Anda di halaman 1dari 21

On limiting force prole, slip depth and response of lateral piles

Wei Dong Guo


*
School of Engineering, Grith University, Gold Coast 9726, Qld, Australia
Received 11 February 2005; received in revised form 30 January 2006; accepted 7 February 2006
Available online 31 March 2006
Abstract
A wealth of numerical analyses have been performed for lateral piles using nite dierence, boundary element, and nite element
methods, etc. Their essence, especially for a free-head pile, is to simulate the mobilisation of the maximum (limiting) force along the pile.
This paper presents new closed-form solutions for a free-head pile embedded in an elasticplastic, non-homogeneous soil by simulating
pilesoil interaction using a series of springs distributed along the pile shaft. The stiness of each spring is theoretically related to soil
modulus, pilesoil relative stiness, and loading properties; and the limiting force is catered for by a new generic expression. The solu-
tions permit non-linear response of the pile to be readily estimated right up to failure. They compare well with continuum-based 3D nite
element analysis of a pile embedded in stratied soils, irrespective of whether the limiting force (or pilesoil relative slip) is progressively
mobilised downwards as the solutions assume or not. Presented in explicit forms, the solutions allow the dominant limiting force prole
(LFP) to be back-estimated against measured data, and may be used as a boundary element for simulating beamsoil interaction under
lateral resistance. Ranges of input parameters are provided, in light of predictions carried out to date against 62 tested piles in clay and
sand. Finally, study on three typical piles is presented to elaborate the calculation procedure.
2006 Elsevier Ltd. All rights reserved.
Keywords: Piles; Closed-form solutions; Non-linear response; Lateral loading; Soilstructure interaction
1. Introduction
A wealth of theoretical studies on response of laterally
loaded piles have been undertaken using a load transfer
(py curves) model [1]. In the model, the pilesoil interac-
tion is simulated using a series of independent (uncoupled)
springs distributed along the pile shaft [2]. With the py
curves at any depths, solution of the response of the piles
is normally obtained using numerical approaches such as
nite dierence method (e.g., COM624 [3]). This model,
however, often oers notable dierent predictions against
continuum-based nite element analysis (FEA) (e.g. [4]).
Derived from a displacement mode for soil around the pile,
a new coupled load transfer model has been recently devel-
oped [5] that allows the interaction among the springs to be
captured. Characterised by modulus of subgrade reaction
(k), and a ctitious tension (N
p
), the coupled model com-
pares well with nite element analysis. However, it is con-
ned to elastic state.
With increase in lateral loads, maximum (limiting) resis-
tance along a pile is gradually mobilised. This limiting force
is the sum of the passive soil resistance acting on the face of
the pile in the direction of soil movement, and sliding resis-
tance on the side of the pile, less any force due to active
earth pressure on the rear face of the pile. Variation of
the net limiting force per unit length p
u
with depth (referred
to as LFP) is critical to pile design. Signicant research
eort has been made to date to construct the LFP, using
mainly the following three techniques of: force equilibrium
on a passive soil wedge [6,7]; upper-bound method of plas-
ticity on a conical soil wedge [8]; and a strain wedge mode
of soil failure [9]. Around a pile, the wedge was assumed to
have developed near ground level; below which, lateral
ow was assumed. The LFP is generally non-uniform with
depth even along piles embedded in a homogeneous soil, as
0266-352X/$ - see front matter 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.compgeo.2006.02.001
*
Tel.: +610755528803; fax: +610755528065.
E-mail address: w.guo@grith.edu.au.
www.elsevier.com/locate/compgeo
Computers and Geotechnics 33 (2006) 4767
it was taken for granted [10]. The most popularly used p
u
(thus the LFP) was that derived from the force equilibrium
on a wedge [7,11]. However, it is noted that the scope of the
wedge should associate with rotation center for a rigid pile
at one extreme, while the wedge may never occur around a
batter (or capped) pile at the other extreme. A universal
expression of LFP should be applicable to any mode of soil
failure.
Nomenclature
A
L
coecient for the LFP;
s
u
undrained shear strength of soil;
s
u
average undrained shear strength over pile
embedment;
d diameter of an equivalent solid cylinder pile;
E
p
Youngs modulus of an equivalent solid cylinder
pile;
e eccentricity, i.e., the height from the loading
location to the mudline;
G
s
average shear modulus of the soil;
G
*
average soil shear modulus, G
*
= (1 + 0.75m
s
)G
s
;
I
p
moment of inertia of an equivalent solid cylin-
der pile;
J empirical factor lying between 0.5 and 3 for esti-
mating N
g
;
k modulus of subgrade reaction;
k
1
parameter for estimating the load transfer fac-
tor, c;
K
i
(c) modied Bessel function of second kind of ith
order;
K
p
tan
2
(45 + /
0
/2), the passive earth pressure coef-
cient;
L embedded pile length;
L
c
critical embedded pile length, beyond which the
pile is classied as innitely long;
LFP net limiting force prole;
M(x) moment induced in a pile element at a depth of
x, also written as M;
M
A
(x), M
B
(z) moment induced in a pile element, at
depth x and z, respectively;
Mx Mxk
n2
=A
L
, normalised bending moment at
depth x;
M
max
maximum bending moment within a pile;
n power for the LFP;
N blow count of the SPT;
N
co
lateral capacity factor correlated soil undrained
strength with the limiting pilesoil pressure at
mudline;
N
c
equivalent lateral capacity factor correlated
average soil undrained strength with average
limiting force per unit length on pilesoil inter-
face by p
u
s
u
N
c
d;
N
g
gradient correlated soil undrained strength with
the limiting pilesoil pressure;
N
p
ctitious tension for a stretched membrane used
to tied together the springs around the pile shaft;
P lateral load applied at a distance of e above
mudline;
P
e
value of the lateral load applied P when the slip
depth is just initiated at mudline;
p force per unit length;
p
u
limiting force per unit length;
P Pk
n1
=A
L
, normalised pile-head load;
p
u
average limiting force per unit length over the
pile embedment;
r
0
radius of an equivalent solid pile;
x depth measured from mudline;
x
max
depth at which maximum bending moment
occurs (x
max
= x
p
+ z
max
when wx
p
P0; x
max
x
max
when wx
p
< 0 );
x
p
slip depth from the elastic to the plastic state;
w lateral deection;
w
A
lateral deection in the upper plastic zone;
w
IV
A
, w
000
A
, w
00
A
, w
0
A
fourth, third, second and rst deriva-
tives, respectively, of deection w with respect
to the depth x;
w
B
lateral deection in the lower elastic zone;
w
IV
B
, w
000
B
, w
00
B
, w
0
B
fourth, third, second and rst deriva-
tives, respectively, of deection w with respect
to the depth x;
w
g
, w
0
g
lateral pile deection, and rotation angle (in
radian) at mudline, respectively;
w
g
w
g
kk
n
=A
L
, normalised mudline deection;
w
p
p
u
/k, lateral deection at the slip depth of x
p
;
w
IV
p
, w
000
p
, w
00
p
, w
0
p
values of fourth, third, second and rst
derivatives, respectively, of deection w with
respect to the depth x at depth x
p
;
w
t
lateral deection;
z depth measured from the slip depth;
z
max
depth below the slip depth, where maximum
bending moment occurs;
a, b stiness factors for elastic solutions using load
transfer model;
a
c
ratio of the shear modulus over the undrained
strength, G
s
/s
u
;
a
N
, b
N
a/k, and b/k, normalised a and b by k, respec-
tively;
a
0
equivalent depth to account for mudline limiting
force;
c load transfer factor;
c
0
s
eective density of the overburden soil;
k the reciprocal of characteristic length;
m
s
Poissons ratio of soil;
/
0
eective frictional angle of soil.
48 W.D. Guo / Computers and Geotechnics 33 (2006) 4767
With the k deduced from the coupled model, and the p
u
from LFP, an ideal elasticplastic load transfer (py) curve
at any depth can be constructed. Transition from the initial
elastic to the ultimate plastic state may be assumed in other
forms of non-linear curves, such as those proposed for soft
clay, sti clay, and sand, etc. [6,7,11]. In terms of instru-
mented piles embedded in uniform soils [7], these forms
of py curves are proved to be very useful. However, to
synthesise the curves, a number of parameters are required
to be properly determined [12], which is often time-con-
suming, wearisome, and only warranted for large projects.
In contrast, a simplied elasticplastic py curve is su-
ciently accurate [13] and normally expedite.
A rigorous closed-form expression should be developed
even though there are numerical solutions, as the former
can then be used as a new boundary element to advance
the latter. Elastic, perfectly plastic solutions (typically for
free-head piles) have been proposed [14,15], using (uncou-
pled) Winkler model. Valid for either a uniform or a line-
arly increase limiting force prole, they were nevertheless
not rigorously linked to properties of a continuum medium
such as shear modulus etc. Thus, a new approach is
required to render response of a pile (e.g., loaddeection,
and loadmaximum bending moment relationships) to be
simulated, which should be consistent with continuum-
based analysis. It should also permit a new LFP to be
back-estimated using measured responses of a pile regard-
less of mode of soil failure.
This paper addresses the elasticplastic solutions of lat-
erally loaded, free-head piles that are embedded in non-
homogeneous medium, for which limiting force per unit
length varies monotonically with depth. The following tar-
gets are set:
Propose a generic expression to describe the distribution
of limiting force per unit length with depth, and nd its
valid depth (maximum slip depth).
Establish new closed-form solutions for piles in elastic
plastic, non-homogenous soil.
Examine the eect of each input parameter by conduct-
ing parametric analysis and case studies.
The study uses the 3D FE analysis for a pile in two strat-
ied soils to verify the closed-form solutions. The solu-
tions are implemented into a spreadsheet program to
facilitate the prediction of response of a number of tested
piles, and parametric analysis. Selection of input parame-
ters is discussed at length. Three typical examples are
elaborated.
2. Solutions for the pilesoil system
The problem addressed here is schematically depicted in
Fig. 1, where a laterally loaded pile is embedded in a non-
homogenous elastoplastic medium. No constraint is
applied at the pile-head and along the eective pile length,
except for the soil resistance. The free length (eccentricity)
measured from the point of applied load, P to the ground
surface is written as e. The pilesoil interaction is simulated
by the model shown in Fig. 2(a). The uncoupled model
indicated by the p
u
is utilised to represent the plastic inter-
action, and the coupled load transfer model indicated by
the k, and N
p
to portray the elastic pilesoil interaction,
respectively. The two interactions occur, respectively, in
regions above and below the slip depth, x
p
. In other
words, the following hypotheses are adopted:
(i) Each spring is described by an idealised elasticplas-
tic py curve (y being written as w in this paper,
Fig. 2(b)).
(ii) In elastic state, equivalent, homogenous and isotropic
elastic properties (modulus and Poissons ratio) are
used to estimate the k and the N
p
.
(iii) In plastic state, the interaction among the springs is
ignored by taking the N
p
as zero.
(iv) Pilesoil relative slip occurs down to a depth where
the displacement, w
p
is just equal to p
u
/k and net
resistance per unit length p
u
is fully mobilised.
(v) The slip (or yield) can only occur from ground level,
and progress downwards.
All of the ve assumptions are adopted to establish
closed-form solutions presented in this paper. Inuence
e
L
x
p
x , or z
Plastic zone
Elastic zone
P
Pressure distribution
(L > L
c
)
Fig. 1. The problem addressed.
(a) Coupled load transfer model (b) Normalised p -y (w) curve
P
Plastic
zone, x
p

Elastic
zone
Membrane, N
p

Spring, k
p
u
0
0.5
1
0 1 2 3
Normalised deflection, w/d (%)
N
o
r
m
a
l
i
s
e
d

f
o
r
c
e
,

p
/
p
u


Plastic zone
Elastic zone
Transition zone
Fig. 2. Coupled load transfer analysis.
W.D. Guo / Computers and Geotechnics 33 (2006) 4767 49
of deviation from these assumptions are assessed and com-
mented upon in the later sections.
2.1. Coupled load transfer model (elastic state)
Applying a variational approach to simulate elastic
response of the coupled pilesoil system shown in
Fig. 2(a) with x
p
= 0, the following conclusions drawn pre-
viously [5] are directly adopted herein.
Innitely long. A free-head pile is dened as innitely
long, should the pilesoil relative stiness, E
p
/G
*
be less
than a critical ratio, (E
p
/G
*
)
c
[16], where (E
p
/G
*
)
c
=
0.052(L/r
0
)
4
/(1 + 0.75m
s
), G
*
= (1 + 0.75m
s
)G
s
, G
s
, m
s
is
the shear modulus, and Poissons ratio of an equivalent,
homogenous and isotropic elastic soil, respectively; L is
the embedded pile length; E
p
and r
0
are the Youngs
modulus, and radius of an equivalent solid cylinder pile,
respectively. In other words, unless a slenderness ratio
L/r
0
is less than L
c
/r
0
, the laterally loaded pile can be
treated as if it were innitely long.
Modulus of subgrade reaction k. The modulus k [FL
2
]
may be estimated by Eq. (1)
k 1:5pG
s
f2cK
1
c=K
0
c c
2
K
1
c=K
0
c
2
1g
1
where K
i
(c) is the modied Bessel function of second
kind of ith order (i = 0, 1). This k is not the custom-
ary one [5,17]. The factor c for a free-head, innitely
long pile is a power function of the relative stiness,
i.e.,
c k
1
E
p
=G

0:25
2
where k
1
= 1.0, and 2.0, respectively, for a lateral load
applied at ground surface, and at an innitely high
eccentricity (i.e., a pure moment loading). k
1
lies in be-
tween 1.0 and 2.0.
Fictitious tension N
p
. The tension, N
p
[F] of the mem-
brane linking the springs is determined by
N
p
pr
2
0
G
s
K
1
c=K
0
c
2
1 3
2.2. Generic net LFP (plastic state)
A monotonic variation in the net limiting force per unit
length, p
u
with depth x (LFP) may be expressed by a gen-
eric expression of
p
u
A
L
x a
0

n
4
where A
L
is the coecient for the LFP [FL
1n
]; x is the
depth below mudline [L]; a
0
is an equivalent depth to cover
the force at x = 0 [L]; p
u
is the net limiting force per unit
length [FL
1
]; n is a constant (<3). As shown later on,
the dimensional parameters A
L
, a
0
, and n allow non-dimen-
sional response of a pile to be presented in compact form.
To facilitate practical use, A
L
, and a
0
are themselves
re-expressed through two non-dimensional parameters
N
g
, and N
co
A
L
s
u
or c
0
s
dN
g
d
1n
; any n
a
0
N
co
=N
g

1=n
d n 6 0
5a
where c
0
s
is the average eective density of the overburden
soil [FL
3
] (i.e., unit weight above water table, and buoy-
ant weight below); c
0
s
d is the eective overburden stress at
a depth of the pile diameter [FL
2
]; s
u
is the average un-
drained shear strength of the soil over the slip depth under
maximum load that initially may be taken as 8d [FL
2
]. In
the expression for A
L
, s
u
and c
0
s
d are used for cohesive soil,
and cohesionless soil, respectively. Eq. (4) indicates (1) at
n = 0, p
u
for cohesive soil is reduced to s
u
N
g
d; and (2) at
mudline, p
u
becomes s
u
N
co
d or c
0
s
d
2
N
co
. To allow compar-
ison with existing solutions for n = 0, in particular, an
equivalent N
c
is dened as N
c
p
u
=s
u
d for cohesive soil
with p
u
being an average limiting force per unit length over
the maximum slip depth.
A maximum p
u
along a laterally loaded pile in a uniform
cohesive soil was deduced as N
c
ds
u
(N
c
9:1411:94) [18].
However, a supercially high value may result for a pile in
layered soil when an average s
u
is adopted for an increasing
strength prole with depth [8]. Eq. (4) along with L > L
c
(innitely long pile) means that the p
u
is generally fully
mobilised from mudline to the slip depth, x
p
that increases
with lateral loads, as indicated in Fig. 1. Thus, the LFP
may be conveniently plotted using Eq. (4), but it is eective
only to the maximum x
p
.
In determining the parameters N
g
, N
co
, and n for Eq. (4),
six options may be considered.
Option 1. For piles in a cohesionless soil, should the LFP
be taken as that suggested by [10], referred to as Broms
LFP later on, the parameters should be adopted as
N
g
3K
p
; N
co
0; n 1 5b
where K
p
= tan
2
(45 + /
0
/2), the passive earth pressure
coecient; /
0
is the eective friction angle of the soil.
Option 2. For piles in a cohesive soil, if the LFP pro-
posed by [6,7] is adopted, then the parameters should
be given by
N
g
c
0
s
d=s
u
J; N
co
2; n 1 5c
where J is a factor lying between 0.5 and 3 [6]. The
LFP obtained using Eqs. (4) and (5c) will be referred
to as Matlock using J = 0.5; and as Reese (C) if
J = 2.8 [7].
Option 3. An available LFP [19] may be expressed by
choosing a set of parameters N
g
, N
co
and n. For
instance, the LFP for a pile in sand employed in
COM624P, which is referred to as Reese (S), was well
tted using N
g
K
2
p
; N
co
0, and n = 1.7 [20].
Option 4. Should a set of py curves be available,
numeric value of p
u
may be acquired from the curves
for each depth, hence the LFP is resulted.
50 W.D. Guo / Computers and Geotechnics 33 (2006) 4767
Option 5. For a layered soil prole, the LFP may be con-
structed using the interim procedure as follows. Firstly,
assuming that the entire soil consists of the clay or the
sand only, LFP of Reese (C) or (S) is obtained, respec-
tively. Secondly, the p
u
within a zone of 2d above or
below an interface should be increased in average by
$40% for a weak (clay) layer adjoining a sti (sand)
layer; and decreased by $30% for a sti layer adjoining
a weaker layer [21]. Thirdly, the increased and the
decreased p
u
of the two adjacent layers is represented
averagely by Eq. (4), with the n being gauged visually,
as an exact shape of the LFP (thus n) makes little dier-
ence to the nal predictions (shown later in Case stud-
ies). And nally, a LFP is created for a two-layered
soil. For a multiple layered soil, the same principles
apply, but the n (thus the LFP) should permit the overall
limiting force to be represented. Any layer located more
than 2d below the maximum slip depth is excluded in
this process. Lastly,
Option 6. Should measured pile response be available,
the parameters N
g
, N
co
, and n may be back-gured
through matching predicted with measured responses
of a pile as elaborated in the section of back-estimation
of LFP.
Use of options 15, and 6 are illustrated in sections of
Validation, and Case studies, respectively.
2.3. Elasticplastic solutions
For convenience, response of the pile is denoted by the
subscripts A and B for the upper plastic, and the lower
elastic zones, respectively (see Fig. 2(a)). Above the depth
x
p
, using the uncoupled model for the plastic zone (x 6 x
p
),
the governing equation for the pile is
E
p
I
p
w
IV
A
A
L
x a
0

n
0 6 x 6 x
p
6
where w
A
is the deection of the pile at depth x that is
measured from ground level; w
IV
A
is the fourth derivative
of w
A
with respect to depth x; I
p
is the moment of inertia
of an equivalent solid cylinder pile. Below the depth x
p
,
using the coupled model for the elastic zone (x
p
6 x 6 L),
the governing equation for the pile may be written as
[5,22]
E
p
I
p
w
IV
B
N
p
w
00
B
kw
B
0 0 6 z 6 L x
p
7
where w
B
is the deection of the pile at depth z
(=x x
p
) that is measured from the slip depth. w
IV
B
,
w
00
B
is the fourth, and second derivatives of w
B
with re-
spect to depth z. Taking N
p
= 0, Eq. (7) reduces to the
one based on Winkler model. By invoking the deection
and slope (rotation) compatibility restrictions at x = x
p
(z = 0) for the innitely long pile, Eqs. (6) and (7) were
solved as elaborated in Appendix I. The solutions allow
response of the pile at any depth to be predicted readily.
In particular, three key responses were recast in dimen-
sionless forms, which are:
Normalised load, P (=Pk
n+1
/A
L
), where P is a lateral
load applied at a distance of e above mudline; k is the
reciprocal of characteristic length, k

k=4E
p
I
p

4
_
.
Normalised mudline deection, w
g
(=w
g
kk
n
/A
L
), where
w
g
is the pile deection at mudline.
Normalised bending moment, Mx (=M(x)k
n+2
/A
L
),
where M(x) is the bending moment in the pile at depth
x.
The key responses are described as follows.
2.3.1. Lateral load
Using Eq. (A-26), an expression for the normalised lat-
eral load is derived as
P
F 1; 0x
p
a
N

x
p
a
N
e

F 2; x
p
F 2; 0 a
N
F 1; x
p
F 0; x
p
=2
x
p
a
N
e
8
where x
p
kx
P
, normalised slip depth using k; e ke, nor-
malised eccentricity; and
a
N
a=k

1 N
p
=

4E
p
I
p
k
_
_
a

k=4E
p
I
p

_
N
p
=4E
p
I
p

_
9a
and
F m; x
p
x
p
a
0
k
nm

m
j1
n j
_
10
where

m
j1
n j n m n 2n 1, m is the
integer (6 4). For instance, when m = 2, F m; x
p
becomes
F 2; x
p
, and

2
j1
n j n 2n 1. In particular, at
m = 0,

0
j1
n j is taken as 1.
2.3.2. Groundline deection
Substituting Eqs. (A-21) and (8) into Eq. (A-23), the
normalised pile deection at ground level is obtained as
w
g
4F 4; x
p
x
p
F 3; x
p
F 4; 0 C
x2
F 2; x
p

C
x1
F 1; x
p
C
x0
F 0; x
p
C
o2
F 2; 0
C
o1
F 1; 0 11
where
C
x2
41 a
2
N
C
m
; C
x1
2 4a
2
N
x
p
a
N
C
m
;
C
x0
1 2a
N
x
p
C
m
=2; C
o2
2x
2
p
C
x2
;
C
o1
4x
3
p
=3 2x
p
x
p
a
N
C
m
;
C
m

4
3
x
3
p
2x
p
2x
2
p
4a
2
N
4e
_ _
1
x
p
a
N
e
12
At a relative small eccentricity, e, pile-head deection, w
t
may be approximately taken as w
g
e w
0
g
, where w
0
g
is
the mudline rotation angle (in radian) obtained using Eq.
(A-22).
W.D. Guo / Computers and Geotechnics 33 (2006) 4767 51
2.3.3. Maximum bending moment
The maximum bending moment, M
max
occurs at a depth
x
max
(or z
max
) at which shear force is equal to zero. The
depth could locate in plastic or elastic zone, thus it needs
to be determined by using either Q
A
(x
max
) = 0 in Eq. (A-
1), or Q
B
(z
max
) = 0 in Eq. (A-30). Which zone the M
max
lies
in depends on a function, wx
p
that itself is derived from
Eq. (A-30)
wx
p
b
N
=a
N
2kw
00
P
=w
000
P
13
where
b
N
b=k

1 N
p
=

4E
p
I
p
k
_
_
b

k=4E
p
I
p

_
N
p
=4E
p
I
p

_
9b
w
000
P
kk
n
=A
L
4k
3
F 1; x
p
F 1; 0 P 14
w
00
P
kk
n
=A
L
4k
2
F 2; x
p
F 2; 0 F 1; 0x
p
x
p
eP
15
The depth z
max
may be obtained from Eq. (A-30) as
z
max
tan
1
wx
p
=b 16
which expression to be used for determining the M
max
de-
pends on the wx
p
.
(a) If wx
p
> 0, then z
max
> 0. The maximum moment
M
max
should occur in elastic zone, thus it locates at a
depth (measured from mudline) that is equal to x
p
+
z
max
. The value of M
max
may be estimated using Eq.
(17) by replacing z with z
max
, i.e., M
max
= M
B
(z
max
).
M
B
z e
az
C
1
x
p
cos bz C
2
x
p
sin bz 17
where C
1
x
p
E
p
I
p
w
00
P
k
n2
=A
L
, C
2
x
p
E
p
I
p
w
000
P
aw
00
P
k
n2
=bA
L
. M
B
(z) is the bending moment at
depth z that is derived from Eq. (A-29).
(b) If wx
p
< 0, then z
max
< 0, which is expected at a rel-
atively high value of x
p
. The M
max
should locate in
plastic zone, and at depth x
max
. The normalised depth
x
max
(=x
max
k) is obtained, using Eq. (A-1), as
x
max
k a
0
k
n1
n 1P
1=n1
a
0
k 18
The M
max
may be calculated using Eq. (19), derived
from Eq. (A-2) by replacing x with x
max
.

M
max
A
L

1
n 2
a
n1
0
n 1
P
A
L
_ _
n2=n1

a
n2
0
n 2

a
0
P
A
L
_ _

Pe
A
L
19
The M
max
of Eq. (19) may be recast in form of the
normalised slip depth as well by replacing the P with
that given by Eq. (8). The form of Eq. (19) is to stress
its independence of the characteristic length but A
L
,
a
0
, n (hence the LFP), and the load P. However, the
normalised form, M
max
k
n+2
/A
L
will be used later to
provide a consistent presentation (e.g., Figs. 7 and
8) from elastic through to plastic state.
In summary, responses of the laterally loaded piles are
presented in explicit expressions of the slip depth, at which
the maximum p
u
normally occurs. The expressions are
valid for innitely long (L > L
c
) piles that are embedded
in a soil of a constant modulus (k) with depth. This con-
stant may be regarded as an average value over the maxi-
mum slip depth expected (initially taken as 8d) below
ground level. Derived from Eq. (4), the new solutions are
not based on mode of soil failure. Inuence of the failure
mode is catered for by selecting dierent parameters and/
or types of solutions, such as the current solutions for a
free-head pile, and those for a xed-head pile [23] etc.
2.4. Some extreme cases
The normalised slip depth under lateral loads may be
estimated using Eq. (8) that associates with the LFP (via
A
L
, a
0
, n), and the pilesoil relative stiness (via k). The
minimum load, P
e
required to initiate the slip at mudline
(x
p
0) is given by
P
e
k
n1
=A
L
a
0
k
n
=2a
N
e 20
The current solutions can be reduced to available solutions
for some special cases.
(a) Imposing n = e = 0, a
N
= 1, and a
0
= 0, the norma-
lised load by Eq. (8) reduces to
Pk=A
L
1 x
p
=2 21
And the normalised deection by Eq. (11) reduces to
w
g
k=A
L
x
4
p
=6 2x
3
p
=3 x
2
p
x
p
1 22
Eqs. (21) and (22) are essentially identical to those
brought forwarded previously [24] using Winkler
model (N
p
= 0) for a pipeline that is embedded in a
homogenous soil, and has a constant (n = 0) limiting
force (resistance) along its length.
(b) Setting a
0
= 0, and e = 0, M
max
obtained from Eq.
(19) then reduces to P
2
/(2A
L
), and

8P
3
=9A
L
_
, respec-
tively, for n = 0, and 1; and the corresponding depth
x
max
derived from Eq. (18) becomes P/A
L
, and

2P=A
L
_
. These results are what were put forwarded
previously [25].
(c) Introducing x
p
0 (elastic state), and at e = 0, Eq.
(15) oers that w
00
P
0. Furthermore, using N
p
= 0
in Eqs. (16) and (17), z
max
and M
max
obtained, respec-
tively, are virtually identical to the results obtained
using Winkler model [24].
2.5. Numerical calculation and back-estimation of LFP
In spite of the complicated appearance the expressions
such as Eqs. (8), (11), (13) and (17) can be readily estimated
using modern mathematical packages. In this paper, they
have been implemented into a spreadsheet operating in
52 W.D. Guo / Computers and Geotechnics 33 (2006) 4767
EXCEL called GASLFP, along with Eqs. (1)(5) and the
expressions shown in Appendix I. At a
0
= 0, and a
N
= 1,
simplied forms of Eqs. (8), (11), (13) and C
1
x
p
; C
2
x
p

for Eq. (17) have also been obtained and provided in


Appendix II.
Using Eq. (8), the slip depth, x
p
may be obtained itera-
tively using a purpose- designed macro under a lateral load,
P, as is adopted in GASLFP. The slip depth, x
p
may also
be assumed to estimate the load, P when the simplied
expression is employed. In either approach, the normalised
slip depth, x
p
may be calculated with the k, which allows
the ground-level deection, w
g
to be computed, the wx
p

to be calculated. The latter in turn permits the maximum


bending moment, M
max
and its depth, x
max
to be evaluated.
The calculation is repeated for a series of P or x
p
, each
oers a set of load, deection, bending moment and its
depth, thus the P w
g
, M
max
w
g
, and x
max
M
max
curves etc are determined. All of the numerical values of
the current solutions presented subsequently are obtained
using GASLFP, except where specied. For comparison,
the predictions using simplied expressions are sometimes
provided as well.
Using GASLFP, the three input parameters N
g
, N
co
and
n (or A
L
, a
0
, and n) may also be adjusted until the predicted
w
g
, M
max
and x
max
across all load levels agree well with the
corresponding measured data (if available), respectively.
The values thus obtained should reect the overall pilesoil
interaction rather than a real limiting force prole in the
case of a layered soil. They are unique, so is the corre-
sponding LFP (option 6). This back-estimation may also
be carried out using other expressions given in Appendix
I, for instance, Eq. (A-22) for rotation. If only one mea-
sured (normally w
g
) response is available, N
g
may be
back-estimated by taking N
co
as 04, and n as 0.7 and
1.7 for clay and sand, respectively. Should two measured
(say w
g
and M
max
) responses be available, both N
g
and n
may be back-estimated by an assumed value of N
co
. Mea-
sured responses should encompass the integral eect of all
intrinsic factors on piles in a particular site, so should the
corresponding back-gured LFP. In this manner, LFP
can be gradually updated to reect the eect for future
design.
3. Validation against FEA
Yang and Jeremic [4] reported a 3D nite element anal-
ysis (FEA) of a laterally loaded pile. The square aluminium
pile was 0.429 m in width, and 13.28 m in length. It had a
exural stiness E
p
I
p
of 188.581 MN m
2
. The pile was
installed to a depth of 11.28 m in a deposit of claysand
clay prole, and sandclaysand prole, respectively. Lat-
eral loads were applied at 2 m above ground level. The
claysandclay prole refers to a uniform clay layer that
has a uniform interlayer of sand between a depth of
1.72 m (4d, d = width of the pile) and 3.44 m (i.e. 8d). A
contrasting case was a clay layer sandwiched between
two sand layers and is refereed to as the sandclaysand
prole. In either prole, the clay has a undrained shear
strength of 21.7 kPa, Youngs modulus of 11 MPa, Pois-
sons ratio of 0.45, and unit weight of 13.7 kN/m
3
. The
medium dense sand has an internal friction angle of
37.1, shear modulus of 8.96 MPa at the level of pile base,
Poissons ratio of 0.35, and unit weight of 14.5 kN/m
3
.
The FE analysis of the pile oered the following results
for either soil prole: (1) py curves at depths up to 2.68 m;
(2) pile-head load and mudline displacement relationship;
(3) mudline displacement and maximum bending moment
curve; and (4) proles of bending moment under 10
selected load levels. In this study, with the py curves,
the limiting p (i.e., the p
u
) at each depth was approximately
evaluated, thus the variation of p
u
with depth (i.e., LFP)
was obtained (option 4), and is shown in Figs. 3 and 4(a)
as FEA, respectively. Using the bending moment proles,
the depths of maximum bending moment were estimated,
and are shown in Figs. 3 and 4(d), respectively.
In the current predictions of the pile embedded in the
claysandclay prole, an average shear modulus of the
soil G
s
was calculated as 759.5 kPa (=35s
u
[26]). This ren-
ders k, and N
p
to be estimated as 2.01 MPa, and
892.9 kN using Eqs. (1) and (3), respectively. Following
option 5 for constructing LFP for a stratied soil, LFP
of Reese (C) or (S) was obtained, respectively, using prop-
erties of the clay or sand layer. The p
u
at depth (24)d of
the clay layer was approximately increased in average by
30% due to the underlying sti sand layer, while at depth
(46)d, it was approximately reduced by 20%, due to the
overlying weak clay layer [21]. A smooth transition from
the upper to the low layer allows the n to be gauged visually
as 0.8 (Fig. 3). The bottom clay layer is ignored as justied
later. Thus, the current LFP for the pile in the stratied soil
is described by n = 0.8, N
g
= 6, and a
0
= 0.
The predictions using the current LFP compare well
with the FEA results, in terms of mudline displacement
(see Fig. 3(b)), and maximum bending moment (see
Fig. 3(c)). Nevertheless, the depth of the M
max
, x
max
is
overestimated by up to 20%, because the current predic-
tions use an equivalent homogenous medium through the
constant k [2], and the stratied prole reduces the depth
x
max
. The maximum slip depth x
p
was calculated as 4.63d
under the maximum pile-head load. Thus, the bottom clay
layer located at 3.37d (=8d 4.63d) below the maximum
x
p
can be ignored. The current p
u
between 1.5d and 8d
slightly exceeds what would be obtained using the previous
instruction [27] for a layered soil that is not shown herein.
However, it is quite compatible with the overall trend of
FEA result within the maximum slip depth. Interestingly,
replacing the Current LFP with the Reese (C), the current
solutions still oer good predictions to 30% of the maxi-
mum load, to which the response is dominated by the
upper layer.
For the pile in the sandclaysand prole, shear mod-
ulus, G
s
was obtained as 1.206 MPa by averaging that of
the sand at mid-depth (=0.5 8.96 MPa) and of the clay
(=0.759 MPa). This leaded to that k = 3.33 MPa and
W.D. Guo / Computers and Geotechnics 33 (2006) 4767 53
N
p
= 5.731 MN. Using option 5, LFPs of both Reese (S)
and Broms were ascertained using the sand properties;
and that of Reese (C) using the clay ones. Importantly,
the eect of the bottom sand layer on LFP is considered
in this case, as the interface (located at depth of 8d) is less
than 2d from the maximum slip depth x
p
of 6.75d
obtained subsequently. In comparison with the Broms
LFP, the p
u
was maintained unchanged in depth of (0
4)d, and it was reduced approximately by 20% in the
depth of (68)d. For instance, p
u
=c
0
s
d
2
at depth of 8d
was found to be 70.0. An overall t to the LFP of the
top layer and the point (70, 8) allows the Current LFP
to be described by using n = 0.8, N
g
= 16.32
[=14.5 tan
4
(45+37.1/2)], and a
0
= 0. The current pre-
dictions compare well with the FE analysis in terms of
the mudline displacement, w
g
(see Fig. 4(b)), and maxi-
mum bending moment, M
max
(see Fig. 4(c)). Only the
depth of the M
max
was overestimated at a load less than
270 kN, probably due to the same reason as explained
previously for claysandclay prole. The maximum slip
depth x
p
was found to be 6.75d at a load of 400 kN.
The current LFP is close to the overall trend of that
obtained from FEA within the x
p
.
For the pile in sandclaysand prole, it is noted that
at the depth of 4d, the w
p
is calculated as 25.2 mm
(=9 21.7 0.429/3.33) using the clay strength; but it is
39.7 mm (=85.71 (4 0.429)
0.8
/3330) using the sand
properties and A
L
= 85.71 kN/m
1.8
. This implies that
around the depth of the interface, pilesoil relative slip
occurred in the clay before it did in the overlying sand.
This sequence is unlike the assumption of slip
progressing downwards. The very good comparisons
noted above indicate that (1) response of the pile is
mainly aected by the overall trend of the LFP within
the maximum slip depth x
p
; and (2) Current solutions
are suciently accurate for the stratied soil by employ-
ing n = 0.8 to describe the LFP; and using x
p
to repre-
sent a transition zone that should occur if non-linear
py curves [4] are employed. To decide whether a layer
comes into eect, initial x
p
may be based on an educated
guess, say, 8d. These conclusions are further enhanced by
the facts that:
(a) (b)
(c) (d)
Fig. 3. Comparison between the current predictions and FEA results [4] for a pile in claysandclay layers.
54 W.D. Guo / Computers and Geotechnics 33 (2006) 4767
Should the LFP be replaced with that of Broms or Reese
(S), excellent predictions up to a load of 400 kN would
also be achieved. Beyond the load level, the displace-
ment, w
g
and the moment, M
max
would be gradually
underestimated.
Should the gradient of the LFP reduce by 10%, better
predictions against the FEA results would be achieved.
Should the Georgiadis proposal be adopted, an overes-
timation of w
g
and M
max
would be expected, as a much
lower gradient LFP than the current one would be
resulted.
The later two analyses are not shown herein.
4. Pile response versus slip depth
The previous section indicates that responses of a later-
ally loaded pile are generally dominated by the LFP within
the maximum slip depth. Taking e = 0 (thus w
t
= w
g
), nor-
malised responses such as P, w
g
, M
max
, and so forth are pri-
marily aected by the three non-dimensional parameters n,
a
0
k and x
p
. The ranges of these parameters were identied
later in Comments on use of current solutions as: n = 0.5
2.0, a
0
k < 0.3, and x
p
< 2:0, for which parametric analysis
was carried out using GASLFP and presented below.
Fig. 5(a), and (b) indicate that:
(1) The normalised load, P is generally below 2.
(2) With n > 0, and x
p
< 2:0, P normally increases with
decrease in n, and/or increase in a
0
k.
(3) As P increases, the slip at n > 0 expands at a gradually
slowrate [The rate refers tonet increase inthe slipdepth
over corresponding increase in the load], as implied by
the increasing gradients of the concave curves.
Fig. 6(a), and (b) indicate that
(1) w
g
is 1020 times the value of P, andnormallyshouldbe
below50. For instance, at x
p
2:0 (n = 0.5, a
0
k = 0), it
is noted that P 1:38 and w
g
21:1, respectively.
(2) w
g
follows an opposite trend to P in response to n and
a
0
k.
(b)
(d) (c)
(a)
Fig. 4. Comparison between the current predictions and the FEA results [4] for a pile in sandclaysand layers.
W.D. Guo / Computers and Geotechnics 33 (2006) 4767 55
In comparison with Fig. 6, Fig. 7 demonstrates that
response of the normalised moment, M
max
to the load,
P resembles that of the deection w
g
to the load P. Fur-
thermore, the gure together with Fig. 8 demonstrates
that:
(1) M
max
and x
max
are below 5.0, and 3.0, respectively.
(2) As x
p
increases from 2 to 3 (n = 1.0), x
max
increases by
35%, and M
max
increases by 250%.
(3) At x
p
2, and n = 1, as a
0
k increases from 0 to 0.2, P
increases by 20% (from 1.47 to 1.77), Fig. 7(b), and
M
max
by 13% (from 1.68 to 1.90), Fig. 8(b). It seems
that w
g
and M
max
are more susceptible to x
p
than P
and x
max
are, which are also demonstrated in Table
4 discussed next.
5. Case studies
Predictions using the current solutions are presented for
three typical piles embedded in a two-layered silt, a sand-
silt layer, and a sti clay, respectively. Highlighted in bold,
solid lines in the subsequent gures, the cases of best pre-
dictions or match with measured data are obtained using
a b
Fig. 6. Normalised load and groundline deection (e = 0).
a b
Fig. 5. Normalised load versus slip depth (e = 0).
56 W.D. Guo / Computers and Geotechnics 33 (2006) 4767
the parameters N
g
, N
co
, and n (thus the LFPs) that are
provided in Table 1, along with six other input parameters.
These predictions are elaborated individually in this
section.
5.1. Piles tested in two-layered silt, and sand-silt layer
Single piles A, and C were driven into two dierent
types of two-layered soils, respectively. They were tested
a b
Fig. 7. Normalised load versus maximum bending moment (e = 0).
a b
Fig. 8. Normalised moment versus depth x
max
(e = 0).
Table 1
Parameters for the predictions of bold lines in Figs. 911
Figures s
u
(kPa) or (N)
*
a
c
or (G
s
/N)
*
E
p
I
p
(MN m
2
) e (m) L (m) r
0
(m) N
g
/N
co
N
c
=a
0
k n
9 15 121 298.2 0.1 17.4 0.305 4.53/3.28 5.8/.08 0.5
10 (12)
*
(640)
*
169.26 0.2 23.3 0.305 8.3/4.67 1.0
11 153 545.1 493.7 0.31 14.9 0.32 0.85/0.0 1.67/0 1.5
NB, indicated by
*
, the values refer to N (blow count of SPT); and G
s
/N (ratio of shear modulus and blow count) in kPa, respectively.
W.D. Guo / Computers and Geotechnics 33 (2006) 4767 57
individually and instrumented to measure the bending strain
down the depth [28]. Pile Awas 17.5 min length, and 0.61 m
in diameter. It had a exural stiness E
p
I
p
of 298.2 MN m
2
.
The pile was driven into a two-layered silt, with a uniform
undrained shear strength, s
u
of 15 kPa to a depth of 4.8 m,
and 22.5 kPa below the depth. G
s
was taken as 121s
u
(s
u
=
15 kPa) [28]. Thus, critical slenderness ratio, L
c
/r
0
was esti-
mated as 26.06, and the pile was classied as innitely long.
With the shear modulus, G
s
, pilesoil relative stiness,
E
p
/G
*
was calculated; factor c was obtained using Eq.
(2); modulus of subgrade reaction, k using Eq. (1); norma-
lised ctitious tension, N
p
/(2E
p
I
p
) using Eq. (3); stiness
factors, a and b, using Eqs. (9a) and (9b); and k using its
denition. All of these values for the elastic state are sum-
marised in Table 2(a). The LFP for the upper layer
(s
u
= 15 kPa) was determined using the prole of Reese
(C). Similarly, for the lower layer (x P7.87d = 4.8 m),
the p
u
was found to be a constant of 9d 22.5 kN/m, which
gives a p
u
/(s
u
d) of 13.5 using the upper layer s
u
of 15 kPa as
the normaliser. Using option 5, the overall LFP for the
two-layered soil should be close to the Reese (C) near the
ground level, and pass through point (13.5, 7.87), as indi-
cated by n = 0.5 in Fig. 9(a). The LFP may be expressed
by using n = 0.5, A
L
= 53.03 kPa/m
0.5
, and a
0
= 0.32 m in
Table 2(a)
Parameters for the bold lines, Case (I) elastic state
Piles E
P
/G
*
c k (MPa) N
P
/(2E
p
I
p
) a (m
1
) b (m
1
) k (m
1
)
A 18587.7 0.0856 5.378 0.0169 0.2750 0.2422 0.2591
C 2739.2 0.1382 26.400 0.0674 0.4808 0.4047 0.4444
Notes. G
s
= 1.82 and 7.68 MPa for piles A and C, respectively. If N
p
= 0, then a = b = k.
(a) (b)
(d)
(c)
Fig. 9. Comparison between the calculated and measured [28] response of Pile A.
58 W.D. Guo / Computers and Geotechnics 33 (2006) 4767
Eq. (4). The three parameters are thereby obtained using
Eq. (5a) as N
g
= 4.53, N
co
= 3.28, and n = 0.5. The higher
strength s
u
of 22.5 kPa of the lower layer renders 050%
(an average of 25%) increase in the p
u
in the depth of (4
7.87)d, which is similar to the eect of an interlayer of sand
on its overlying clay deposit [8,21].
The above-mentioned parameters oer excellent predic-
tions of the pile response as shown in Fig. 9. They allow the
responses shown in Table 2(b) to be obtained as well, which
encompass:
(a) Minimum lateral load, P
e
for initiating the slip.
(b) Lateral load, P
**
for the slip depth shown in Fig. 9.
(c) Maximum imposed lateral load, P
max
.
(d) Slip depth, x
p
under the P
max
.
Those critical loads (P
e
, P
**
, and P
max
) and the slip
depth are useful to examine the depth of inuence of each
soil layer. Pilesoil relative slip occurred along the pile A at
a low load, P
e
of 54.6 kN, but it touched the second layer
(i.e., x
p
= 4.8 m) at a rather high load, P
**
of 376.9 kN.
Inuence of the lower layer on the pile A seems to be well
catered for by the LFP. The above study is referred to as
Case (I).
To examine the eect of the parameters on the predic-
tions, against Case (I) the following investigations (see
Table 3) are made:
Case (II). Taking N
p
as 0, the responses obtained, and
shown in Fig. 9 as n = 0.5 (N
p
= 0), are slightly higher
than those obtained otherwise, indicating the elastic
coupled interaction is limited.
Case (III). Assuming a
0
= 0, the LFP reduces to the tri-
angular dots shown in Fig. 9(a). For this LFP, using the
simplied expressions (N
p
= 0, Appendix II), the pre-
dicted responses are indicated by (S. Eqs) in Fig. 9.
They are close to those bold lines obtained earlier
(a
0
6 0, N
p
6 0). Furthermore, assuming e = 0, the
responses for x
p
= 1, and 5 are hand-calculated as illus-
trated in Appendix II. The results are summarised in
Table 4 along with those obtained for x
p
= 3, and 8.
Case (IV). A new LFP of n = 0.4 described by
N
g
= 4.79, N
co
= 3.03 and n = 0.4, is utilised, which
oers an equivalent resistance force to that estimated
using the Reese (C) prole along the pile within the max-
imum x
p
of 8.36d. This LFP oers very good predictions
against the measured data as well. Thus, the n value
has limited eect on the predictions of response of the
pile, and use of visual assessment of n value is deemed
suciently accurate.
Pile C was 23.3 m in length, and 0.61 m in diameter, and
had an E
p
I
p
of 169.26 MN m
2
. It was driven through a
sand layer that extended to 15.4 m from the ground level,
and subsequently into a underlying silt layer with a s
u
of
Table 2(b)
Parameters and predictions for the bold lines, Case (I) plastic state
Piles Input parameters Predictions
N
g
or c
0
s
(kN/m
3
), / (deg) A
L
(kN/m
n+1
) a
0
(m) P
e
(kN) P
**
(kN) P
max
(kN) x
p
x
p
at P
max
A N
g
= 4.53 53.03 0.320 54.57 376.9 393.0 8.36d(1.32)
C c
0
s
16:5 /
0
= 28 83.58 0.563 48.90 244.4 440.0 4.72d(1.28)
Notes. P
e
= P at x
p
= 0; P

P at the slip depth x


p
shown in Figs. 9 and 10.
Table 3
Sensitivity of current solutions to k(N
p
), LFP and e (see Fig. 9)
Cases Limiting force proles Remarks References
(I) N
g
, N
co
and n provided in Table 1 Using the current LFP n = 0.5
(II) N
g
, N
co
and n provided in Table 1 N
p
= 0 n = 0.5(N
p
= 0)
(III) N
g
= 4.53, N
co
= 0, and n = 0.5 Using e = N
co
= N
p
= 0 n = 0.5(S. Eqs)
(IV) N
g
= 4.79, N
co
= 3.03, and n = 0.4 Total force is equal to that obtained using Reese(C)s LFP over the max. x
p
n = 0.4
Table 4
Response of pile A using simplied expressions and e = 0 (see Fig. 9)
x
p
(m) x
p
Pk
n1
AL
wgkk
n
AL

Mmaxk
n2
AL
P (kN) w
g
(mm) M
max
(kN m) x
max
(m)
1 0.2591 0.2792 0.6797 0.1079 112.3 13.2 167.4 3.09
3 0.7774 0.5851 2.3435 0.3218 235.2 45.4 501.1 3.65
5 1.2957
a
0.8982 6.1526 0.6574 361.1 119.2 1020.0 4.71
8 2.0730
a
1.4187 20.327 1.3716 570.4 393.8 2185.1 6.38
a
When x
p
exceeds 1.118, M
max
occurs at the upper plastic zone (i.e., x
max
< x
p
).
W.D. Guo / Computers and Geotechnics 33 (2006) 4767 59
55 kPa. The blow count of SPT, N of the sand layer was
found as: 12 (in depth of 011.0 m), 8 (11.013.8 m), and
16 (13.815.4 m), respectively. The eective unit weight c
0
s
was about 16.5 kN/m
3
. The shear modulus, G
s
, and angle
of friction of the sand, /
0
were correlated with the blow
count by the expression of [28] G
s
= 640N kPa, and
/
0

8N 4
_
20, respectively. Thus, with N = 12 for
the top 11 m, G
s
and /
0
were estimated to be 7.68 MPa,
and 28, respectively. This allowed L
c
/r
0
to be estimated
as 15.8, and the pile to be classied as innitely long. The
eective pile length L
c
is within the top layer, thus the prob-
lem becomes a pile in a single layer.
Similar to pile A, relevant parameters for elastic state
were estimated and are shown in Table 2(a). An apparent
cohesion was reported in the wet sand near the ground
level around the driven pile. This is represented by a
N
co
of 4.67 (of similar magnitude to N
c
). N
g
was obtained
as 8.31 using Eq. (5b). Thus, with n = 1.0, A
L
was com-
puted as 83.58 kPa/m, and a
0
as 0.563 m. The LFP is then
plotted as n = 1 in Fig. 10(a). The above-mentioned
parameters oer close predictions of the pile responses
to the measured data. The responses at a typical slip
depth of 2.5d (=1.52 m) are also highlighted. Table 2(b)
tabulates the critical values explained before. The P
e
was 48.9 kN, the x
p
extended to 2.88 m at a P
max
of
440.1 kN.
In the study, the n = 1 and N
co
> 0 are dissimilar to
n = 1.31.7 and N
co
= 0 normally used for sand. Thus their
eects are examined (Table 5) as follows:
Case (II). Should the apparent cohesion be ignored,
the n = 1.0 LFP then reduces to the Broms prole
(Fig. 10(a)). This new prole leads to overestimation
of displacement (Fig. 10(b)) against measured data.
The Reese (S) LFP after depth corrections was found
nearly identical to the Broms one for this case,
thereby the over-estimation reported previously [28]
using the former LFP is anticipated, so are the over-
estimations of maximum bending moment, M
max
,
and depth of the M
max
(not shown herein).
(c)
(d)
(b)
(a)
Fig. 10. Comparison between the calculated and measured [28] response of Pile C.
60 W.D. Guo / Computers and Geotechnics 33 (2006) 4767
Case (III). Taking a
0
(N
co
) = 0 and n = 2, a lower lim-
iting force than that derived from the n = 1 LFP is seen
in Fig. 10(a) above a depth of 1.8d, and vice versa. Con-
sequently, deection, Fig. 10(b) and bending moment,
Fig. 10(c) are slightly overestimated up to a load level
of 380 kN, to which the total limiting force in the slip
zone reaches that for n = 1 case, Fig. 10(a). Thereafter,
an underestimation of each response is observed.
Case (IV). Employing an identical LFP to that of Case
(III), but taking e = 0, the predictions using the simple
expressions are slightly higher than those obtained
earlier.
5.2. A pipe pile tested in sti clay
A steel pipe pile was tested in a sti clay near Manor [7].
The pile was 14.9 m in length, and 0.641 m in diameter. The
moment of inertia, I
p
was 2.335 10
3
m
4
, and the exural
stiness, E
p
I
p
was equal to 493.7 MN m
2
. The undrained
shear strength, s
u
of the clay increases linearly from
25 kPa at the ground level to 333 kPa at depth 4.11 m.
The submerged unit weight, c
0
s
was 10.2 kN/m
3
. No infor-
mation was available about the a
c
, so it was back-estimated
as 545.1 by substituting k of 331.3 MPa [7] into Eq. (1).
L
c
/r
0
was estimated as 10.8, thus the pile was innitely
long. The p
u
prole was estimated by using Eq. (5c) with
J = 0.92, which was then modied using the depth factor
provided by Reese [7]. The LFP thus obtained can be
described by using Eqs. (4) and (5a) with N
g
= 0.961,
N
co
= 0.352, and n = 1.0, which is plotted in Fig. 11(a) as
Sti clay (J = 0.92). In presenting this p
u
/(ds
u
), however,
an average s
u
of 153 kPa was used as the normalisation
value. Using this LFP, good predictions of the pile
responses to a load level of 450 kN are achieved against
the measured data (see Fig. 11).
To improve the overall predictions, an excellent back-
estimation was undertaken which rendered the three param-
eters N
g
, N
co
, and n to be adjusted to 0.854, 0, and 1.5,
respectively. The n = 1.5 unfolded is quite close to n =
1.7 used for piles in sand, as is the strength (s
u
) prole.
The slip was initiated upon loading, thus the back-gured
LFP is rational. Bending moment proles were computed
for lateral loads of 179.7, 317.7, 485.6, and 606.2 kN, respec-
tively, using Eqs. (A-2) and (A-9) provided in Appendix I.
As depicted in Fig. 11(d), there is an excellent agreement
between the calculated and the measured proles, although
the transferring depths of the bending moments of 57 m
predicted are up to 1 m higher than the measured ones.
Using a constant k in the current solutions, the limiting
deection w
p
should increase with depth at a power, n of
1.5, starting with zero at mudline. In contrast, a linearly
increasing k is adopted in the COM624P. The eect of
these dierences was examined by assuming a conservative
k of 150x (MPa) [7] so that the average k over the maxi-
mum slip depth of 2.71 m is 203.3 MPa (a
c
= 348.4). Using
this k value, the predicted P w
g
is shown in Fig. 11(b) as
Dierent a
c
. Only slight overestimation of mudline deec-
tion is noted in comparison with those obtained from the
n = 1.5 case utilising k = 331.3 MPa (see Table 6). Thus,
the eect of k on the predictions is generally not obvious.
The pronounced overestimation [29] using Characteristic
load method (based on COM624P) for this case may thus
be attributed to the LFP.
6. Comments on use of the current solutions
The current solutions were used to predict response of
12 innitely long single piles tested in clay, and sand due
to lateral loads, or soil movements. In particular, using
the Matlocks LFP via Eq. (5c), good predictions were
made for three laterally loaded single piles tested [30] in
Shanghai clay and two single piles due to lateral soil move-
ment [31,32]. Good comparison with measured response
for a pile embedded in sand due to soil movement was also
noticed previously using Bartons LFP [32,33]. In particu-
lar, the analyses of 10 piles in clay showed that:
(a) N
g
= 0.34.79 (clay).
(b) N
co
= 04.67; [or a
0
k < 0.8 (all cases), and a
0
k < 0.3
(full-scale piles)].
(c) n = 0.52.0 with n = 0.50.7 for a uniform strength
prole, n = 1.31.7 for a linearly increase strength
prole (similar to sand).
(d) x
p
0:51:69 at maximum loads [or x
p
= (48.4)d].
(e) a
c
= 50340 (clay), and 556 (sti clay).
These magnitudes are consistent with those summarised
below for piles in clay:
(a) N
g
= 24 for n = 0 [34].
(b) N
co
= 2 for a smooth shaft [35], 3.57 for a rough shaft
[36], and 0.0 for a pile in sand.
(c) The values of n tentatively obtained by tting the
reported p
u
proles [19] with Eq. (4). Particularly,
the case of n > 1 is consistent with the theoretical
solution [37], and the upper bound method under-
taken for layered soil proles [8].
Table 5
Sensitivity of current solutions to k(N
p
), LFP and e (Fig.10)
Cases Limiting force proles Remarks References
(I) N
g
, N
co
and n provided in Table 1 Using the current LFP n = 1
(II) N
g
= 3K
p
, N
co
= 0, and n = 1.0 Using Broms LFP Broms LFP
(III) N
g
= 8.3, N
co
= 0, and n = 2 e = 0.31 m, & G
s
= 7.68 MPa n = 2
(VI) N
g
= 8.3, N
co
= 0, and n = 2 identical to (III) Using e = N
co
= N
p
= 0 n = 2(S. Eqs)
W.D. Guo / Computers and Geotechnics 33 (2006) 4767 61
(d) Finally, a
c
being equal to 80140 [26]; 210280 [13];
175360 [38], 330550 [39], and those summarised
previously [40].
The analyses also showed that each combination of n,
N
g
, and N
co
can produce a special LFP. The existing LFPs
such as those of Matlock, Reese (C) and (S) may work well
for relevant predictions. However, the special factors such
as a layered soil prole (see Fig. 9), an apparent cohesion
around a driven pile in sand (see Fig. 10) and so forth
can only be readily accommodated through the current
LFP. By means of an equivalent, homogeneous modulus
(b) (a)
(c)
(d)
Fig. 11. Comparison between the calculated and measured [7] response of the Manor test.
Table 6
Eect of elastic parameter a
c
on the Manor test (Fig. 11)
Cases Input parameters Calculated elastic parameters
n a
0
(m) A
L
(kN/m
n+1
) a
c
E
p
/G
*
k (MPa) N
P
/(2E
p
I
p
)
Sti clay
a
1.0 0.234
b
147.0 545.1 551.20 331.28 0.1683
n = 1.5
c
1.5 0 163.3 545.1 551.16 331.31 0.1683
Dierent a
c
1.5 0 163.3 348.4 862.44 203.30 0.1234
Notes for all cases: c
0
s
10:2 kN=m
3
.
a
LFP for sti clay (J = 0.92)
0
.
b
Corresponding P
e
is 24.5 kN.
c
Maximum x
p
= 4.23d, and x
p
1.73.
62 W.D. Guo / Computers and Geotechnics 33 (2006) 4767
for elastic state, and the generic LFP for plastic state, the
current solutions are suciently accurate for analysing
overall response of lateral piles in layered soil if a specic
distribution prole of limiting force is not a major concern.
The above-mentioned analysis, together with our recent
study on another 22 piles in clay indicates that
(a) k = (2.73.92)G
s
with an average of 3.04G
s
; G
s
= (25
340)s
u
with an average of 92.3s
u
.
(b) Given an equivalent, uniform strength prole, the
LFP may simply be obtained using n = 0.7, a
0
=
0.050.2 m (average of 0.11 m) and N
g
= 0.64.79
(1.6).
Study so far on 20 piles installed in sand shows that
(a) k = (2.43.7)G
s
with an average of 3.2G
s
; G
s
= (0.25
0.62)N (MPa) with an average of 0.5N (MPa).
(b) Given an equivalent uniform sand, it follows that
n = 1.7, a
0
= 0, and N
g
0:42:5K
2
p
.
The LFP can be more rigorously deduced from soil
strength parameters. All of these results are ready for
publication.
Finally, the current (CF) solutions are compared with
the numerical program COM624P, as can be seen from
Table 7. Although both are capitalised on load transfer
model, only the CF solutions are linked to soil modulus
via Eq. (1), which allow continuum-based pilesoil interac-
tion to be simulated. COM624P can incorporate various
forms of non-linear py curves, but the resulting overall
pile response is negligibly dierent from that obtained
using the current solutions. COM624P and the CF solu-
tions actually oer predictions for a linearly increasing,
and a uniform prole of k, respectively, which should
bracket non-homogeneous k normally encountered.
Reecting overall pilesoil interaction, only the parameters
for the LFP is dependent of mode of soil failure. To date, it
oers very good to excellent prediction of pile response in
comparison with measured data of 62 tested piles. Thus
the current solutions may be used as a boundary element
to simulate beamsoil interaction due to lateral force
within a complicated reinforced soil structure.
The current solutions are based on ve assumptions
mentioned previously.
Contradictory to the assumption (i), the py curve may
be a parabola [6] or a hyperbola [41]. A transition zone
(not depicted in Fig. 2(a)) may thus form in between the
upper plastic zone and the lower elastic zone. Neverthe-
less, adopting the idealised elasticplastic py(w) curve
instead, response of the pile is negligibly aected [13,42].
Assumption (ii) may lead to a 20% overestimation of
x
max
.
Assumption (iii) has negligible inuence, as slip gener-
ally occurs under a very low load level, but it extends
to a limited depth under a maximum load level.
Assumption (iv) is automatically satised, with the
introduction of assumption (i), as the latter renders the
transition zone to be reduced to a single slip depth
(see Fig. 2(a)).
Assumption (v) is ensured by two conditions: rstly, the
pile should be innitely long with L > L
c
, otherwise,
another slip may be initiated from a short, rigid pile base
at a rather high load level [43]. And secondly, Eq. (4) is
used.
Along with assumption (ii), Eq. (4) allows a gradual
increase in w
p
with depth. This is not always true in a strat-
ied soil, as a deeply embedded weak layer may have a
Table 7
Salient features of COM624P and the current CF solutions
Item COM624P [48] GASLFP (CF solutions)
Pilesoil interaction model Uncoupled, inconsistent with continuum-based numerical
analysis
Coupled, consistent with continuum-based numerical
analysis
Subgrade modulus, k Increase linearly with depth at a gradient of n
h
(k = n
h
x). A constant calculated from an average modulus, G
s
over
the maximum slip depth, x
p
and using Eq. (1)
Empirically related to soil properties Theoretically related to soil and pile properties, pile-head,
and base conditions
Limiting force per unit
length (LFP)
1. Many parameters, dierent expressions and procedures
for dierent soils
2. Parameters are mainly derived from soil failure modes
of wedge type and lateral plastic ow
1. Three parameters n, N
c
(or a
0
), and N
g
, a unied
expression of Eq. (4), and procedure for all kinds of
soils
2. Parameters are deduced from overall pile response,
regardless of mode of failure
py curve Consisting of four piecewise curves, dashed line in
Fig. 2(b)
An elasticperfectly plastic curve, solid line in Fig. 2(b)
Computation Finite dierence method Explicit expressions of the x
p
using spreadsheet program
GASLFP or by hand
Advanced use In form of numerical program; Other use has not been
specied
In form of explicit expressions; Reecting overall pilesoil
interaction by LFP, and indicating the eective depth by
x
p
. May be used as a boundary element for advanced
numerical simulation
W.D. Guo / Computers and Geotechnics 33 (2006) 4767 63
lower w
p
than a shallow, sti layer may. However, the use
of an overall LFP is suciently accurate as demonstrated
previously against 3D FEA results. This is also valid for
a very sti, upper layer, since in such a case, pilesoil rela-
tive slip may never extend to an underlying weak layer. Pile
response would be mainly aected by the sti layer. In
brief, assumption (v) is generally acceptable. Deriving from
the normal and shear stresses, respectively, on the pilesoil
interface [44,45], the resistance in elastic state may be su-
ciently accurately evaluated using elastic theory [5], and in
the plastic (slip) state by Eq. (4).
In rare cases, the non-homogeneous modulus may aect
markedly the pile response, for which the previous numer-
ical results [46,47] may be consulted along with the current
predictions.
7. Conclusions
This paper put forwards new elasticplastic solutions for
laterally loaded, innitely long, free-head piles. They have
been calibrated against FEA results for a pile in two dier-
ent types of stratied soils. The solutions permit non-linear
response of the piles to be readily estimated right up to fail-
ure. Presented in explicit expressions of slip depth and
LFP, the current solutions may be used as a boundary ele-
ment to represent beamsoil interaction in the context of
analysing a complicated soilstructure interaction. In terms
of analysis of 62 pile tests to date, the ranges of input
parameters are provided. In particular the following con-
clusions are drawn:
The generic expression of Eq. (4) is applicable to all
types of soils. It can generally accommodate existing
LFPs through selecting a suitable set of parameters.
Response of free-head piles is dominated by the LFP and
the maximum slip depth. Thus, non-linear predictions
may be made by selecting a series of slip depth x
p
, using
GASLEP or the simplied expressions provided.
By maintaining total resistance within a maximum slip
depth x
p
, pile response is insensitive to the shape of
the LFP. Available, veried procedures may be used
to construct LFP for current solutions.
To generate LFP, the new procedure proposed herein
along with n = 0.52.0 may be used. A low value of n
may correspond to a uniform strength prole, and a
high one to a sharply changed strength prole. For a
layered soil, the generated LFP may not necessarily
reect a detailed distribution prole of limiting force
along a pile but an overall trend.
LFP should be back-estimated using current solutions
along with measured data, as it then can account for
overall pilesoil interaction rather than sole soil failure
mechanism. In this manner, LFP may be updated to
cater for various inuence factors.
The current study has been limited to static loading, and
linear elastic, free-head piles, but it can be extended to
complicated loading. The current solutions have been used
successfully for non-linear piles. They have been extended
to xed-head piles as well. All of these results will be pub-
lished in due course.
Acknowledgments
The work reported herein was sponsored by Australian
Research Council research fellowship (F00103704) and
Discovery Grant (DP0209027). It was initiated in Singa-
pore through the sponsor of the (Singapore) National Sci-
ence and Technology Board. This nancial assistance is
gratefully acknowledged. The author also would like to
acknowledge the reviewers comments.
Appendix I. Development of elasticplastic solutions
In this appendix, derivation of the elasticplastic solu-
tions for the pile mentioned in the paper is elaborated.
All of the symbols used are of identical meanings to those
dened in the paper.
Integrating Eq. (6) for plastic state yields expressions for
shear force, Q
A
(x), bending moment, M
A
(x), rotation,
w
0
A
x, and deection, w
A
(x) of the pile at depth x, as
detailed below
Q
A
x E
p
I
p
w
000
A
x
A
L
x a
0

n1
a
n1
0
n 1

P
A
L
_ _
A-1
M
A
x E
p
I
p
w
00
A
x
A
L
x a
0

n2
a
n2
0

2
j1
n j

a
n1
0
n 1

P
A
L
_ _
x
Pe
A
L
_ _
A-2
w
0
A
x
A
L
E
p
I
p
x a
0

n3

3
j1
n j

a
n1
0
n 1

P
A
L
_ _
x
2
2
_

a
n2
0

2
j1
n j

Pe
A
L
_ _
x
_
C
3
A-3
w
A
x
A
L
E
p
I
p
x a
0

n4

4
j1
n j

a
n1
0
n 1

P
A
L
_ _
x
3
6
_

a
n2
0

2
j1
n j

Pe
A
L
_ _
x
2
2
_
C
3
x C
4
A-4
where C
3
, C
4
= constants. In the integration, the condi-
tions for a free-head pile (at x = 0) are adopted as follows
Q
A
0 E
p
I
p
w
00
A
0 P; M
A
0 E
p
I
p
w
00
A
0
Pe A-5
Eq. (7) for the elastic state may be solved as
(N
p
< 2

kE
p
I
p
_

w
B
z e
az
C
5
cos bz C
6
sin bz A-6
64 W.D. Guo / Computers and Geotechnics 33 (2006) 4767
whereas C
5
, C
6
= constants; a, b are given by Eqs.
(9a),(9b). From Eq. (A-6), it follows that
w
IV
B
z e
az
a
4
6a
2
b
2
b
4
C
5
4aba
2
b
2
C
6
cos bz
e
az
4aba
2
b
2
C
5
a
4
6a
2
b
2
b
4
C
6
sinbz
A-7
w
000
B
z e
az
aa
2
3b
2
C
5
b3a
2
b
2
C
6
cos bz
e
az
3a
2
b
2
bC
5
aa
2
3b
2
C
6
sinbz
A-8
w
00
B
z e
az
fa
2
b
2
C
5
2abC
6
cos bz
2abC
5
a
2
b
2
C
6
sinbzg A-9
w
0
B
z e
az
aC
5
bC
6
cos bz bC
5
aC
6
sinbz
A-10
The constants C
i
(i = 36) are determined using the com-
patibility conditions at the slip depth x = x
p
(z = 0) from
elastic to plastic state, which require
w
IV
A
x x
p
w
IV
B
z 0 w
IV
P
A-11
w
000
A
x x
p
w
000
B
z 0 w
000
P
A-12
w
00
A
x x
p
w
00
B
z 0 w
00
P
A-13
w
0
A
x x
p
w
0
B
z 0 w
0
P
A-14
w
A
x x
p
w
B
z 0 w
P
A-15
With Eq. (A-8), Eq. (A-12) may be written as an expression
of unknown constants C
5
and C
6
. In terms of Eq. (A-9),
Eq. (A-13) oers another expression for the two constants.
Solution of these two expressions oers
C
5

E
p
I
p
k
2aw
000
P
3a
2
b
2
w
00
P
A-16
C
6

E
p
I
p
kb
a
2
b
2
w
000
P
aa
2
3b
2
w
00
P
A-17
Utilising Eqs. (A-3) and (A-10), Eq. (A-14) can be ex-
panded, thus C
3
is determined as
C
3

4A
L
k
1n
k
F 3; x
p
F 2; 0x
p
F 1; 0 P
x
2
p
2
Pex
p
_ _
aC
5
bC
6
A-18
In the same manner, Eqs. (A-4) and (A-6) allow
Eq. (A-15) to be expanded, which gives C
4
as
C
4

4A
L
kk
n
F 4; x
p
F 1; 0 P
x
3
p
6
F 2; 0 Pe
x
2
p
2
_ _
C
3
x
p
k
C
5
A-19
Substituting Eqs. (A-16) and (A-17) into Eq. (A-18), a
normalised C
3
is derived as
C
3
kk
n1
A
L
4F 3; x
p
x
p
F 2; 0 4a
N
F 2; x
p
F 2; 0
2F 1; x
p
21 2a
N
x
p
x
2
p
F 1; 0
2x
2
p
1 2a
N
x
p
2a
N
x
p
eP A-20
In light of Eqs. (A-17) and (A-20), the normalised form of
C
4
is obtained
C
4
kk
n
A
L
4F 4; x
p
x
p
F 3; x
p
41 a
2
N
F 2; x
p

F 2; 0 2x
2
p
F 2; 0 2x
p
1 2a
2
N
F 1; x
p

1 2a
N
x
p
F 0; x
p
4x
2
p
=3 2x
p
F 1; 0
4x
3
p
=3 2x
p
2x
2
p
4 4a
2
N
eP A-21
At ground level, Eqs. (A-3) and (A-4) permit the rotation,
w
0
g
and deection, w
g
to be expressed, respectively, as
w
0
g
kk
n1
A
L
4F 3; 0
C
3
kk
n1
A
L
A-22
w
g
kk
n
A
L
4F 4; 0
C
4
kk
n
A
L
A-23
Eq. (A-23) has been rewritten as Eq. (11). Substituting
Eq. (A-6) into Eqs. (A-14), (A-15), the normalised rota-
tion and deection at the slip depth are written, respec-
tively, as,
w
0
p
kk
n
A
L
a
C
5
kk
n
A
L
b
C
6
kk
n
A
L
A-24
w
p
kk
n
A
L

C
5
kk
n
A
L
A-25
Eqs. (A-11)(A-13) render the following relationship at the
slip depth to be established
0:5w
IV
P
aw
000
P
k
2
w
00
P
0 A-26
In terms of Eqs. (6), (14) and (15), Eq. (A-26) can be re-
written in the explicit form of Eq. (8), correlating the load
to the normalised slip depth.
The slip depth under a given load may be computed
using Eq. (8), which is then used to calculate the pile
responses. Particularly, the mudline rotation of the pile
may be predicted using Eq. (A-22), as it becomes impor-
tant to predictions of pile response due to soil movement.
Responses of the pile along its length are predicted sepa-
rately using elastic and plastic solutions. Within plastic
state of x < x
p
, Eqs. (A-4), (A-3), (A-2), and (A-1) are
used for deection, rotation, moment, and shear force,
respectively. Otherwise, in the elastic state of x Px
p
,
Eqs. (A-27)(A-30) should be employed, which were
derived from Eqs. (A-7)(A-10) using Eqs. (A-16) and
(A-17).
w
B
z
e
az
E
p
I
p
k
_
2aw
000
P
3a
2
b
2
w
00
P
cos bz:

a
2
b
2

b
w
000
P

a
b
a
2
3b
2
w
00
P
_ _
sin bz
_
A-27
w
0
B
z
e
az
2k
2
_
w
000
P
2aw
00
P
cos bz

aw
000
P
a
2
b
2
w
00
P
b
_ _
sin bz
_
A-28
W.D. Guo / Computers and Geotechnics 33 (2006) 4767 65
M
B
z E
p
I
p
w
00
B
z
E
p
I
p
e
az
w
00
P
cos bz
w
000
P
aw
00
P
b
_ _
sin bz
_ _
A-29
Q
B
z E
p
I
p
w
000
B
z
E
p
I
p
e
az
w
000
P
cos bz
aw
000
P
2k
2
w
00
P
b
sin bz
_ _
A-30
At depth z
max
, the shear force, Q
B
(z
max
) is zero, and the
maximum bending moment should occur. Therefore, Eq.
(A-30) allows the function wx
p
to be dened as Eq.
(13), and z
max
to be derived as Eq. (16), respectively.
Appendix II. Simplied expressions (at a
0
= 0, and a
N
= 1)
Provided that a
0
= 0, and a
N
= 1 (i.e., b
N
= 1), Eq. (*)
provided in the paper may be simplied to Eq. (*s) given
below. Thus, for instance, Eq. (8) may be replaced with
Eq. (8s).
Pk
n1
A
L

0:5x
n
p
n 1n 2 2x
p
2 n x
p

x
p
1 en 1n 2
8s
w
g
kk
n
A
L

2
3
x
n3
p
2x
2
p
2n 10x
p
n
2
9n 20
x
p
1 en 2n 4

2x
2
p
2x
p
1x
n
p
x
p
1 e

2x
4
p
n 4x
p
12x
2
p
n 1x
p
1
x
p
1 en 1n 4
ex
n
p
11s
wx
p

2x
p
2 ne 21 nx
2
p
n 2n 1
2n 2x
p
1 ne 2x
2
p
22 nx
p
n 2n 1
13s
Under the above-mentioned conditions, the constants
C
1
x
p
, and C
2
x
p
in Eq. (17) may be replaced with
C
1
x
p

0:5x
n
p
x
p
1 e
1 n 2x
p
1 n
e
2 n 2x
p
2 n
x
p
_ _
C
2
x
p
0:5x
n
p
A-31
From Eq. (13s), the normalised slip depth, x
p
(rewritten as
x
0
) at wx
0
0 is obtained as
x
0

0:52 n
1 n
e 0:5

2 n
1 n
e
2
2n 2
_
A-32
The condition of x
p
> x
0
will lead to wx
p
< 0, thus maxi-
mum bending moment should occur above the slip depth.
The estimation using above-mentioned expressions is
referred to as S. Eqs. It can be readily undertaken in a
spreadsheet, similar to the form shown in Table 4 for pile
A. Here provides the calculation for two typical values of
x
p
. From Table 2(a), k = 0.2591/m, given x
p
= 1 m, x
p
is
computed to be 0.2591. Using n = 0.5 (Table 1), and
e = 0, x
0
is computed as 1.118, thus
As x
p
< x
0
wx
p
> 0, the maximum bending moment
should occur below the slip depth. Substituting
wx
p
0:602, and b = k into Eq. (16), z
max
is computed
to be 2.091 m, thus, tan(k z
max
) = 0.602, and cos(kz
max
) =
0.8567. Also from Eq. (A-31)
C
1
x
p

0:5 2:5 2 0:2591 0:2591
1:5
2:5 1:2591
0:06323;
C
2
x
p
0:5 0:2591
0:5
0:25451
These values of C
1
and C
2
allow the normalised moment to
be estimated using Eq. (17) as
M
B
z
max
k
n2
=A
L
e
0:54188
0:06323 0:25451 0:602
0:8567 0:10785
Assuming x
p
= 5 m, and e = 0, x
p
is found to be 1.2957.
Following the above-mentioned procedures, it follows that
Pk
1.5
/A
L
= 0.8982, and w
g
kk
0.5
/A
L
= 6.1526. As x
p
> x
0
,
the depth of x
max
may be obtained directly from Eq. (18) as
x
max
1:5 0:8982
1=1:5
=0:2591 4:71m
The moment, M
0
(=P e) is zero. Thus, the normalised
maximum moment may be estimated from Eq. (19) as
M
max
k
n2
=A
L
1:5 0:8982
2:5=1:5
=2:5 0:6574
References
[1] Reese LC. Soil modulus for laterally loaded piles. Trans ASCE
1958;123:10714.
[2] Matlock H, Reese LC. Generalized solutions for laterally loaded
piles. J Soil Mech Found Eng Div ASCE 1960;86(5):6391.
wx
p

2 1:5 0:2591
2
2:5 1:5
2 0:2591
2
2 2:5 0:2591 2:5 1:5
0:6020
Pk
1:5
A
L

0:5 0:2591
0:5
1:5 2:5 2 0:2591 2:5 0:2591
1:2591 1:5 2:5
0:2792
w
g
kk
0:5
A
L

2
3
0:2591
3:5
2 0:2591
2
2 0:5 10 0:2591 0:5
2
9 0:5 20
1:2591 2:5 4:5

2 0:2591
2
2 0:2591 1 0:2591
0:5
1:2591
0:6796
66 W.D. Guo / Computers and Geotechnics 33 (2006) 4767
[3] Reese LC. Laterally loaded piles: program documentation. J Geotech
Eng ASCE 1977;103(4):287305.
[4] Yang Z, Jeremic B. Numerical analysis of pile behaviour under lateral
load in layered elasticplastic soils. Int J Numer Anal Meth Geomech
2002;26(14):1385406.
[5] Guo WD, Lee FH. Load transfer approach for laterally loaded piles.
Int J Numer Anal Meth Geomech 2001;25(11):110129.
[6] Matlock H. Correlations for design of laterally loaded piles in soft
clay. In: Proceedings of the 2nd annual oshore technology confer-
ence OTC 1204, Oshore Technology Conference. Dallas, Texas
1970.
[7] Reese LC, Cox WP, Koop FD. Field testing and analysis of laterally
loaded piles in sti clay. In: Proceedings of the 7th annual oshore
technology conference OTC. 2312, Oshore Technology Conference.
Dallas, Texas 1975.
[8] Mur JD, Hamilton JM. P-Ultimate for undrained analysis of
laterally loaded piles. J Geotech Eng ASCE 1993;119(1):91107.
[9] Ashour M, Norris G. Modeling lateral soilpile response based on
soilpile interaction. J Geotech Geoenviron Eng Div ASCE
2000;126(5):4208.
[10] Broms B. Lateral resistance of piles in cohesionless soils. J Soil Mech
Found Eng Div ASCE 1964;90(3):12356.
[11] Reese LC, Cox WR, Koop FD. Analysis of laterally loaded piles in
sand. In: Proceedings of the 6th annual oshore technology conference
OTC. 2080, Oshore Technology Conference. Dallas, Texas 1974.
[12] ReeseLC, AllenJD, HargroveJQ. Laterallyloadedpiles inlayeredsoils.
In: Proceedings of the 10th international conference on soil mechanics
and foundation engineering. Stockholm: A.A. Balkema; 1981.
[13] Poulos HG, Hull TS. The role of analytical geomechanics in
foundation engineering. In: Foundation engineering: current princi-
ples and practices. Evanston, IL: ASCE; 1989.
[14] Scott RF. Foundation analysis. NJ: Prentice-Hall, Englewood Clis;
1981.
[15] Alem A, Gherbi M. Graphs for the design of laterally loaded piles in
clay. In: Proceedings of the international conference on geotechnical
and geological engineering, GeoEng2000. Melbourne, Australia 2000.
[16] Randolph MF. The response of exible piles to lateral loading.
Geotechnique 1981;31(2):24759.
[17] Guo WD. Subgrade modulus for laterally loaded piles. In: Proceed-
ings of the 8th international conference on civil and structural
engineering computing. Eisenstadt, NR Vienna, Austria: Civil-Comp
Press, Stirling, United Kingdom; 2001.
[18] Randolph MF, Houlsby GT. The limiting pressure on a circular pile
loaded laterally in cohesive soil. Geotechnique 1984;34(4):61323.
[19] Briaud JL, Smith TD, Meyer B. Laterally loaded piles and the
pressuremeter: comparison of existing methods. In: Langer JA,
Mosley ET, Thompson CD, editors. Laterally loaded deep founda-
tions: analysis and performance. ASTM, STP 835 Kansas City, MO,
1983.
[20] Guo WD, Zhu BT. Laterally loaded xed-head piles in sand. In:
Proceedings of the 9th ANZ conference on geomechanics. Auckland,
New Zealand: New Zealand Geotechnical Society Inc.; 2004.
[21] Yang Z, Jeremic B. Study of soil layering eects on lateral loading
behaviour of piles. J Geotech Geoenviron Eng Div ASCE
2005;131(6):76270.
[22] Hetenyi M. Beams on elastic foundations. Ann Arbor: University of
Michigan Press; 1946.
[23] Guo WD. Nonlinear behaviour of laterally loaded xed-head piles.
Int J Numer Anal Meth Geomech [under review].
[24] Rajani BB, Morgenstern NR. Pipelines and laterally loaded piles in
elastoplastic medium. J Geotech Eng ASCE 1993;119(9):143147.
[25] Ito T, Matsui T, Hong WP. Extended design method for multi-row
stabilizing piles against landslide. Soils Found 1981;21(1):2137.
[26] Poulos HG, Davis EH. Pile foundation analysis and design. New
York: Wiley; 1980.
[27] Georgiadis M. Development of py curves for layered soil. In:
Proceedings of the geotechnical practice in oshore engineering,
ASCE 1983.
[28] Kishida H, Nakai S. Large deection of a single pile under horizontal
load. In: Proceedings of the 9th International Conference on Soil
Mechanics and Foundation Engineering. Speciality session 10.
Tokyo, 1977.
[29] Duncan JM, Evans LT, Ooi PSK. Lateral load analysis of single piles
and drilled shafts. J Geotech Eng ASCE 1994;120(5):101833.
[30] Wu D, Broms BB, Choa V. Design of laterally loaded piles in
cohesive soils using py curves. Soils Found 1999;38(2):1726.
[31] Guo WD. A simplied approach for piles due to soil movement. In:
Proceedings of the 12th Panamerican conference on soil mechanics
and geotechnical engineering. Cambridge, MA, USA: Verlag Glu-
ckauf GMBH, Essen (Germany); 2003.
[32] Guo WD, Ghee EH. Response of free-head piles due to lateral soil
movement. In: Proceedings of the 9th ANZ conference on geome-
chanics. Auckland, New Zealand: New Zealand Geotechnical Society
Inc.; 2004.
[33] Barton YO. Laterally loaded model piles in sand: centrifuge tests and
nite element analysis. University of Cambridge; 1982.
[34] Viggiani C. Ultimate lateral load on piles used to stabilise landslide.
In: Proceedings of the 10th International Conference Soil Mechanics
and Foundation Engineering, 1981. Stockholm, Sweden.
[35] Fleming WGK, Weltman AJ, Randolph MF, Elson WK. Piling
engineering. 2nd ed. New York: Surrey University Press, Glasgow,
Halst Press; 1992.
[36] Mayne PW, Kulhawy FH. Loaddisplacement behaviour of laterally
loaded rigid shafts in clay. In: Proceedings of the 4th international
conference on piling and deep found. Stresa, Italy: A.A. Balkema,
Rotterdam; 1991.
[37] Hansen BJ. The ultimate resistance of rigid piles against transversal
forces. The Danish Geotechnical Institute Bulletin No. 12: Copenha-
gen, Denmark, 1961. p. 59.
[38] DAppolonia DJ, Lambe TW. Performance of four foundations on
end-bearing piles. J Soil Mech Found Eng Div ASCE
1971;97(1):7793.
[39] Budhu M, Davies TG. Analysis of laterally loaded piles in soft clays. J
Geotech Eng ASCE 1988;114(1):2139.
[40] Kulhawy FH, Mayne PW. Manual on estimating soil properties for
foundation design. In: Electric Power Research Institute: Palo Alto,
Califoria. Report EL-2870. 1990.
[41] Jimiolkwoski M, Garassino A. Soil modulus for laterally loaded piles.
In: Proceedings of the 9th international conference soil mechanics and
foundation engineering. Speciality session 10. Tokyo, 1977.
[42] Castelli F, Maugeri E, Motta ME. Discussion on Design of laterally
loaded piles in cohesive soils using py curves. by Wu D, Broms BB,
Choa V, 1998. Soils Found 1999;39(6):1334.
[43] Guo WD. Response of laterally loaded rigid piles. In: Proc 12th
Panamerican conference on soil mechanics and geotechnical engi-
neering. Cambridge, MA, USA: Verlag Gluckauf GMBH, Essen
(Germany); 2003.
[44] Baguelin F, Frank R, Said YH. Theoretical study of lateral reaction
mechanism of piles. Geotechnique 1977;27(3):40534.
[45] Briaud JL, Smith TD, Tucker LM. A pressuremeter method for
laterally loaded piles. In: Proceedings of the 10th international
conference on soil mechanics and foundation engineering. San
Francisco: A.A. Balkema; 1985.
[46] Davisson MT, Gill HL. Laterally-loaded piles in a layered soil system.
J Soil Mech Found Eng Div ASCE 1963;89(3):6394.
[47] Pise P. Laterally loaded piles in a two-layer soil system. J Geotech
Eng ASCE 1982;108(9):117781.
[48] FHWA, COM624P-laterally loaded pile analysis program for the
microcomputer, in No. FHWA-SA-91-048. Washington (DC), USA,
1993.
W.D. Guo / Computers and Geotechnics 33 (2006) 4767 67

Anda mungkin juga menyukai