Vous êtes sur la page 1sur 20

INTERNATIONAL JOURNAL FOR NUMERICAL AND ANALYTICAL METHODS IN GEOMECHANICS

Int. J. Numer. Anal. Meth. Geomech., 2003; 27:1235–1254 (DOI: 10.1002/nag.319)

Inclined load capacity of suction caissons

C. P. Aubeny1,n,y,z, S. W. Han1,} and J. D. Murff2,}


1
Texas A&M University, College Station, TX 77843–3136, U.S.A.
2
Tierra, Inc., Raleigh, NC 27615, U.S.A.

SUMMARY
A simplified method of analysis for estimating lateral load capacity of suction caisson anchors based on an
upper bound plasticity formulation is presented. The simplification restricts the analysis to caissons in
uniform and linearly varying undrained strength profiles; nevertheless, its computational efficiency permits
quick evaluation of a number of parameters affecting load capacity. The validity and limitations of the
simplified formulation are demonstrated through comparisons to more rigorous finite element solutions.
A series of sensitivity studies demonstrate the effects of various soil conditions and loading parameters.
Copyright # 2003 John Wiley & Sons, Ltd.

KEY WORDS: suction caissons; anchors; lateral loading; plasticity; offshore; plastic limit analysis

INTRODUCTION

Suction caissons are large, vertical pipes, closed at the top and open at the bottom, for offshore
facilities. They are installed by pressure drawdown within the cylinder, referred to as ‘suction’,
after partial penetration of the cylinder due to its dead weight. They can have significant
advantages over conventional piles and have been used as both fixed foundations and anchors
for a variety of offshore structures. In deep water petroleum exploration and production
operations, floating systems moored to anchors in the seafloor are typically the alternative of
choice. The loading directions on the mooring anchors vary with concept from primarily vertical
uplift force for the tension leg platform (TLP), to horizontal loading for a catenary moored
floating production system, to inclined loading for a taut moored spar. In deep and ultra-deep
waters, taut mooring systems are increasingly attractive due to the excessive lateral distances

n
Correspondence to: Prof. C. P. Aubeny, Texas A&M University, CE/TTI Building Room 709A, College Station,
TX 77843, USA
y
E-mail: caubeny@civilmail.tamu.edu
z
Assistant Professor.
}
Geotechnical Staff Professional.
}
Visiting Professor.
Contract/grant sponsor: Department of the Interior Minerals Management Service; contract/grant number: 1435-01-99-
CA-31003
Contract/grant sponsor: Offshore Technology Research Center
Contract/grant sponsor: University of Texas

Received 7 September 2001


Revised 20 August 2002
Copyright # 2003 John Wiley & Sons, Ltd. Accepted 5 June 2003
1236 C. P. AUBENY, S. W. HAN AND J. D. MURFF

associated with catenary mooring systems; hence, the behaviour of anchors subjected to inclined
loads is becoming of increasing importance.
This paper presents a plastic limit approach for estimating the capacity of suction caissons
subjected to inclined loads. The formulation is subject to the restrictions of undrained loading
conditions, a linearly varying strength profile, and isotropic strength characteristics. The goal in
developing this model was to provide a relatively straightforward analysis method accessible to
practicing engineers while retaining much of the rigor of more complex analytical and numerical
approaches.

BACKGROUND

An upper bound plastic limit approach to lateral load capacity of piles and caissons was first
proposed by Murff and Hamilton [1]. Their model assumes a collapse mechanism (Figure 1)
comprised of a surface failure wedge, a plane strain flow-around zone, and a hemispherical
failure surface at the pile tip. For the surface wedge and hemispherical mechanisms, Murff and
Hamilton [1] derive dissipation functions based on three-dimensional strain fields, von Mises
and Tresca yield criteria, and an associated flow rule. For the flow-around zone, they used a
simplified energy dissipation function using the bearing factors derived by Randolph and
Houlsby [2] for plane strain flow around an infinitely long cylinder. By optimizing four
geometric parameters, Murff and Hamilton obtain a minimum pile lateral load capacity
associated with the above assumed failure mechanism. The interfaces between the three
mechanisms in Figure 1(a) are not kinematically compatible and the flow-around zone does not
necessarily deform purely in a plane strain mode except for the case of a translating caisson.
Hence, the Murff–Hamilton solution is not strictly speaking a rigorous upper bound, although it
employs a general upper bound approach. This model has been used extensively within the
industry and results have been validated by a number of investigators (e.g. References [3–5])
using more rigorous numerical solutions as well as experimental results.

1. Surface Failure
Applied
Force, F
2. Flow-Around Zone
Center of
Rotation

(a) 3. Spherical Failure

Applied 1. Unit Lateral


Force, F Resistance,
Np (z)
Center of
Rotation

(b) 2. Spherical Failure Surface

Figure 1. Full three-dimensional and simplified plastic limit models of suction Caissons: (a) failure
mechanism assumed by Murff and Hamilton (1993); and (b) simplified analysis by Aubeny et al. (2001).

Copyright # 2003 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2003; 27:1235–1254
LOAD CAPACITY OF SUCTION CAISSONS 1237

Murff and Hamilton [1] also derive relationships for a dimensionless lateral bearing capacity
factor Np relating unit resistance on the side of a caisson Pl to soil strength Su ; Pl ¼ Np Su : This
Np factor is a function of depth (Figure 1(b)) and depends on the soil strength profile. Murff and
Hamilton [1] present proposed functions Np ðzÞ for uniform and linearly varying strength
profiles. They found the resistance factor Np to be essentially independent of the deformation
mode (rotation or translation) of the caisson. Using these findings, Aubeny et al. [5] proposed a
simplified upper bound model, in which energy dissipation is computed by integrating the
product of unit resistance Pl times the local displacement over the projected side area of the
caisson. This model makes use of the concept of generalized yield conditions where the
resistances are generalized stresses and displacements and rotations are generalized strains. As
discussed by Prager [6] this is a valid basis for upper bound calculations as long as the forces and
displacements (or moments and rotations) are work conjugate pairs. The simplified model
retains the original formulation for energy dissipation at the bottom of the caisson; i.e. the
dissipation per unit area is integrated over the hemispherical slip surface. In the simplified
model, the displacements and rotations are a function of a single optimization variable, the
depth to the center of rotation of the caisson; hence, obtaining a solution (optimizing) is greatly
simplified. The reason for developing this model was to greatly simplify its use in practice and to
make it more accessible to practicing engineers than the more rigorous original model.
As noted above, versions of the original Murff–Hamilton model were formulated for both the
Tresca and von Mises yield criteria. The work by Aubeny et al. [5] and the present study utilize
Np profiles based on the latter; hence, a von Mises yield criterion is implicit in the analyses.

OVERVIEW OF PROPOSED FORMULATION FOR INCLINED LOADING

Basic elements of the proposed approach are as follows:


1. The proposed formulation extends the lateral load capacity analysis proposed by Aubeny
et al. [5] to general conditions of inclined loading. The simplified failure mechanism shown
in Figure 1(b) is modified to incorporate vertical rigid body displacements as shown in
Figure 2. The load F is inclined at an angle c measured from horizontal and the load
attachment depth Li is the location at which the line of action of the load F intersects the
centreline of the caisson.
2. Two interaction functions are developed and utilized in the plastic limit formulation. The
first characterizes the interaction between the ultimate skin resistance and the ultimate
average lateral resistance acting on the sides of the caisson under conditions of inclined
loading. Three-dimensional finite element [7] solutions are used to develop this interaction
diagram. The second characterizes the interaction among ultimate resistance to vertical,
horizontal and moment loading at the bottom of the caisson. A simplified form of an
interaction relationship proposed by Bransby and Randolph [8] is applied to the latter
bottom of the caisson.
3. A generalized method of plastic limit analysis [6,9] is employed, in which forces and
moments are characterized as generalized stresses while displacements and rotations are
characterized as generalized strains. In this framework, the interaction diagrams described
above are considered as generalized yield surfaces. By invoking an associated flow rule,
generalized displacements are related to generalized stresses. Displacements are derived

Copyright # 2003 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2003; 27:1235–1254
1238 C. P. AUBENY, S. W. HAN AND J. D. MURFF

ξV0
V0
F

Li

Li
ψ V

Lo

Lo
Lo-Li

Lo-Li
Lf

Lf
β β

z
Suction Caisson Velocity Field
Figure 2. Suction Caisson rigid body motions under inclined loading.

from collapse mechanisms corresponding to simultaneous rotation and upward translation


of the caisson. The case of lateral translation is a special case of caisson rotation with the
center of rotation at an infinite vertical distance above or below the caisson.
4. A minimum caisson load capacity is obtained by optimizing the kinematics of the caisson
rigid body motions with respect to two parameters: the center of rotation of the rotating
caisson, L0 ; and a parameter relating the magnitude of vertical to horizontal motion, x:
Incorporation of the vertical capacity in this manner was suggested by Randolph.
The development of the proposed formulation is presented below. The plastic limit formulation
is evaluated through comparisons with finite element simulations of suction caissons subjected
to inclined loads.

AXIAL-LATERAL RESISTANCE ON SIDES OF CAISSON

The proposed formulation expresses the ultimate unit axial and lateral resistance per unit
projected area along the side of the caisson, Pa and Pl respectively, as
Pa ¼ Nas ðcÞSu ð1aÞ

Pl ¼ Nps ðc; zÞSu ð1bÞ


where Su is the undrained strength of the soil at depth z; the axial resistance factor Nas is a
function of the load inclination angle c (measured from horizontal), and the lateral resistance
factor Nps is a function of c and the depth z below the mudline.
For the case of purely vertical loading, Nas ð908Þ will equal p for a circular caisson, while
Nps ð908; zÞ reduces to 0 for all z: In this case Equation (1) corresponds to the a-method for
estimating skin resistance for axially loaded piles, with an adhesion factor a ¼ 1: While the
proposed formulation can accommodate values of a less than unity, the remainder of this paper
will consider only the special case implied by Equation (1a). For the case of purely horizontal
loading, Nas ð08Þ equals 0 while Nps ð08; zÞ becomes the lateral resistance factor associated with a

Copyright # 2003 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2003; 27:1235–1254
LOAD CAPACITY OF SUCTION CAISSONS 1239

P -ultimate analysis of laterally loaded piles, e.g., References [10,11]. Nps in general varies with
depth z for all cases except for the limiting case of purely vertical loading. An analytical
approach to determination of Nps ð08; zÞ is given by Murff and Hamilton [8]. They present an
upper bound analysis based on a collapse mechanism comprised of a three-dimensional wedge
mechanism near the free surface and a flow around failure [2] at depth. Based on their analysis,
Murff and Hamilton [1] propose an empirical expression for Nps ð08; zÞ applicable to uniform and
linearly varying soil strength profiles.
The side resistance factors ðNas ; Nps Þ in Equation (1) are therefore defined for the limiting cases
of purely vertical and purely horizontal loading, c ¼ 90 and 08; respectively. For intermediate
cases, 05c5908; an interaction relationship between Nas and Nps is required. To define this
relationship, a series of three-dimensional FEM analyses were performed. The analyses
employed 8-node brick (linear displacement) elements using a Prandtl–Reuss material model;
i.e. a linearly elastic, perfectly plastic material obeying a von Mises yield criterion with an
associated flow rule. The FEM model employed elements having dimensions roughly equal to
the caisson radius R in the vicinity of the caisson–soil interface, with proportionately larger
elements at greater distances from the caisson. The Young’s modulus of the soil will not effect
the predicted ultimate lateral resistance of a translating caisson [5], although it will affect the
lateral displacement level at which the ultimate resistance is mobilized. These studies utilized a
rigidity E=Su ¼ 500; which resulted in ultimate capacities being mobilized at about d=D ¼ 0:1
(Figure 3). The FEM simulations assumed a rough (no slippage) soil–caisson interface and
complete contact on the back side of the translating caisson. In the FEM analyses the caisson is
idealized as a rigid cylinder translating with a constant ratio of vertical to horizontal
displacement. This approach assumes that the resistance factors ðNas ; Nps Þ are independent of
rotation, an assumption supported by the findings of Murff and Hamilton [1] and Matlock [10].
The computed nodal reaction forces corresponding to these imposed displacements are then
summed at a given elevation and related to axial and lateral resistance factors by:

Nas ¼ ðSRa Þc =Su DDL ð2aÞ

Nps ¼ ðSRl Þc =Su DDL ð2bÞ

Figure 3. Finite element predictions of unit lateral resistance on laterally translating caisson.

Copyright # 2003 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2003; 27:1235–1254
1240 C. P. AUBENY, S. W. HAN AND J. D. MURFF

where SRa and SRl are vertical and horizontal nodal reaction forces summed around the
circumference of the caisson at a selected elevation z ¼ c; DL is the tributary vertical dimension
corresponding to these reaction forces, and D ¼ caisson diameter:
The FEM simulations were conducted for a relatively long caisson (length-to-diameter ratio,
Lf =D ¼ 10). The Nas  Nps interaction relationships are assumed to be solely a function of
normalized depth z=D; i.e. the load distribution for a shorter caisson is assumed to be the same
as that in the upper portions of the Lf =D ¼ 10 caisson. This is again consistent with the results
of Murff and Hamilton [1].
Figure 3 presents a summary of an initial series of FEM predictions of normalized lateral unit
resistance (Equation (2b)) versus displacement d=D at various elevations z=D for purely
horizontal loading conditions. The FEM predictions show the load–displacement curves
coalescing at depths z=D of about 6 or greater; hence, conditions below this depth were
considered to correspond to the Randolph–Houlsby [2] flow-around plane strain solution. FEM
predictions of the ultimate normalized lateral unit resistance for all depths in this flow-around
region was about Nps ¼ 13:2 versus the Randolph–Houlsby value of 11.94; hence, the FEM
predictions over-estimate the actual solution by about 10 per cent. The FEM analyses also
predicted a lateral resistance factor near the mudline Nps ¼ 7:5; in contrast to the expected lateral
resistance factor near the free surface of a rough wall in a purely cohesive soil Nps ¼ 5:62: The
larger discrepancy between the FEM and reference values near the free surface is attributed to
the fact that the FEM value actually reflects average conditions down to the half-depth ðz=D ¼
1=2Þ of the first element rather than exactly at z=D ¼ 0; and Nps increases relatively rapidly at
this location (see Equation (4)).
A FEM study of an axially loaded caisson using the same three dimensional finite element
mesh showed the axial bearing factors Nas (Equation (2a)) to be equal to about 3.48 at all depths
z=D compared to a theoretical value of 3.14. This again amounts to the FEM solution over-
predicting reference values by about 10 percent.
Since the FEM predictions of ultimate resistance under conditions of pure lateral and upward
translation consistently over-estimated reference values by about 10 per cent, subsequent FEM
predictions for caisson translation in intermediate directions were assumed to be high by a like
amount. Accordingly, FEM ultimate load capacity estimates were reduced by 10 percent in the
development of the Nas  Nps interaction relationships.
Interpretation of the FEM predictions as described above gives the Nas  Nps interaction
relationships shown in Figure 4. Each data point shown represents the results of a FEM
analysis. Based on these studies the Nas  Nps interaction relationship at depth ðz=D > 0:8Þ was
formulated as follows:
Nps 58 Nas ¼ p ð3aÞ

Nps > 8 ðNps  I3 Þ2 þ ðNas Þ2 =I1  I2 ¼ 0 ð3bÞ


where I1 ; I2 ; and I3 are interaction coefficients equal to 0.68, 15, and 8, respectively. Equation
(3a) arises from the essential independence between vertical and lateral resistance indicated by
the FEM studies for Nps 58: Equation (3b) is an elliptical curve fit to the FEM resistance
predictions depicted in Figure 4.
Near the free surface the lateral resistance factor decreases. For the cases of uniform and
linearly increasing soil strength profiles, with Su0 being the soil strength at mudline and Su1 being
the vertical strength gradient, the reduction factor Rf can be expressed using the relations

Copyright # 2003 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2003; 27:1235–1254
LOAD CAPACITY OF SUCTION CAISSONS 1241

Figure 4. Axial and lateral resistance factors on side of caisson.

proposed by Murff and Hamilton [1]:


N1  N2 expðZz=DÞ
Rf ¼ ð4Þ
N1
where N1 is the limiting value for pure horizontal loading at depth, 11.94 and 9.42 for rough and
smooth boundaries, respectively. N2 is selected such that ðN1  N2 Þ is the intercept at the soil
surface; the intercept being 2.82 and 2 rough and smooth boundaries, respectively. If full
adhesion (no gap) is assumed to occur in the active wedge on the back side of the caisson, the
computed value of Rf in Equation (4) should be doubled, subject to the restriction that Rf does
not exceed one. The parameter Z in Equation (4) characterizes the effect of the soil strength
profile as follows:
Z ¼ 0:25 þ 0:05r for r56

Z ¼ 0:55 for r > 6

r ¼ Su0 =Su1D
Applying the reduction factor Rf to Equation (3) gives
Nps 58Rf Nas ¼ p ð5aÞ

Nps > 8Rf ðNps  I3 Rf Þ2 þ ðRf Nas Þ2 =I1  ðRf Þ2 I2 ¼ 0 ð5bÞ

DISSIPATION FUNCTION FOR WALLS OF CAISSON

The yield condition defined by Equation (5b) can be expressed in terms of resisting force per unit
length of caisson, Fas ¼ Pa D; Fls ¼ Pl D:
ðRf Fas =Su DÞ2
f ¼ ðFls =Su D  I3 Rf Þ2 þ  I2 R2f ¼ 0 ð6Þ
I1
The particle velocity at any point on the wall of the caisson is defined as proposed by Randolph.
A failure mechanism is postulated (Figure 2) comprising a rigid body rotation about a center of
rotation at a depth L0 below the mudline and an upward translation va ¼ xv0 ; where is v0 is the

Copyright # 2003 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2003; 27:1235–1254
1242 C. P. AUBENY, S. W. HAN AND J. D. MURFF

virtual lateral velocity at the mudline. L0 and x are optimization parameters used to obtain a
minimum upper bound solution for the horizontal and vertical components of the caisson load
capacity. The virtual velocity of the caisson at any depth is then
vl ¼ ð1  z=L0 Þv0 ð7aÞ

va ¼ xv0 ð7bÞ
A flow rule can now be expressed in terms of generalized stresses (force per unit depth, Fls ; and
Fas ) and generalized strain increments (velocities, vl and va )
 
@f Fls 1
vl ¼ l ¼ 2l  Rf I 3 ¼ b’ ðLo  zÞ ð8aÞ
@Fls Su D Su D
!
@f R2f Fas 1
va ¼ l ¼ 2l ¼ vo x ð8bÞ
@Fas Su D I 1 S u D

where l is a scalar controlling the magnitude of plastic strain and b’ ¼ v0 =L0 : Appendix C
obtains a solution of Equations (8a) and (8b) for load resistance per unit depth ðFls ; Fas Þ to yield
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Fls ¼ A4 þ A24  A5 ð9aÞ

Fas ¼ C2 Fls  C3 ð9bÞ


where the coefficients A4 ; A5 ; C2 and C3 are defined in Appendix C.
Finally, the forces per unit length times the local velocities (dissipation rates) are integrated
along the entire length of the caisson to obtain the energy dissipation rate due to side resistance:
Z
D’ s ¼ ðFas va þ Fls vl Þ dz ð10Þ

RESISTANCE INTERACTION AT THE CAISSON TIP

Soil resistance at the caisson tip will contribute to the anchor load capacity in the form of
rotational resistance to the moment loading and axial uplift resistance to vertical loading.
Bottom resistance can comprise a significant component of overall load capacity for relatively
short (small Lf =D) caissons. As the caisson length increases the relative contribution of the tip
resistance diminishes. The proposed formulation is a modification of the collapse mechanism
proposed by Murff and Hamilton [1]. This mechanism assumes a rigid rotating spherical soil
mass at the bottom of the caisson failing on a compatible slip surface. The center of rotation of
the mechanism is located at a depth L0 below the soil surface as previously defined for the lateral
resistance. The geometry of the postulated spherical surface, in terms of L0 ; is shown in Figure 5.
By invoking virtual work principles, the ultimate moment resistance at the base of the caisson
to pure rotation, Mb0 ; is derived [1] in terms of the radius of the caisson, Rð¼ D=2Þ; and the
distance from the axis of rotation to the caisson tip, Rl ð¼ Lf  L0 Þ: The subscript ‘‘0’’ indicates
the maximum capacities with no other components present. It is convenient to normalize
the moment capacity Mb0 by the moment capacity of a hemispherical slip surface centred at the
bottom of the caisson, ð¼ 0:5p2 R3 Suavg Þ: Further, by curve fitting numerical evaluations of

Copyright # 2003 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2003; 27:1235–1254
LOAD CAPACITY OF SUCTION CAISSONS 1243

Vo

Lo
ω R2
R1

R
Spherical End Cap
Figure 5. Failure Mechanism at Bottom of caisson.

the Murff–Hamilton [1] solution for Mb0 ; a simplified dimensionless expression for moment
capacity can be formulated:
   
Mb0 Rl I 4 Rl
¼2 þ exp ð11Þ
0:5p2 R3 Suavg pR R
where the fitting parameter, I4 ¼ 1:118: For large values of Rl =R; the failure approaches a
translational mode with the limiting moment approaching the asymptote of Equation (11)
pD2
ðMb0 Þtrans ¼ Rl ðSu0 þ Su1 Lf Þ ð12Þ
4
Two comments are in order prior to explaining the modifications of the above-mentioned
mechanism. First, for the simplified spherical mechanism there is a single optimization
parameter, the depth of the axis of rotation, L0 : As such, the interaction of the horizontal
resistance and the moment resistance (yield condition) is implicit in the optimized upper bound
solution. Second, a primary objective of the paper is to maintain simplicity in the analysis. As
such, no attempt will be made to modify the kinematics of the spherical failure mechanism to
account for the vertical displacement as this would add significant complexity. Rather, a
simplified empirical interaction equation between the moment capacity and the axial load
capacity will be incorporated. This will effectively decrease the overall capacity in the general
case where mechanisms combining vertical displacement and rotation are considered. The
following briefly describes this approach. It is well known that shallow footings with eccentric
loading have reduced bearing capacity. Various approaches have been taken to incorporate this
effect into classical bearing capacity theory [12, 13]. These usually assume that no uplift
resistance can be developed. Bransby and Randolph [8] have recently published results for
‘skirted foundations’ which are capable of sustaining uplift loads. They employed a ‘scoop’
mechanism similar to the spherical one used here but somewhat more elaborate. These
foundations effectively have some embedment and are similar to relatively short caissons. The
authors proposed an empirical interaction relationship which (including vertical and horizontal

Copyright # 2003 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2003; 27:1235–1254
1244 C. P. AUBENY, S. W. HAN AND J. D. MURFF

load and moment) of the form


  2 "    #1=2
Vb Mb 2:5 Hb 5
f ¼ þ þ 1 ¼ 0 ð13Þ
Vb0 Mb0 Hb0

where V and H are the vertical and horizontal load, respectively. The moment Mb is the moment
about the centre of rotation of the mechanism. For the purposes here we are seeking an equation
to represent the interaction between vertical load and moment. Again, interactions of these
components with the horizontal load are implicit in the mechanism. A simpler equation which is
a reasonable approximation for the above case where Hb ¼ 0 is
 2  
Vb Mb 2
f ¼ þ 1 ¼ 0 ð14Þ
Vb0 Mb0
Figure 6 shows a comparison of the two. This equation clearly captures the essential
characteristics of the interaction and should be quite adequate, particularly for the deeper
caissons where the tip contribution becomes less significant. In a later section this assumption
will be evaluated by comparing results with finite element analyses.
The maximum vertical resistance at the bottom of the suction caisson for purely vertical
loading is Vb0 ¼ pR2 Suavg Nab : A series of two-dimensional axisymmetric finite element analyses,
with element dimensions adjacent to the caisson being approximately one-quarter the caisson
diameter, predicted tip resistance bearing factors Nab of 6.5 at the mudline and 11.5 at depths
Lf =D between 3 and 4. Reducing the finite element estimates by 10 per cent for the reason
discussed earlier, the authors selected a profile of Nab that varies linearly from 6.0 for an aspect
ratio of zero (surface loading) to 10.5 for aspect ratios Lf =D greater than 3.

DISSIPATION FUNCTION FOR BOTTOM RESISTANCE

Now we will use the above-mentioned approach in which forces are treated as generalized
stresses and velocity and angular velocity are treated as generalized velocities. Since the effect of
horizontal load is implicit in the spherical slip surface mechanism, only that portion of the yield

Figure 6. Axial load–moment interaction.

Copyright # 2003 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2003; 27:1235–1254
LOAD CAPACITY OF SUCTION CAISSONS 1245

function defined by Equation (14) is relevant to energy dissipation at the bottom of the caisson;
hence
lð@f =@Mb Þ b’ b’ 1
¼ ¼ ¼ ð15Þ
lð@f =@Vb Þ v xvo xLo
2
1 Mb0
Mb ¼ 2
Vb ¼ C5 Vb ð16Þ
xLo Vb0
By substituting into the failure function and manipulating terms, we can find the vertical force,
Vb :
sffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1
Vb ¼ 2
Vb0 ð17Þ
C4 þ 1
where
2
1 Mb0 1 Mb0
C4 ¼ and C5 ¼ 2
Lo x Vb0 xLo Vb0
The final dissipation at the bottom of the suction caisson is then
D’ b ¼ Mb v0 =L0 þ Vb xvo ð18Þ

LOAD CAPACITY

If the total applied external force F e acting at the centreline of the caisson is inclined at an angle
c measured from horizontal, the axial and lateral components are related by
V
tan c ¼ ð19Þ
H
where, V is the externally applied load in vertical direction, H the externally applied load in
horizontal direction.
The external work rates performed by applied lateral and axial loads are:
W’ l ¼ H jLo  Li jb’ ð20aÞ

W’ a ¼ V xvo ¼ VLo xb’ ð20bÞ


where L0 is the depth to the centre of rotation and Li is the depth of the resultant applied load at
the centreline of the caisson. The depth of load application at the caisson centreline, Li ; is related
to the depth of an attachment point on the side of the caisson as illustrated by Figure 2.
The total external work is therefore
  
 Li 
W’ t ¼ H b’ Lo x tan c þ 1   ð21Þ
Lo
Equating the external work in Equation (21) to the energy dissipation from side resistance
(Equation (10)) and from bottom resistance (Equation (18)) leads to
R
ðFls j1  ðz=Lo Þj þ Fas xÞ dz þ ðMb =Lo Þ þ Vb x
H¼ ð22Þ
ðx tan c þ j1  Li =Lo jÞ

Copyright # 2003 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2003; 27:1235–1254
1246 C. P. AUBENY, S. W. HAN AND J. D. MURFF

The lateral component of load capacity is obtained by minimizing with respect to the
optimization parameters x and Lo : The vertical component of load capacity is then computed
from:
V ¼ H tan c ð23Þ

COMPARISONS TO FINITE ELEMENT PREDICTIONS

The upper bound plastic formulation developed in the preceding paragraphs was evaluated
through comparisons to a series of FEM simulations. In contrast to the FEM analyses described
earlier in the development of the vertical-lateral unit resistance relationships, these analyses
involved prescribed loading rather than prescribed displacements. Hence, the loaded caisson
was free to seek its own optimal displacement configuration. All FEM analyses utilized a
Prandtl-Reuss material model and the caisson was subjected to loads inclined from horizontal at
angles of c ¼ 0; 15; 30; 45; 60; 75; and 908: Simulations were made for load attachment points at
the mudline ðLi =Lf ¼ 0Þ and at the mid-depth of the caisson ðLi =Lf ¼ 0:5Þ: The FEM analyses
(Figure 7) considered a uniform soil strength profile ðSu1 ¼ 0Þ and three caisson aspect ratios
were considered, Lf =D ¼ 2; 6; 10:
Based on the comparisons of FEM solutions to reference values discussed earlier, the FEM
predictions were reduced by 10 per cent. It should be noted that this reduction will actually be
somewhat high when the motions of the caisson involve rotation about a point located within
the boundaries of the caisson, since the ultimate lateral resistance cannot be fully mobilized in
the vicinity of the center of rotation. Regarding this effect, Aubeny et al. [5] found that the 10
per cent error on the high side associated with their FEM approximation was offset by a 5–15
per cent under-estimate of capacity due to lack of full mobilization of resistance about the center
of rotation. Unfortunately, the latter error is dependent on the length Lf =D of the caisson}the
effect is most significant for short caissons}so applying a single correction factor to all FEM
predictions, which is reasonable when considering translating caissons, is not strictly
appropriate when caisson rotation occurs. Thus the 10 per cent corrective reduction to FEM
predictions can be considered appropriate when the caisson experiences pure translation (e.g.
pure vertical uplift), but this reduction may be excessive when the caisson displacement
configuration involves rotation.
Notwithstanding the difficulties described above in directly comparing FEM to plastic limit
analyses, the interaction relationships between vertical and horizontal load capacity presented in
Figure 7 indicate reasonable agreement for slender caissons (Lf =D ¼ 6 and 10). For these
slender caissons the load capacity estimates from the FEM and limit analysis solutions for
purely vertical loading are in excellent agreement. Larger discrepancies occur between the FEM
and limit analysis solutions when considering loads with large horizontal components,
particularly for the case of Lf =D ¼ 6: A plausible explanation for the differences is found in
the preceding paragraph; namely, the 10 per cent corrective reduction applied to the FEM
solutions is likely to be somewhat high, especially for shorter caissons.
For the shorter caissons, Lf =D ¼ 2; the discrepancies between the FEM and limit analysis
solutions are more serious, particularly when considering loads with a large vertical component.
These differences may be attributed in large measure to the simplified limit analysis model of
caisson vertical load capacity. This model was formulated assuming that side resistance is

Copyright # 2003 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2003; 27:1235–1254
LOAD CAPACITY OF SUCTION CAISSONS 1247

Figure 7. Plastic limit analyses and FEM solutions for uniform soil strength profiles.

comprised of the full shear strength of the soil acting along the entire side area of the caisson,
with no reductions to account for stress interactions occurring near the bottom of the caisson.
For slender caissons (Lf =D ¼ 6 and 10) this is apparently a reasonable assumption, while for
short caissons the side-bottom stress interaction effects appear to be of relatively greater
importance.

PARAMETRIC STUDY

With the upper bound plastic limit model thus validated, a series of parametric studies were
performed to assess the influence of load inclination angle c; depth of the load attachment point
Li =Lf ; caisson aspect ratio Lf =D; and soil strength profile on load capacity. With regard to the

Copyright # 2003 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2003; 27:1235–1254
1248 C. P. AUBENY, S. W. HAN AND J. D. MURFF

latter, linearly varying undrained shear strength profiles were considered having a strength Su0 at
the mudline with a strength gradient Su1 : Load interaction diagrams are conveniently expressed
by normalizing vertical and horizontal load capacity by Hmax corresponding to purely horizontal
translation. Plots of Vmax and Hmax for a number of strength profiles and caisson aspect ratios
are presented in Figure 8. It should be noted that the maximum capacities in Figure 8
correspond to the following conditions: full adhesion at the soil–caisson interface, full
mobilization of reverse end bearing at the base of the caisson, full suction (no gap) on the back
side of the caisson.
Figure 9 shows anchor capacity interaction relations for caissons with aspect ratios Lf =D ¼
2; 6; and 10 under uniform soil strength conditions ðSu1 D=Su0 ¼ 0Þ: For this strength profile, the
optimum depth of load attachment is about Li =Lf ¼ 0:5: It should be noted that, if no suction is
assumed to occur on the back of the failure wedge, the optimum load attachment point will be
significantly lower ðLi =Lf  0:75Þ; these conditions are considered more fully by Aubeny et al.
[5]. The characteristics of the interaction diagrams depend on the caisson aspect ratio and the
depth of the load attachment. For long caissons (ratios Lf =D ¼ 6 and 10) with the load
attachment depth near its optimum location, vertical-horizontal load interaction effects are
significant (i.e., the interaction diagram is curved) for load inclination angles c between 15 and
308: In contrast, for shorter caissons ðLf =D ¼ 2Þ; significant interaction effects occur for
inclination angles c between 30 and 458: Locating the load attachment point at depths other
than optimum will also effect on the characteristics of the interaction diagram. For example, for
a load attachment at the mud line ðLi =Lf ¼ 0Þ; interaction effects are most significant at load
inclinations ranging between 30 and 608:

Figure 8. Maximum Caisson capacity under vertical and horizontal loading.

Copyright # 2003 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2003; 27:1235–1254
LOAD CAPACITY OF SUCTION CAISSONS 1249

Figure 9. Vertical–horizontal load capacity interaction diagrams for uniform soil strength profiles.

Figure 10 illustrates the effect of the soil strength profile on the load capacity interaction
diagram for caissons with aspect ratio Lf =D ¼ 6: As expected, when the soil strength increases
with depth, the optimum load attachment point deepens. For example, for the case of linear
strength increase with zero strength at the mudline ðSu0 =Su1 D ¼ 0Þ the optimum load attach-
ment depth is at approximately three-fourths of the caisson depth compared to one-half
of the caisson depth for a uniform soil strength profile. With the load attachment point at its
optimum location, the range of load inclination angles c for which vertical–horizontal load
interaction effects are significant is generally between 15 and 308 for all soil strength profiles. As
noted earlier, when the load attachment point is not at optimum, significant interaction effects
occur at larger load inclination angles, generally between 45 and 758 for the cases shown in
Figure 10.

Copyright # 2003 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2003; 27:1235–1254
1250 C. P. AUBENY, S. W. HAN AND J. D. MURFF

Figure 10. Vertical–horizontal load capacity interaction diagrams for non-uniform soil strength profiles.

The effects of load attachment depth are further illustrated in Figure 11 where the resultant of
the vertical and horizontal capacities obtained from the upper bound analysis, F ¼ ðV 2 þ
H 2 Þ1=2 ; is plotted as a function of Li =Lf and load inclination c: The maximum capacity Fmax
corresponds to upward and lateral translation of the caisson with no rotation. Fmax is computed
by (1) direct integration of Equations (1a) and (1b) along the entire depth of the caisson, and (2)
assuming negligible interaction between vertical and horizontal resistance along the sides or at
the base of the caisson. From the second assumption, the horizontal resistance Hb mobilized at
the base of the caisson will be simply the soil shear strength at that depth times the cross-
sectional area of the base, while the vertical resistance Vb will be the soil shear strength at the
depth of the base times the axial load bearing factor ðNab ¼ 10:5Þ times the cross-sectional area
of the base.

Copyright # 2003 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2003; 27:1235–1254
LOAD CAPACITY OF SUCTION CAISSONS 1251

Figure 11. Effect of load attachment point on total load capacity.

Figure 11 again shows that the optimum load attachment depth depends on the soil strength
profile; when the soil strength increases with depth the optimum load attachment point also
deepens. The figure also shows that the optimum attachment depth is insensitive to load
inclination; however, at larger load inclination angles ðc > 458Þ maximum capacity can be
achieved over a relatively wide range of load attachment depths. For example, Figure 11(a)
(uniform soil strength) shows that for a c ¼ 458 load inclination angle any load attachment
depth in the range 0:35Li =Lf 50:7 can mobilize the maximum capacity. Figure 11 also shows
that for small load inclination angles ðc5308Þ load capacity F =Fmax is highly sensitive to load
attachment depth in the vicinity of the optimum attachment location. Hence, the transition from
a rotational to a translational failure mode is quite abrupt, with the load capacity in a rotational
failure mode being significantly below the translation capacity ðF =Fmax ¼ 80–90 per cent) even
for load attachment depths in close proximity to the optimum attachment depth.

Copyright # 2003 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2003; 27:1235–1254
1252 C. P. AUBENY, S. W. HAN AND J. D. MURFF

CONCLUSIONS

This paper presents a simplified upper bound solution for estimating load capacity of
suction caisson anchors under inclined loading conditions. The analysis is restricted to
undrained loading in clays in which the soil strength is uniform or increases linearly with
depth. The formulation can evaluate anchor load capacity for anchor load orientations
ranging from horizontal to vertical. Anchor line attachment points anywhere along the
depth of the caisson may be considered. The formulation can be implemented in a simple
spreadsheet or personal computer format; hence, it is well suited for conceptual and prelimi-
nary design calculations. Three-dimensional finite element solutions compare favorably
to the proposed limit analyses when slender (Lf =D ¼ 6 and 10) caissons are considered.
When short ðLf =D ¼ 2Þ caissons are considered, comparisons to FEM predictions suggest
the limit analysis formulations presented in this paper become less reliable on the
unconservative side.
A parametric study was presented in this paper illustrating the effects of load inclination c;
suction anchor aspect ratio Li =D; anchor line attachment depth Li =Lf ; and soil strength profile
characteristics. This study led to the following conclusions:

* In general, horizontal anchor load capacity is not significantly affected by vertical


components of load for load orientation up to 158 (often more) from horizontal. Vertical
load capacity is usually not significantly affected by horizontal load components for load
orientations of 15–308 from vertical. Vertical–horizontal load interaction effects will
depend somewhat on specific details of the anchor line attachment point and the soil
strength profile as illustrated in Figures 9 and 10.
* Load capacity is very sensitive to anchor line attachment depth for load inclination
angles less than 308: The optimum line attachment depth depends on the soil strength
profile characteristics and can range from mid-depth ðLi =Lf ¼ 0:5Þ in a uniform strength
profile to the three-quarter point ðLi =Lf ¼ 0:75Þ for the more realistic condition of
linearly increasing strength with depth and zero strength at the mudline. At larger
load inclination angles ðc > 458Þ load capacity becomes less sensitive to anchor line
attachment depth and full horizontal capacity can be mobilized over an increasingly
broader range of Li =Lf :

APPENDIX A: WALL SIDE RESISTANCES AT WALL SIDES OF SUCTION CAISSON

From Equation (8) the ratio of lateral to vertical velocity is


 , 2 !
vl Fls Rf Fas 1  ðz=Lo Þ
¼ I1  Rf I3 ¼ ðA1Þ
va Su D Su D x

By manipulating the terms of the equation above,

Fls Su D
Fas ¼ I1 2
 I3 * I1 ¼ C2 Fls  C3 ðA2Þ
Rf C1 Rf C1

Copyright # 2003 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2003; 27:1235–1254
LOAD CAPACITY OF SUCTION CAISSONS 1253

where,
1  ðz=Lo Þ
C1 ¼
x
I1
C2 ¼ 2
Rf C1
Su D
C3 ¼ I3 * I1
Rf C1
By substituting Equation (A2) into yield function, Equation (6), the quadratic equation is
derived as follow:
A1 Fls2  2A2 Fls þ A3 ¼ 0 ðA3aÞ

Fls2  2A4 Fls þ A5 ¼ 0 ðA3bÞ


where
1 þ ðI1 =R2f C12 Þ
A1 ¼
ðSu DÞ2
!
Rf I3 I1
A2 ¼ 1þ 2 2
ðSu DÞ Rf C1
!
I 2 I1
A3 ¼ R2f I32  I2 þ 23 2
Rf C1

A4 ¼ A2 =A1
A5 ¼ A3 =A1
Now we can find the solution about Fls from Equation (9a) and Fas from Equation (9b):
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Fls ¼ A4 þ A24  A5 ðA4aÞ

Fas ¼ C2 Fls  C3 ðA4bÞ

APPENDIX B: NOMENCLATURE

Ai ; Ci curve fit coefficients for side resistance dissipation function


D caisson diameter
D’ h ; D’ s dissipation at bottom and on sides of caisson
F total caisson load capacity
Fls ; Fas lateral and axial force per unit depth on side of caisson
H; V horizontal and vertical caisson load capacity
Hb ; Vb ; Mb mobilized horizontal, vertical, and moment resistance at bottom of caisson
I1 ; I2 ; I3 ; Rf curve fitting coefficients for Nas  Nps interaction relation

Copyright # 2003 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2003; 27:1235–1254
1254 C. P. AUBENY, S. W. HAN AND J. D. MURFF

Li load attachment depth


Lf length of caisson
L0 optimization parameter, center of rotation of caisson
Nas ; Nps axial and lateral resistance factors on side of caisson
Nab axial resistance factor on bottom of caisson
N1 ; N2 factors characterizing lateral resistance near the free surface
Su undrained shear strength
Su0 undrained shear strength at mudline
Su1 rate of strength increase per unit depth

Greek letters
v velocity
v0 velocity at mudline
W work due to external load on caisson
z depth below mudline
c angle of load inclination from horizontal
x optimization parameter controlling vertical velocity of caisson

ACKNOWLEDGEMENTS

The authors would like to acknowledge the support of Department of the Interior Minerals Management
Service (Cooperative Agreement No. 1435-01-99-CA-31003), the Offshore Technology Research Center
and their colleagues at Texas A&M University and the University of Texas.

REFERENCES
1. Murff JD, Hamilton JM. P-Ultimate for undrained analysis of laterally loaded piles. ASCE Journal of Geotechnical
Engineering 1993; 119(1):91–107.
2. Randolph MF, Houlsby GT. The limiting pressure on a circular pile loaded laterally in cohesive soil. Geotechnique
1984; 34(4):613–623.
3. Randolph MF, O’Neill MP, Stewart DP, Erbich C. Performance of suction anchors in fine-grained calcareous soils.
Proceedings of the 30th Offshore Technology Conference, Houston (Paper 8831).
4. Sukamaran B, McCarron W. Total and effective stress analysis of suction caissons for Gulf of Mexico conditions.
Analysis, Design, Construction, and Testing of Deep Foundations. ASCE Geotechnical Special Technical Publication,
New York: 1999; Vol. 88:247–260.
5. Aubeny CP, Moon SK, Murff JD. Lateral undrained resistance of suction caisson anchors. International Journal of
Offshore and Polar Engineering 2001; 11(2):95–103.
6. Prager W. An Introduction to Theory of Plasticity. Addison Wesley: Reading, MA, 1959.
7. HKS. ABAQUS Version 6.1 User’s Manual 2000. Hibbitt, Karlson and Sorensen, Inc: Pawtucket, Rhode Island.
8. Bransby MF, Randolph MF. Combined loading of skirted foundations. Geotechnique 1998; 48(5):637–655.
9. Murff JD. The mechanics of pile foundation collapse. Analysis, Design, Construction and Testing of Deep
Foundations. OTRC Geotechnical Special Publication, Austin, TX, 1999; Vol. 88:76–95.
10. Matlock H. Correlations for design of laterally loaded piles in soft clay. Proceedings of the 2nd Offshore Technology
Conference, Houston, 1970; 577–594.
11. Reese LC, Cox WR, Koop RD. Field testing and analysis of laterally loaded piles in stiff clay. Proceedings of the 7th
Offshore Technology Conference, Houston, 1975; 473–483.
12. Brinch Hansen J. A revised and extended formula for bearing capacity, Bulletin No. 28, Danish Geotechnical
Institute, Copenhagen, 2000; 5–11.
13. Meyerhof GG. The bearing capacity of foundations under eccentric and inclined loads. Proceedings of the 3rd
International Conference on Soil Mechanics and Foundation Engineering, 1953; Vol. 1:440–445.

Copyright # 2003 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2003; 27:1235–1254

Vous aimerez peut-être aussi