Friction fatigue on displacement piles in sand by nikeborome

VIEWS: 104 PAGES: 14

White, D. J. & Lehane, B. M. (2004). Geotechnique 54, No. 10, 645–658

                        Friction fatigue on displacement piles in sand
                                             D. J. W H I T E * a n d B. M . L E H A N E †

Experiments with instrumented displacement piles have                                                               ´
                                                                            Les essais avec des piles de deplacement instrumentees           ´
shown that the ultimate shaft friction that can develop in                                 ´
                                                                            ont montre que la friction d’arbre ultime qui peut
a given sand horizon decreases as the pile tip penetrates                             ˆ
                                                                            apparaıtre dans un horizon sableux donne diminue a     ´             `
to deeper levels. This phenomenon, which is now com-                                                                     ´ `
                                                                            mesure que la pointe de la pile penetre a des niveaux`
monly referred to as friction fatigue, is investigated here                                        ´                     ´    `
                                                                            plus bas. Nous etudions ici ce phenomene qui est au-
using centrifuge model piles equipped with lateral stress                                        ´            ´
                                                                            jourd’hui appele communement fatigue de friction, en
sensors, and by drawing on other experimental data from                                                         `
                                                                            utilisant des piles en modeles centrifuges equipees de   ´ ´
the laboratory and the field. It is shown that the primary                     ´                                   ´
                                                                            detecteurs de contrainte lateraux et en puisant dans
mechanism controlling friction fatigue is the cyclic his-                                ´               ´
                                                                            d’autres resultats experimentaux en laboratoire et sur le
tory imparted during pile installation to soil elements at                                                            ´
                                                                            terrain. Nous montrons que le mecanisme primaire con-
the pile–sand interface. For a given installation method                        ˆ
                                                                            trolant la fatigue de friction est l’histoire cyclique impar-
the stationary lateral stress acting at any given level on a                                                                    ´´
                                                                            tie pendant l’installation des piles sur les elements de sol
displacement pile can be described as a relatively unique                   `                                              ´
                                                                            a l’interface pile-sable. Pour une methode d’installation
function of the cone penetration test end resistance and                            ´                       ´
                                                                            donnee, la contrainte laterale stationnaire agissant a un    `
the number of cycles imposed during installation. The                                        ´                          ´
                                                                            niveau donne sur une pile de deplacement peut etre             ˆ
strong influence of cycling, which is also seen in cyclic                      ´
                                                                            decrite comme une fonction relativement unique de la
constant normal stiffness interface shear tests, is attribu-                  ´
                                                                            resistance finale CPT et du nombre de cycles imposes                ´
ted to contraction of a narrow shear zone at the shaft–                     pendant l’installation. La forte influence du cyclage qui
soil interface that is surrounded by soil with a relatively                       ´                  ´
                                                                            est egalement notee dans les essais de cisaillement sur
high lateral stiffness.                                                                        `          ´           `
                                                                            une interface a rigidite normale a constante cyclique est
                                                                                      ´ `
                                                                            attribuee a la contraction dans une zone de cisaillement
                                                                            ´          `                                           ´
                                                                            etroite a l’interface arbre-sol qui est entouree par un sol
KEYWORDS: piles; sands                                                                            ´    ´
                                                                            ayant une rigidite laterale relativement elevee. ´ ´

INTRODUCTION                                                                techniques are to become more widely accepted, research is
There are significant uncertainties associated with the pre-                 needed to investigate the influence of the installation method
diction of the axial capacity of displacement piles in sand.                on pile behaviour. Since conventional design methods have
Recent experiments involving field-scale instrumented piles                  been developed based on historic experience with dynami-
(e.g. Lehane, 1992; Chow, 1997) have successfully reduced                   cally installed piles, caution is required when these methods
the level of this uncertainty, leading to the development of                are applied to jacked piles.
improved approaches such as those proposed by Lehane &
Jardine (1994), Randolph et al. (1994) and Jardine & Chow
(1996). These methods have a greater prediction reliability                 BACKGROUND
primarily because of the improved understanding of shaft                       Field measurements of the distribution of horizontal and
friction characteristics obtained through measurement of                    shear stress acting on a pile shaft have highlighted the
horizontal stresses acting on the pile shaft (Poulos et al.,                deficiencies of conventional ‘earth pressure’ design ap-
2001). However, a number of significant facets require                       proaches in which the in situ vertical effective stress profile
further investigation to enable formulation of more generally               is multiplied by assumed earth pressure and interface friction
accepted design approaches with reduced reliance on empiri-                 coefficients to estimate a distribution of the available ulti-
cal correlations.                                                           mate shaft shear stress (ôf ) with depth; ôf values derived in
   A second motivation for research into shaft friction on                  this way are often restricted to limiting maximum values
displacement piles lies in the development of new pile                      (e.g. API, 2000). The ‘earth pressure’ approach for displace-
installation techniques that involve jacking rather than con-               ment piles in sand has been largely replaced in the onshore
ventional dynamic techniques, which are becoming less                       environment by methods that relate ôf to an in situ test
acceptable in urban areas because of noise, vibration and                   parameter such as the cone penetration test (CPT) end
emission restrictions. White et al. (2002) and Lehane et al.                resistance (qc : e.g. Bustamante & Gianeselli, 1982). The
(2003) describe two novel pile-jacking systems with capaci-                 design method of Jardine & Chow (1996) also assumes a
ties greater than 4 MN for the installation of large pre-                   direct relationship between ôf and qc but is shown to have
formed displacement piles. These machines offer an alter-                   greater reliability primarily because it incorporates the trend,
native to bored piling for the construction of deep founda-                 seen in field experiments, for the available shaft friction at a
tions in urban areas. However, if these new jacking                         given soil horizon to decrease with increasing penetration of
                                                                            the pile tip (e.g. Lehane et al., 1993). Heerema (1980)
                                                                            referred to this phenomenon as friction fatigue and demon-
Manuscript received 17 February 2004; revised manuscript accepted
                                                                            strated its influence on pile driveability.
20 September 2004.
Discussion on this paper closes on 1 June 2005, for further details            The Randolph et al. (1994) and Jardine & Chow (1996)
see p. ii.                                                                  design methods account for friction fatigue using relation-
* Department of Engineering, University of Cambridge, UK.                   ships that can be divided into (a) the calculation of a
† School of Civil and Resource Engineering, University of Western           maximum available shaft friction (ôf ) from a sand horizon
Australia, Crawley, Western Australia.                                      just above the pile tip and (b) the calculation of a degrada-

646                                                     WHITE AND LEHANE
tion, or fatigue, of ôf as the pile tip penetrates further.         gests a possible influence of differences between the respec-
Randolph et al. (1994) propose a reduction in lateral stress        tive jacking procedures. For example, the ICP was installed
(and hence shaft friction) towards active earth pressure            in jacking strokes with a typical length of 0.23 m (2.3D),
conditions in proportion to eÀì h=D , where h is the distance       and therefore the number of cycles encountered at z ¼ 3 m
above the pile tip, D is the pile diameter, and ì is an             prior to arrival of the three instruments was about 2, 6 and
empirical degradation factor. They note that ì is likely to be      17 respectively. A 40% reduction in ó hc (or ôf ) occurs
influenced by a number of factors, including soil compressi-         between h/D of 4 and 37 for this case, whereas no friction
bility, pile roughness and incremental driving energy. Jardine      fatigue is measured between h/D of 4 and 35 for the CPT,
& Chow (1996) refer to friction fatigue as the ‘h/R effect’         for which the number of cycles experienced at a soil horizon
(R being pile radius), and propose that the horizontal stress       less than 1 m behind the cone tip is either zero or one, and
acting on the pile shaft reduces in proportion to (h/R)À r ,        between 1 m and 2 m is either one or two.
where an r value of 0.38 was found to provide a best fit to a           The final example of friction fatigue is from a full-scale
large database of high-quality load tests.                          instrumented tubular pile installed in dense sand for the
   The mechanisms governing friction fatigue are unclear,           EURIPIDES project, described by Fugro (1996). A heavily
and there is no particular behavioural model that provides a        instrumented 762 mm diameter open-ended tubular pile was
basis for the derivation of factors such as ì and r. Further-       load-tested at three penetrations between 30 and 47 m below
more, while both of these friction fatigue approaches choose        ground level. The test site comprised dense sand below 22 m
to normalise the distance h by pile diameter (or radius), this      depth, with CPT qc values in the range 45–70 MPa. The
normalisation remains to be justified by a governing me-             pile was driven using a 90 kJ hammer reaching a blowcount
chanism. Few field records include measurements of local             of $400 blows/m at the final embedded depth.
shaft friction distribution, and these are often unreliable            Figure 1(c) shows the distribution of external ôf values at
owing to residual loads, instrument damage during driving,          a pile head settlement of D/10 calculated from the axial
and the difficulty of separating internal and external shaft         force distribution in the pile and ignoring internal shaft
friction on open-ended piles.                                       friction (which Randolph et al., 1991, and others, show to
   To illustrate friction fatigue, profiles of ultimate shaft        be negligible at more than about 4D above the tip). Some
friction from three different types of pile or penetrometer         scatter is evident, as would be expected considering the
are shown in Fig. 1. These have been selected to demon-             demands placed on the instrumentation during such hard
strate the extreme cases of low and high friction degradation       driving. It is also noted that the external shaft frictions
during installation, and show that the rate of degradation is       inferred over the lower two diameters are influenced by high
strongly influenced by the method of installation.                   internal shaft friction mobilised on the plug (shown dotted),
   The results from four soundings using a CPT instrument           as evidenced from internal horizontal stress measurements.
equipped with four friction sleeves located above the penet-        However, there is a clear trend for significant friction
rometer tip are shown in Fig. 1(a) (DeJong, 2001; DeJong &          fatigue. For example, over the depth range 30–35 m, the
Frost, 2002). Three configurations of the CPT allow meas-            mean ôf value was 330 kPa during the load test when the
urements of sleeve friction at distances of between 167 mm          pile tip was at 38.7 m. After a further 8.2 m (11D) of
and 1517 mm behind the shoulder of the cone tip. During             driving, a subsequent re-test indicated that this value of ôf
the soundings, each of which involved jacking strokes with a        reduced to 130 kPa, which is only 40% of the original value.
length of 1 m (or 23D), the second and third sleeve indicate        During the period of driving between load tests, approxi-
frictions that are $8% below and above those registered by          mately 3000 hammer blows were applied to the pile head.
the first and last sleeves respectively. Post-test measurements         The three examples described above suggest that friction
of the sleeve diameters revealed the second and third sleeves       fatigue does not occur in the absence of loading cycles, and
to be 0.05 mm smaller and larger than the first and last,            that a greater number of cycles imposed during installation
explaining the observed consistent minor difference in re-          leads to a larger reduction in shaft friction at a given soil
corded friction. Aside from these minor discrepancies, the          horizon. This observation suggests that the distribution of
three soundings indicate no reduction in shaft friction be-         available shaft friction after pile installation might be better
tween h/D ¼ 4 and 35 (or between 167 and 1517 mm behind             predicted by linking friction fatigue to the loading cycles
the penetrometer shoulder).                                         induced by the installation method rather than employing an
   Measurements of stationary horizontal effective stress, ó hc ,
                                                              9     empirical relationship with h/D.
acting on the shaft of the 101 mm diameter, Imperial College           Prompted by trends such as those inferred from the data
instrumented model pile (ICP) at three stages during installa-      in Fig. 1, this paper describes a systematic investigation into
tion in dense sand are shown in Fig. 1(b) (Chow, 1997; test         the effects of the installation method on friction fatigue and
DK2). The term ‘stationary’ is employed as these values were        shaft capacity of displacement piles in sand. The investiga-
recorded between jacking strokes when the pile was under            tion was conducted at reduced scale using instrumented
zero head load. Four horizontal stress transducers located          model piles in a geotechnical drum centrifuge so that a large
along the pile shaft provide a series of observations at a given    number of pile installations could be performed in a uniform
soil horizon as each instrument passes that depth, z. A clear       soil sample at a cost that was approximately two orders of
trend for ó hc measured at a given z to reduce with increasing
             9                                                      magnitude lower than that of an equivalent series of field-
penetration is evident. For example, close to z ¼ 3 m, ó hc is
                                                            9       scale tests. The paper focuses on the distribution of lateral
recorded as 169 kPa when the first instrument, located               stresses acting on the centrifuge piles during installation and
406 mm (4D) behind the pile tip, passes that level. A 36%           subsequent cyclic load tests to assist understanding of the
reduction occurs prior to the arrival of the second instrument      friction fatigue mechanism.
after a further 965 mm (9.5D) of penetration. On arrival of
the fourth instrument, located 3760 mm (37D) behind the pile
tip, the registered ó hc value had fallen to 98 kPa, which is
                      9                                             DESCRIPTION OF EXPERIMENTS
less than 60% of the value measured at h ¼ 4D.                      Drum centrifuge
   Measured ôf values showed a virtually identical rate of            This investigation comprised 18 pile installations con-
degradation with h to that of the ó hc values (see Chow,
                                         9                          ducted in the drum centrifuge at the University of Western
1997). This degradation is in marked contrast to that shown         Australia (UWA). The ring channel of this machine has an
in the multiple friction sleeve CPT experiments, which sug-         outer diameter of 1.2 m, an inner diameter of 0.8 m, and a
                                                                                         FRICTION FATIGUE ON DISPLACEMENT PILES IN SAND                                                                                                                                                                                           647
                                                                                                                                                                                                                                                    Stationary horizontal stress, σ′

                                                                                                                                                                                                                                                0      50            100       150      200
                                                                         40                                                                                                                                                                 0                                              0

              Normalised instrument distance behind cone shoulder, h/D
                                                                                                                          S1        1600

                                                                                                                                            Instrument distance behind cone shoulder: mm
                                                                                                                                                                                                                                            1                                             10
                                                                                                                          S3        1400

                                                                                                                                                                                                                                                                                               Normalised instrument depth, z/D
                                                                         25                                                                                                                                                                 2                                             20

                                                                                                                                                                                                                   Instrument depth, z: m

                                                                                                                                    800                                                                                                     3                                             30

                                                                                                                                                                                                                                            4                                             40

                                                                         5        Conventional
                                                                                     f riction                                      200                                                                                                     5                                             50
                                                                                     sleeve                                                                                                                                                                  z tip    3·26 m = 32 D
                                                                                                                                                                                                                                                             z tip    4·36 m = 43 D
                                                                         0                                                         0
                                                                              0        25          50                75         100                                                                                                                          z tip    5·77 m = 57 D
                                                                                      Mean shear stress, τf: kPa                                                                                                                            6                                             60

                                                                                                   (a)                                                                                                                                                               (b)

                                                                                        Ultimate shaft friction, τf: kPa

                                                                                  0     200      400           600        800       1000

                                                                                                                z tip     30·45 m
                                                                                                                z tip     38·70 m
                                                                          25                                    z tip     46·90 m

                                                                                                                                                                                           Normalised depth, z/D
                                         Depth, z: m





                                                                          50                                                           65


              Fig. 1. Observed distributions of shaft friction on: (a) a multi-sleeve CPT instrument (D    43.7 mm)
              (DeJong, 2001); (b) a jacked instrumented model pile (D            101.6 mm) (Chow, 1997); (c) a long
              offshore pile (D   762 mm) (Fugro, 1996)

channel height of 0.3 m. The key advantage of a drum                                                                                                                                                                  A key feature of the centrifuge is its independently
centrifuge over a beam centrifuge is the increased plan area                                                                                                                                                       rotating central shaft and tool table, which can be driven
of the soil sample, which permits a large number of tests to                                                                                                                                                       relative to the ring channel by a hollow stepper motor, and
be conducted in a uniform soil model. Compared with the                                                                                                                                                            brought to a halt independently of the channel. An actuator
1.8 m radius beam centrifuge at UWA, a sample in the drum                                                                                                                                                          is mounted on the tool table, onto which instrumented tools
centrifuge has three times more plan area, and so is ideally                                                                                                                                                       can be attached and controlled. Twin stepper motors allow
suited to parametric studies of the kind reported in this                                                                                                                                                          precise vertical and radial movement of the tool, while a
paper.                                                                                                                                                                                                             further stepper motor controls a counterbalance. Two per-
648                                                    WHITE AND LEHANE
sonal computers (PCs) are mounted on the tool table: one        pile, after which the slots were backfilled with clear epoxy
controls the actuator stepper motors, and the other is for      (Fig. 2). The head of the pile was attached to a cap that was
data acquisition. The actuator control PC receives commands     connected via a 10 kN axial load cell to the tool table. The
from a PC in the control room via a serial link across the      wiring for the pressure cells passed up the body of the pile
slip rings. Feedback between the onboard data acquisition       in a 6 mm diameter hole, giving the pile a net cross-
PC and the control room PC allows the actuator to be            sectional area of 52.7 mm2 . Noting that the ratio of the
operated in a load-controlled mode. A full technical descrip-   Young’s modulus of steel to that of concrete is $7, the axial
tion of the facility is provided by Stewart et al. (1998).      stiffness of the model pile is about 4.5 times higher than
                                                                that of a solid concrete pile.
                                                                   After construction of the pile, the pressure cells were
Model pile                                                      exercised under air in a pressurised chamber. No hysteresis
   To measure the horizontal stress acting on the shaft, a      was evident in the range 0–500 kPa, and linear calibration
model pile was fabricated with miniature total pressure cells   factors, including those to correct for a small cross-sensitiv-
(Kyowa PS-5KA) embedded on the surface (Fig. 2). A              ity to axial load in the range 10–20 kPa/kN, were estab-
square section (9 mm 3 9 mm) pile was used so that these        lished for each sensor. The possibility of cell action effects
flat circular cells could be mounted flush with the pile          influencing the operative calibration factor in soil was con-
surface. The pile was fabricated from stainless steel and       sidered, although it will be shown later that cell action
machined to a mean centreline surface roughness, RCLA , of      effects were, at worst, consistent between instruments, per-
0.55 ìm.                                                        mitting comparison of horizontal stress measurements at
   Six pressure cells were mounted at four locations between    different locations.
9 and 108 mm behind the pile tip (i.e. h/B ¼ 1, 3, 6 and 12,
where h is the height above the pile tip and B is the pile
width). By instrumenting both sides of the pile close to the    Soil properties
tip, a degree of redundancy was incorporated and the repeat-       The centrifuge tests employed uniform rounded fine silica
ability of the measurements could be checked. The pressure      sand, with properties summarised in Table 1. A series of 30
cells were glued into slots machined into the face of the       constant normal load direct interface shear box tests were
                                                                performed at a range of relative densities in excess of $50%
                                                                and vertical stresses between 50 and 250 kPa. Although the
                                                                interface would normally be considered smooth, having a
                                                                relative roughness, Rn % 2RCLA /D50 of 0.006, all samples
                                                                dilated by between 0.01 and 0.02 mm and registered a peak
                                                                strength. The average peak and constant-volume friction
                                                                angles were 168 and 128 respectively, showing no significant
                                                                variation with initial density or confining stress over the
                                                                range investigated.

                                                                Soil model preparation
                                                                   A single bed of sand was used for the entire testing
                                                                programme. This bed was prepared in three stages. First, the
                                                                channel was filled to the maximum depth of 200 mm with
                                                                dry sand using an automatic pourer while the ring channel
                           Cell                                 was spinning at 20g. The sand was then saturated with water
                           B4                                   and left to drain overnight through the base, leaving the
                                                                material slightly damp and with sufficient suction to hold
                                         108 mm                 the sample in position when the ring channel was halted.
                                         (12B)                  Finally, the channel was brought to a halt and the surface
                                                       185 mm   was screeded flat to a nominal sample depth of 180 mm
                                                                using a rotating cutter attached to the tool table. The actual
                                                                variation in sample depth around the channel, as shown
                                                                by the touchdown position of each pile test, was less than
                                                                Æ 1 mm. The channel was then accelerated to 50g, and
                           Cell                                 remained spinning for 9 days while the test programme was
                           B3                                   conducted. The bottom channel drain remained open
                                                                throughout testing, and no outflow of water was evident after
                                         54 mm                  the first 24 h of spinning. However, during emptying of the
                                         6B                     channel after testing, the lower 50 mm of the sand bed
                           Cell                                 remained slightly damp, indicating that completely dry con-
 F2                        B2                                   ditions had not been achieved.

                                         27 mm
 Cell                      Cell          (3B)                   Table 1. Soil properties
 F1                        B1                                   Property                                          Value
                                         9 mm (B)               D10 particle size: mm                             0.100
                                                                D50 particle size: mm                             0.180
                                                                Maximum voids ratio, emax                         0.762
                9 mm (B)      9 mm (B)
                                                                Minimum voids ratio, emin                         0.493
                                                                Specific gravity, Gs                               2.65
Fig. 2. Schematic diagram of instrumented model pile
                                     FRICTION FATIGUE ON DISPLACEMENT PILES IN SAND                                                            649
SOIL PROFILE CHARACTERISATION                                                      distribution of Dr may be attributed to the reducing ‘fall
   A total of 12 CPT soundings were conducted to character-                        height’ of the sand during deposition.
ise the sand bed, using a 6 mm diameter CPT probe inserted
into the soil at a rate of 1 mm/s. These tests were conducted
at different positions around the sand bed to assess the                           TEST PROGRAMME
homogeneity of the sample, and after each change in g-level.                          The full test programme included tests at acceleration
The CPT soundings conducted during the initial phase of                            levels of 50g and 150g. This paper will consider only the
testing at 50g are shown in Fig. 3. Good agreement is found                        initial series of tests conducted at 50g (tests T1–T13, Table
between all tests in the upper 80 mm of the sample, with                           2). The tests were located at 108 intervals around the drum
deviation of up to Æ15% from the average evident between                           perimeter, and staggered between the upper and lower parts
80 and 120 mm. The average of soundings 1A, 1B, 2A and                             of the channel. This arrangement was selected to maximise
2B has been used as the ‘design’ profile throughout the                             the separation of the test locations and the channel walls.
back-analysis described in the paper.                                              Adjacent tests were located 108 mm (12B) apart at the sand
   Using the Lunne & Christofferson (1983) correlation be-                         surface, increasing to 123 mm (13.7B) apart at the final
tween qc , vertical effective stress, ó v , and relative density,
                                        9                                          installation depth of 120 mm owing to the radial direction of
Dr, it is estimated that the sample’s relative density (Dr )                       installation.
increases with depth from $20% at the sample surface to
$80% at a depth of 60 mm and $90% at 120 mm. This
                                                                                   Installation methods
                              Cone resistance, qc: MPa                                The pile was installed to a final depth, L, of 120 mm in
                                                                                   all 18 pile installations performed. Installation was generally
                     0   10             20               30              40
                                                                                   paused after 60 mm, L/2, of penetration so that a static
                                                                                   compression test could be performed before installation
                                                                                   continued to the full depth (Table 2). After reaching full
                                                          Average                  depth, the pile was unloaded before undergoing a sequence
                                                          CPT 1A
                                                                                   of static and/or cyclic load tests. Three techniques were used
                                                                                   to install the test pile in order to investigate the influence of
                                                          CPT 1B
                                                                                   cycling during penetration:
                                                          CPT 2A
               40                                         CPT 2B                   (a) monotonic installation, comprising a monotonic push at
                                                                                       0.2 mm/s to half of the final pile depth followed by
                                                                                       another monotonic push to the final pile length
                                                                                   (b) jacked installation, comprising cycles of fixed down-
                                                                                       ward displacement (2 mm at 0.2 mm/s, i.e. one ‘jack-
   Depth: mm

                                                                                       stroke’) followed by unloading to zero head load at
                                                                                       0.005 mm/s
               80                                                                  (c) ‘pseudo-dynamic’ installation, comprising two-way
                                                                                       cycles of fixed downward (2 mm at 0.2 mm/s) and
                                                                                       upward (1.5 mm at 0.2 mm/s) displacement.

                                                                                   Load tests
               120                                                                    Each test included a sequence of load tests as shown in
                                                                                   Table 2. The procedure for each type of load test presented
                                                                                   in this paper is as follows:

               140                                                                 (a) compression load test, where the pile was jacked
                                                                                       downwards at 0.005 mm/s for a distance of at least
Fig. 3. Cone resistance against depth                                                  1.5 mm

Table 2. Summary of test programme
Test number                                      T1           T2    T3        T4    T5     T6     T7      T8     T9     T10    T11    T12    T13*
Installation method
Monotonic installation
Jacked installation
Load testing
Compression load test at L/2
Compression load test at L
Tension test on extraction L
1-way cyclic compression test
2-way cyclic compression/tension test
1-way cyclic tension test
* Test halted after compression load test at 60 mm penetration.
650                                                 WHITE AND LEHANE
(b) cyclic compression load test, where the pile head load      installation on the density, stress history, and hence stiffness
    was cycled between a nominally zero pile head load of       of the soil beneath the pile tip.
    50 N and a maximum of 750 N at a velocity of                   It should be noted that the relative stiffness of the shaft
    0.01 mm/s                                                   and the base response of model piles does not mimic field
(c) cyclic compression-tension load test, where the pile        piles. Base response scales primarily with diameter, since it
    head load was cycled between –50 N and 750 N at a           is a continuum failure mechanism, whereas shaft response
    velocity of 0.01 mm/s.                                      does not scale, since it is governed partly by the shear
                                                                stress–displacement response of the interface and partly by
                                                                continuum downdrag of the soil surrounding the pile, as
                                                                required to generate this shear stress in the far field. As a
                                                                result, the settlement required to mobilise ultimate base
   Although the primary aim of this paper is to examine
                                                                resistance during model tests scales correctly as a fraction of
mechanisms controlling friction fatigue for displacement
                                                                pile diameter, typically D/5 or D/10. In contrast, the settle-
piles in sand, it is instructive to first examine trends indi-
                                                                ment to mobilise shaft resistance does not scale, and remains
cated by the pile head load measurements obtained during
                                                                similar to the absolute field values of a few millimetres,
pile installation and subsequent load testing.
                                                                rather than reducing in proportion to the model scale to a
                                                                fraction of a millimetre. Therefore, although the extraction
                                                                strokes (1.5 mm) represent a realistic rebound and reversal
Installation                                                    of shaft resistance for modelling dynamic installation, the
   The pile head loads recorded during each installation        base rebound is very high, at D/6. This is an unrealistic
mode are plotted in Fig. 4. The monotonic installation data     model to examine the effect of installation method on base
are collated in Fig. 4(a) and show excellent repeatability,     response, and therefore the resulting differences in base
with a maximum deviation between the four tests of Æ10%.        stiffness (and hence head stiffness) should not be considered
Although the pile was not equipped with a base load cell,       representative of the field case.
some estimate of the base resistance can be made by
subtracting the pullout load, which was typically 200 N, or
10% of the final installation resistance. Since the head load    HORIZONTAL STRESS DURING MONOTONIC
is dominated by base resistance, any difference between         INSTALLATION
compressive and tensile shaft capacity can be overlooked in        The six total stress cells provide measurements of hori-
an approximate estimate of base resistance. The mean final       zontal stress acting on the pile shaft at four different
installation force of 2100 N therefore indicates a mean unit    locations behind the pile tip (h/B ¼ 1, 3, 6 and 9). The
base resistance, qb , of about 23.5 MPa, which is slightly      measurements recorded during monotonic installation (test
below the mean CPT resistance at the same depth (Fig. 3).       T10) are shown in Fig. 6. These are seen to mirror the CPT
   The pile head loads recorded during jacked installation at   qc profile, apart from interruptions corresponding to a tip
the end of each 2 mm jacking stroke are shown in Fig. 4(b).     depth of 60 mm (when the pile head was unloaded prior to a
Between strokes, the pile was extracted until the head load     static compression load test) and the final tip depth of
reduced to a nominal zero value (of 50 N) to mimic the one-     120 mm (when the pile was unloaded prior to further load
way cycling at the pile head imposed by pile jacking. The       testing).
pseudo-dynamic installation data are shown in Fig. 4(c) and,       Stresses recorded when the pile is moving are referred to
although showing greater scatter than the other two installa-   as ó hm . All ó hm data recorded during the four monotonic
                                                                      9         9
tion methods, exhibit a deviation of less than 15% between      installations are normalised by the corresponding qc values
piles. It is clear that the 1.5 mm of extraction during each    and plotted against instrument depth in Fig. 7; the error bars
cycle mobilises significant negative shaft friction, particu-    shown in this figure indicate one standard deviation above
larly as the final embedment is approached. This two-way         and below the average value. It is evident that:
cycling of shaft friction mimics the conditions during dy-
namic pile installation in an approximate manner.               (a) The normalised horizontal stress, ó hm /qc , remains
   The maximum pile head loads for each installation mode           approximately constant throughout installation at
are very similar and, since the head load is dominated by           0.016, with possibly a very slight decrease with
base resistance, this close agreement indicates that the            increasing depth (or stress level). This value corre-
installation procedure has minimal influence on the ultimate         sponds to a friction ratio, ôf /qc , of 0.34%, using the
base resistance.                                                    constant-volume friction angle of 128 recorded during
                                                                    interface testing. This ratio is at the lower end of the
                                                                    range typically measured during CPTs in dense uniform
                                                                    sands (Lunne et al., 1997), reflecting the higher
Monotonic compression load tests
                                                                    roughness and therefore interface friction angle of a
   The pile head load–displacement data recorded for each
                                                                    CPT (e.g. Rcla ¼ 0.18–6.85 ìm; DeJong et al., 2001)
monotonic compression test conducted after installation to
                                                                    compared with the model pile. It may therefore be
120 mm are shown in Fig. 5 for the three installation meth-
                                                                    assumed that no gross error exists in the horizontal
ods employed. Although plunging failure is not reached
                                                                    stress measurements.
during these tests, all curves approach similar maximum
                                                                (b) ó hm /qc ratios are independent of the distance behind the
loads of 2100 Æ 300 N after 2.5–3 mm pile head displace-
                                                                    pile tip: that is, at a given soil horizon, approximately
ment. However, the initial stiffness of the load–displacement
                                                                    equal horizontal stress is recorded by each instrument
response varies with the installation method. Both the mono-
                                                                    as it passes that point.
tonic and jacked piles show an initial constant pile head
stiffness of about 2000 N/mm, whereas the pseudo-dynamic        The independence of ó hm /qc during monotonic installation
piles, which are load tested after ending installation with a   from the distance h is emphasised in Fig. 8, which plots the
1.5 mm extraction stage, show a lower initial stiffness of      ratio of the normalised horizontal stress recorded at h/B ¼ 1
1500 N/mm. The difference in stiffness of the monotonic         (i.e. the instrument level closest to the pile tip) to that for
and jacked piles compared with the pseudo-dynamic piles         each of the other three instrument levels. No systematic
must arise primarily from the influence of the final stage of     friction fatigue is evident between h/B ¼ 1 and 9. This
                                                     FRICTION FATIGUE ON DISPLACEMENT PILES IN SAND                                                                                  651
                                                 Pile-head load: N
                              0     500          1000         1500         2000          2500
                          0                                                                                                         Pile-head load at end of jacking stroke: N

                                                                                                                                0      500        1000         1500      2000     2500
                         40                                                                                                                                                T2
                                                                                  T10                                      40
   Pile-tip depth: mm

                         60                                                                                                                                                T1 1

                                                                                                     Pile-tip depth: mm


                        100                                                                                               100

                        120                                                                                               120


                                                        (a)                                                                                              (b)

                                        Pile-head load at each end of ‘blow’: N
                              500   0          500       1000        1500         2000      2500



   Pile-tip depth: mm






Fig. 4. Pile-head load against tip depth during installation: (a) monotonic installations; (b) jacked installations; (c) pseudo-dynamic

result confirms the observation shown in Fig. 1(a) through                                          incompatible with these observations. The observations con-
direct measurement of the horizontal stress acting on a pile.                                      trast with those of Klotz & Coop (2001) and Vesic (1970),
The proposition that friction fatigue is associated with relief                                    who show maximum ó hm values at between 5 and 10
from the highly stressed region close to the pile tip is                                           diameters behind the pile tip. Fellenius & Altaee (1995)
652                                                                                         WHITE AND LEHANE
                                                                 Pile-head load: N                                                             Horizontal stress,σ′hm: kPa
                                   0                  500        1000        1500    2000    2500
                                                                                                                                  0   100           200          300            400         500
  Pile-head settlement: mm

                                          T4                                                                                20
                             1·5                                                                                                                                         h/B        1

                                                                                                                                                                         h/B        3
                                             T10                                                                                                                         h/B        6
                                                                                                                            40                                           h/B        9


                                                                                                     Instrument depth: mm

                                                                 Pile-head load: N
                                   0                  500        1000        1500    2000    2500

  Pile-head settlement: mm

                                                            T2                                                              80
                             1·5                 T7
                                                 T1 1
                             2·0                                                                                            100


                                                                       (b)                                                  120

                                                                                                    Fig. 6. Horizontal stress measurement during monotonic instal-
                                                                 Pile-head load: N                  lation (test T10)
                                   0                  500        1000        1500    2000    2500

  Pile-head settlement: mm

                                                                                                                                       Normalised horizontal stress, σ′hm/qc

                             1·0                                                                                                  0     0·01              0·02               0·03           0·04
                                                 T3                                                                           0
                             2·0       T9

                                                                                                    Instrument depth: mm

Fig. 5. Pile-head load–settlement during compression load tests:
(a) monotonic installations; (b) jacked installations; (c) pseudo-
dynamic installations

suggest that residual loads, ignored during the interpretation,
may have led Vesic to falsely conclude that the maximum
value of unit shaft resistance occurs some distance above the
pile base, although Klotz & Coop’s observations cannot be
explained in this way.
                                                                                                                                                                              h/B       1

HORIZONTAL STRESS DURING CYCLIC                                                                                             100                                               h/B       3
INSTALLATION                                                                                                                                                                  h/B       6
   The jack stroke length was insufficient to mobilise full
friction during the pseudo-dynamic and jacked cyclic instal-
lation methods. This became evident from the results of                                                                     120
static compression tests, which showed that lateral stresses
and pile head loads continued to increase until the pile head                                       Fig. 7. Normalised horizontal stress during monotonic installa-
displacement reached between 5 and 8 mm. The ó hm data9                                             tion (mean of all four tests, error bars 1 std dev.)
                                                                 FRICTION FATIGUE ON DISPLACEMENT PILES IN SAND                                                                                                                              653
                                             Horizontal stress reduction: σ′hm, h     nB /σ′
                                                                                           hm, h B                                      stress, ó hc , recorded during each installation cycle. For
                                 0                 0·5              1·0                                    1·5                 2·0      jacked installation, this corresponds to the value acting when
                            0                                                                                                           the pile is unloaded to nominally zero head load (actually
                                                                                                                                        50 N). For pseudo-dynamic installation, this is the minimum
                                                                                                                                        value recorded during each cycle, and occurs close to the
                                                                                                                                        moment of zero pile head load. The two unloading stages at
                                                                                                                                        pile tip penetrations of 60 and 120 mm provide two meas-
                                                                                                                                        urements of ó hc for the monotonic installation method.
                                                                                                                                           The profiles of ó hc with depth for each installation
                                                                                                                                        method are shown in Fig. 9, grouped by instrument position
                            30                                                                                                          and averaged over four tests for each method. Friction
                                                                                                                                        fatigue is clearly evident in these data: that is, at a given
                                                                                                                                        depth, ó hc decreases as each instrument passes. The reduc-
                            40                                                                                                          tion in ó hc due to two-way cycling during installation is
                                                                                                                                        demonstrated by the very low values recorded on the
   Depth: mm

                                                                                                                                        pseudo-dynamic piles compared with the monotonic installa-
                            50                                                                                                          tions. The upper half of the pseudo-dynamically installed
                                                                                                                                        piles has typically only 10% of the stationary horizontal
                                                                                                                                        stress of the monotonic case. The one-way cycling induced
                            60                                                                                                          by jacking leads to ó hc values that lie between those devel-
                                                                                                                                        oped during pseudodynamic and monotonic installation.
                                                                                                                 n    3                    The reduction in ó hc at a given depth with increasing
                                                                                                                                        distance from the pile tip (h) is quantified in Fig. 10. The
                                                                                                                 n    6                 measured values of ó hc have been normalised by qc and
                            80                                                                                                          averaged over the entire installation (the error bars show Æ1
                                                                                                                 n    9                 standard deviation). Two methods of presenting these nor-
                                                                                                                                        malised data have been used. In Fig. 10(a), the decay in
                            90                                                                                                          ó hc /qc behind the pile tip is plotted against the distance h.
                                                                                                                                        The high lateral stress close to the pile tip matches the
                                                                                                                                        trends observed in the field (Lehane, 1992; Chow, 1997) but,
                        100                                                                                                             evidently, the data from the two installation methods do not
                                                                                                                                        overlie each other. As seen in Fig. 10(b), by plotting ó hc /qc
Fig. 8. Reduction in horizontal stress between instruments                                                                              against number of cycles at that soil horizon, a better
during monotonic installation (mean of all four tests, error                                                                            agreement between the two sets of data is found. In a given
bars 1 std dev.)
                                                                                                                                        soil horizon, the number of cycles (N) experienced at each
                                                                                                                                        lateral stress sensor position is simply 2h for the pseudo-
                                                                                                                                        dynamic installation since the net penetration is 0.5 mm per
recorded during these installation methods cannot, therefore,                                                                           cycle. For jacked installation, the value of N required to
be compared directly with the ó hm measurements recorded
                                    9                                                                                                   bring a given sensor to a specific soil horizon increases
during monotonic installation (i.e. those in Figs 6 and 7).                                                                             slightly with depth since the rebound during unloading in-
Instead, the influence of installation method on shaft friction                                                                          creases as the pile head load increases. The mean set per
can be examined by considering the stationary horizontal                                                                                jacking cycle is approximately 1.5 mm, and error bars show-

                                     Stationary horizontal stress, σ′hc: kPa                                         Stationary horizontal stress, σ′ : kPa
                                                                                                                                                                                                    Stationary horizontal stress, σ′ : kPa

                            0        20    40    60      80   100 120 140 160                                   0         10    20     30     40      50       60                               0       2         4          6          8     10
                       0                                                                                    0                                                                              0

                                                               Jacked                                                                        Jacked                                                                         Jacked
                                                               Pseudo-                                                                       Pseudo-                                                                        Pseudo-
                       20                                      dynamic                                     20                                dynamic                                       20                               dynamic
                                                               Monotonic                                                                     Monotonic                                                                      Monotonic

                       40                                                                                  40                                                                              40
Instrument depth: mm

                                                                                                                                                                    Instrument depth: mm
                                                                                    Instrument depth: mm

                       60                                                                                  60                                                                              60

                       80                                                                                  80                                                                              80

                 100                                                                                 100                                                                             100

                                                                                                     120                                                                             120
                                                      (a)                                                                             (b)                                                                             (c)

Fig. 9. Variation of stationary horizontal stress with installation method: (a) h/B                                                                           1; (b) h/B                            3; (c) h/B        6
654                                                                                                WHITE AND LEHANE
                                             70                                                                                       140

                                                                                      Jacked                                                                                  Jacked
                                             60                                                                                       120
                                                  N      108                          Pseudo-dynamic                                                                          Pseudo-dynamic
                                                               Nmean    35
                                                                                                                                                  h/B    6
                                             50                                                                                       100
            Distance above pile tip, h: mm

                                                                                                                Number of cycles, N
                                             40                                                                                       80

                                             30 N        54                                                                           60
                                                                       Nmean   20                                                                 h/B    3

                                                                                                                                                   h/B   6
                                             20                                                                                       40

                                                                                      Nmean    6                                                         h/B   3
                                             10 N        18                                                                           20
                                                                                                                                                   h/B   1
                                                                                                                                                                           h/B   1
                                                      N: Number of cycles during installation
                                                      Mean value shown for jacked installation
                                             0                                                                                         0
                                                  0          0·002         0·004         0·006         0·008                                0          0·002        0·004        0·006          0·008
                                                       Normalised stationary horizontal stress, σ′ /qc
                                                                                                                                                Normalised stationary horizontal stress, σ′ /qc

                                                                                (a)                                                                                     (b)

           Fig. 10. Influence of loading cycles during installation on stationary horizontal stress: (a) normalised
           horizontal stress against distance above pile tip; (b) normalised horizontal stress against number of cycles

ing the variation throughout installation are shown in Fig.                                                    of the loading type imposed during installation. All four
10(b).                                                                                                         piles were first subjected to one cycle of loading in the form
   A comparison of Fig. 10(a) with Fig. 10(b) indicates that                                                   of a static compression test to a pile settlement of 1.5 mm
the variation of ó hc along the pile shaft is better related to N
                   9                                                                                           before initiation of the scheduled component of the cyclic
than to h (or h/B). A direct comparison is possible between                                                    tests.
the instrument located at h/B ¼ 3 during jacked installation                                                      The variation of horizontal stresses at h/B ¼ 1 during
and h/B ¼ 1 during pseudo-dynamic installation (Fig. 10(b)).                                                   selected cycles throughout tests T2 and T8 is presented in
These instruments encounter 18 and 20 cycles respectively,                                                     Figs 11 and 12. The increase in stress during loading
and register approximately equal normalised stationary hor-                                                    followed by a sharp reduction after changing direction is a
izontal stresses. In this case, although jacked installation                                                   characteristic of interface shear under conditions of constant
involves one-way cycling of the pile head load and pseudo-                                                     normal stiffness or constrained dilation. Comparable patterns
dynamic installation involves two-way cycling, the degrada-                                                    of normal stress–shear displacement are widely observed in
tion after around 20 cycles is comparable. It is shown later,                                                  interface shear box testing under constant normal stiffness
from cyclic load tests, that over a greater number of cycles,                                                  (CNS) conditions (e.g. Airey et al., 1992; Fakharian &
two-way head loading leads to greater degradation. It is also                                                  Evgin, 1997; DeJong et al., 2003).
observed that one-way loading of the pile head leads to a                                                         Under one-way pile head loading there is some evidence
degree of two-way loading along the pile shaft due to                                                          of two-way shear stress cycling at h/B ¼ 1 since the mini-
rebound.                                                                                                       mum horizontal stress does not coincide with the minimum
                                                                                                               pile head load. For interpretation of these cyclic load test
                                                                                                               data, ó hc is defined as the minimum value within each cycle.
HORIZONTAL STRESS DURING CYCLIC LOAD                                                                           This definition differs slightly from that used by Lehane
TESTING                                                                                                        (1992) and Chow (1997), who measure ó hc under zero head
   The effects on shaft friction of cyclic loading during                                                      load, and is arguably more fundamental since it is the
installation and a cyclic ‘working load’ may be compared by                                                    minimum value of ó hc within a particular loading cycle,
examining the horizontal stress measurements during the                                                        rather than the value under an arbitrary amount of residual
cyclic load tests. Four tests will be highlighted (see Table                                                   shear stress. The reduction in ó hc due to cycling is clearly
2):                                                                                                            evident in both Fig. 11 and Fig. 12. In the case of one-way
                                                                                                               cycling (of the pile head load), a limiting value of ó hc is
(a) the cyclic compression test following jacked installation
                                                                                                               reached after 30 cycles whereas, under more arduous two-
    of pile T2
                                                                                                               way cycling, a progressive reduction towards zero is re-
(b) the cyclic compression–tension test following pseudo-
    dynamic installation of pile T8
                                                                                                                  The progressive reductions of ó hc observed at h/B ¼ 1
(c) the cyclic compression test following monotonic
                                                                                                               throughout the cyclic load tests are plotted in Fig. 13. It
    installation of pile T1
                                                                                                               should be noted that if ó hc was defined during the cyclic
(d) the cyclic compression–tension test following mono-
                                                                                                               load tests as the value under zero head load, a lower
    tonic installation of pile T4.
                                                                                                               degradation would be evident. Figs 11 and 12 show a small
The cyclic tests on piles T2 and T8 represent an extension                                                     but increasing discrepancy between ó hc at Phead ¼ 0 and the
                                                                                   FRICTION FATIGUE ON DISPLACEMENT PILES IN SAND                                                               655
                                                                                                                                                   Phead          750 N
                                                                                                                                           End of cycle
                                                                          140                                              Cycle 1         Phead 50 N
                                                                                                                                               Start of cycle
                                                                                                                                               Phead 50 N
                                                                                                                             Cycle 5
                                            Horizontal stress, σ′h: kPa

                                                                          100                                                 Cycle 30              Cycle 55
                                                                                                                                                                      Cycle 100





                                                                               119·6   119·7      119·8    119·9   120·0     120·1      120·2       120·3         120·4     120·5    120·6
                                                                                                                     Pile tip depth: mm

                                     Fig. 11. Horizontal stress degradation during one-way cyclic compression load test (test T2,
                                     jacked installation, h/B    1)

                                                                                                                                         Phead       50 N                  Phead    750 N

                                                                                                                                                   Start of cycle
                                                           140                                                                                     Phead 0 N


                                                                                                                                           σ′hc          End of cycle
                  Horizontal stress, σ′ : kPa

                                                           100                                 Cycle 1                                                   Phead 0 N
                                                                                                                           Cycle 5

                                                                                                                                                  Cycle 30
                                                                                                                                                                               Cycle 100




                                                                           122·0          122·2           122·4      122·6             122·8           123·0              123·2         123·4
                                                                                                                      Pile tip depth: mm

                 Fig. 12. Horizontal stress degradation during two-way cyclic compression-tension load test (Test
                 T8, pseudo-dynamic installation, h/B    1)

minimum value. Regardless of the chosen definition there is                                                                     Hence the starting offsets of 11 and 18 cycles adopted for
a reducing trend in ó hc with number of cycles.
                       9                                                                                                       the jacked and pseudo-dynamically installed piles (T2 and
   The one-way cyclic load tests on the monotonic and                                                                          T8) enable direct comparisons with the monotonic base
jacked installations (T1 and T2) are compared in Fig. 13(a),                                                                   cases (T1 and T4). This representation leads to closely
and the two-way cyclic load tests on the monotonic and                                                                         comparable ó hc variations with N for tests T1 and T2 (in
pseudo-dynamic installations (T4 and T8) are compared in                                                                       Fig. 13(a)) and for tests T4 and T8 (in Fig. 13(b)), suggest-
Fig. 13(b). The number of cycles includes those experienced                                                                    ing that the cycling is the only mechanism leading to
during installation (and the extra cycle due to the subsequent                                                                 degradation of ó hc .
static compression tests) in addition to the load test cycles.                                                                   It is apparent in Fig. 13(a) that the relatively high values
656                                                                                 WHITE AND LEHANE
                                  Normalised stationary horizontal stress, σ′ /qc
                                                                            hc                                                   Normalised stationary horizontal stress, σ′ /qc
                             0        0·002            0·004            0·006           0·008                               0        0·002           0·004            0·006            0·008
                       140                                                                                            140
                                                                                                                                T4 (monotonic) two-way load test (σ′hc)
                                       T1 (monotonic) one-way load test (σ′ )

                                       T1 (jacked) one-way load test (σ′hc) (Fig. 11)                                           T8 (Pseudo-dynamic) two-way load test (σ′hc) (Fig. 12)

                       120             Jacked installation (σ′hc/qc, Fig. 10)                                         120       Pseudo-dynamic installation (σ′hc/qc, Fig. 10)

                       100                                                                                            100
 Number of cycles, N

                                                                                                Number of cycles, N
                       80                                                                                             80

                       60                                                                                             60

                       40                                                                                             40

                       20                                                                                             20

                        0                                                                                              0
                             0   25     50      75      100      125     150    175     200                                 0   25     50      75     100      125     150       175   200

                                      Stationary horizontal stress, σ′hc: kPa                                                        Stationary horizontal stress, σ′hc: kPa
                                                        (a)                                                                                           (b)

Fig. 13. Degradation of stationary horizontal stress with cycling at h/B                                               1 during load tests: (a) one-way compression load test; (b)
two-way compression-tension load test

of ó hc that exist at h/B ¼ 1 after a low number of cycles
      9                                                                                         the number of cycles and their amplitude on friction fatigue.
reduce to a constant value of about 50 kPa after 30 one-way                                     Such sensitivity was also observed by Kelly (2001), who
cycles. A similar degradation of ó hc with number of cycles
                                      9                                                         showed in CNS interface shear tests that larger-amplitude
is apparent in Fig. 13(b) for two-way cycling, and ó hc      9                                  cycles lead to a higher rate of degradation.
continues to reduce beyond 30 cycles to a value close to
zero. This reduction in ó hc does not, however, mean that the
available shaft friction is negligible. For example, it may be                                  DISCUSSION
inferred from Fig. 12 that dilation at the interface can lead                                   Implications for design
to an increase in normal stress at ultimate conditions to well                                     The key observation from this investigation is that friction
in excess of 100 kPa. This surprising recovery of normal                                        fatigue arises from cycles of loading. During continuous
stress is also observed in CNS interface shear testing (e.g.                                    penetration there is no reduction in horizontal stress at a
Airey et al., 1992; Shahrour et al., 1999).                                                     given soil horizon as the pile penetrates deeper: that is,
   The agreement seen in Fig. 13 between the degradation of                                     lateral stresses are independent of distance, h. Friction
ó hc during each cyclic load test, when modified to account
  9                                                                                             fatigue cannot therefore be attributed to the departure of a
for the cycles induced during installation, supports the link                                   zone of high stress around the pile tip as the pile penetrates
between friction distribution and cyclic history. To examine                                    deeper. Instead, it has been found that the reduction in
this link more closely, in addition to the two cyclic load test                                 stationary horizontal stress at a given soil horizon is better
curves shown in each of Figs 13(a) and 13(b), the normal-                                       linked to the cyclic history. On-pile measurements of hor-
ised stationary horizontal stresses, ó hc /qc , recorded at
                                            9                                                   izontal stress have shown trends of behaviour that agree with
h/B ¼ 1, 3 and 6 during jacked and pseudo-dynamic instal-                                       observations during constant normal stiffness (CNS) cyclic
lation (i.e. as shown in Fig. 10) are also plotted in Figs                                      interface shear testing (e.g. Tabucanon, 1997; Kelly, 2001;
13(a) and 13(b). There is reasonable agreement between the                                      DeJong et al., 2003). A mechanism of net contraction with
installation data and the cyclic load test data, supporting the                                 cycling of a thin interface layer that is confined by the far
link between cycling and friction fatigue, which is indepen-                                    field soil explains the behaviour observed in both cases.
dent of the distance h. For the pseudo-dynamic installation,                                       These results suggest that, for design, an appropriate non-
in which the cyclic load test amplitude (Fig. 12) is compar-                                    dimensional quantity to govern the reduction in horizontal
able to the amplitude of the installation cycles, the agree-                                    effective stress from an initial ‘unfatigued’ value behind the
ment is better than that seen for jacked installation, possibly                                 pile tip with continued pile penetration is number of cycles,
because the jacked installation procedure involves larger                                       N, rather than h/D. This observation is supported by the
cycles of displacement than that induced in the one-way                                         similar relationship between ó hc /qc and N during both
cyclic tests. This discrepancy highlights the influence of both                                  installation and load testing. However, the centrifuge tests
                              FRICTION FATIGUE ON DISPLACEMENT PILES IN SAND                                                                 657
support inferences made from interface shear tests by de-                                           qc: MPa            Jacking force: kN
monstrating that the rate of degradation depends not only on                                    0     20          40   0      2500         5000
N, but also on the mode and amplitude of cycling. Two-way                                   0
cycling leads to a greater degradation than one-way cycling
during both installation and load testing.                                                           Predrilled
   Although this investigation has demonstrated that friction                               1        for CPT
fatigue is not linked per se to h or to the normalised
distance, h/D, some dependence on diameter may remain,
since the confining stiffness, which governs the reduction in                                2
ó hc for a given contraction of the interface layer, depends on
  9                                                                    Fine to
4G/D (G being the operational shear stiffness of the soil              medium
surrounding the installed pile).                                                            3
   The relative success of the Jardine & Chow (1996) design
method, which relates friction fatigue to h/D, is likely to be
primarily because the chosen power law for the h/R effect                                   4
reflects the degradation in friction caused by contraction of

                                                                                 Depth: m
the interface during the number of installation cycles of
typical driven piles (forming the database used to validate                                 5
Jardine & Chow’s design method), under the corresponding
confining stiffness. The examples highlighted in Fig. 1 warn
against extrapolation beyond this range. The sharp reduc-
tions in ó hc with cycling seen during this investigation may
be attenuated by the lower confining stiffness (/ 1/D)
around field piles; for a given interface contraction a lower
drop in ó h would result. However, the mechanism of behav-
           9                                                                                7

iour is scale-independent, and cycling would still lead to
friction fatigue.
   Finally, it is noted that this paper does not address the                                8
influence of cycling, and of ó hc reduction, on the subsequent
increase in horizontal effective stress observed during static
loading to failure. Any increase in ó h during loading
                                             9                                              9
contributes an additional component of normal stress at
failure and hence increases ôf . Therefore any reduction in       Fig. 14. Jacking record for precast concrete pile in sand
ó hc may not cause a proportional reduction in ôf . However,
CNS interface shear box tests, which can be considered
analogous to elements of a pile–soil interface, show that any     pleting installation with far fewer cycles than dynamic
cyclic contraction and loss of normal stress is not fully         methods. Furthermore, optimisation of dynamic methods can
recovered during subsequent shearing to failure (Ghionna et       reduce the total blowcount during installation. It should be
al., 2004), so any drop in ó hc will have an influence on ôf .
                             9                                    noted, however, that any additional friction on a jacked pile
                                                                  may degrade more quickly under a cyclic working load than
                                                                  on a ‘pre-degraded’ dynamically installed pile.
Implications for construction                                        Considerable further research is needed in order to capture
   This investigation has demonstrated that the stationary        this behaviour in a prediction method. However, since site-
horizontal stress acting on a pile shaft is strongly influenced    specific pile load testing is usually required for major
by the number of installation cycles. Although some recov-        projects, the value of these findings lies equally in the
ery in lateral stress is possible during subsequent loading       suggestion that optimisation of the pile installation process
(e.g. Figs 11 and 12), significant load cycles reduce the          can yield higher shaft friction.
available pile shaft friction. This behaviour is illustrated in
Fig. 14, which plots the jacking record of a 350 mm square
precast concrete pile installed using a jacking machine of        CONCLUSIONS
the type described by Lehane et al. (2003). The site con-            Drum centrifuge tests have been conducted to examine
sisted of medium dense sand with a CPT qc value of                more closely the distribution of horizontal stress acting on
20 MPa between depths of 3 and 9 m. For experimental              the pile shaft during installation and subsequent cyclic
purposes, the contractor jacked the pile in 1.5 m increments      loading. During monotonic installation no friction fatigue
to a depth of 6.3 m and then completed the installation to a      was recorded, which is in agreement with field data from
final pile tip depth of 8 m using 22 no. 75 mm jacking             multi-sleeve CPT installation. The variation of available
increments. It can be assumed that the base resistance            shaft friction with depth followed the CPT profile, indicating
(following the qc profile) was unchanged between 6.3 m and         that the normalisation of lateral stresses by qc provides a
8 m, and therefore that the mobilisation of a constant jacking    useful basis for design.
resistance over this depth interval, as seen in Fig. 14, was         Cyclic installation methods have been seen to cause sig-
due to friction fatigue induced by cyclic loading. Clearly, in    nificant degradation of shaft friction. One-way and two-way
this example the additional time spent during installation        installation methods lead to differing profiles of stationary
from 6.3 m to 8 m and the additional cost of pile length was                                  9
                                                                  lateral effective stress (ó hc ), when plotted against the relative
entirely negated by the additional loading cycles.                position of the pile tip, h. The centrifuge cyclic load tests
   This link between the cycles during installation and the       also showed that, for a given installation method, the degra-
shaft friction during first loading suggests that opportunities    dation of ó hc during cyclic loading followed the same decay
may exist for improving pile capacity, and increasing design      pattern as that during installation. These measurements agree
efficiency, if the loading cycles during installation are mini-    with trends indicated in CNS interface shear tests. A
mised. Pile-jacking machines offer the possibility of com-        mechanism of net contraction with cycling of a thin inter-
658                                                         WHITE AND LEHANE
face layer that is confined by the far field soil explains the                came into being and why it does not exist. Proc. Inst. Civ.
behaviour observed in both cases, and provides a rational                   Engrs Geotech. Engng 113, No. 2, 107–111.
basis for improved design.                                               Fugro (1996). EURIPIDES database report, Vols 1–5. Leidschen-
   Based on these centrifuge test data, and corroboratory                   dam, The Netherlands: Fugro BV.
evidence from field-scale test results, it is concluded that the          Ghionna, V. H., Mortara, G. & Vita, G. P. (2004). Sand-structure
                                                                            interface behaviour under cyclic loading from constant normal
degradation of available shaft friction at a given soil horizon             stiffness direct shear tests. Proceedings of the international
during installation and subsequent cyclic loading is better                 symposium on deformation characteristics of geomaterials,
characterised by the number of cycles experienced at that                   Lyon, pp. 231–238.
point, than by the non-dimensional distance from the pile                Heerema, E. P. (1980). Predicting pile driveability: heather as
tip, h/D. This conclusion has implications for construction,                an illustration of the friction fatigue theory. Ground Engng 13,
as well as for design. Modern installation techniques of pile               15–37.
jacking involve reduced cycling, and may therefore yield                 Jardine, R. J. & Chow, F. C. (1996). New design methods for
higher shaft friction than conventional dynamic installation                offshore piles, MTD96/103. London: Marine Technology Direc-
methods.                                                                    torate.
                                                                         Kelly, R. (2001). Development of a large diameter ring shear
                                                                            apparatus and its use for interface testing. PhD thesis, Univer-
                                                                            sity of Sydney.
ACKNOWLEDGEMENTS                                                         Klotz, E. U. & Coop, M. R. (2001). An investigation of the effect
   The support provided by the Australian Research Council                                                                             ´
                                                                            of soil state on the capacity of driven piles in sands. Geotech-
(ARC) for this research project is gratefully acknowledged.                 nique 51, No. 9, 733–751.
                                                                         Lehane, B. M. (1992). Experimental investigations of pile behaviour
The ARC Centre for Offshore Foundation Systems at UWA
                                                                            using instrumented field piles. PhD thesis, Imperial College,
also provided support. The authors also acknowledge the                     University of London.
excellent technical assistance provided by Mr Bart Thomp-                Lehane, B. M. & Jardine, R. J. (1994). Shaft capacity of driven
son and the staff of the UWA civil engineering workshop.                    piles in sand: a new design method. Proc. 7th Int. Conf. on the
We also thank Dr Peter Mitchell (formerly of PPK Interna-                   Behaviour of Offshore Structures, Boston 1, 23–36.
tional) for his permission to present the data shown in Fig.             Lehane, B. M., Jardine, R. J, Bond, A. J. & Frank, R. (1993).
14.                                                                         Mechanisms of shaft friction in sand from instrumented pile
                                                                            tests. ASCE J. Geotech. Engng 119, No. GT1, 19–35.
                                                                         Lehane, B. M., Pennington, D. & Clark, S. (2003). Jacked end-
                                                                            bearing piles in the soft alluvial sediments of Perth. Aust.
REFERENCES                                                                  Geomech. J. 38, No. 3, 123–134.
Airey, D. W., Al-Douri, R. & Poulos, H. G. (1992). Estimation of         Lunne, T. & Christofferson, H. P. (1983). Cone penetrometer
    pile friction degradation from shearbox tests. ASTM Geotech.            interpretation for offshore sands. Proc. Offshore Technology
    Test. J. 15, No. 4, 388–392.                                            Conf., Houston, OTC4464, 181–192.
API (2000). RP2A: Recommended practice of planning, designing            Lunne, T., Robertson, P. K. & Powell, J. J. M. (1997). Cone
    and constructing fixed offshore platforms: Working stress design,        penetration testing in geotechnical practice. London: Blackie
    21st edn, pp. 59–61. Washington: American Petroleum Institute.          Academic & Professional.
Bustamante, M. & Gianeselli, L. (1982). Pile bearing capacity            Poulos, H. G., Carter, J. P. & Small, J. C. (2001). Foundations and
    prediction by means of static penetrometer CPT. Proc. 2nd Eur.          retaining structures: research and practice. Proc. 15th Int. Conf.
    Symp. on Penetration Testing, Amsterdam, 1, 493–500.                    Soil Mech. Found. Engng, Istanbul 4, 2527–2606.
Chow, F. C. (1997). Investigations into the behaviour of displace-       Randolph, M. F., Dolwin, J. & Beck, R. (1994). Design of driven
    ment piles for offshore foundations. PhD thesis, Imperial Col-                           ´
                                                                            piles in sand. Geotechnique 44, No. 3, 427–448.
    lege, University of London.                                          Randolph, M. F., Leong, E. C. & Houlsby, G. T. (1991). One-
DeJong, J. T. (2001). Investigation of particulate-continuum inter-                                                               ´
                                                                            dimensional analysis of soil plugs in pipe piles. Geotechnique
    face mechanisms and their assessment through a multi-friction           41, No. 4, 587–598
    sleeve penetrometer attachment. PhD thesis, Georgia Institute of     Shahrour, I., Rezaie, F. & Nauroy, J.-F. (1999). Experimental study
    Technology, Atlanta, GA.                                                of the behaviour of calcareous sand: structure interface. Proc.
DeJong, J. T. & Frost J. D. (2002). A multi-friction sleeve attach-         2nd Int. Conf. on Engineering for Calcareous Sediments,
    ment for the cone penetrometer. ASTM Geotech. Test. J. 25, No.          Bahrain, 69–77.
    2, 111–127.                                                          Stewart, D. P., Boyle, R. S. & Randolph, M. F. (1998). Experience
DeJong, J. T., Frost, J. D. & Cargill, P. E. (2001). Effect of surface      with a new drum centrifuge. Proc. Int. Conf. Centrifuge ’98,
    texturing on CPT friction sleeve measurements. ASCE J. Geo-             Tokyo, 1, 35–40.
    tech. Geoenviron. Engng 127, No. 2, 158–168.                         Tabucanon, J. T. (1997). Shaft resistance of piles in sand. PhD
DeJong, J. T., Randolph, M. F. & White, D. J. (2003). Interface load        dissertation, University of Sydney, Australia.
    transfer degradation during cyclic loading: a microscale investi-    Vesic, A. S. (1970). Tests on instrumented piles, Ogeechee River
    gation. Soils Found. 43, No. 4, 81–93.                                  site. J. Soil Mech. Found. Div. ASCE 96, No. SM2, 561–584
Fakharian, K. & Evgin, E. (1997). Cyclic simple-shear behaviour of       White, D. J., Finlay, T. C. R., Bolton, M. D. & Bearss, G. (2002).
    sand-steel interfaces under constant normal stiffness condition.        Press-in piling: ground vibration and noise during pile installa-
    ASCE J. Geotech. Geoenviron. Engng 123, No. 12, 1096–1105.              tion. Proc. Int. Deep Foundations Congress, Orlando, ASCE
Fellenius, B. H. & Altaee, A. A. (1995). Critical depth: how it             Special Publication 116, 363–371.

To top