UPLIFT CAPACITY OF PILES 6.6.1 Piles in Soil Some published test results (e.g.

Radhakrishnan & Adams, 1973; Broms & Silberman, 1964; O'Neill, 2001) indicate
that the uplift resistance in the pile shaft is less than the corresponding shaft
resistance in compression, possibly by up to 50% less in a granular soil. O'Neill
(2001) suggested that this may be due to the influence of the reduction in vertical
effective stress in the ground and Poisson's ratio effect under tension loading.
Kulhawy (1991) examined the pile test data for bored piles and found no discernible
difference 118 between shaft resistance in uplift and compression. While both
loading cases develop shaft resistance along a cylindrical shear surface, a breakout
of soil cone may occasionally develop in the uplift loading cases. Fellenius (1989) &
Fleming et al (1992) considered that the interpretation of many pile loading tests
took insufficient account of the residual stresses, which existed after pile
installation. Consequently the end-bearing capacity of the pile was under-estimated
and the shaft resistance over-estimated. They suggested that there is no systematic
difference in the shaft resistance that may be mobilised by an unstressed pile
loaded either in tension or compression. Premchitt et al (1988) observed that the
pattern of residual stresses developed after pile driving was complex and erratic.
Therefore, it is difficult to generalise for design purposes. It was noted by Premchitt
et al that the residual shaft resistance and end-bearing resistance locked in after
pile driving were not associated with well-defined displacements or an applied
loading. Furthermore, the consideration of the shaft resistance associated with the
applied loading in a loading test (i.e. zeroing the instrumentation immediately prior
to a loading test) represents the condition of actual working piles supporting
superstructure loads. With driven piles, a number of researchers have also
emphasized the importance of the dependence of radial horizontal stresses and
shaft resistance on the relative position of the pile tip as the pile is advanced, based
on observations made in instrumented piles (e.g. Lehane, 1992; Lehane et al, 1993,
Jardine et al, 1998). Nicola & Randolph (1993) suggested that the ratio of uplift
resistance and compression can be determined based on the relative
compressibility and Poisson's ratio of the pile. The ratio typically ranges between
0.7 and 0.9 for piles installed in medium dense to dense sand. For design purposes,
it is recommended that the shaft resistance of bored piles under tension may be
calculated in the same way as for shaft resistance for compression piles (Sections
6.4.4.3 & 6.4.4.5). For driven piles, in view of the uncertainties associated with the
distribution of residual stresses after driving and the available capacity having
already been partially mobilised, it is recommended that the shaft resistance under
tension be taken conservatively as 75% of that under compression (Sections 6.4.4.4
& 6.4.4.6), unless higher values can be justified by a sufficient number of loading
tests. For relatively slender piles, such as mini-piles, contraction in the shaft under
tension load may become significant. This leads to the reduction of radial stress and
shaft resistance on the pile. Fleming et al (1992) estimated that this reduction may
amount to 10% to 20%. Any possible suction effects that may develop at the base
of a pile should be disregarded for prudence as this may not be reliable. The
working load under tension loading, Qwt is given by the following : Qs Qwt = + Wp'
[6.9] Fs where Qs = ultimate shaft resistance under tension Fs = factor of safety
Wp' = effective self weight of the pile 119 It is recommended that a minimum factor
of safety of 2.0 to 3.0 (Table 6.1) should be provided on the ultimate shaft

resistance in tension. For piles with an enlarged base, Dickin & Leung (1990)
reviewed existing design methods and investigated the uplift behaviour of such
piles embedded in sand using a centrifuge (Figure 6.13). For dense sand, they found
reasonable agreement with earlier research on anchor plates and published field
data. It was concluded that the best prediction for pile capacity in dense sand when
compared with the centrifuge test results is that given by Vermeer & Sutjiadi (1985).
For loose sand, the existing methods appear to over-predict the ultimate resistance
to uplift with the exception of the simple vertical slip surface model proposed by
Majer (1955). In the absence of relevant field data from instrumented piles, it is
suggested that the above recommendations may be adopted for preliminary design.
However, the design methods are based on model test results with embedded
lengths less than seven times the pile diameter. The design should be confirmed by
a pull-out test. Due consideration should be given to the difficulty in enlarging the
base of a bored pile in soil to form a bell-out section. The uplift resistance also
depends on the integrity of the bell-out section under tension. The possibility of
breaking off of the bell-out section along the pile shaft should be considered. 6.6.2
Rock Sockets Kulhawy & Carter (1992b) observed that there is no significant
difference in shaft resistance between piles under tension and compression,
provided that the piles are relatively rigid when compared to the rock mass. They
defined a rigidity factor as Ec/Em (Ds/Ls) 2 , in which Ec and Em is the Young's
modulus of the concrete in pile shaft and the rock mass respectively, Ds is the pile
diameter and Ls is the pile embedment length in rock. A pile is considered as rigid if
the rigidity factor is greater than 4. In case where this is less than 4, the shaft
resistance developed in a rock socket under tension should be taken as 0.7 of the
shaft resistance in compression. The pile data presented in Figure 6.12 include
bored piles socketed into rock, which were subject to tension and compression loads
in successive loading stages. The results also indicated that there is no significant
difference between shaft resistances mobilised in either tension or compression
loads. The rigidity factor of the test piles are generally greater than 4. For designing
rock-socketed piles to in resisting uplift load, the correlation given in Figure 6.12 can
be used to estimate the shaft resistance, provided that the rigidity factor is greater
than 4. Otherwise, a reduction of 30% of the shaft resistance in compression should
be assumed, unless a higher value is justified by loading tests. The cone failure
mode of a rock mass is normally the governing criterion under pull out. The actual
shape of the mass of rock lifted depends on the degree of jointing, fissuring and the
inclination of the bedding planes of the rock. For a heavily jointed or shattered rock,
a cone with a half angle of 30 will give a conservative estimate for the pull-out
resistance (Tomlinson, 1994). Shear at the interface between the cone surface and
the surrounding rock should be neglected. For rock mass with steeply inclined joint
sets, the weight of the rock cone should be conservatively assessed. 120 Ds L Ds Db
Db L (a) For Pile in Loose Sand (Majer, 1955) (b) For Pile in Dense Sand (Vermeer
& Sutjiadi (1985) L L Breakout factor, Nu = 1 + 2 Ks tan ' Breakout factor, Nu = 1
+ 2 tan ' cos ' D cv b Be where equivalent width of bell, where Ks = coefficient of
earth pressure Db = diameter of base Be = Ds = diameter of shaft 'cv = critical
state angle of shearing ' = angle of shearing resistance resistance of soil of soil
= angle of dilation of soil The ultimate shaft resistance for a belled pile in tension is
given by : Qs = Nu Ab 's L where Ab = area of pile base L = embedment length of

pile 's = effective unit weight of soil Figure 6.13 Failure Mechanisms for Belled
Piles in Granular Soils Subject to Uplift Loading (Dickin & Leung, 1990) Bonding at
the base of the socket will be governed by the tensile strength of the weaker of the
rock or concrete. However, given the potential construction problems due to
difficulties in achieving proper base cleanliness, possible intermixing of tremie
concrete and water and bentonite, etc, it is suggested that this should be
conservatively ignored in design. Rock anchors are sometimes provided for tension
piles to increase their uplift capacity. The uplift resistance of the rock anchors
depends on the permissible stress in the anchor, bond strength between the anchor,
the grout, and the rock, and the weight of rock mass and overlying soil lifted by the
anchor or a group of anchors (Tomlinson, 1994). 6.6.3 Cyclic Loading Cyclic loading
leads to at least three aspects of soil response that are not encountered