Professional Documents
Culture Documents
A Model Scale Investigation Into Setup of Displacement Piles in Sand
A Model Scale Investigation Into Setup of Displacement Piles in Sand
A Model Scale Investigation Into Setup of Displacement Piles in Sand
ABSTRACT: A model scale investigation into set-up of displacement piles in sand is described. Thirty-two
model piles were installed into two test chambers, and repeatedly load tested over a period of 100 days. Cor-
rodible and inert piles were used, and in one test chamber the water table was cycled. Consistent profiles of
penetration resistance and initial load test capacity confirmed the repeatability of the modelling procedure.
Corrosion of the small model piles led to dramatic set-up. In the absence of corrosion, modest set-up was ob-
served. Cycles of the water table significantly increased the amount of set-up, providing some suggestion of
the governing mechanism, and indicating a possible technique for the enhancement of set-up.
1 INTRODUCTION 1.3 Mechanisms of set-up
1.1 ‘Set-up’ of displacement piles in sand From the initial stress conditions shown in Figure 1,
any creep-induced equalization of stresses will tend
The shaft capacity of displacement piles in sand is
to increase the radial stress at the pile wall, and
often observed to increase with time, even after dis-
hence cause set-up. Creep of a granular material can
sipation of installation-induced excess pore pres-
be attributed to small perturbations in loading caus-
sure- this phenomenon is known as set-up. Set-up
ing sliding at points of contact. Sources of load per-
rates of 20%-170% per log cycle of time and in-
turbations in the field include thermal effects, traffic
creases in capacity by factors of 5 or more have been
loading, precipitation and tides. Deliberate small cy-
reported, and trend lines have been proposed (Skov
cles of loading will accelerate this mechanism of
& Denver, 1988; Fellenius et al, 1989; Chow et al
creep by increasing the frequency that contacts slide
1998; Bullock et al 2005). However, the governing
and loads redistribute. These cycles of loading could
mechanisms are not well understood.
be applied to the soil near the pile either by the pile
itself, or by ambient changes in either the total stress
1.2 Effect of installation or the pore pressure.
Other proposed mechanisms of set-up include (i)
Since set-up is not observed on bored piles in sand,
an increase in soil dilatancy with time due to aging,
some aspect of the installation process of displace-
leading to a greater increase in radial stress on load-
ment piles is a necessary condition for set-up to oc-
ing, (ii) kinematically restrained dilatant creep,
cur. The installation-induced stress history of a soil
which causes an increase in radial stress with time
element adjacent to a displacement pile in sand in-
(Bowman & Soga 2005), and (iii) chemical or corro-
volves (i) loading to failure at a stress level compa-
sion effects. Bowman & Soga (2005) conducted tri-
rable to the cone resistance qc as the pile tip ap-
axial tests that mimicked the stresses created by pile
proaches the soil element, (ii) a reduction in stress
installation. They observed a dilatant response dur-
by approximately two-orders of magnitude as the
ing creep at constant stress and a high stress ratio.
pile tip moves past, and (iii) cyclic shearing along
Greater dilation was observed when small loading
the pile shaft, leading to contraction of the interface
cycles were imposed. The mechanisms of set-up are
zone and ‘friction fatigue’ (Poulos 1989, Randolph
reviewed in more detail by Chow et al. (1998),
2003, White 2005).
Axelsson (2000) and Bowman & Soga (2005).
This loading history causes particle breakage and
This paper describes a model-scale investigation
results in a stress distribution close to the pile shaft
into set-up. The hypothesis that small loading cycles
that can be idealized as cavity expansion followed
can increase set-up is tested by imposing cycles of
by a slight cavity contraction (White et al. 2005).
ambient effective stress via changes in the water ta-
The final unloading step creates a zone close to the
ble level.
pile in which the circumferential (hoop) stress is
greater than the radial stress (Figure 1).
sand which originates from the west coast of Ireland.
Coarse particles were removed by a 1.18 mm sieve
before the sand was pluviated into the test chambers
from a constant drop height at a constant rate, form-
ing a uniform medium dense bed of depth 330 mm.
Coop (1990) notes that carbonate sands “exhibit
all the principal features expected of conventional
soils”, and presents triaxial test data for Dog’s Bay
sand within the framework of critical state soil me-
chanics. Yielding, compression and the critical state
line all take the same form as for silica sands, but
shifted to a lower stress level, reflecting the weaker
carbonate particles. The low ambient stress level in
this investigation (compared to prototype condi-
tions) may be partly compensated for by the low
strength and stiffness of the model sand: the ob-
served trends may be representative of stronger soil
Figure 1. Idealized stress distribution close to pile, based on
cavity expansion followed by contraction (White et al 2005)
and deeper depths. White (2005) shows that the shaft
resistance of displacement piles in ‘normal’ silica
and ‘unusual’ carbonate sand show similar trends. A
2 EXPERIMENTAL METHODOLOGY key difference is that greater friction fatigue occurs
in carbonate sands, leading to a sharper reduction in
2.1 Model-scale testing radial stress above the pile tip.
The testing programme was conducted at model
scale in the laboratory. Although stress similitude 2.4 Model piles
could be achieved using the centrifuge, set-up is not
necessarily accelerated at high g-levels in the same The model piles were fabricated from steel rod, with
manner as consolidation. If set-up is due to creep, a diameter of 10 mm. Most of the piles were made
and if creep arises from random perturbations of from mild steel, which can corrode, but 4 additional
load, these random loads must be replicated within piles were made from stainless steel to separate this
the centrifuge and accelerated. Since such modelling influence. The piles were initially smooth.
is not practically achievable, this investigation was The piles were uninstrumented, so the relative
conducted in the laboratory at 1-g. contributions of base and shaft resistance to the pile
The use of a pluviated soil bed and small piles al- capacity could not be identified. However, since it is
lowed a large number of tests to be conducted in generally accepted that set-up of base resistance
uniform conditions. By not using the centrifuge, the does not occur, any changes in compression capacity
piles could be left undisturbed for almost 100 days. are assumed to arise from changes in shaft resis-
tance.
0.4 The four stainless steel piles did not corrode dur-
ing the test program, and exhumation revealed that
0.6 no soil had adhered to the shafts, which remained
smooth. Figure 6 compares the repeated load testing
0.8 of one stainless steel pile in each chamber. There is
negligible set-up in chamber 1, but a steady increase
in capacity in chamber 2. As for the mild steel piles,
1
cycles of pore pressure increase the amount of set-
up; in this case in the absence of corrosion.
1.2 The trends in each of Figures 4-6 were matched
by another pile of the same type in the same cham-
1.4 ber undergoing the same sequence of tests, confirm-
(a) Settlement zeroed at start of load test ing the repeatability of the results. All of these re-
sults are collated in Figure 7. The D/10 capacity in
Pile head load (N) each load test has been normalized by the capacity
0 100 200 300 400 measured during the initial load test on that pile.
0
Pile head load (N)
1 day
0 50 100 150
Cumulative pile head settlement (mm)
1
0
5 days
Pile head settlement (mm)
2 0.25
14 days 0 days
3 0.5 2 days
22 days 9 days
4 0.75 16 days
26 days 27 days
1
5
33 days 1.25
6 (a) Pile 14, chamber 1: fixed water table
(b) Settlement cumulated during load test sequence Pile head load (N)
0 50 100 150
Figure 5. Tests on pile 16, chamber 2 (mild steel, cyclic WT)
0
Although the capacity of pile 10 did increase be-
Pile head settlement (mm)
yond the day 0 value, the later tests (after 20 days) 0.25
indicate a reduction in capacity that can be attributed
to degradation of the shaft resistance due to the re- 0 days
0.5
peated cycling- an effect known as friction fatigue 6 days
(Figure 4b) (White & Lehane 2004). 18 days
Pile 16 in chamber 2 showed more dramatic set- 0.75 26 days
up, with a four-fold increase in Q0.1D over the 30 day
test period (Figure 5). The cycles of pore water pres- 1
sure enhanced the rate and magnitude of set-up.
However, post-test exhumation of the mild steel
piles revealed that crushed sand had been bonded to 1.25
the pile shafts by corrosive action. Greater corrosion (b) Pile 18, chamber 2: cyclic water table
and stronger bonding was observed in chamber 2,
presumably since the corrosion was accelerated by Figure 6. Tests on stainless steel piles
4.5 seasonal climatic conditions create significant cycles
Cyclic w ater table
4
Pile 1, chamber 1 of ambient pore pressure. Cycles of pore pressure ar-
Normalised pile capacity
1.5
Pile 13, chamber 1
REFERENCES
Normalised pile capacity