Beruflich Dokumente
Kultur Dokumente
Abstract: Rectangular barrettes are increasingly being used to support large-size and heavy-duty structures, but the in-
teraction among barrettes, soil, and cap has rarely been studied theoretically. This paper presents an approximate three-
dimensional semi-analytical method for the analysis of load–displacement behaviour of single barrettes, barrette groups,
and barrette–soil–cap interaction systems. A unique feature of a barrette, which distinguishes it from a circular pile, is
its nonaxisymmetrical mechanical behaviour. To take into account this feature, both the barrette–soil and the cap–soil
interfaces are discretized. Mindlin’s solution is adopted to define the load–displacement relationship of the soils next to
the barrette and the cap. By assuming the deformation compatibility at the barrette–soil and cap–soil interfaces, the
load–displacement relationship of the soils is incorporated into the static force equilibrium conditions in the interior of
the barrette and cap structures. In this way, governing equations in finite difference form are derived for obtaining the
load–displacement response of the barrette–soil–cap system. The proposed method is verified by comparing the calcu-
lated results for a group of square piles using other existing methods. In addition, some factors such as barrette shape,
barrette spacing, and barrette group layout and finite-layer depth, which influence the response of the barrette–soil–cap
system, are investigated.
cause of their brief history (only about four decades) and the of elements counted consistently from a corner and a longer
complexities and difficulties in analysing the nonaxisym- side of the barrette, j = 1, …n and I = 1, …M. A base element
metrical mechanical behaviour. Without design parameters is denoted as bi, where i represents the number of elements,
from costly and time-consuming full-scale loading tests, bar- i = 1, …m. The vertical load applied at the top of the barrette
rettes have been commonly designed as drilled shafts or bored is P0. The shaft resistance (i.e., the side shear stress) on an
piles of equal cross-sectional area by ignoring the effect of arbitrary shaft element Lk at the Lth layer and the end-bearing
the geometrical shape on the load-carrying performance. resistance on a base element bk are designated by pLk and
To improve the understanding of the load-carrying behav- pbk, respectively.
iour of rectangular barrette–soil–cap interaction, this paper According to Mindlin’s solution (Mindlin 1936) and the
presents a boundary element method for the analysis of the principle of superposition, the vertical displacements, w IjJ
stresses and displacements at the barrette–soil and cap–soil and w biJ , at the corresponding centrepoints of shaft element Ij
interfaces. The nonaxisymmetrical mechanical feature of the and base element bi of an arbitrary Jth barrette of a group of
barrette is taken into account by discretizing the interfaces N barrettes may be calculated as follows:
three-dimensionally. Mindlin’s displacement solution (Mindlin
1 N M n m
1936) is adopted to define the load–displacement relation- [1] w IjJ =
Es
∑ ∑ ∑ Is IjJ , LkK p LkK + ∑ Is IjJ , bkK p bkK
ships of the discretized interface elements. Combining these K =1 L =1 k =1 k =1
relationships with the static force equilibrium conditions of
the barrette and cap structures, governing equations are derived 1 N M n m
in finite difference form for obtaining the load–displacement
[2] w biJ =
Es
∑ ∑ ∑ Is biJ , LkK p LkK + ∑ Is biJ , bkK p bkK
K =1 L =1 k =1 k =1
relationship of the barrette–soil–cap system, together with
the stresses and displacements at the barrette–soil and cap– where Es is the Young’s modulus of the soils next to the bar-
soil interfaces. Firstly the governing equations for a single rette; IsIjJ,LkK and IsbiJ,LkK, and IsIjJ,bkK and IsbiJ,bkK are the soil
barrette are derived, followed by those for a freestanding displacement influence factors, which represent the vertical
barrette group and a barrette–soil–cap system. Then the pro- displacements at the centrepoints of elements Ij and bi of the
posed method is verified by comparing with other existing K
Jth barrette induced by a unit shaft resistance, p Lk , on ele-
methods for a square pile–soil–cap system. Finally the K
ment Lk and a unit end-bearing resistance, p bk , on element
effects of barrette shape, barrette spacing, barrette arrange- bk of the Kth barrette, respectively; these influence factors
ment, and finite-layer depth on the response of the barrette– can be obtained by double integration of Mindlin’s equation
soil–cap system are investigated. as given in Appendix A.
Therefore, the vertical displacements at the centrepoints
Load–displacement relationship at the of all the elements at the barrette–soil interface of the group,
barrette–soil interface {ws}G, can be written in matrix form as
For a barrette embedded in an isotropic homogeneous soil, 1
the barrette–soil interface can be discretized three-dimensionally [3] {w s} G = [ Is ] G {p s} G
Es
as illustrated in Fig. 1. The length, width, and depth of the
barrette are designated by l, w, and d, respectively. The depth in which
of the barrette is equally divided into M layers. For each
layer, the longer and shorter sides of the barrette shaft are [4] {w s} G = {w 11
1
K w 11n K w 1M1 K w 1Mn w 1b1 K w 1bm
equally divided into ml and mw number of elements, re-
spectively. Therefore, the total number of the elements are K w 11
N
K w 1Nn K w M
N
1 K w Mn w b1 K w bm }
N N N T
Is11J,11K K Is11J,1nK K Is11J, M1K K Is11J, MnK Is11J, b1K K Is11J, bmK
M M M K M M M M M M
Is1nJ,11K K Is1nJ,1nK K Is1nJ, M1K K Is1nJ, MnK Is1nJ, b1K K Is1nJ, bmK
K K K K K K K K K K
Is M1J,11K K Is M1J,1nK K Is M1J, M1K K Is M1J, MnK Is M1J, b1K K Is M1J, bmK
[6] [ Is ] J, K =
M M M K M M M M M M
Is MnJ,11K K Is MnJ,1nK K Is MnJ, M1K K Is MnJ, MnK Is MnJ, b1K K Is MnJ, bmK
Is K Is b1J,1nK K Is b1J, M1K K Is b1J, MnK Is b1J, b1K K Is b1J, bmK
b1J,11K
M M M K M M M M M M
Is K Is bmJ,1nK K Is bmJ, M1K K Is bmJ, MnK Is bmJ, b1K K Is bmJ, bmK
bmJ,11K
Fig. 1. Three-dimensional discretization of a barrette–soil interface at: (a) barrette shaft, and (b) barrette base.
[7] {p s } G = {p11
1
K p11n K p1M1 K p1Mn p1b1 K p1bm Fig. 2. Stresses acting on a layer of barrette shaft.
K N
p11 K p1Nn K N
pM 1 K N
p Mn p bN1 K N T
p bm }
2 ( ml + m w )
where wI is the vertical displacement at the midpoint of the
+ δw ∑ p Ij = − p I Ith layer, and δz is the thickness of that layer (i.e., δz = d/M).
At the first layer of the barrette (i.e., I = 1) the normal
j = 2 ml + m w +1
stress is
∂w z σ P
[10] = − I [13] σ1 = 0
∂z I EP lw
© 2007 NRC Canada
784 Can. Geotech. J. Vol. 44, 2007
To obtain the finite difference formula for the first layer, Eplw
an imaginary layer of thickness δz, which has a midpoint dis- [16] pM = ( − 0.2wM − 2 + 2wM −1 − 5wM + 3.2 w b )
δ 2z
placement w0, is considered above the first real layer. There-
fore, from eqs. [10] and [13], the midpoint displacement of where wb is the vertical displacement of the barrette base.
the imaginary layer may be related to that of the first real To obtain the final equation required for the solution of
layer, w1, as the problem, eq. [10] may be applied to the barrette base, us-
δz P0 ing again a finite difference expression for an uneven inter-
[14] w 0 = w1 + val spacing (Mattes and Poulos 1969), as given by
Eplw
Eplw
By substituting eq. [14] into eq. [12] for I = 1, the shaft [17] pb = ( −1.33 f wM −1 + 12 f wM − 10.67 f w b )
δ 2z
resistance at the first layer may be derived as follows:
Eplw P0 where pb is the end-bearing resistance at the barrette base,
[15] p1 = (w 2 − w 1 ) + and f = δz /(4lw).
δ 2z δz Equations [12] and [15]–[17] can be written in matrix form as
For the bottom layer of the barrette (i.e., I = M) the shaft Eplw
[18] {p p} = [ Ip]{w p} + {Y}
resistance may be related to the vertical displacements of the δ 2z
midpoints of the (M–2)th, (M–1)th, Mth layers and the base
of the barrette. Using a finite difference formula for points in which
with nonuniform spacing (Mattes and Poulos 1969), eq. [11]
becomes [19] {p p} = {p1 p2 K pM p b} T
−1 1 0 0 0 K 0 0 0 0 0
1 −2 1 0 0 K 0 0 0 0 0
0 1 −2 1 0 K 0 0 0 0 0
[20] [ Ip] = K K K K K K K K K K K
0 0 0 0 0 K 0 1 −2 1 0
0 0 0 0 0 K 0 − 0.2 2 −5 3.2
0 0 0 0 0 K 0 0 −1.33 f 12 f −10.67 f
[21] {w p} = {w 1 w2 K wM w b} T
T
P
[22] {Y} = 0 0 K 0
δz
where {pp} is a vector of the shaft resistance at all the layers of the barrette and the end-bearing resistance at the base; {wp} is
a vector of the midpoint displacements of the layers and the base; [Ip] is a matrix of the displacement influence factors for the
barrette; {Y} is a coefficient vector; {pp}, {wp}, and {Y} are column vectors of the order of (M + 1); and [Ip] is a matrix of the
order of (M + 1) by (M + 1).
where [A] is a coefficient matrix of the order of MP by (M + 1); the symbol under a flat-lying brace represents the number of terms
involved in the coefficients over the brace; and a matrix superscripted by T represents the transpose of the matrix.
Equation [9] can be written in matrix form as
[25] {p p} = [ B ]{p s }
in which
1δlL δ L δlL δ L 0L 0 0L 0L 0L 0L 0L 0
23 1w23 12 3 1w23
ml mw ml mw
L L L L L L L L L L
[26] [ B] = 0L 0L 0L 0L 0L 0 δlL
123
δ L
1w23
δlL
123
δ L
1w23
0L 0
ml mw ml mw
1 1
0L 0L 0L 0L 0L 0 0L 0L 0L 0L L
m m
123
m
4P0 δz − p1 δ2z 1 N J δz N
[29] S 0 = w1 + [35] SG = ∑ w1 + 4 PG − δ z ∑ p1J
8Eplw N J =1 8 Eplw J =1
From eq. [31], S 0K in eq. [34] equals SG in eq. [35], and
A barrette group hence
PG 2 Eplw 1 N
Force equilibrium equations of a barrette group with a [36] P0K =
N
+
δ z N
∑ w1J − w1K
rigid cap not in contact with soil J =1
For a barrette group with a rigid cap not in contact with
δz 1 N
soil, the displacements at the top of all of the barrettes are −
4 N
∑ p1J − p1K
equal to the displacement of the cap, SG, and the load carried J =1
by the cap is a summation of the loads carried by all the bar-
rettes, PG. These relationships can be expressed as Substituting eq. [36] into eq. [33] yields
1 3
0 L 0
Therefore, eq. [37], instead of eq. [15], should be used as the
first row of the matrix in eq. [32]. For other layers, their fi- 4N + 4
nite difference formulae are the same as eqs. [12], [16], and 0 1 L 0
when J = K
[17]. Based on eqs. [12], [16], [17], [32], and [37], the force
equilibrium equation of each layer of all of the barrettes in M M M M
0 0 L 1
the group can be expressed in matrix form as follows: [40] [U ] J, K =
L 0
Eplw 1
0
[38] [U] G {p p} G = [ Ip] G {w p} G + {Y} G 4N
δ 2z 0 1 L 0 when J ≠ K
M M M M
in which
0 0 L 1
[U ]1,1 L [U ]1, N
[39] [U ] G = M O M [41] {p p} G = {p11 L p1M p1b L p1N L p M
N
p bN } T
[U ] N,1 L [U ] N, N
[ Ip]1,1 L [ Ip]1, N
[42] [ Ip] G = M O M
[ Ip] N,1 L [ Ip] N, N
2
N −3 1 0 0 L 0 0 0 0
1 −2 1 0 L 0 0 0 0
0 1 −2 1 L 0 0 0 0
M M M M M M M M M when J = K
0 0 0 0 L 1 −2 1 0
0 0 0 0 L −0.2 2 −5 3.2
[43] [ Ip] J, K = 4 32
0 0 0 0 L 0 − f 12 f − f
3 3
2
0 0 0 L 0 0 0 0
N
0 0 0 0 L 0 0 0 0
when J ≠ K
M M M M M M M M M
0 0 0 0 L 0 0 0 0
0 0 0 0 L 0 0 0 0
[44] {wp} G = {w 11 L w 1M w 1b L w 1N L w M
N
w bN } T displacement influence factors for the barrettes, whose
submatrix [Ip] J,K represents the displacement of the Jth
barrette induced by a unit shaft or end-bearing resistance
T
at the Kth barrette–soil interface; {U} G and {Y} G are co-
PG PG efficient matrices; {p p} G, {w p} G, and {Y} G are column
[45] {Y} G = 0 L 0 L 0 L 0 vectors of the order of [(M + 1) × N]; the submatrices
1 Nδ z Nδ z
442443 1442443 [Ip]J,K and [U]J,K are of the order of (M + 1) by (M + 1); and
M +1 M +1 [Ip]G and [U]G are matrices of the order of [(M + 1) × N] by
[(M + 1) × N].
where {p p} G is a vector of the shaft resistance at all of
the layers and the end-bearing resistance at all of the Force equilibrium equations of a freestanding barrette
bases of the barrettes in the group; {w p} G is a vector of group
the midpoint displacements of the layers and the dis- A freestanding barrette group refers to a group of barrettes
placements of the bases; the superscript of each coeffi- without the cap. In this case, the load at the top of each bar-
cient of vectors {p p} G and {w p} G denotes the identity rette is known. Therefore, for an arbitrary Kth barrette, the
number of a given barrette; [Ip]G is a matrix of the force equilibrium equations are the same as those derived
from a single barrette, namely eqs. [18]–[22]. By combining Fig. 3. Discretization of a cap–soil interface.
eqs. [18]–[22] for each barrette, the force equilibrium equations
of a freestanding barrette group can be expressed as equations
in the same form as eqs. [38]–[45], except that [U]G is an
identity matrix of the order of [(M + 1) × N] by [(M + 1) × N]
and eqs. [42] and [45] shall be replaced, respectively, by the
following two equations:
[ Ip] [ 0] L [ 0]
[ 0] [ Ip] L [ 0]
[46] [ Ip] G =
M M M M
[ 0] [ 0] L [ Ip]
T
A barrette–soil–cap interaction system
P01 P0N
[47] {Y} G = 0 L 0 L L 0 L 0 For a barrette group with a rigid cap in contact with soil,
1 δz δ
44244 3 1z442443 the cap–soil interface can be discretized as illustrated in
M +1 M +1 Fig. 3, where the Arabic numerals represent the identities of
the barrettes. The lengths of the overhang of the cap in the x
Load–displacement governing equation of a barrette and y directions are designated by ox and oy, respectively.
group The side-to-side spacing among barrettes in the x and y di-
Provided that displacement is compatible at the barrette– rections is designated by sx and sy, respectively. The total
soil interface, the following equation can be derived for a number of the discretized cap–soil interface elements is MC.
barrette group with no cap connected or with a rigid cap not Since the cap is rigid, the vertical displacement at the top of
in contact with soil: the cap, S, will be the same as that at the top of each barrette
in the group. The total load carried by the barrette–soil–cap
[48] {w s} G = [ A] G {wp} G system, P, is the summation of the contact pressure at the
cap–soil interface, PC, and the load carried by the barrettes,
in which PG. These relationships can be expressed as
[ A] [ 0] L [ 0]
[ 0] [ A] L [ 0] [53] P = PC + PG
[49] [ A] G =
[54] S = S 10 = S 02 = K = S 0K = K = S 0N
M M M M
[ 0] [ 0] L [ A]
Load–displacement relationship at the barrette–soil and
where [A]G is a coefficient matrix of the order of MG by cap–soil interfaces
[(M + 1) × N]. Similar to the derivation procedures for a single barrette
In addition, the shaft and end-bearing resistance on each and a barrette group, the load–displacement relationship at
layer of the barrettes can also be related to those on each el- the barrette–soil and cap–soil interfaces may be expressed as
ement at the barrette–soil interface as follows:
{w s} C 1 [ Is ] C [ Is ] C , G {p s} C
[50] {p p} G = [ B] G {p s} G [55] = [ Is ] G {p s} G
{w s} G E s [ Is ] G , C
in which where {ws}C is a vector of the vertical displacements at the
[ B] [ 0] L [ 0] centrepoints of the elements at the cap–soil interface; {ps}C
[ 0] [ B] L [ 0] is a vector of the contact pressure on those elements; [Is]C
[51] [ B] G = and [Is]G,C are matrices of displacement influence factors
M M M M for soils next to the cap and the barrettes induced by a unit
[ 0] [ 0] L [ B] contact pressure at the cap–soil interface, respectively;
[Is]C,G and [Is]G are matrices of displacement influence fac-
where [B]G is a coefficient matrix of the order of [(M + 1) × N] tors for soils next to the cap and the barrettes induced by a
by MG. unit shaft or end-bearing resistance at the barrette–soil in-
From eqs. [3], [38], [48], and [50], it can be shown readily terface, respectively; {ws}C and {ps}C are column vectors
that with MC rows; [Is]C is of the order of MC by MC, [Is]C,G is
of the order of MC by MG, and [Is]G,C is of the order of MG
−1 by MC.
Eplw
[52] {ps}G = [ A] G [ Ip] G
−1
[U] G [ B] G − [ Is ] G
E s δ 2z
Force equilibrium equations of barrettes and cap
−1 Since the cap is rigid, the displacements at the centre-
× [ A] G [ Ip] G {Y}G points of the elements at the cap–soil interface are equal to
© 2007 NRC Canada
788 Can. Geotech. J. Vol. 44, 2007
S; hence the first row of the matrix in eq. [55] can be rear- placement at the top of each barrette can be derived as fol-
ranged as lows:
1 N J δz N 4P K δ − p1K δ 2z
[60] ∑ w1 + 4P − δ z ∑ p1J + 4{C}[ Is ] −C1 [ Is ] C , G {∑
ps} G = w 1K + 0 z
N0 J =1 8 Eplw J =1 8 Eplw
Rearranging eq. [60] yields the load at the top of the Kth barrette as
P 2 Eplw 1 N δz 1 N 1
[61] P0K =
N0
+
δ z N 0
∑ w1J − w1K −
4 N 0
∑ p1J − p1K + N {C}[ Is ] C−1 [ Is ] C , G {p s} G
J =1 J =1 0
Substituting eq. [61] into eq. [33] yields the shaft resistance at the first layer of the Kth barrette as
1 N
3 1 Ep l w K 2 N P
[62] ∑ p1J + 4 p1K − N0 δz
{C}[ Is ] C−1 [ Is ] C , G {p s} G = 2 w 2 + ∑ w1J − 3w1K + N
4 N0 J =1 δz N0 J =1 0 δz
Similar to the derivation procedures for eq. [38], the force Load–displacement governing equation of a barrette–
equilibrium equations for the barrettes in the barrette–soil–cap soil–cap interaction
interaction system can be expressed in matrix form as follows: Substituting eq. [56] into the second row of the matrix in
eq. [55] yields
Eplw
[63] [V1 ]{p s}G = [ Ip] C {w p} G + {Y} C
δ 2z 1
[66] {w s } G = S[ Is ] G , C [ Is ] −C1 {Va} + ([ Is ] G
Es
in which − [ Is ] G , C [ Is ] C−1 [ Is ] C , G ){p s} G
1
[64] [V1 ] = [U ] C [ B] G − {V b}{C}[ Is ] −C1 [ Is ] C , G In addition, eq. [58] can be rewritten in matrix form as
N0 δ z
1 δz
T [67] S= {Vc }{w s } G + ( 4P − δ z {Vd }{p s } G
N0 8 EplwN 0
[65] {V b} = 1 0 L 0 L 1 0 L 0
14243 14243
M +1 M +1 + 4{C}[ Is ] −C1 [ Is ] C , G {p s} G )
where [U]C, [Ip]C, and {Y}C are matrices in the same form in which
and of the same order as [U]G, [Ip]G, and {Y}G in eqs. [39]
and [40], [42] and [43], and [45], respectively, except that N
in these equations should be replaced with N0; [V1] is a coef- [68] {Vc} = 1 0 L 0 L L 1 0 L 0
ficient matrix of the order of [(M + 1) × N] by MG; {Vb} is a 14243 14243
M ×n+ m M ×n + m
coefficient vector of the order of [(M + 1) × N].
[69] {Vd} = δ l L δw L δlL δw L 0 L 0 L L δlL δw L δlL δw L 0 L 0
123 12 3 12 3 12 3 123 12 3 123 12 3
ml mw ml mw
14444444 4 244444444 3 1m4444444
l mw ml
4 24444444mw
4
3
M ×n + m M ×n + m
where both {Vc} and {Vd} are vectors of the order of MG. can be calculated using eq. [34]. For a barrette–soil–cap in-
Substituting eq. [67] into eq. [66] yields teraction system, substituting the solved {ps}G into eq. [63]
P δz yields the displacements at the midpoints of all the layers of
[70] [V2 ]{p s } G + {V3} = [V4 ]{w s} G the barrettes, namely {wp}G and S. Subsequently, the contact
2 Ep lwN 0 pressure on the elements at the cap–soil interface can be ob-
tained by substituting the derived {ps}G and S into eq. [56].
in which Following these solution procedures, a FORTRAN program
δz was developed to perform the numerical calculation. With
[71] [V2 ] = {V3}( 4{C}[ Is ] −C1 [ Is ] C , G − δ z {Vd }) the proposed method, not only the load–displacement rela-
8Ep lwN 0 tionship but also the shaft and end-bearing resistance can
1 be calculated for a single barrette, a barrette group, and a
+ ([ Is ] G − [ Is ] G , C [ Is ] −C1 [ Is ] C , G ) barrette–soil–cap system.
Es
The problem of finite-layer depth, df, which exists below
[72] {V3} = [ Is ] G , C [ Is ] −C1 {Va} the barrette base can be treated approximately using the
method proposed by Poulos and Davis (1968), that is, modi-
1 fying the displacement influence factors for a semi-infinite
[73] [V4 ] = [ I ] − {V3}{Vc} mass to those for a finite-layer depth. In addition, the pre-
N0
ceding analysis can be modified to take into account local
where [V2] and [V4] are matrices of the order of MG by MG; yielding at the barrette–soil interface in a similar manner to
{V3} is a column vector of the order of MG; and [I] is an that described by Mattes and Poulos (1969) and Poulos and
identity matrix of the order of MG by MG. Davis (1980). When local yielding occurs at an interface ele-
Equations [63] and [70] define the relationships of {ps}G ment, the compatibility of the barrette and soil displace-
versus {wp}G and {ps}G versus {ws}G, respectively. There- ments for that element will no longer hold. Any increase in
fore based on the relationship between {ws}G and {wp}G, de- load will cause a redistribution of stress among the remain-
fined by eq. [48], the following equation can be derived: ing elastic elements, and the new stress distribution along
the barrette can be analysed by setting the known ultimate
−1
δ δ2 values on the yielded elements and considering the compati-
[74] {p s } G = z z [ A] G [ Ip] −C1 [V1 ] − [V4 ] −1 [V2 ] bility of the barrette and soil displacements at the elastic ele-
E plw E plw
ments. In numerical implementation, when the calculated
shaft or end-bearing resistance on an element reaches its
P
× δ z [ A] G [ Ip] −C1 {Y} C + [V4 ] −1 {V3} predefined ultimate value, the displacements at the barrette–
2N 0 soil interface, {ws}G, are given by the original elastic dis-
placement matrix, eq. [3], by setting the ultimate value on
that element. The displacements of the layers of the barrettes,
Numerical implementation and modification {wp}G, are calculated in a similar manner to the fully elastic
Equations [28], [52], and [74] are the load–displacement case, except that eq. [10], which is applied to the first and
governing equations for a single barrette, a barrette group bottom layers in the elastic analysis, must now be applied to
with no cap connected or with a rigid cap not in contact with the layers nearest the top and bottom layers that have not al-
soil, and a barrette–soil–cap interaction system, respectively. ready yielded. By equating the interface and layer displace-
These explicit equations can be solved straightforwardly to ments at the elements that have not yielded, equations similar
obtain the shaft or end-bearing resistance of all of the ele- to eqs. [28], [52], and [74] are obtained, and these can be
ments at the barrette–soil interface, that is, {ps} and {ps}G. solved to determine a new stress distribution. This procedure
For a single barrette and a barrette group, substituting the can be repeated until no additional elements have yielded.
solved {ps} and {ps}G into corresponding eqs. [25] and [50] Apart from the above considerations, it may also be modi-
yields the shaft resistance of all of the layers of the barrettes fied to take into account vertically nonhomogeneous soils or
and the end-bearing resistance at the bases, namely {pp} and multilayered soils using the method proposed by Lee and
{pp}G. Subsequently, the vertical displacements at the mid- Poulos (1990), but this is beyond the scope of this paper.
points of the layers of the barrettes, {wp} and {wp}G, can be Without exception, the calculated results from the proposed
obtained by substituting the derived {pp} and {pp}G into cor- finite difference method are sensitive to the number of
responding eqs. [27] and [38]. For a barrette group with a discretized layers of the barrettes and elements of the barrette–
rigid cap not in contact with soil, the load at the top of each soil and cap–soil interfaces. In this study, the trial and error
barrette can be calculated using eq. [36]. For a freestanding method is used to find out the satisfied number of layers and
barrette group, the displacement at the top of each barrette elements for achieving convergent results (Hong 2004).
Verification of method Table 1. Load sharing (%) by cap and individual piles for a
3 × 3 pile group.
To verify the validity of the proposed method, a 3 × 3
square pile–soil–cap interaction system was analysed, and Pile 2
the results from the analysis were compared with those ob- Cap Pile 1 or 3 Pile 4
tained using a truly three-dimensional finite-element method Ep /Es = 200
by Ottaviani (1975), a boundary element method by Proposed method 22.7 10.0 7.9 5.7
Kuwabara (1989), and a variational method by Shen et al. Ottaviani (1975) 22.0 8.0 8.0 8.0
(2000) (the last two being derived from circular piles of
equal cross-sectional area). In this study, the pile cross- Ep /Es = 400
section is a square with side length l or w of 1 m; the pile Proposed method 17.5 11.0 8.3 5.6
depth d below the ground surface is 40 m; the spacing sx or Ottaviani (1975) 15.0 10.0 9.0 7.0
sy is 3 m; the cap is in contact with soil at a depth, dc, of Shen et al. (2000) 17.3 11.2 8.2 5.0
3 m; the cap overhang ox or oy is 0.5 m; the finite-layer
depth, df, is 60 m; the Young’s modulus of the pile, Ep, is Ep /Es = 800
20 GPa; and the Poisson’s ratio of the soil, νs, is 0.45. The Proposed method 14.5 11.8 8.4 4.9
square pile was discretized into M = 15 layers and ml = 5 el- Ottaviani (1975) 11.0 11.0 9.0 7.0
ements and mw = 5 elements, and the cap–soil interface was Kuwabara (1989) 11.0 12.0 9.0 5.0
discretized into MC = 108 elements. The loads taken by the
cap and individual piles calculated by the proposed method Ep /Es = 2000
are summarized in Table 1, where piles 1, 2 and 3, and 4 Proposed method 12.6 12.6 8.3 3.9
represent the corner, midside, and centre piles, respectively, Ottaviani (1975) 9.0 12.0 9.0 6.0
as shown in Fig. 3. It can be seen that the uniformly distrib-
Shen et al. (2000) 11.8 13.0 8.2 3.3
uted load sharing by the piles in the group for the case of
Ep/Es = 200 by Ottaviani (1975) is questionable. The results
calculated using the proposed method are closer to those ob- Nonuniform shaft resistance distribution across a single
tained by Kuwabara (1989) and Shen et al. (2000) than to barrette
those obtained by Ottaviani (1975). This is likely because Using the proposed method, the shaft resistance distribu-
the former two methods were also derived on the basis of tion across a single barrette was investigated. For simplicity,
Mindlin’s solution. The little differences seen here are the ultimate shaft resistance is limited to 45 kPa along the
mainly due to the inherent assumptions of the different entire depth of the barrette, and the ultimate end-bearing re-
methods involved, for example, the square piles were sistance is limited to 800 kPa.
treated as circular piles of equal cross-sectional area by Figure 4 shows the calculated shaft resistance distribution
Kuwabara (1989) and Shen et al. (2000). Generally reason- across one quarter of the barrette at its mid-depth. At other
able agreement is obtained in the results between these dif- depths throughout the barrette, similar shaft resistance distri-
ferent methods. butions were also observed. When the load applied to the
barrette head is less than 8000 kN, the shaft resistance is not
uniformly distributed along the perimeter of the barrette.
The shaft resistance at the corner is larger than that mobi-
Results of analyses lized around the centre of the longer and shorter sides. The
shaft resistance at the centre of the shorter side is about
Many analytical methods have been applied to investigate twice the shaft resistance as that of the longer side. For the
the load-carrying behaviour of circular pile–soil–cap interac- loads less than 8000 kN, the shaft resistance distributions are
tion and its influencing factors, such as the pile length, pile the results of the elastic response of the barrette–soil inter-
spacing, pile-to-soil stiffness ratio, etc. (Poulos and Davis face. After the load is greater than 8000 kN, the shaft resis-
1980; Kuwabara 1989; Shen et al. 2000). By using the tance starts to become increasingly uniformly distributed
method proposed in this paper, these factors were also inves- across the barrette with increasing load, as a result of local
tigated, and similar results were obtained and are presented yielding at the barrette–soil interface. Such phenomena are
in detail in Hong (2004). In the following, some unique re- consistent with the results from a three-dimensional finite
sponses of a barrette–soil–cap system related to the difference analysis (Lei 2001) using FLAC3D (Itasca Con-
nonaxisymmetrical mechanical feature of the barrette are in- sulting Group, Inc. 1997).
vestigated. The length, width, and depth of the barrettes ana- From the above analyses, it can be concluded that a uni-
lysed are 2.8, 0.8, and 40 m, respectively. The Young’s formly distributed shaft resistance is not mobilized across
moduli of the barrette and soil are 20 GPa and 20 MPa, re- the barrette until failure of the barrette–soil interface has
spectively. It has been found that the Poisson’s ratio of the been reached. At the elastic stage, the shaft resistance mobi-
soil, νs, has a relatively small influence on the calculated re- lized at the corner of the barrette is greater than that around
sults (Hong 2004). The value of νs was chosen as 0.5, so that its shorter side, which in turn is greater than that around its
the analysis is particularly applicable to barrettes in un- longer side. The reason for this may be approximately ex-
drained soil where the load transfer takes place mainly by plained by the analytical solution for a rigid rectangular
interface adhesion. The barrette was discretized into M = 15 foundation, which shows similar distribution characteristics
layers and ml = 25 elements and mw = 7 elements. (of contact pressure) (Milovi 1992).
Effect of the arrangement of a barrette group sides and the shorter the shorter sides of the barrettes will
To study the effect of barrette arrangement on the load- be. When the barrettes are in transversal array, the opposite
carrying behaviour of a barrette group, a 1 × 2 group and a sides between the barrettes are the longer sides of the bar-
2 × 1 group were analysed using the proposed method. Each rettes, and hence the group effect increases with increasing
group has two barrettes with various side-to-side spacing of aspect ratio. When the barrettes are in longitudinal array,
0.1, 1.0, 5.0, and 20.0 m. The barrettes in the two groups the opposite sides between the barrettes are the shorter
are, respectively, in longitudinal and transversal arrays. Each sides of the barrettes, and hence the group effect decreases
barrette is subjected to a vertical load of 2000 kN. with increasing aspect ratio. From the analyses of the group
Figure 5 shows the calculated shaft resistance distribution effect it can be inferred that at any given displacement the
across one half of a given barrette at its mid-depth in the maximum load applied to the barrette group increases with
groups. It can be seen that the shaft resistance at the decreasing group effect.
barrette–soil interface is much smaller on the opposite sides It can also be seen from Fig. 6 that with increasing bar-
between the barrettes than on the other sides because of the rette spacing, the ratio of SG/S0 decreases, which means that
group effect. With increasing barrette spacing, the shaft re- the group effect decreases. For the barrettes considered, the
sistance on the sides inside and outside the group gradually group effect starts to attenuate appreciably after the spacing
increases and decreases, respectively. When the barrette is greater than 2.0 m.
spacing is far enough to eliminate the group effect, the dif-
ference between them becomes negligible, and the shaft re- Load-carrying behaviour of a barrette–soil–cap
sistance distribution approaches that for a single barrette. interaction
These results are consistent with those obtained by Ottaviani Using the proposed method, a 3 × 3 barrette–soil–cap sys-
(1975). tem was analysed to investigate its load-carrying behaviour.
Figure 6 shows the calculated settlements of different The overhang ox or oy and embedment depth dc of the cap
groups, SG, normalized by the corresponding settlement of are 0.5 m and 0 m, respectively. Table 2 shows the calcu-
a single barrette, S0. The barrettes in these groups are of lated loads taken by the cap and individual barrettes of vari-
equal cross-sectional area of 2.24 m2 but with different as- ous side-to-side spacing sx and sy of 3, 5, 10, and 20 m. It
pect ratios of l/w = 1.0, 3.5, and 10.0. The calculated values can be seen that the corner and centre barrettes take the
of SG/S0 are plotted against the barrette spacing on a loga- maximum and minimum loads, respectively. This is consis-
rithmic scale. The magnitude of SG/S0 reflects the group ef- tent with the results from the elastic analysis using other
fect. The higher the ratio of SG/S0, the more significant the methods (Kuwabara 1989). With increasing barrette spacing,
group effect will be, and vice versa. It can be seen from namely increasing area of the cap–soil interface, the load
Fig. 6 that at a given barrette spacing, the group effect for taken by the cap increases, as expected; the loads taken by
barrettes in transversal array (full curves) increases with in- the corner and midside barrettes (c.f., Fig. 3) decrease ac-
creasing aspect ratio; but the group effect for barrettes in cordingly; and the load taken by the centre barrette de-
longitudinal array (broken curves) decreases with increas- creases very slightly. For the barrette group with cap, the
ing aspect ratio. This is because on an equal cross-sectional difference between the loads taken by the corner, midside,
area basis, the higher the aspect ratio, the longer the longer and centre barrettes is relatively larger than that for the bar-
Fig. 5. Shaft resistance distribution across barrettes: (a) in transversal array, and (b) in longitudinal array.
rette group without cap. In other words, with the presence of Effect of finite-layer depth
the cap, the degree of nonuniformity of the load distribution Figures 7 and 8 show the calculated percentage of the loads
on the barrettes of the group is higher than that without the taken by the cap and individual barrettes and the calculated
cap. This indicates that besides minimizing the differential settlement of the cap in the barrette–soil–cap interaction sys-
settlement of the pile group with a flexible cap (Randolph tem with finite-layer depths df varying from 1.1d to infinity. It
2003), an optimized design would also be necessary to de- can be seen that when the finite-layer depth is greater than
termine the layout and geometry of the barrettes by mini- twice the barrette depth d, the finite-layer depth has little in-
mizing the differential load sharing of a barrette group with fluence on the load sharing and the settlement of the barrette–
a rigid cap. soil–cap system. Therefore, for practical problems, it appears
Fig. 6. Influence of barrette spacing on the group effect. Fig. 7. Effect of finite-layer depth on load sharing.
ox, o y cap overhangs in the x and y directions wIjJ centrepoint displacement of shaft element Ij of Jth barrette
pb end-bearing resistance at barrette base {wp} vector of displacements of the layers of a barrette
pbk end-bearing resistance of the kth base element {wp}G vector of displacements of the layers of barrettes
pI shaft resistance at the Ith layer of a barrette {wp}K vector of displacements of the layers of Kth barrette
pIj shaft resistance of the jth element at the Ith layer of a {ws} vector of displacements of the elements at the barrette–
barrette soil interface
pIK shaft resistance at the Ith layer of the Kth barrette {ws}C vector of displacements of the elements at the cap–soil
K
pbk end-bearing resistance of element bk of the Kth barrette interface
pIjK shaft resistance of element Ij of the Kth barrette {ws}G vector of displacements of the elements at the barrette–
pLk shaft resistance of the kth element at the Lth layer of a soil interface in a group
barrette (x,y,z) co-ordinates of the centrepoint of element Ij
P load at the top of a cap {Y} coefficient vector related to a single barrette
P0 load at the top of a barrette {Y}C coefficient vector related to barrette–soil–cap interaction
P0K load at the top of the Kth barrette {Y}G coefficient vector related to a barrette group
PC load carried by the cap–soil interface {Y}K coefficient vector related to the Kth barrette
PG load carried by a barrette group Z1, Z2 temporary variables
{pp} vector of shaft or end-bearing resistance of the layers of
a barrette
Appendix A. Expression for the soil
{pp}G vector of shaft or end-bearing resistance of the layers of
barrettes displacement influence factor
{pp}K vector of shaft or end-bearing resistance of the layers of Let (x, y, z) and (u, v, c) represent the co-ordinates of the
Kth barrette centrepoints of elements Ij and Lk at the barrette shaft, re-
{ps} vector of shaft or end-bearing resistance of the elements spectively, as shown in Fig. A1. According to the Mindlin
at the barrette–soil interface solution (Mindlin 1936), the vertical displacement dw IjJ , of
{ps}C vector of contact pressure of the elements at the cap– the centrepoint of element Ij of the Jth barrette induced by a
soil interface K
differential load, dp Lk , acting on element Lk of the Kth bar-
{ps}G vector of shaft or end-bearing resistance of the elements rette may be expressed as (Vaziri et al. 1982):
at the barrette–soil interface in a group
K
R1, R2 temporary variables dp Lk
[A1] dw IjJ = Iθ
S displacement at the top of a cap Es
S0 displacement at the top of a barrette
S 0K displacement at the top of Kth barrette in which
SG displacement at the top of a barrette group with cap not
in contact with soil (1 + v s ) 1 Z 22 2cz Z 12
sx, sy side-to-side spacing among barrettes in the x and y di- [A2] Iθ = (3 − 4v s ) + 3 − 3 + 3
8π (1 − v s ) R1 R2 R2 R1
rections
σ1 normal stress at barrette head 6czZ 22 8 (1 − v s ) 2 − (3 − 4v s )
σI normal stress at the Ith layer of a barrette + +
σz normal stress along the depth of a barrette R25 R2
(u,v,c) co-ordinates of the centrepoint of element Lk
[U ]C coefficient matrix related to a barrette–soil–cap interaction [A3] Ri = ( x − u) 2 + ( y − v) 2 + Z i2 (i = 1, 2)
[U ]G coefficient matrix related to a barrette group
[U ]J , K submatrix of [U]G where Z1 = z – c; Z2 = z + c; and νs is Poisson’s ratio of the
[V1 ] coefficient matrix soil.
[V2 ] coefficient matrix If the plane of element Lk is parallel to the yz plane in
{V3} coefficient vector Fig. A1, then
[V4 ] coefficient matrix c2 v2
{Va} coefficient vector 1
[A4] Is IjJ, LkK = ∫ ∫ E s Iθ dvdc
{Vb} coefficient vector
c1 v1
{Vc} coefficient vector
{Vd} coefficient vector If the plane of element Lk is parallel to the xz plane, then x
w width of barrette and y in eq. [A3] should be interchanged (Vaziri et al. 1982).
wb displacement of barrette base In addition, the displacement influence factor, IsbiJ,LkK, for the
wbi centrepoint displacement of ith base element soil next to element bi at the base of the Jth barrette induced
wI midpoint displacement at the Ith layer of a barrette by a unit shaft resistance of element Lk of the Kth barrette can
wIj centrepoint displacement of the jth element at the Ith also be calculated by substituting z = d into eqs. [A2]–[A4].
layer of a barrette Similarly, the displacement influence factor for soil next
wz vertical displacement along the depth of a barrette to element Ij of the Jth barrette induced by a unit end-
wIK midpoint displacement at the Ith layer of Kth barrette bearing resistance on element bk at the base of the Kth bar-
J
wbk centrepoint displacement of base element bk of Jth barrette rette may be calculated by
Fig. A1. Three-dimensional discretization of a barrette–soil interface at: (a) barrette shaft, and (b) barrette base.
u2 v2
1 References
[A5] Is IjJ, bkK = ∫ ∫ E s Iθ dvdu
u1 v1 Mindlin, R.D. 1936. Force at a point in the interior of a semi-
infinite solid. Physics, 7: 195–202.
The results of integrating eqs. [A4] and [A5] can be ob- Vaziri, H., Simpson, B., Pappin, J.W., and Simpson, L. 1982. Inte-
tained from Vaziri et al. (1982). grated forms of Mindlin equation. Géotechnique, 32(3): 275–278.