develop guidelines for limit equilibrium and py lateral
spreading analyses of pile groups, following methodologies developed
at RPI, USC and other research centers in previous years;
 Conducting centrifuge Tests 7 and 8, plus four additional centrifuge
experiments, as needed, to fill gaps in knowledge and generalize
further the guidelines (these additional tests may include a repeat of
some pile group tests with a viscous pore fluid; spacings other than
3d; pile groups including batter piles; rectangular pile group tests;
and testing of friction rather than endbearing pile groups);
 Reevaluation of the JRA limit equilibrium procedure for pile groups.
Technical Challenges
The primary technical challenge will be to determine how best to
combine centrifuge results and case history data to formulate a design
procedure for pile groups subjected to liquefactioninduced lateral ground
deformation.
Products
 Report documenting the results of all centrifuge tests, including
comparisons, interpretations, and analyses.
 Design guidelines on the application and use of of py and limit
equilibrium analyses, consistent with the examined case histories and
the centrifuge test results.
