You are on page 1of 12

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/243756114

The response of pile groups under cyclic lateral loads

Article  in  International Journal of Physical Modelling in Geotechnics · January 2010


DOI: 10.1680/ijpmg.2010.10.2.47

CITATIONS READS

11 271

3 authors, including:

Stuart Kenneth Haigh Malcolm D. Bolton


University of Cambridge University of Cambridge
122 PUBLICATIONS   874 CITATIONS    236 PUBLICATIONS   7,981 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Thermal behaviour of soils View project

Soil liquefaction View project

All content following this page was uploaded by Stuart Kenneth Haigh on 17 April 2015.

The user has requested enhancement of the downloaded file.


Li, Z. et al. (2010). International Journal of Physical Modelling in Geotechnics 10, No. 2, 47–57

The response of pile groups under cyclic lateral loads


Z. LI*, S. K. HAIGH* and M. D. BOLTON*

Numerous pile groups are subjected to significant cyclic lateral loads due to wind, waves or
earthquakes, and many have failed catastrophically. In this research, centrifuge modelling of a pile
group subjected to cyclic lateral loads has been conducted to investigate the interaction effect in pile
groups and the influence of cyclic lateral loads on the performance of pile groups. Different pile
installation methods were also applied to capture the full range of construction-induced soil conditions
available in the field. Lateral permanent displacements of the pile group were seen to be induced by
one-way cyclic lateral loads but not by two-way symmetric cyclic lateral loads. The lateral secant
stiffness of the pile group increases slightly with increasing number of cycles, and leading piles attract
greater loads than trailing piles. Furthermore, permanent settlements of the pile group accumulate,
which can be attributed to the swaying motion of the pile cap induced by cyclic lateral loads.

doi: 10.1680/ijpmg.2010.10.2.47

KEYWORDS: centrifuge modelling; piles & piling; waves & wave loading

INTRODUCTION piles in a group have less resistance than the sum of


Piled foundations are the most prevalent class of deep individual pile capacities if the pile spacing is less than eight
foundations; they are often designed with the intention of times the pile diameter in the direction of lateral load.
resisting dead load and to demonstrate a factor of safety Similar conclusions were drawn for driven pile groups from
during their working life. Numerous piled foundations are the test results of Brown et al. (1987) and Rollins et al.
subjected to significant cyclic lateral loads because of their (2005).
fluid environment, particularly in high-rise buildings and Nowadays, monotonically jacked piles are popular for
offshore platforms, which may suffer wind and wave use in urban construction owing to the minimal noise and
actions. Pile heads may also be subjected to large- ground vibration caused during their installation.
magnitude cyclic loads arising from the swaying and Compared to bored and driven pile groups, jacked pile
rocking motions of superstructures during earthquakes. groups exhibit a stiffer axial response resulting from
These significant cyclic lateral loads lead to permanent enhanced normal resistance of the soil surrounding pile
deformations of the soil surrounding the pile shafts, shafts (White & Deeks, 2007). However, the behaviour of
causing significant accumulated permanent displacements. monotonically jacked pile groups under cyclic lateral loads
Consequently, many piled foundations have failed cata- is not yet fully understood.
strophically owing to these cyclic loads, causing over- In field tests, it is difficult to exert cyclic loading on piles
turning or collapse of major pile-supported structures. with large diameters because of the limitations of test
In order to understand such behaviour, many research- facilities and high costs. Additionally, few real-time failures
ers (e.g. Long & Vanneste, 1994) have investigated the of piled foundations subjected to live loads in service are
response of single piles to cyclic lateral loads. However, well monitored and recorded. This lack of well-documented
piles are rarely constructed in isolation but work together soil–pile response cases has hindered the progress of
as pile groups. The response of a pile group under cyclic research on the effects of cyclic loads and earthquakes on
lateral loads is generally different from that of an the behaviour of piled foundations.
individual pile because of the interaction between neigh- Fortunately, centrifuge modelling offers an effective
bouring piles, resulting in a reduction of pile group method to understand the influence of cyclic loads on
capacity. piled foundations. Compared with full-scale field tests,
The leading piles in a group will carry significantly centrifuge modelling tests are more convenient and efficient
higher loads than the trailing piles at the same deflection. to conduct with significantly reduced costs. Centrifuge
The piles in trailing rows are thought to exhibit less lateral modelling can offer physical conditions similar to those in
resistance because of interference (shadowing) with the the field and can reproduce the response of a piled
failure surface of the row of piles in front of them. This foundation subjected to cyclic loading. In centrifuge tests,
shadowing or group interaction effect is expected to small-scale model pile loading tests are conducted in
become less significant as the spacing between piles acceleration fields of magnitude many times the earth’s
increases because there is less overlap between adjacent gravity, such that full-scale prototype stresses are correctly
reproduced. This ensures that the highly non-linear soil
failure zones (Ochoa & O’Neill, 1989).
behaviour can also be replicated realistically. Thereby,
Prakash (1962) and Franke (1988) suggest that bored
observations from small-scale model piles can be related to
the full-scale prototype situation by appropriate scaling
Manuscript received 26 March 2010, revised manuscript laws.
accepted 17 June 2010. The objective of the current work was to investigate a
Discussion on this paper should reach the editor by 15 February range of pile installation procedures. Three different
2011. jacking protocols were used: ‘Cyclic jacking’ corresponds
* Department of Engineering, University of Cambridge, quite closely to conventional pile jacking in the field, such
Cambridge, UK. as by the Giken method. Model piles are pressed into the

47
Delivered by ICEVirtualLibrary.com to:
IP: 129.169.110.108
On: Tue, 26 Oct 2010 14:12:00
48 Li, Haigh and Bolton

Motor

3D load cell Adaptor


2D actuator

3D load cell Pile cap


Strain gauges
Strain gauges

Pile cap Lasers

Model piles
Encoder

Lasers

P1 and P2 P3 and P4

Dry sand 350 mm


Load cells

850 mm

(a) (b)

Fig. 1. The centriguge test package: (a) side view of the test package; (b) front view of the test package

soil in flight at high ‘g’, with unload/reload cycles actuator and an introduction to the facility can be found in
representing the successive repositioning of the driving Haigh et al. (2010).
head as the pile is driven in the field. ‘Monotonic jacking’
represents an ideal procedure more easily achievable in a
model than in the field, in which piles are installed in flight Model pile group
without intermediate unloading cycles. ‘Pre-jacking’ refers A 2 6 2 pile group configuration of stainless steel tubular
to the jacking-in of the model piles prior to centrifuging, at piles spaced at four times the pile diameter has been
1g. Although this insertion procedure remains one of soil manufactured and instrumented, and placed in the test rig,
displacement, the consequential locked-in stresses must be as shown in Fig. 1. Stainless steel tubes of 10 mm (0?5 m at
very small, so the outcome may be more similar to the prototype scale) outer diameter and 2 mm (0?1 m at
construction of a bored pile in the field. prototype scale) wall thickness were used as model piles.
Regarding conventional pile driving with a hammer, it is The length of each pile was 220 mm (11 m at prototype
known that hammer blows tend to reduce the axial capacity scale), and the embedment depth was 180 mm (9 m at
of piles compared with monotonic jacking. This mechanism prototype scale). Pile shaft surfaces were smooth, and pile
is known as friction fatigue, which was attributed by White tips were flat and close-ended. An aluminium plate with
& Lehane (2004) to densification under cyclic shearing dimensions of 70 6 70 6 35 mm (3?5 6 3?5 6 1?75 m at
causing radial stress relief immediately around the pile. prototype scale) was fabricated as a pile cap to give a fixed-
This effect might be seen in the differences between head condition.
monotonically and cyclically jacked piles in the centrifuge All the piles were installed with tip load cells to measure
tests. pile base resistance and instrumented with strain gauges on
Accordingly, pile group cyclic lateral load tests were
conducted using a 1 in 50 scale model pile group tested at
50g in the centrifuge. Individual pile responses and group
interaction effects are reported. The influence of cyclic
lateral loads on pile group lateral secant stiffness and Axial force
accumulated displacements are described. Moreover, since strain gauges
cyclic lateral loads can give rise to swaying, which induces Bending moment
cyclic axial loading of the piles, the effect of cyclic lateral 65 70
strain gauges
loads on the axial performance of the pile group is
demonstrated.

5
METHODOLOGY
Test apparatus 45
Pile group tests were conducted at 50g on the Turner beam 50
centrifuge at the Schofield Centre, Cambridge University. 70
A two-axis servo actuator was used to install the pile group
as an integrated assembly and then to apply vertical and Fig. 2. Strain gauge arrangement at front and back sides of
horizontal loads, as shown in Fig. 1. The design of the the 3D load cell

Delivered by ICEVirtualLibrary.com to:


IP: 129.169.110.108
On: Tue, 26 Oct 2010 14:12:00
The response of pile groups under cyclic lateral loads 49

2D actuator rotated 90° Axial force at prototype scale: kN


0 10 000 20 000 30 000
0 0

Penetration depth at prototype scale: m


40 2

Penetration depth at model scale: mm


80 4

120 6
260 mm 240 mm

160 8

200 10
0 2000 4000 6000 8000 10 000 12 000
Axial force at model scale: N

Fig. 4. The penetration resistance of the monotonically jacked


pile group
Fig. 3. Centrifuge test arrangement

their external shaft surface at levels above the sand surface and the ratio of the smallest side boundary separation to the
to record pile head axial force, lateral force and bending pile diameter was approximately 26, as was the base boundary
moment. separation. Thus, the boundary effects should be negligible
A new three-dimensional (3D) load cell was designed and based on the conclusions of Gui et al. (1998).
manufactured to measure the pile cap axial forces, lateral All test locations are presented in Fig. 3. The distance
forces and bending moments applied to the pile group as a between successive tests was 28 times the pile diameter to
whole, as shown in Fig. 2. A full wheatstone bridge, nullify any effects owing to locked-in effective stresses
comprising four 5 V d.c. powered 350 V strain gauges, was induced in the sand by the preceding tests.
attached on the external surface of the cylinder to measure
axial forces. Another two bridges of strain gauges were
attached close to each end of the cylinder to measure TEST RESULTS
bending moments. Shear forces were calculated based on Based on the scaling laws for centrifuge tests, in order to
the difference in bending moments measured at the top and express test results at prototype scale the test data from
bottom of the load cell.
Axial force at prototype scale: kN
0 10 000 20 000 30 000
Instrumentation 0 0
Vertical displacements at the leading and trailing edges of
the pile cap and lateral displacements of the pile cap were
measured using three lasers mounted under the two-
Penetration depth at prototype scale: m

40 2
Penetration depth at model scale: mm

dimensional actuator. Additionally, vertical and horizontal


displacements of the actuator were recorded by two
encoders.
80 4

Sand and container


Dry fraction E silica sand was used in this project. The sand
has a d50 of 218 mm measured using the single-particle optical 120 6
sizing (SPOS) technique (White, 2003). Since the ratio of pile
diameter to average grain size was about 85, larger than the
limiting value of 20 suggested by Gui et al. (1998), the sand
should have behaved like a continuum and scaling effects 160 8
should be negligible. The sand was pluviated into a cylindrical
steel tub (850 mm diameter and 400 mm deep) using an
automatic sand-pouring machine (Madabhushi et al., 2006). 200 10
A dense homogeneous sand specimen with a relative density 0 2000 4000 6000 8000 10 000 12 000
of 83% was achieved. The cylindrical model container was Axial force at model scale: N
designed to be strong enough to sustain the large soil
pressures acting during the centrifuge tests. The ratio of Fig. 5. The penetration resistance of the cyclically jacked
container diameter to individual pile diameter was about 85, pile group

Delivered by ICEVirtualLibrary.com to:


IP: 129.169.110.108
On: Tue, 26 Oct 2010 14:12:00
50 Li, Haigh and Bolton
Table 1. Installation forces of the pile group

Test number ZL07-2 ZL07-4 ZL08-2 ZL08-4 Average


Installation type MJ CJ MJ MJ —
Installation force: N 10 800 11 100 9800 9900 10 400

Note: MJ is monotonic jacking; CJ is cyclic jacking.

these centrifuge tests were multiplied by 50 for displace- During the cyclic jacking process, a loading/unloading
ments, 50 for pile group stiffness, and 502 for forces. Test loop was conducted in each stroke. Thus the whole
results at both model and prototype scale are plotted in all installation is equivalent to a set of loading tests conducted
of the following figures. It should be noted that the axes at at different embedded depths. The initial pile group vertical
left and bottom sides of graphs represent test data at the stiffness (Kc,install), in other words the initial gradient of the
model scale, while axes at right and top sides represent pile cap axial load–settlement curve, increases almost
corresponding data at the prototype scale. All numerical linearly with increasing embedded depth, as shown in
values quoted in the following text and in the tables are Fig. 6. The initial pile group base stiffness (Kb,install), that is
given at model scale. the sum of the four individual pile base stiffnesses, is also
plotted here to exclude the benefit from the pile shaft
friction. Kb,install is similar to Kc,install at shallow depths
Pile group installation owing to the negligible shaft friction; while Kb,install is
The model pile group was pre-jacked at 1g, monotonically smaller than Kc,install at depths over 30 mm, this difference
jacked at 50g or cyclically jacked at 50g into dry dense increases with increasing depth since the pile shaft friction
sand. becomes more considerable at depth.
Two typical penetration resistance curves for the Moreover, the ratio of the Kb,install to the mobilised pile
monotonically and cyclically jacked pile groups are group base capacity reduces from 3?5 to 1?4 mm21 as the
presented in Fig. 4 and Fig. 5 respectively. The mono- embedded depth increases from 15 mm to 180 mm, as
tonically jacked pile group was pushed into the sand with a shown in Fig. 7. This phenomenon is similar to test results
monotonic stroke in-flight, while the cyclically jacked pile from Deeks (2008), indicating that the Kb,install is propor-
group was successively penetrated 1 mm and then unloaded tional to the soil small strain shear modulus, which is
to zero head load, with 180 strokes being used during the approximately proportional to the square root of the soil
installation process. At a final penetration depth of vertical effective stress (McDowell & Bolton, 2001).
180 mm, the maximum penetration forces of the mono- Table 1 lists the maximum installation force of the pile
tonically jacked and cyclically jacked pile group are group at the final penetration depth for all centrifuge tests.
10 800 N and 11 100 N respectively, as shown in Table 1. The average value of the installation force is 10 400 N, and
It seems that the cyclic jacking installation obtained a the coefficient of variation is 5?4%; therefore, these pile
similar penetration force to the monotonic jacking installa- group centrifuge tests are very repeatable.
tion. During the cyclic jacking process, although the Additionally, during the jacking process, individual piles
amplitudes of cyclic axial loads were very large, and the of a pile group shared the pile cap force equally, and had
maximum relative displacements between pile shaft and soil the same force distribution, as shown in Fig. 8. This
are up to 16% of the pile diameter per cycle, the friction indicates that the pile group was jacked vertically without
fatigue effect is not evident. any obvious eccentric force.

Pile initial stiffness at prototype scale: kN/m Stiffness ratio at prototype scale: 1/m
250 000 500 000 750 000 20 40 60 80
0 0
Pile cap
Pile bases

40 2 40 2
Penetration depth at prototype scale: m

Penetration depth at prototype scale: m


Penetration depth at model scale: mm

Penetration depth at model scale: mm

80 4 80 4

120 6 120 6

160 8 160 8

200 10 200 10
0 5000 10000 15000 0 1 2 3 4
Pile initial stiffness at model scale: N/mm Stiffness ratio at model scale: 1/mm

Fig. 6. The initial pile group stiffness variation during cyclic Fig. 7. Ratio of pile group base stiffness to the mobilised pile
jacking installation group base capacity during cyclic jacking installation

Delivered by ICEVirtualLibrary.com to:


IP: 129.169.110.108
On: Tue, 26 Oct 2010 14:12:00
The response of pile groups under cyclic lateral loads 51
Pile head axial force at prototype scale: kN

0 2500 5000 0 2500 5000


0 0 0 0
Pile 1 Pile 2
40 2 40 2

80 4 80 4

120 6 120 6

Penetration depth at prototype scale: m


Penetration depth at model scale: mm

160 8 160 8

200 10 200 10
0 500 1000 1500 2000 2500 0 500 1000 1500 2000 2500

0 2500 5000 0 2500 5000


0 0 0
0
Pile 3 Pile 4
40 2 40 2

80 4 80 4

120 6 120 6

160 8 160 8

200 10 200 10
0 500 1000 1500 2000 2500 0 500 1000 1500 2000 2500

Pile head axial force at model scale: N


(a)
Pile base axial force at prototype scale: kN

0 2500 5000 0 2500 5000


0 0 0 0

Pile 1 Pile 2
40 2 40 2

80 4 80 4

120 6 120 6 Penetration depth at prototype scale: m


Penetration depth at model scale: mm

160 8 160 8

200 10 200 10
0 500 1000 1500 2000 2500 0 500 1000 1500 2000 2500

0 2500 5000 0 2500 5000


0 0 0 0

Pile 3 Pile 4
40 2 40 2

80 4 80 4

120 6 120 6

160 8 160 8

200 10 200 10
0 500 1000 1500 2000 2500 0 500 1000 1500 2000 2500

Pile base axial force at model scale: N


(b)

Fig. 8. Penetration resistances of individual piles for a jacked pile group: (a) pile head axial force of individual piles during
penetration process; (b) pile base resistance of individual piles during penetration process

Delivered by ICEVirtualLibrary.com to:


IP: 129.169.110.108
On: Tue, 26 Oct 2010 14:12:00
52 Li, Haigh and Bolton
In Fig. 8, at the final penetration depth, the average

ZL08-2
individual pile head axial force, base resistance and shaft friction

One-way

60
1000
are 2475 N, 1965 N and 510 N respectively. The ratio of the pile
base resistance to the pile head axial force is around 80%.

MJ
The load at the pile cap reduced to zero after the
installation. However, residual stresses were locked in the
piles. This is because the surrounding sand prevents the pile

ZL08-1-2

One-way
from rebounding upwards, producing negative (down-
wards) skin friction on the upper part of the shaft and

1000
positive (upwards) skin friction on the lower part of the

PJ

5
shaft. At the same time, a portion of the base resistance still
remains at the pile tip, as shown in Fig. 8(b). The ratios of
the residual force to the ultimate capacity of the base

ZL08-1-1
resistance and shaft friction are about 8% and 30%

One-way
respectively. These ratios are not as large as those from

600
other test results (Poulos, 1987; Jacobsz, 2002) owing to the

PJ

11
smooth surface of the pile shaft used in these tests.

ZL08-4-2
Lateral performance of pile groups

Two-way
In the current research, force-controlled lateral load cycles

¡750
were conducted at the pile cap. The cyclic load ratio RH, that

MJ

10
is the minimum load/maximum load in a cycle, is considered
to be an important characteristic of cyclic loads. Two typical
RH values are used here, RH 5 0 representing one-way cyclic

ZL08-4-1
loads and RH 5 21 representing two-way cyclic loads. The

Two-way
cyclic load frequency is not considered here since the loading

¡600
rate was chosen to be around 0?1 mm/s to obtain quasi-static

MJ

10
behaviour. The parameters of these cyclic lateral loads are
listed in Table 2.

ZL07-
Figure 9(a) presents the pile cap lateral response of the

¡160
Two-
4

way
monotonically jacked and cyclically jacked pile groups in

CJ

20
the first two-way lateral load cycle. The cyclic lateral load–
displacement curves of both jacked pile groups in the
following load cycles are similar to those in the first cycle,

Two-way
ZL07-2

without any permanent lateral displacement or pile group

¡160
lateral stiffness variation. It is seen that a cyclically jacked
MJ

20
pile group has a similar response to a monotonically jacked
pile group under subsequent cyclic lateral loads. Therefore,
only the monotonically jacked pile group is used to
represent the behaviour of the jacked pile group in this
ZL08-3-4

Two-way

paper.
¡1000

Figure 9(b) presents pile cap lateral force–displacement


curves for the pre-jacked pile group subjected to cyclic
PJ

lateral load with a small amplitude of ¡200 N. It should be


noted that an axial load was applied on the pile cap to
model the dead weight of the structure before cyclic lateral
ZL08-3-3

Two-way

loading. A small lateral load was induced because the pile


PJ, pre-jacking; MJ, monotonic jacking; CJ, cyclic jacking.
¡875

group was not perfectly vertical, causing a non-zero initial


lateral load at the onset of lateral cycling. This initial lateral
PJ

load seems to have negligible effect on the pile group


behaviour under cyclic lateral loads. These lateral force–
displacement curves form narrow hysteresis loops, with the
ZL08-3-2

Two-way

secant stiffness reducing marginally with increased lateral


displacement. It is seen that no permanent lateral displace-
¡600
Table 2. Cyclic lateral load information

ment of the pile group was accumulated during the cyclic


PJ

loading process. Under small-amplitude two-way cyclic


lateral loads, the deformation of the soil surrounding the
pile shaft, especially at depth, is very small. Thus the pile
ZL08-3-1

cap cyclic lateral load–displacement response is quasi-


Two-way

elastic and repeatable without any pile cap permanent


¡200

displacement.
PJ

10

Figure 9(c) presents pile cap lateral force–displacement


curves for the pre-jacked pile group subjected to cyclic
lateral load with a large amplitude of ¡1000 N. Although
amplitude: N
Test number

Cyclic load

Cyclic load

Cyclic load

the cyclic lateral loads induced a significant pile cap lateral


number

displacement of ¡2?1 mm (21% of pile diameter), these


cyclic force–displacement curves are also repeatable. No
type

type
Pile

permanent lateral displacement accumulated during the

Delivered by ICEVirtualLibrary.com to:


IP: 129.169.110.108
On: Tue, 26 Oct 2010 14:12:00
The response of pile groups under cyclic lateral loads 53

Lateral displacement at prototype scale: mm Lateral displacement at prototype scale: mm


_10 _5 0 5 10 _20 _10 0 10 20
200 500
Monotonic-jacked The 1st cycle

Lateral force at prototype scale: kN


Cyclic-jacked The 2nd_10th cycles
Lateral force at model scale: N

Lateral force at prototype scale: kN


200 500

Lateral force at model scale: N


100 250

100 250

0 0
0 0

_100 _250 _100 _250

_200 _500
_200 _500
_0.2 _0.1 0.1 0.2 _0.4 _0.2
0 0 0.2 0.4
Lateral displacement at model scale: mm Lateral displacement at model scale: mm
(a) (b)
Lateral displacement at prototype scale: mm Lateral displacement at prototype scale: mm
_100 _50 _50 _25 0 25 50
0 50 100
1200 3000 800 2000
The 1st cycle The 1st cycle

Lateral force at prototype scale: kN


Lateral force at prototype scale: kN

The 2nd_5th cycles The 2nd_5th cycles


Lateral force at model scale: N
Lateral force at model scale: N

600 1500 400 1000

0 0 0 0

_600 _1500 _400 _1000

_1200 _3000 _800 _2000


_2 _1 0 1 2 _1.0 _0.5 0 0.5 1.0
Lateral displacement at model scale: mm Lateral displacement at model scale: mm
(c) (d)

Fig. 9. Lateral force–displacement curves of pile groups under cyclic lateral loads: (a) monotonic-jacked and cyclic-jacked pile
groups in the first two-way lateral load cycle in test ZL07-2 and ZL07-4; (b) pre-jacked pile group under two-way cyclic lateral loads
in test ZL08-3-1; (c) pre-jacked pile group under two-way cyclic lateral loads in test ZL08-3-4; (d) jacked pile group under two-way
cyclic loads in test ZL08-4-1

large-amplitude two-way cyclic lateral loading test, similar group passes its original position. Moreover, the response
to those curves under small-amplitude cyclic lateral loads. of the monotonically jacked pile group under two-way
However, these curves displayed local softening with a cyclic lateral loads is similar to that of the pre-jacked pile
reduced tangent stiffness of the pile cap whenever the pile group, although the local softening behaviour in the force–
displacement curves is even more evident, as shown in
Fig. 9(d).
2.0 Figure 10 shows the overall pile cap lateral secant
Pre-jacked ± 200 N stiffnesses in each load cycle, normalised by the stiffness
value in the first cycle. The secant stiffness is calculated
1.8 Pre-jacked ± 1000 N
Normalised pile secant stiffness

based on the position of the pile cap at the beginning of


Jacked ± 600 N that cycle, not the initial pile cap position. The pile secant
1.6 Jacked + 1000 N stiffness is seen to increase slightly with increasing number
of lateral load cycles. The best explanation is that the cyclic
1.4 lateral movement of the pile group induces local densifica-
tion of the sand surrounding the pile shafts.
This local densification due to lateral cycling will reduce
1.2
the radial effective stresses on the pile as it passes its neutral
position, where the lateral load is zero, rather in the fashion
1.0 of White & Lehane’s (2004) friction fatigue model for axial
cycling. This would explain the local tangent stiffness
0.8 reduction. The response is analogous to the ‘post-holing’
1 10 effect described by Fleming et al. (1992) as involving a gap
Cycle number opening up around piles under combined axial and cyclic
lateral loading, leading to the failure of piles close to their
Fig. 10. Pile cap normalised lateral secant stiffness in cyclic neutral position. Soil densification could also explain overall
lateral load tests secant stiffness enhancement when the pile has been pushed

Delivered by ICEVirtualLibrary.com to:


IP: 129.169.110.108
On: Tue, 26 Oct 2010 14:12:00
54 Li, Haigh and Bolton
Lateral displacement at prototype scale: mm scale, its value arriving at 1?25 mm after 60 cycles. As before,
_50 0 50 100 150 the pile cap lateral secant stiffness in each load cycle increased
and the displacement amplitude reduced slightly with
The 1st cycle increasing number of force-controlled lateral load cycles, as
The 2nd_60th cycles shown in Fig. 11(c). Evidently, the effect of cyclic lateral loads

Lateral force at prototype scale: kN


on the lateral accumulated displacement and pile cap secant
Lateral force at model scale: N

1000 2500
stiffness is greatest for the first load cycle, with the effect of
cyclic loads diminishing as cycling continues.
With one-way cyclic shearing, the induced lateral stress
500 1250
states will be ‘large’ at maximum displacement and ‘small’
at the end of the return leg. This lack of symmetry is
thought to lead to the progressive lateral displacements, in
contrast to symmetrical two-way loading.
0 0

_1 0 1 2 3
Interaction effect within pile groups
Lateral displacement at model scale: mm Figure 12 presents the lateral force–displacement curves of
(a)
individual piles of the pre-jacked pile group under two-way
cyclic lateral loads with an amplitude of ¡1000 N. Piles P1
3.5

Lateral displacement at prototype scale: mm


175 and P2 are in the same row; piles P3 and P4 are in the other
Lateral displacement at model scale: mm

Maximum displacement row. The spacing of individual piles is four times the pile
3.0 Minimum displacement 150 diameter. It is evident that the lateral responses of individual
piles in different rows are quite different from each other.
2.5 125
During the first quarter-cycle, the pile group moves
forward in the positive direction. The piles P1 and P2 are
2.0 100
leading piles, and their lateral secant stiffnesses are much
1.5 75 larger than those of the trailing piles P3 and P4, while the
behaviours of individual piles in the same row are similar to
1.0 50 each other. This is because the soil restraining the leading
edges of the trailing piles is affected by the movement of the
0.5 25 leading piles, and thus the stress level and the stiffness of
the soil is reduced by this pile group interaction effect,
0 0 causing the behaviour of the trailing piles to be less stiff
100 101 102
Cycle numbers
than those of the leading piles. The maximum force exerted
on pile P1 or P2 is 330 N, while that on pile P3 or P4 is
(b)
100 N, as shown in Fig. 12. When the pile group is
2.0 100 unloaded and loaded in the opposite direction, piles P1 and
Displacement magnitude at prototype scale: mm
Displacement magnitude at model scale: mm

P2 become trailing piles, and piles P3 and P4 are leading


piles. The responses of piles P3 and P4 are much stiffer than
1.5 75 those of piles P1 and P2. The minimum force exerted on
pile P1 or P2 is 2170 N, while that on pile P3 or P4 is
2400 N. Although the extreme forces of individual piles in
different rows are different, the force amplitudes of these
1.0 50
individual piles are similar, indicating that the globe secant
stiffnesses of the individual piles of the pile group under
two-way cyclic lateral loads are similar to each other.
0.5 25 Moreover, the interaction effect of individual piles is also
evident for the pile group under one-way cyclic lateral
loads, as shown in Fig. 13. The responses of the leading
0 0 piles P1 and P2 are around 27% stiffer than the trailing
10 101 102
piles P3 and P4 owing to the pile group interaction effect.
Cycle numbers
(c)
Axial performance of pile groups
Fig. 11. The response of a jacked pile group under one-way During the cyclic lateral loading process, a constant axial
cyclic lateral loads in test ZL08-2: (a) cyclic lateral force– force of approximately 1800 N was applied to the pile cap
displacement curves; (b) maximum and minimum displace- to model the dead load of the superstructure. However, a
ment; (c) displacement amplitude swaying motion of the pile group is induced by the pile cap
lateral movement, leading to individual pile head cyclic
far enough from its neutral position to regenerate large axial forces and displacements. Fig. 14 shows the variation
lateral stresses on the leading edge of the pile. of individual pile head axial forces of the monotonically
On the other hand, Fig. 11(a) presents the response of the jacked pile group subjected to cyclic lateral loads with an
monotonically jacked pile group subjected to one-way cyclic amplitude of ¡600 N.
lateral loads with an amplitude of 1000 N. Unlike two-way The maximum compression axial force is beyond 1200 N
cyclic lateral loads, one-way cyclic lateral loads induce and the maximum pull-out axial force is around 200 N. A pile
significant accumulated permanent lateral displacements of cap permanent settlement of 0?42 mm accumulated cycle by
the pile group, as shown in Fig. 11(b). This accumulated cycle during this cyclic lateral loading process, as shown in
permanent lateral displacement increases approximately Fig. 15. The maximum accumulated permanent settlements
linearly with increasing number of cycles on a logarithmic in all tests are listed in Table 3. The settlement increases with

Delivered by ICEVirtualLibrary.com to:


IP: 129.169.110.108
On: Tue, 26 Oct 2010 14:12:00
The response of pile groups under cyclic lateral loads 55

Pile lateral displacement at prototype scale: mm


_100 _50 0 50 100 150 _100 _50 0 50 100 150
400 400

Pile 1 Pile 2
200 500 200 500

Pile head lateral force at prototype scale: kN


0 0 0 0
Pile head lateral force at model scale: N
_200 _500 _200 _500

_400 _400
_2 _1 0 1 2 3 _2 _1 0 1 2 3

_100 _50 0 50 100 150 _100 _50 0 50 100 150


400 400

Pile 3 Pile 4
200 500 200 500

0 0 0 0

_200 _500 _200 _500

_400 _400
_2 _1 0 1 2 3 _2 _1 0 1 2 3
Pile lateral displacement at model scale: mm

Fig. 12. Lateral force–displacement curves of individual piles under two-way cyclic lateral loads

increasing amplitude of cyclic lateral loads. Additionally, the performance of pre-jacked pile groups and jacked pile
settlement of the pre-jacked pile group is much larger than groups subjected to cyclic lateral loads were investigated.
that of the monotonically jacked pile group. This is because The installation forces of monotonically jacked pile
the pile group jacked in at 50g has a stiffer base response than groups and cyclically jacked pile groups are similar to each
the pre-jacked pile group owing to the stresses induced during other, the friction fatigue effect not being evident during
insertion. The jacked pile group also has a higher shaft the cyclic jacking process.
resistance and therefore higher stiffness caused by the larger In the cyclic jacking installation, the initial pile group
radial stress in the soil surrounding the pile shafts induced vertical stiffness and the initial pile group base stiffness in
during the jacking process. each loading stroke increased with increasing embedded
depth. The ratio of the initial pile group base stiffness to the
mobilised pile group base capacity reduced with increasing
CONCLUSIONS embedded depth, indicating that the pile group base
Cyclic lateral loads with different amplitudes were applied stiffness is proportional to the small strain shear modulus
to model pile groups after installation in the centrifuge. The of soil.

Pile lateral displacement at prototype scale: mm


50 75 100 125 50 75 100 125
300 750 300 750

Pile 1 Pile 2
200 500 200 500
Pile head lateral force at prototype scale: kN
Pile head lateral force at model scale: N

100 250 100 250

0 0 0 0

0.5 1.0 1.5 2.0 2.5 3.0 0.5 1.0 1.5 2.0 2.5 3.0

50 75 100 125 50 75 100 125


300 750 300 750

Pile 3 Pile 4
200 500 200 500

100 250 100 250

0 0 0 0

0.5 1.0 1.5 2.0 2.5 3.0 0.5 1.0 1.5 2.0 2.5 3.0
Pile lateral displacement at model scale: mm

Fig. 13. Lateral force–displacement curves of individual piles under one-way cyclic lateral loads

Delivered by ICEVirtualLibrary.com to:


IP: 129.169.110.108
On: Tue, 26 Oct 2010 14:12:00
56 Li, Haigh and Bolton
Piles 1 and 2 1.0 50
1500 3750

Pile head axial force at prototype scale: kN


Piles 1 and 2

Axial displacement at prototype scale: mm


1000 2500 Piles 3 and 4
Axial displacement at model scale: mm

Axial displacement at model scale: mm


Average
500 1250
0 0 0.5 25
_500 _1250
0 200 400 600 800 1000 1200

Piles 3 and 4
1500 3750
0 0
1000 2500
500 1250
0 0
_500 _1250 _0.5 _25
0 200 400 600 800 1000 1200
Elapsed time: s 0 200 400 600 800 1000 1200
Elapsed time: s

Fig. 14. Axial force variation of individual piles during the cyclic
lateral load test ZL08-4-1 Fig. 15. Axial displacements accumulated during the cyclic
lateral load test ZL08-4-1

Table 3. Permanent settlement accumulated during cyclic lateral load tests

Test number ZL08-3-1 ZL08-3-2 ZL08-3-3 ZL08-4-1 ZL08-4-2 ZL08-1-1 ZL08-1-2 ZL08-2
Pile type PJ PJ PJ MJ MJ PJ PJ MJ
Cyclic load ampli- ¡200 ¡600 ¡875 ¡600 ¡750 +600 +1000 +1000
tude: N
Cyclic load number 10 5 5 10 10 11 5 60
Permanent settle- 1?6 2?4 3?0 0?42 1?23 2?5 1?3 2?05
ment: mm

PJ, pre-jacking; MJ, monotonic jacking.

Significant permanent lateral displacements of the pile groups, offering some insights regarding appropriate
group accumulate under one-way cyclic lateral loading but construction and design methods.
not with two-way symmetric cyclic loading. This accumu-
lated permanent lateral displacement increases approxi-
mately linearly with the logarithm of the number of one- NOTATION
way cycles.
The pile group lateral secant stiffness increases slightly d50 average particle size
under cyclic lateral loads, possibly caused by the local Kcinstall initial pile group vertical stiffness
densification of sand surrounding individual pile shafts. Kbinstall initial pile group base stiffness
RH cyclic load ratio
However, when the pile group is close to its original
location, the stress level of the sand at the pile–soil interface
can drop, inducing local softening of the pile group. Both
ACKNOWLEDGEMENTS
observations are consistent with the effects of cyclic
This research was funded by Atkins Geotechnics and by
densification.
Giken Seisakusho Ltd. The first author is also supported
When the spacing of individual piles is four times the pile
by the China Scholarship Council and the Cambridge
diameter, interaction effects on the pile group are
Overseas Trust.
significant. Under cyclic lateral loads, the soil within the
pile group in front of the trailing piles is affected by the
movement of leading piles, and thus the stress level and the REFERENCES
stiffness of the soil reduces. The leading piles have larger Brown, D. A., Reese, L. C. & O’Neill, M. W. (1987). Cyclic lateral
pile head secant stiffnesses and carry higher loads than the loading of a large-scale pile group. J. Geotech. Engng ASCE
trailing piles. 113, No. 11, 1326–1343.
A constant axial force was applied on the pile cap, but a Deeks, A. D. (2008). An investigation into the strength and stiffness
swaying motion of the pile group was induced by the cyclic of jacked piles in sand. PhD thesis, University of Cambridge,
lateral loads. Individual piles were thus subjected to cyclic Cambridge, UK.
axial loads induced by the lateral loads, causing large Franke E. (1988). Group action between vertical piles under
accumulated permanent settlements. The settlement horizontal loads (ed. W. F. Van Impe), pp. 83–93. Rotterdam:
increases with increasing amplitude of cyclic lateral loads. Balkema.
Additionally, the settlement of the pre-jacked pile group is Fleming W. G. K., Weltman A. J., Randolph M. F., and Elson W. K.
(1992). Piling Engineering, 2nd edn. Blackie (Halsted Press),
much larger than that of the monotonically jacked pile Glasgow, 1992, pp. 168.
group, because the more highly stressed soil created by in- Gui, M. W., Bolton, M. D., Garnier, J. et al. (1998). Guidelines
flight jacking installation causes an increased stiffness in for cone penetration tests in sand. Proceedings International
the soil surrounding individual piles. Conference Centrifuge ’98, Tokyo, 1, pp. 155–160.
These tests provide a better understanding of the Haigh, S. K., Houghton, N. E., Lam, S. Y., Li, Z. & Wallbridge,
influence of cyclic lateral loads on the performance of pile P. J. (2010). Development of a 2D servo-actuator for novel

Delivered by ICEVirtualLibrary.com to:


IP: 129.169.110.108
On: Tue, 26 Oct 2010 14:12:00
The response of pile groups under cyclic lateral loads 57
centrifuge modelling. Proceedings of the 7th International factors in submerged sand. J. Geotech. Engng ASCE 115, No.
Conference on Physical Modelling in Geotechnics 2010 (eds. 3, 359–378.
S. Springman and J. Laue), pp. 239–244. Zurich: Taylor & Poulos, H. G. (1987). Analysis of residual effects in piles.
Francis. J. Geotech. Engng ASCE 113, No. 3, 216–229.
Jacobsz, S. W. (2002). The effect of tunnelling on pile foundations. Prakash, S. (1962). Behaviour of pile groups subjected to lateral
PhD thesis, University of Cambridge, Cambridge, UK. loads. PhD thesis, University of Illinois at Urbana, Urbana,
Long, J. H. & Vanneste, G. (1994). Effects of cyclic lateral loads IL, USA.
on piles in sand. J. Geotech. Engng ASCE 120, No. 1, 225– Rollins, K. M., Gerber, T. M., Lane, J. D. & Ashford, S. A.
243. (2005). Lateral resistance of a full-scale pile group in liquefied
Madabhushi, S.P.G., Houghton, N.E., & Haigh, S.K. (2006). A sand. J. Geotech. Geoenviron. Engng, ASCE 131, No. 1, 115–
new automatic sand pourer for model preparation at 125.
University of Cambridge. Proceedings of the 6th International White, D. J. (2003). PSD measurement using the single particle
Conference on Physical Modelling in Geotechnics 2006 (eds optical sizing method. Géotechnique 53, No. 3, 137–326.
C. W. W. Ng, C. Schaur and Y. H. Wang), pp. 217–222. Hong White, D. J. & Lehane, B. M. (2004). Friction fatigue on
Kong: Taylor & Francis. displacement piles in sand, Géotechnique 54, No. 10, 645–658.
McDowell, G. R. & Bolton, M. D. (2001). Micro-mechanics of White, D. J. & Deeks, A. D. (2007). Recent research into the
elastic soil. Soils Found. 41, No.6, 147–152. behaviour of jacked foundation piles. International workshop on
Ochoa, M. & O’Neill, M. W. (1989). Lateral pile interaction recent advances in deep foundations, The Netherlands, pp. 3–26.

Delivered by ICEVirtualLibrary.com to:


IP: 129.169.110.108
On: Tue, 26 Oct 2010 14:12:00
View publication stats

You might also like