Load-Settlement Response of Rectangular and Circular Piles

in Multilayered Soil
H. Seo, S.M.ASCE1; D. Basu, A.M.ASCE2; M. Prezzi, A.M.ASCE3; and R. Salgado, M.ASCE4

Abstract: Traditionally, analyses developed for circular piles have also been used for rectangular piles by replacing in calculations the
rectangular pile with a circular pile of equivalent area. In this paper, we present a settlement analysis that applies to piles with either
rectangular or circular cross sections installed in multilayered soil deposits. The analysis follows from the solution of the differential
equations governing the displacements of the pile-soil system obtained using variational principles. The input parameters needed for the
analysis are the pile geometry and the elastic constants of the soil and pile. Pile displacements and vertical soil displacements calculated
using this analysis match well those from finite-element analysis. A parametric study highlights some important insights for rectangular
and circular piles in multilayered soil. A user-friendly spreadsheet program 共ALPAXL兲 was developed to facilitate the use of the analysis.
Examples illustrate the use of the analysis in design.
DOI: 10.1061/共ASCE兲1090-0241共2009兲135:3共420兲
CE Database subject headings: Elastic analysis; Piles; Settlement; Layered soils; Variational principles.

Introduction and Davis 1968; Mattes and Poulos 1969; Butterfield and Baner-
jee 1971; Poulos 1979; Rajapakse 1990; Lee and Small 1991兲.
Most of the pile settlement analyses available in the literature Recently, a continuum-based analysis that produces pile dis-
共Poulos and Davis 1968; Randolph and Wroth 1978; Guo and placements and soil displacements using closed-form solutions
Randolph 1997; Guo 2000兲 apply to piles with circular cross sec- was developed for circular piles 共Vallabhan and Mustafa 1996;
tions; however, these analyses are often used in the design of piles Lee and Xiao 1999; Seo and Prezzi 2007兲. The analysis, which is
with rectangular cross sections by converting the rectangular pile based on energy principles, was extended to piles with rectangular
cross section to an equivalent circle. The available settlement cross section 共Basu et al. 2008兲. The advantage of this continuum-
analyses for circular piles either assume that the soil resistance based analysis is that it captures the three-dimensional nature
can be represented by a series of independent springs 共the spring of the pile-soil interaction and produces pile load-settlement
stiffness is determined through theoretical, experimental, or em- responses in seconds. In this paper, we compare the meth-
pirical means兲 or treat the soil as a continuum. The approach with odologies developed for both circular and rectangular piles and
the corresponding load-settlement responses. We also perform a
springs 共Seed and Reese 1957; Coyle and Reese 1966; Murff
parametric study and highlight some important aspects of the
1975; Randolph and Wroth 1978; Kraft et al. 1981; Armaleh and
load-settlement response of rectangular and circular piles in mul-
Desai 1987; Kodikara and Johnston 1994; Motta 1994; Guo and
tilayered soil. The results presented in this paper are applicable to
Randolph 1997; Guo 2000兲 has the advantage that approximate
the initial stages of pile loading because the analysis is based on
analytical or simple numerical solutions of pile settlement can be
linear elasticity. We provide design examples that illustrate the
obtained 共Randolph and Wroth 1978; Armaleh and Desai 1987;
applicability of the analysis in practical problems.
Motta 1994兲. The continuum approach, although more appealing
conceptually, has traditionally required expensive numerical tech-
niques, such as the boundary integral method, the finite layer Analysis
method, or the finite-element method, to obtain solutions 共Poulos
Problem Definition
1
Ph.D. Candidate, School of Civil Engineering, Purdue Univ., 550
Stadium Mall Dr., West Lafayette, IN 47907. E-mail: seoh@purdue.edu The analysis considers a single pile embedded vertically into a
2
Assistant Professor, Dept. of Civil and Environmental Engineering, multilayered elastic soil deposit 共Fig. 1兲. There are altogether N
Univ. of Connecticut, 261 Glenbrook Rd., Unit 2037, Storrs, CT 06269. discrete soil layers, and the bottom 共base兲 of the pile rests at the
E-mail: dbasu@engr.uconn.edu interface of the mth and 共m + 1兲th layer 共with m ⬍ N兲. The pile has
3
Associate Professor, School of Civil Engineering, Purdue Univ., 550 a length L p and is subjected to an axial force Qt at the pile head
Stadium Mall Dr., West Lafayette, IN 47907 共corresponding author兲. 共the pile head is assumed to be flush with the ground surface兲.
E-mail: mprezzi@purdue.edu The pile can be of circular or rectangular cross section. A Carte-
4
Professor, School of Civil Engineering, Purdue Univ., 550 Stadium
sian 共x-y-z兲 coordinate system is used for rectangular piles, and a
Mall Dr., West Lafayette, IN 47907. E-mail: rodrigo@purdue.edu
cylindrical 共r-␪-z兲 coordinate system is used for circular piles. In
Note. Discussion open until August 1, 2009. Separate discussions
must be submitted for individual papers. The manuscript for this paper both cases, the z axis coincides with the pile axis, and the positive
was submitted for review and possible publication on November 28, z direction points downward. The rectangular pile cross section is
2007; approved on April 28, 2008. This paper is part of the Journal of described by its dimensions Bx and By in the x and y directions,
Geotechnical and Geoenvironmental Engineering, Vol. 135, No. 3, respectively, while the circular cross section is described by its
March 1, 2009. ©ASCE, ISSN 1090-0241/2009/3-420–430/$25.00. diameter Br 共=2r p, where r p is the pile radius兲.

420 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / MARCH 2009

Downloaded 11 Jun 2012 to 180.211.192.67. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
 Soil Displacement
The vertical displacement uz at any point within the soil mass
共caused by settlement of the pile兲 is assumed to be a product of
separable variables; for rectangular piles, there are three functions
u共z兲, ␾x共x兲, and ␾y共y兲 accounting for each of the three directions;
for circular piles, there are two functions u共z兲 and ␾r共r兲 account-
ing for the vertical and radial directions 共there is no tangential
variation of soil displacement for axially loaded circular piles兲.
For rectangular piles, uz is given by

uz共x,y,z兲 = u共z兲␾x共x兲␾y共y兲 共2兲

while, for circular piles, it is given by

uz共r,z兲 = u共z兲␾r共r兲 共3兲

where uz共x , y , z兲 and uz共r , z兲 = vertical displacements within the
soil mass at a point 共x , y , z兲 or 共r , z兲 caused by the displacements
Fig. 1. Axially loaded pile with rectangular or circular cross section of the rectangular or circular pile, respectively; u共z兲
in multilayered soil = displacement function that describes the axial pile displacement
with depth; and ␾x共x兲, ␾y共y兲, and ␾r共r兲 = dimensionless displace-
ment functions varying along x, y, and r, respectively; these ␾
Hi denotes the vertical distance from the ground surface to the functions describe the decrease in the soil displacement with in-
bottom of any soil layer i 共the subscript i denotes the ith layer兲; creasing horizontal distance from the pile axis. All the ␾ func-
thus, the thickness of layer i is given by Hi − Hi−1 with H0 = 0. All tions are assumed to be equal to one at the pile-soil interface. This
soil layers extend to infinity in the horizontal direction, and the ensures proper pile-soil contact. The displacements in the soil
bottom 共Nth兲 layer extends to infinity downward in the vertical must vanish at infinite horizontal distances from the pile; there-
direction. The soil medium is assumed to be elastic and isotropic, fore, ␾x共x兲, ␾y共y兲, and ␾r共r兲 are assumed to be zero at x = ⫾ ⬁,
homogeneous within each layer, with elastic properties described y = ⫾ ⬁, and r = ⬁, respectively. Soil displacements in the horizon-
by Lame’s constants ␭si and Gsi. The pile is assumed to behave as tal directions 共i.e., ux and uy or ur兲 are considered negligible in
an elastic column 共i.e., an elastic axial compression element兲 with this analysis.
Young’s modulus E p. There is no slippage or separation between
the pile and the surrounding soil or between the soil layers. The Principle of Minimum Potential Energy
horizontal soil displacements in the soil mass due to the axial load
With the assumed displacement fields 关Eqs. 共2兲 or 共3兲兴, strains are
Qt are neglected in the analysis since, in general, these are small
calculated and subsequently related to stresses using elasticity
compared with the vertical soil displacements.
theory. Thereafter, the soil potential energy density is expressed in
terms of the elastic constants and strains. Since the strains can be
Potential Energy expressed in terms of u共z兲, ␾r共r兲, du共z兲 / dz, and d␾r共r兲 / dr 共for
Since the pile-soil system is assumed to be elastic in this analysis, circular piles兲 or u共z兲, ␾x共x兲, ␾y共y兲, and the derivatives of u共z兲,
the principle of minimum potential energy can be used to obtain ␾x共x兲, and ␾y共y兲 共for rectangular piles兲, the expression of the
the equilibrium configuration of the system. The principle of potential energy ⌸ contains these functions and their derivatives
minimum energy states that a conservative system attains its equi- 共Seo and Prezzi 2007; Basu et al. 2008兲. Applying the principle of
librium configuration when its total potential energy is at a mini- minimum potential energy 共i.e., ␦⌸ = 0兲 yields an equation of the
mum. The total potential energy ⌸ of the pile-soil system, form
including both the internal and external energies, is given by
A共u兲␦u + B共␾x兲␦␾x + C共␾y兲␦␾y = 0 共4a兲

冕冉 冊 冕
Lp 2 for rectangular piles, and of the form
1 du 1
⌸ = E pA p dz + ␴ pq␧ pqdV − Qt兩u兩z=0 共1兲
2 dz 2
0 R A共u兲␦u + D共␾r兲␦␾r = 0 共4b兲
where A p = area of the pile cross section 共equal to ␲r2p for circular for circular piles. Since the functions u, ␾x, ␾y, and ␾r are not
piles and to BxBy for rectangular piles兲; u = u共z兲 = vertical pile dis- known a priori, their variations ␦u, ␦␾x, ␦␾y, and ␦␾r are not
placement at a depth z; ␴ pq and ␧ pq = stress and strain tensors in zero. Therefore, Eqs. 共4a兲 and 共4b兲 are satisfied if and only if the
the soil 共summation is implied by the repetition of the indices p differential equations: A共u兲 = 0, B共␾x兲 = 0, C共␾y兲 = 0, and D共␾r兲
and q in the ␴ pq␧ pq term兲; and R = region occupied by the soil = 0 are satisfied. These equations represent the governing differ-
mass surrounding the pile, over which the integration of the soil ential equations of the functions u共z兲, ␾x共x兲, ␾y共y兲, and ␾r共r兲,
potential energy density ␴ pq␧ pq / 2 is performed. For circular piles, respectively; the equilibrium configuration of the pile-soil system
the limits of integration for the integrand ␴ pq␧ pq / 2 are z = 0 to z is obtained by solving these equations.
= ⬁; r = r p to r = ⬁ 共for 0 艋 z 艋 L p兲 and r = 0 to r = ⬁ 共for L p 艋 z
⬍ ⬁兲; and ␪ = 0 to ␪ = 2␲. For rectangular piles, the integration
Soil Displacement Decay Functions
limits are z = 0 to z = ⬁, x = −⬁ to x = ⬁ and y = −⬁ to y = ⬁, with
subsequent subtraction of the energy in the region 共0 艋 z 艋 L p, For rectangular piles, solution of the differential equation of ␾x
−Bx / 2 艋 x 艋 Bx / 2, −By / 2 艋 y 艋 By / 2兲 in which the pile is placed. yields 共Basu et al. 2008兲

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / MARCH 2009 / 421

Downloaded 11 Jun 2012 to 180.211.192.67. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
 冋 冉 冊册
冦 冧
of the pile in the circular pile analysis. The soil displacements in
␥x Bx Bx the case of rectangular piles exhibit an exponential decay, while
exp x+ for − ⬁ ⬍ x 艋 −
rq 2 2 those for circular piles follow a decay described by a modified
Bx Bx Bessel function. The parameters ␥x, ␥y, and ␥r, present in the
␾x = 1 for − 艋x艋 共5兲 exponential and Bessel functions, describe the rate at which the
2 2

冋 冉 冊册
displacement in the soil decreases with horizontal distance from
␥x Bx Bx the pile axis; the greater the values of these parameters are, the
exp − x− for 艋x⬍⬁
rq 2 2 faster the rate of decay is.
where the dimensionless parameter ␥x is given by

冑 冉 冊 冉 冊
Pile Displacement
␥y rq
ms + ns + By The form of the differential equation governing pile displacement
␥x rq ␥y

冉 冊
= 共6兲 u is the same for both the rectangular and circular piles
rq rq
ms + By d 2u i
␥y − 共EiAi + 2ti兲 + k iu i = 0 共14兲
dz2
with
where Ai = A p; Ei = E p for i = 1 . . . m; Ei = ␭si + 2Gsi for i = m + 1 . . . N;
rq =
1
冑BxBy 共7兲 and

冋冉 冊冉 冊 册
2
␥x rq rq ␥y
ki = Gsi + By + + Bx 共15兲

冕 冕 rq ␥ y ␥x
⬁ N Hi rq
ms = Gs 2
u dz = 兺 Gsi u2i dz 共8兲
0 i=1 Hi−1 1
ti = r2q共␭si + 2Gsi兲
1
冉+
By
+
Bx
␥ x␥ y ␥ xr q ␥ y r q
冊 共16兲

冕冉 冊 冕 冉 冊
2
⬁ 2 N Hi 2
du dui
ns = 共␭s + 2Gs兲
0 dz
dz = 兺
i=1
共␭si + 2Gsi兲
Hi−1 dz
dz for rectangular piles, and
关K1共␥r兲 + ␥rK0共␥r兲兴2 − 共␥r2 + 1兲关K1共␥r兲兴2
共9兲 ki = ␲Gsi 共17兲
关K0共␥r兲兴2
Similarly, for the y direction, we get

冋 冉 冊册
冦 冧
␥y By By 1 关K1共␥r兲兴2 − 关K0共␥r兲兴2
for − ⬁ ⬍ y 艋 − ti = ␲r2p共␭si + 2Gsi兲 共18兲
exp
rq
y+
2 2 2 关K0共␥r兲兴2
By By for circular piles. In the equations above, K1共·兲 = modified Bessel
␾y = 1 for − 艋y艋 共10兲 function of the second kind of first order. The constants ki and ti
2 2

冋 冉 冊册
represent the shear and compressive resistances offered by the
␥y By By soil mass against pile settlement.
exp − y− for 艋y⬍⬁
rq 2 2 The general solution of Eq. 共14兲, which is a second-order lin-
ear differential equation, is given by

冑
where

冉冊 冉 冊
ui共z兲 = Bie␨iz + Cie−␨iz 共19兲
␥x rq
ms + ns + Bx where
␥y rq ␥x

冉 冊 冑
= 共11兲
rq rq ki
ms + Bx ␨i = 共20兲
␥x EiAi + 2ti
For circular piles, solution of the differential equation of ␾r and Bi and Ci = integration constants. The axial force Qi共z兲 in the
gives 共Seo and Prezzi 2007兲 pile shaft at a depth z in the ith layer is obtained from

冉 冊
冦 冧
␥r dui
K0 r Qi共z兲 = − 共EiAi + 2ti兲 = − aiBie␨iz + aiCie−␨iz 共21兲
rp for r p 艋 r ⬍ ⬁ dz
␾r共r兲 = 共12兲
K0共␥r兲
where
1 for 0 艋 r 艋 r p
ai = ␨i共EiAi + 2ti兲 = 冑ki共EiAi + 2ti兲 共22兲
where K0共·兲 = modified Bessel function of the second kind of zero
order, and As there are 2N unknown integration constants
共B1 , C1 , B2 , C2 , . . . , BN , CN兲, we need to identify 2N boundary
␥r
rp
= 冑 ns
ms
共13兲
conditions in order to determine their values. The first condition
follows from the requirement that the vertical soil displacement at
an infinite depth below the pile base must be zero 关兩uN共z兲兩z→⬁
In the case of rectangular piles, the term rq 关Eq. 共7兲兴 is a = 0兴. Also, the magnitude of the load at the pile head must be
representative length used to make the parameters ␥x and ␥y di- equal to the applied external load 关兩Q1共z兲兩z=0 = Qt兴. These bound-
mensionless. This is not an equivalent pile radius mandated by the ary conditions lead to
analysis but rather a quantity chosen to normalize the equations.
In contrast, r p 关appearing in Eq. 共13兲兴 represents the actual radius BN = 0 共23兲

422 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / MARCH 2009

Downloaded 11 Jun 2012 to 180.211.192.67. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
 − a 1B 1 + a 1C 1 = Q t 共24兲
As the displacement and force should be continuous at the
interfaces between adjacent layers, ui共Hi兲 = ui+1共Hi兲 and Qi共Hi兲
= Qi+1共Hi兲. These conditions yield
e␨iHiBi + e−␨iHiCi = e␨i+1HiBi+1 + e−␨i+1HiCi+1 共25兲

− aie␨iHiBi + aie−␨iHiCi = − ai+1e␨i+1HiBi+1 + ai+1e−␨i+1HiCi+1

共26兲
Eqs. 共23兲–共26兲 produce the 2N boundary conditions needed to
determine all the integration constants. Seo and Prezzi 共2007兲
developed a recurrence-based solution scheme for the integration
constants Bi and Ci and presented explicit expressions for these
constants.

Modification of Soil Moduli

The above analysis assumes zero horizontal displacement in the
soil. This assumption does not perfectly capture the nature of the
displacement field, particularly near the pile head, where the
downdrag exerted by the pile on the surrounding soil induces
horizontal displacements that point toward the pile. Thus, re-
straining the horizontal displacement results in pile response that Fig. 2. Flow chart for iterative solution scheme
is stiffer than it is in reality. In fact, the term 共␭si + 2Gsi兲 in Eqs.
共9兲, 共16兲, and 共18兲 represents the soil constrained modulus, which
is an indication that the analysis produces a stiff response. As the 共13兲兴. As a consequence of this interdependence of the soil and
soil Poisson’s ratio ␯si approaches 0.5, the pile load-settlement pile differential equations, an iterative solution scheme is neces-
response becomes increasingly stiffer 共the constrained modulus is sary.
equal to infinity for a Poisson’s ratio of 0.5兲. A comparison with In the first iteration, initial values are assumed for ␥, and pile
finite-element analysis 共FEA兲 using ABAQUS also confirms this displacement and its derivative 共obtained from the axial force兲 are
observation. calculated. At the end of the iteration, new ␥ values are obtained
In order to eliminate the artificial stiffness resulting from the using the calculated pile displacement and its derivative and com-
assumption of zero lateral displacement for high ␯s values, we set pared with the assumed initial values. If the differences are
␭si = Esi␯si / 共1 + ␯si兲共1 − 2␯si兲 = 0, where Esi is the soil Young’s greater than the prescribed tolerances, iterations are continued,
modulus of the ith layer, which is equivalent to making the soil with the calculated values of ␥ taken as the new guesses. Succes-
Poisson’s ratio ␯si = 0 关removal of the artificial stiffness by setting sive iterations are continued until the values of ␥ obtained from
␭si = 0 was proposed for laterally loaded piles by Guo and Lee two consecutive iterations fall below the prescribed limits. This
共2001兲兴, and replace Gsi by a modified shear modulus Gsi*. The iterative solution scheme is provided in the form of a flow chart in
effect of Poisson’s ratio is indirectly taken into account through Fig. 2.
the modified shear modulus Gsi* 关a similar procedure was recom-
mended by Randolph 共1981兲 for laterally loaded piles兴. We pro-
pose the following expressions for the modified shear modulus Results
Gsi* by matching the pile responses obtained from our analyses
with those obtained from FEA 共performed for identical pile and Comparison with Finite-Element Analysis
soil conditions兲 using ABAQUS:
The results of our analysis are compared with those of FEA per-
Gsi* = 0.6Gsi共1 + 1.25␯si2 兲 共27兲 formed using ABAQUS. Twenty-noded brick elements were used
to represent both the pile and the soil. The horizontal extent of the
for rectangular piles, and soil domain was taken to be at least 15 times the pile width or
Gsi* = 0.75Gsi共1 + 1.25␯si2 兲 共28兲 diameter from the pile axis, and the vertical extent of the soil
domain below the pile base was taken as at least the pile length.
for circular piles. The location of the boundaries was found to be sufficient for there
Accordingly, Eqs. 共8兲, 共9兲, and 共14兲–共18兲 are modified by mak- to be no boundary effects; convergence checks were also per-
ing ␭si = 0 共irrespective of the value of Poisson’s ratio兲 and by formed.
replacing Gsi by Gsi*. We consider a barrette 共an excavated rectangular concrete pile兲
with 2.7 m ⫻ 1.2 m cross section and a drilled shaft with 2 m
diameter, both 30 m long, embedded in a four-layered soil. Both
Iterative Solution Scheme
cross sections have the same area. The axial force Qt at the head
The ␥’s 共␥x, ␥y, and ␥r兲 of Eqs. 共15兲–共18兲 must be known in order of the piles is 8,000 kN. The Young’s modulus E p of the piles is
to estimate the parameters ki and ti, without which Eq. 共14兲 for the 25 GPa. Both the piles are embedded in a four-layer deposit with
pile displacement cannot be solved. These ␥’s, in turn, depend on H1 = 2 m, H2 = 12 m, and H3 = 22 m 共the pile base rests in the
the pile settlement u and its derivative du / dz 关Eqs. 共6兲, 共11兲, and fourth layer兲; Es1 = 15 MPa, Es2 = 25 MPa, Es3 = 30 MPa, and Es4

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / MARCH 2009 / 423

Downloaded 11 Jun 2012 to 180.211.192.67. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
 Pile displacement u (mm) shaft 共2 m diameter兲 共note that both the piles have the same cross-
0 2 4 6 8 sectional area and length, and are embedded in the same soil
0 profile兲. The results from our analytical method are in good agree-
ment with the FEA results. Fig. 3共b兲 shows the vertical soil dis-
Qt = 8000 kN Es1= 15 MPa,
s1 = 0.4
placements at the ground surface as a function of the horizontal
5 z=2m distance from the pile center. The soil displacements from the
Es2= 25 MPa,
s2 = 0.3 FEA are similar for both the barrette and the drilled shaft. How-
Ep= 25 GPa ever, the results from our analysis show that the vertical soil
z = 12 m
10 Es3= 30 MPa,
s3 = 0.3 displacement for the barrette diminishes faster in the horizontal
Lp = 30 m direction than those of the drilled shaft. This is due to the differ-
Depth z (m)

z = 22 m
ent analytical forms of the soil displacement decay functions for
15 Es4= 100 MPa,
s4 = 0.15 the rectangular and circular cross sections.
Bx = 2.7 m

By = 1.2 m
2m Comparison with Previous Pile Settlement Studies
20
We now compare results from our study with numerical or ana-
This study (drilled shaft) lytical solutions available in the literature 共Blaney et al. 1976;
FEA (drilled shaft) Poulos and Davis 1980; El-Sharnouby and Novak 1990; Fleming
25
This study (barrette) et al. 1992; Mylonakis 2001兲. The results are presented in terms
FEA (barrette) (a) of the normalized pile head stiffness KN defined as
30
Qt
KN = 共29兲
u tE pB *
Horizontal distance from pile center (m)
0 5 10 15 20 25 30 35
0 where Qt = applied load at the pile head; ut = settlement at the pile
head 共=兩u兩z=0兲; E p = Young’s modulus of the pile; and B*
Bx = 2.7 m = equivalent diameter 共Br for a circular pile and 冑4BxBy / ␲ for a
rectangular pile兲. The equivalent diameter B* for the rectangular
Vertical soil displacement uz(mm)

2 x pile is equal to the diameter of a circular section with the same

By = 1.2 m
cross-sectional area as that of the rectangular pile.
y
Fig. 4 compares the values of normalized pile head stiffness
KN versus normalized pile length L p / B* of ideal end-bearing piles
Br = 2 m
共piles with zero base settlement兲 obtained from this study with
4 those from previous studies for two different pile-soil modulus
ratios 共E p / Gs兲. For comparison purposes, pile head stiffness nor-
This study (drilled shaft) malized with respect to the soil modulus 共Qt / utGsB*兲 is also
FEA (drilled shaft) given on the right vertical axis. The pile base is assumed to rest
This study (barrette - x direction) on a rigid layer; the soil above the rigid layer is homogeneous
6
FEA (barrette - x direction) with Es as its Young’s modulus; and ␯s = 0.5 as its Poisson’s ratio
This study (barrette - y direction)
共i.e., the soil shear modulus Gs = Es / 3兲. It should be noted that,
although we plotted the results obtained from the analysis of
FEA (barrette - y direction) (b) Fleming et al. 共1992兲 in Fig. 4, they did not specifically address
8 the case of ideal end-bearing piles in their analysis. However, by
allowing the shear modulus below the pile base to tend to infinity
Fig. 3. Comparison between displacements obtained using analytical in the equation of the magical radius rm 共Randolph and Wroth
method and FEA for 30-m-long barrette and drilled shaft: 共a兲 pile 1978兲, the results shown in Fig. 4 关corresponding to Fleming et al.
displacement versus depth; 共b兲 vertical soil displacement 共at ground 共1992兲兴 are obtained.
surface兲 versus horizontal distance Solutions for rectangular or square piles are not available in
the literature, so we present the results for square piles 共i.e., for
Bx / By = 1兲 together with those for circular piles in Fig. 4. For
= 100 MPa; ␯s1 = 0.4, ␯s2 = 0.3, ␯s3 = 0.3, and ␯s4 = 0.15 共Esi and ␯si E p / Gs = 300, the pile head stiffness following Blaney et al. 共1976兲
are the Young’s modulus and Poisson’s ratio of the ith soil layer, and Poulos and Davis 共1980兲 first decreases, then increases 关Fig.
respectively兲. The corresponding values of Gs* for the four soil 4共a兲兴 as the pile slenderness ratio L p / B* increases. As pointed out
layers used along with ␭si = 0 in the analysis are Gs1 * = 3.9 MPa,
by El-Sharnouby and Novak 共1990兲 and Mylonakis 共2001兲, this
* * *
Gs2 = 6.4 MPa, Gs3 = 7.7 MPa, and Gs4 = 26.8 MPa for the barrette, reversal cannot occur for ideal end-bearing piles because, no mat-
and Gs1 * = 4.8 MPa, G* = 8.0 MPa, G* = 9.6 MPa, and G* ter how much load is transferred to the pile base, it does not
s2 s3 s4
= 33.5 MPa for the drilled shaft. contribute to the head stiffness because the base is rigid. El-
Fig. 3共a兲 shows the pile displacement as a function of depth, as Sharnouby and Novak 共1990兲 suggested that the likely cause for
obtained from our analysis and FEA. The results from both the this reversal is the small number of pile elements used in these
analytical method and FEA show that the displacement of the analyses. The results from our analyses are in good agreement
barrette 共2.7 m ⫻ 1.2 m兲 is slightly less than that of the drilled with the solutions of Mylonakis 共2001兲 and El-Sharnouby and

424 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / MARCH 2009

Downloaded 11 Jun 2012 to 180.211.192.67. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
 0.1 Qt 30 0.16
(a) Qt
Gs Bx

Normalized pile head stiffness, Qt /(ut Gs B*)

Normalized pile head stiffness, Qt /(ut Ep B*)

Normalized pile head stiffness, Qt /(ut Ep B*)

0.09 Lp Ep
s = 0.5 27 By
Ep/Gs = 300 Gs
Ep
rigid layer 0.12 Lp
s = 0.5
0.08 24
B* Br

0.07 21
0.08
This study (circular)
0.06 This study (square) 18
Mylonakis (2001)
Fleming et al. (1992)
0.05 El-Sharnouby and Novak (1990) 15 0.04 Circular Lp/B* = 100
Poulos and Davis (1980) Square
Blaney et al. (1976) (a) Rectangular
0.04 12 (Bx/By = 5) Lp/B* = 25
25 50 75 100
Normalized pile length, Lp /B* 0
100 1000 10000
Modulus ratio, Ep/Gs
0.04 120
This study (circular)
Normalized pile head stiffness, Qt /(ut Gs B*)
0.16
Normalized pile head stiffness, Qt /(ut Ep B*)

This study (square)

0.035 Mylonakis (2001) 105 (b) Qt Bx

Normalized pile head stiffness, Qt /(ut Ep B*)

Fleming et al. (1992)
El-Sharnouby and Novak (1990) By
0.03 Poulos and Davis (1980) 90 Gs
Ep
Blaney et al. (1976) 0.12 Lp
s

0.025 75 B* Br

Qt
Lp/B* = 25
0.02 Gs 60 0.08

Lp Ep
s = 0.5
0.015 Ep/Gs = 3000 45
s = 0.5

s = 0
rigid layer (b) 0.04
Circular
0.01 30
25 50 75 100 Square
Normalized pile length, Lp /B* Rectangular
(Bx/By = 5)
Fig. 4. Comparison of normalized pile head stiffness versus normal- 0
ized pile length of end-bearing piles with: 共a兲 E p / Gs = 300; 共b兲 100 1000 10000
Modulus ratio, Ep/Gs
E p / Gs = 3,000
Fig. 5. Comparison of normalized pile head stiffness of square and
Novak 共1990兲. In the case of E p / Gs = 3,000, the normalized pile circular piles: 共a兲 effect of pile slenderness ratio; 共b兲 effect of soil
head stiffness decreases with increasing L p / B* for all analyses, as Poisson’s ratio
seen in Fig. 4共b兲.

Comparison between Rectangular and Circular Piles = 25 and E p / Gs = 1,000, the normalized pile head stiffness for the
square and the rectangular piles is, respectively, 8–9% and 12–
To compare the response of piles with different cross sections, the 13% larger than that of the circular pile. For example, for a con-
normalized pile head stiffness KN was obtained for a wide range crete pile 共E p = 30 GPa兲 with B* = 0.6 m 共Br = 0.6 m for a circular
of values of pile-to-soil modulus ratio 共E p / Gs兲 and pile slender- pile, Bx = By = 0.53 m for a square pile, and Bx = 1.19 m and By
ness ratio 共L p / B*兲 for piles in homogeneous soil layers. Fig. 5共a兲 = 0.238 m for a rectangular pile兲 and length L p = 15 m, embedded
shows normalized pile head stiffness versus E p / Gs for two differ- in a layer with Gs = 30 MPa and ␯s = 0.5, the pile head settlements
ent values of L p / B*, representing relatively short 共L p / B* = 25兲 and for a load of 700 kN applied at the head are 1.20, 1.10, and
long 共L p / B* = 100兲 piles for circular, square, and rectangular 1.07 mm for a circular, a square, and a rectangular pile, respec-
共Bx / By = 5兲 cross sections. Square piles show a slightly stiffer re- tively. This additional stiffness is due to the larger perimeter of
sponse than circular piles, except at low values of E p / Gs 共of the the square and rectangular pile for the same cross-sectional area;
order of 100兲. On the other hand, rectangular piles show a slightly the square and the rectangular cross section with Bx / By = 5 have
stiffer response than square piles 共the effect of the aspect ratio 13 and 51% larger perimeter, respectively, than a circular one
Bx / By of the cross section is discussed in detail in the next sec- with the same area 共the perimeters of a circular, a square, and a
tion兲. Similar trends are observed for different soil Poisson’s ra- rectangular pile considered above are 1.885, 2.120, and 2.856 m,
tios as well 关Fig. 5共b兲兴. For a typical practical case, with L p / B* respectively兲.

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / MARCH 2009 / 425

Downloaded 11 Jun 2012 to 180.211.192.67. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
 0.05 50 stiffness increases as the pile length increases and as the aspect
(a)
ratio Bx / By increases. The increase in stiffness due to an increase

Normalized pile head stiffness, Qt /(ut Gs* B*)

Normalized pile head stiffness, Qt /(ut Ep B*)
in the Bx / By ratio is a result of the fact that greater shaft resistance
0.045 45 is obtained as the pile perimeter increases with increasing Bx / By.
Bx/By = 1 However, as seen in Fig. 6共b兲, this increase of the normalized
Bx/By = 2 stiffness with Bx / By is very small. Even for the case of E p / Gs*
0.04 Bx/By = 3 40 = 100, which shows maximum increase in stiffness, the gain in
Bx/By = 4 stiffness is only 6% 共while the perimeter increases by 34%兲 as
Bx/By = 5 Bx / By increases from 1 to 5. Therefore, if we assume the soil to
0.035 Qt 35 be linear elastic and fix the axial pile stiffness 共E pA p兲, then the
Bx effect of the cross-section aspect ratio on the load-settlement re-
Gs* sponse of rectangular piles is practically negligible.
Ep By
0.03 Lp 30
Ep/Gs* = 1000 Effect of Bearing Layer

0.025 25 If weak soil layers overly a stiff soil layer, depending on the depth
25 50 75 100 of the stiff layer, it is often advantageous to extend the pile length
Normalized pile length, Lp/B*
to the stiff layer in order to capitalize on the end bearing resis-
tance available there. We consider the case of a weak soil layer
0.2 with equivalent shear modulus Gs* lying above a stiff layer with
equivalent shear modulus Gsb * . The pile base is assumed to lie on
(b)
Ep/Gs* = 100 the interface of the weak and the strong layer. In practice, we
Normalized pile head stiffness, Qt /(ut Ep B*)

0.16 Qt would embed the pile at least two diameters into the stiff layer in
Bx order to guarantee a proper development of base resistance. In our
Gs*
Ep By analysis, there is no such requirement, as the base resistance will
Lp directly reflect the modulus of the soil underlying the pile base.
0.12 Lp/B* = 25 Ep/Gs* = 200
Fig. 7共a兲 shows the normalized pile head stiffness of a circular
pile as a function of the modulus ratio Gsb * / G* of the two soil
s
layers for different values of E p / Gs and L p / B*. Irrespective of the
*
0.08
Ep/Gs* = 500 pile slenderness ratio L p / B*, the pile head stiffness increases as
E p / Gs* decreases 共i.e., as the stiffness of the weaker soil in-
Ep/Gs* = 1000 creases兲. When the soil layer surrounding the pile shaft becomes
0.04
very stiff 共E p / Gs* = 100兲 or the pile slenderness ratio is large
Ep/Gs* = 10000
共L p / B* = 100兲, as is the case for micropiles, the normalized pile
head stiffness is practically independent of the soil properties
0 below the pile base. If the soil below the pile base is only slightly
1 2 3 4 5
Aspect ratio, Bx/By stiffer than the soil surrounding the shaft 共i.e., for low values of
Gsb* / G*兲, piles with larger L / B* show a stiffer response, but if
s p
Fig. 6. Normalized pile head stiffness versus: 共a兲 normalized pile the base soil is much stiffer than the soil surrounding the shaft
length L p / B* with E p / Gs* = 1,000; 共b兲 aspect ratio Bx / By of rectangu- 共i.e., for large values of Gsb * / G*兲, then piles with smaller L / B*
p
s
lar cross section with L p / B* = 25. have a larger normalized pile head stiffness. Fig. 7共b兲 shows the
normalized pile head stiffness of a square pile for the same cases
discussed above. Similar trends are observed for a square pile as
Effect of Aspect Ratio of Rectangular Cross Sections well, with only slightly larger normalized pile head stiffness than
Although driven precast concrete piles usually have square cross that of a circular pile.
sections, the aspect ratio Bx / By of the cross section of barrettes
are usually greater than 2 共Fellenius et al. 1999; Ng and Lei Piles in Two-Layer Soil
2003兲. We are not aware of any research study available in the
literature on the effect of the aspect ratio Bx / By on the load- We performed a parametric study for cases with two soil layers
settlement response of rectangular piles. In this section, we inves- present along the pile shaft. Fig. 8共a兲 shows the results for five
tigate the effect of the Bx / By ratio by changing the Bx / By ratio different soil modulus ratios Gs1 * / G* = 0.2, 0.5, 1, 2, and 5, with
s2
while maintaining the same cross-sectional area and hence the L p / B* = 25 and E p / Gs2 = 1,000. The thickness h of the top layer
*
same axial pile head stiffness. The case of a homogeneous soil is varies from 0 to L p. Fig. 8共b兲 shows the variation of the normal-
considered. From this point on, all figures present the results with ized pile head stiffness as a function of E p / Gs2* when h = 0.5L . As
p
respect to the modified shear modulus Gs* to avoid including ad- illustrated later in an example, the curves shown in this figure
ditional charts for different Poisson’s ratios 共the effect of Pois- may be used as design charts in early stages of pile design when
son’s ratio is already incorporated in the expressions for Gs*兲. similar soil profiles are encountered.
Fig. 6共a兲 shows the normalized pile head stiffness versus nor-
malized pile length 共or pile slenderness ratio兲 for different Bx / By
Piles in Three-Layer Soil
values and for E p / Gs* = 1,000. Note that the values of the pile head
stiffness normalized with respect to the soil modulus 共Qt / utGs*B*兲 We also consider the case with three-layer soil deposits. It is
is also given on the right vertical axis. The normalized pile head assumed that each soil layer has the same thickness, but has dif-

426 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / MARCH 2009

Downloaded 11 Jun 2012 to 180.211.192.67. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
 0.2 0.16
Lp/B* = 25, 50, 100 Qt
(a)

Normalized pile head stiffness, Qt /(ut Ep B*)

Normalized pile head stiffness, Qt /(ut Ep B*)
Qt 0.14 h Gs1*
0.1 Ep/Gs2* = 1000
Lp
0.08 Lp Ep Gs* 0.12 Ep Gs2*
Lp/B* = 25
0.06 Br
Lp/B*= 25
Gsb* 0.1 Gs1*/Gs2* = 5
0.04 50
100
0.08
Lp/B* = 25
0.02
50 0.06 2
100
0.04 1
0.01
Ep/Gs* = 100
0.008 0.5
Ep/Gs* = 1000 0.02 Circular
0.006 Square 0.2
Ep/Gs* = 10000
(a) 0
0.004
1 10 100 0 0.2 0.4 0.6 0.8 1
Modulus ratio of the soil layer, Gsb*/Gs* Normalized layer depth, h/Lp

0.8
0.2 Gs1*/Gs2* Qt
0.6 (b)
Qt *
Lp/B = 25, 50, 100 =5

Normalized pile head stiffness, Qt /(ut Ep B*)

0.4 0.5Lp
Normalized pile head stiffness, Qt /(ut Ep B*)

Gs1*
0.1 Bx 2
Lp Ep Gs* By
0.2 0.5Lp Ep Gs2*
0.08 1
0.06 Lp/B*= 25 0.1 0.5 Lp/B* = 25
Gsb* 0.08
50 0.2
0.04 100 0.06
0.04
Lp/B* = 25
0.02 50 0.02
100
0.01
0.008
0.01 0.006
Ep/Gs* = 100
0.008 0.004 Circular
Ep/Gs* = 1000
0.006 Square
Ep/Gs* = 10000
(b) 0.002
0.004 100 1000 10000
1 10 100 Modulus ratio, Ep/Gs2*
Modulus ratio of the soil layer, Gsb*/Gs*
Fig. 8. Normalized pile head stiffness in two-layer soil versus: 共a兲
Fig. 7. Normalized pile head stiffness versus modulus ratio of base h / L p; 共b兲 E p / Gs2
* with h = 0.5L
* / G* for: 共a兲 circular pile; 共b兲 square pile p
soil to shaft soil Gsb s

9共b兲 shows the normalized pile head stiffness versus E p / Gs*, with
ferent equivalent shear modulus Gs* such that the average value L p / B* = 25 for all three cases 共I, II, and III兲. Case III shows the
*
Gs,avg 关= 共Gs1* + G* + G* 兲 / 3兴 remains the same for the cases 共I, II,
s2 s3 stiffest behavior but the difference in the observed normalized
and III兲 considered. In Case I, the soil modulus increases with pile head stiffness for the three cases becomes smaller as the soil
depth, with the smallest soil modulus observed for the uppermost becomes weaker 共i.e., as E p / Gs* becomes larger兲. When the soil
layer. In Case III, the soil modulus decreases with depth, with the deposit is very weak 共i.e., for E p / Gs* = 10,000兲, there is no practi-
largest soil modulus observed for the uppermost layer. Case II cal difference in the normalized pile head stiffness for the three
represents a profile with an intermediate weak layer. cases.
Fig. 9共a兲 shows the normalized pile head stiffness as a function
of pile slenderness ratio L p / B*, with E p / Gs* = 1,000 for all the
three cases considered. On the right vertical axis, the pile head Design Examples
stiffness normalized with respect to the average soil modulus
共Qt / utGs,avg
* B*兲 is given. The normalized pile head stiffness de- Although our analysis is limited to elastic soils, it may provide a
creases with L p / B* for end-bearing piles 共Case I兲 and increases reasonable estimation of pile settlement under working loads. In
for floating piles 共Case III兲. These results imply that analyses this section, we illustrate the use of some of the charts presented
considering a single layer with a simple weighted average of the in this paper for design purposes. Furthermore, we developed a
soil modulus of different soil layers with layer thicknesses as user-friendly spreadsheet program ALPAXL to facilitate the use
weights will not produce correct pile head stiffness values. Fig. of our analysis. This program is based on the solution scheme

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / MARCH 2009 / 427

Downloaded 11 Jun 2012 to 180.211.192.67. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
 We first calculate the slenderness ratio of the pile: L p / B*
= 14/ 0.56= 25. The equivalent moduli for the two soil layers are
calculated as follows:
* = 0.6G 共1 + 1.25␯2 兲 = 0.6 ⫻ 10共1 + 1.25 ⫻ 0.22兲 = 6.3 MPa
Gs1 s1 s1

* = 0.6G 共1 + 1.25␯2 兲 = 0.6 ⫻ 50共1 + 1.25 ⫻ 0.22兲 = 31.5 MPa

Gs2 s2 s2

for the square pile, and

* = 0.75G 共1 + 1.25␯2 兲 = 0.75 ⫻ 10共1 + 1.25 ⫻ 0.22兲
Gs1 s1 s1

= 7.875 MPa

* = 0.75G 共1 + 1.25␯2 兲 = 0.75 ⫻ 50共1 + 1.25 ⫻ 0.22兲

Gs2 s2 s2

= 39.375 MPa
for the circular pile. Modulus ratios for the square and circular
piles are E p / Gs2* = 25,000/ 31.5⬇ 795 and E / G* = 25,000/
p s2
39.375⬇ 635, respectively. We use Fig. 8共b兲 to estimate the pile
head settlement. Since the stiffness ratio of the two soil layers is
* / G* = 0.2, the normalized pile head stiffness K = Q /
Gs1 s2 N t
共utE pB*兲 = 0.030 for the square pile and 0.028 for the circular pile.
Finally, the pile head settlement is calculated as
Qt 800 kN
ut = = ⬇ 1.9 mm
KNE pB* 0.030 ⫻ 25 ⫻ 106 kPa ⫻ 0.56 m
for the square pile, and
Qt 800 kN
ut = = ⬇ 2.0 mm
KNE pB* 0.028 ⫻ 25 ⫻ 106 kPa ⫻ 0.56 m
for the circular pile.

Example 2
We consider a case history as the second example. Chang and
Wong 共1987兲 reported the results of instrumented load tests on
drilled shafts installed in weathered sedimentary rocks of the Ju-
Fig. 9. Normalized pile head stiffness in three-layer soil versus: 共a兲 rong Formation in Singapore. The top 11 m of the soil profile
L p / B*; 共b兲 E p / Gs* consists of medium stiff to hard silty clay 关NSPT = 7 – 36, where
NSPT⫽standard penetration test 共SPT兲 blow count兴 with an und-
rained shear strength su ranging from 40 to 200 kPa; underneath
presented in this paper and uses built-in Microsoft EXCEL func- this layer there is a layer of highly weathered siltstone 共NSPT
tions. ALPAXL produces the deformed configuration of the pile- = 50– 145兲. The test pile, which was embedded 13 m into the
soil system and the pile load-settlement curve in seconds. siltstone layer, was 0.9 m in diameter and 24 m in length. It was
ALPAXL can be downloaded at 具http://cobweb.ecn.purdue.edu/ instrumented with five vibrating-wire strain gauges at 7.5, 11.0,
⬃mprezzi典. 15.5, 20.5, and 24.0 m below the ground surface. The represen-
tative Young’s modulus of the pile was 31 MPa. The pile was
designed to carry an axial load of 2,500 kN and tested to four
Example 1
times the design load 1 month after its installation using the slow
We will calculate the settlement of concrete piles 共one with a maintained-load test method.
cross section of 0.5 m ⫻ 0.5 m and another with a diameter equal The elastic properties of the soil and rock layers were not
to 0.56 m兲 subjected to a working load of Qt = 800 kN. We do not available in the original paper by Chang and Wong 共1987兲. For
concern ourselves with the installation effects, focusing instead the rock layer, input values for the Young’s modulus were ob-
on a direct comparison of two piles that are equivalent in all tained from Kim et al. 共1999兲 since they reanalyzed the pile load
aspects and are surrounded by soil with exactly the same proper- test results reported by Chang and Wong 共1987兲 to develop load-
ties but have different cross-section geometries. Both the piles transfer functions for drilled shafts installed in weathered rock.
have the same cross-sectional area and the same length 共L p The Young’s modulus of the weathered siltstone used in the
= 14 m兲; the Young’s modulus of the pile is E p = 25 GPa. There analysis of Kim et al. 共1999兲 was 1,000 MPa. For the silty clay
are two soil layers present at the site. The top layer, with elastic layer, the Young’s modulus was estimated from the undrained
constants Gs1 = 10 MPa and ␯s1 = 0.2, extends from the ground sur- shear strength su. According to Callanan and Kulhawy 共1985兲,
face to a depth of 7 m. The second layer, with Gs2 = 50 MPa and values for the Es / su ratio generally range between 200 and 900,
␯s2 = 0.2, extends from 7 m to large depths. with an average value of 500. Using Es / su = 500, Es values for the

428 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / MARCH 2009

Downloaded 11 Jun 2012 to 180.211.192.67. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
Table 1. Input Values for Analysis of Example 2 共L p = 24 m; E p Summary and Conclusions
= 31 GPa兲
Hi Esi A method of settlement analysis that applies to piles with either
Layer 共m兲 共MPa兲 ␯si rectangular or circular cross sections installed in multilayered soil
1 11 60 0.5 deposits was presented. This analysis is based on the solution of
2 24 1,000 0.15 the governing differential equations for pile and soil displace-
3 50 1,000 0.15 ments obtained using the principle of minimum potential energy
Note: Br = 0.9 m for drilled shaft; Bx = 1.13 m, and By = 0.56 m for bar- and calculus of variations. Comparisons of the load-settlement
rette. responses of piles with rectangular and circular cross sections
were made for various conditions.
The analysis produces pile displacement and axial force as
clay layer range from 20 to 100 MPa; an average value Es,avg functions of depth and vertical soil displacement as a function of
= 60 MPa was used in the analysis. The Poisson’s ratio was as- the horizontal distance from the center of the pile if the following
sumed to be 0.5 for the clay layer and 0.15 for the rock layer. are known: the pile cross-sectional dimensions and length, the
Even though only a drilled shaft was installed at the test site, we thicknesses of the soil layers, Young’s modulus of the pile mate-
consider a barrette with Bx = 1.13 m and By = 0.56 m in addition to rial, the Young’s moduli and Poisson’s ratios of the soils in the
the drilled shaft for comparison purposes 共the barrette and the various layers, and the magnitude of the applied axial force.
drilled shaft have the same cross-sectional area兲. The input values The results 共pile displacement versus depth and vertical soil
used in the analysis are summarized in Table 1. Note that we displacement at the ground surface versus horizontal distance
divided the soil profile into three layers for the analysis, with the from pile center兲 from FEA and our analysis for a multilayered
bottom of the second layer flush with the base of the pile. soil showed good agreement; the results 共normalized pile head
The program ALPAXL was used for all the calculations. The
stiffness versus normalized pile length兲 from our analysis for end-
results from our analysis are compared with measured data for up
to two times the design load because our analysis is elastic and is bearing piles also compared well with results from previous stud-
valid only for the initial stages of loading. Fig. 10共a兲 shows the ies. Square and rectangular piles have slightly stiffer response
predicted and measured load-settlement curves for the test pile than circular piles if the axial stiffness E pA p is the same. How-
and the predicted load-settlement curve for the barrette. The re- ever, for rectangular piles, the increase in the normalized pile
sults from our analysis are in good agreement with the measured head stiffness with the increase in the aspect ratio of the pile cross
data. In particular, the calculated settlement shows very good section is negligible. When there is a stiff upper layer in a profile
agreement with the measured values up to the design load 共Qt with two layers, the condition of the soil below the pile base has
= 2,500 kN兲. The reason for the sudden jump in the measured almost no effect on the pile head stiffness. However, when the
load-settlement curve at 3,000 kN is not mentioned in the original soil surrounding the pile is weak and the pile is short, the pile
paper. The predicted load-settlement curve for the barrette falls on head stiffness depends strongly on the properties of the base soil.
top of the curve for the drilled shaft. Fig. 10共b兲 shows the pre-
In the case of piles in multilayered soil, pile response depends on
dicted and measured load-transfer curves for the test pile and the
soil layering, with the uppermost soil having the most effect on
predicted load-transfer curve for the barrette. The results from
both the load test and our analysis indicate that most of the ap- the pile head stiffness. Two examples illustrated the use of our
plied load was carried by shaft friction, in particular along the analysis in design. In order to facilitate use of our analysis, a
pile-rock interface. Both the drilled shaft and the barrette show user-friendly spreadsheet program ALPAXL was developed.
very similar load-transfer behavior.

Load (kN) Load (kN)

0 1000 2000 3000 4000 5000 6000 0 1000 2000 3000 4000 5000 6000
0 0
(a) (b) Ep = 31 GPa

1 Medium Stiff to
5
11 m Hard Silty Clay
su = 40 ~ 200 kPa
2 NSPT = 7 ~ 36
Settlement (mm)

10
Depth (m)

3 Curves for a drilled shaft

and a barrette fall on top
of each other. 15 Weak Highly
Weathered
4 13 m Siltstone
This study (drilled shaft) This study (drilled shaft) NSPT = 50 ~ 145
20
5 This study (barrette) This study (barrette)
Measured (drilled shaft)
Measured (drilled shaft)
Br = 0.9 m
6 25 (Bx= 1.13 m, By = 0.56 m)

Fig. 10. Comparison between results from present analysis and measured pile load test data: 共a兲 load-pile head settlement curves; 共b兲 load-
transfer curves

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / MARCH 2009 / 429

Downloaded 11 Jun 2012 to 180.211.192.67. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org
References axially loaded piles in rock.” Int. J. Numer. Analyt. Meth. Geomech.,
18, 427–437.
Armaleh, S., and Desai, C. S. 共1987兲. “Load-deformation response of Kraft, L. M., Ray, R. P., and Kagawa, T. 共1981兲. “Theoretical t-z curves.”
axially loaded piles.” J. Geotech. Engrg., 113共12兲, 1483–1500. J. Geotech. Engrg. Div., 107共11兲, 1543–1561.
Basu, D., Prezzi, M., Salgado, R., and Chakraborty, T. 共2008兲. “Settle- Lee, C. Y., and Small, J. C. 共1991兲. “Finite-layer analysis of axially
ment analysis of piles with rectangular cross sections in multi-layered loaded piles.” J. Geotech. Engrg., 117共11兲, 1706–1722.
soils.” Comput. Geotech., 35共4兲, 563–575. Lee, K.-M., and Xiao, Z. R. 共1999兲. “A new analytical model for settle-
Blaney, G. W., Kausel, E., and Roesett, J. M. 共1976兲. “Dynamic stiffness ment analysis of a single pile in multi-layered soil.” Soils Found.,
of piles.” Proc., 2nd Int. Conf. on Numerical Methods in Geomechan- 39共5兲, 131–143.
ics, ASCE, Blacksburg, Va., 1001–1012. Mattes, N. S., and Poulos, H. G. 共1969兲. “Settlement of single compress-
Butterfield, R., and Banerjee, P. K. 共1971兲. “The elastic analysis of com- ible pile.” J. Soil Mech. and Found. Div., 95共1兲, 189–207.
pressible piles and pile groups.” Geotechnique, 21共1兲, 43–60. Motta, E. 共1994兲. “Approximate elastic-plastic solution for axially loaded
Callanan, J. F., and Kulhawy, F. H. 共1985兲. “Evaluation of procedures for piles.” J. Geotech. Engrg., 120共9兲, 1616–1624.
predicting foundation uplift movements.” Rep. to Electric Power Re- Murff, J. D. 共1975兲. “Response of axially loaded piles.” J. Geotech.
search Inst., No. EPRI EL-4107, Cornell Univ., Ithaca, N.Y. Engrg. Div., 101共3兲, 356–360.
Chang, M. F., and Wong, I. H. 共1987兲. “Shaft friction of drilled piers in Mylonakis, G. 共2001兲. “Winkler modulus for axially loaded piles.” Geo-
weathered rocks.” Proc., 6th Int. Conf. on Rock Mechanics, ISRM, technique, 51共5兲, 455–461.
Ng, C. W. W., and Lei, G. H. 共2003兲. “Performance of long rectangular
Montreal, Canada, 313–318.
Coyle, H. M., and Reese, L. C. 共1966兲. “Load transfer for axially loaded barrettes in granitic saprolites.” J. Geotech. Geoenviron. Eng., 129共8兲,
piles in clay.” J. Soil Mech. and Found. Div., 92共2兲, 1–26. 685–696.
El-Sharnouby, B., and Novak, M. 共1990兲. “Stiffness constants and inter- Poulos, H. G. 共1979兲. “Settlement of single piles in nonhomogeneous
action factors for vertical response of pile groups.” Can. Geotech. J., soil.” J. Geotech. Engrg. Div., 105共5兲, 627–641.
27, 813–822. Poulos, H. G., and Davis, E. H. 共1968兲. “The settlement behavior of
Fellenius, B. H., Altaee, A., Kulesza, R., and Hayes, J. 共1999兲. “O-cell single axially loaded incompressible piles and piers.” Geotechnique,
testing and FE analysis of 28-m-deep barrette in Manila, Philippines.” 18共3兲, 351–371.
J. Geotech. Geoenviron. Eng., 125共7兲, 566–575. Poulos, H. G., and Davis, E. H. 共1980兲. Pile foundation analysis and
Fleming, W. G. K., Weltman, A. J., Randolph, M. F., and Elson, W. K. design, Wiley, New York.
共1992兲. Piling engineering, Blackie, Glasgow, U.K. Rajapakse, R. K. N. D. 共1990兲. “Response of an axially loaded elastic pile
Guo, W. D. 共2000兲. “Vertically loaded single piles in Gibson soil.” J. in a Gibson soil.” Geotechnique, 40共2兲, 237–249.
Geotech. Geoenviron. Eng., 126共2兲, 189–193. Randolph, M. F. 共1981兲. “The response of flexible piles to lateral load-
Guo, W. D., and Lee, F. H. 共2001兲. “Load transfer approach for laterally ing.” Geotechnique, 31共2兲, 247–259.
loaded piles.” Int. J. Numer. Analyt. Meth. Geomech., 25共11兲, 1101– Randolph, M. F., and Wroth, C. P. 共1978兲. “Analysis of deformation of
1129. vertically loaded piles.” J. Geotech. Engrg. Div., 104共12兲, 1465–1488.
Guo, W. D., and Randolph, M. F. 共1997兲. “Vertically loaded piles in Seed, H. B., and Reese, L. C. 共1957兲. “The action of soft clay along
non-homogeneous media.” Int. J. Numer. Analyt. Meth. Geomech., friction piles.” Trans. Am. Soc. Civ. Eng., 122, 731–754.
21共8兲, 507–532. Seo, H., and Prezzi, M. 共2007兲. “Analytical solutions for a vertically
Kim, S., Jeong, S., Cho, S., and Park, I. 共1999兲. “Shear load transfer loaded pile in multilayered soil.” Geomech. Geoeng., 2共1兲, 51–60.
characteristics of drilled shafts in weathered rocks.” J. Geotech. Vallabhan, C. V. G., and Mustafa, G. 共1996兲. “A new model for the
Geoenviron. Eng., 125共11兲, 999–1010. analysis of settlement of drilled piers.” Int. J. Numer. Analyt. Meth.
Kodikara, J. K., and Johnston, I. W. 共1994兲. “Analysis of compressible Geomech., 20, 143–152.

430 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / MARCH 2009

Downloaded 11 Jun 2012 to 180.211.192.67. Redistribution subject to ASCE license or copyright. Visit http://www.ascelibrary.org