Marine Structures 24 (2011) 528–550

Contents lists available at ScienceDirect

Marine Structures
journal homepage: www.elsevier.com/locate/
marstruc

Controlled installation of spudcan foundations on loose

sand overlying weak clay
Gang Qiu*, Sascha Henke
Hamburg University of Technology, Institute of Geotechnical Engineering and Construction Management, Harburger Schloßstraße 20,
D-21079 Hamburg, Germany

a r t i c l e i n f o a b s t r a c t

Article history: Offshore jack-up rigs are often used for site exploration and oil
Received 1 December 2010 well drilling. The footings of jack-up rigs are known as spudcan
Received in revised form 3 May 2011 foundations. The risk of rapid uncontrolled penetration of spudcan
Accepted 5 June 2011
in seabed (“punch-through”) exposes jack-ups to significant risk
during installation in strong over weak layered seabeds. An
Keywords:
example for this is a thin loose sand layer overlying a weaker
Jack-up
stratum of clay. To prevent spudcans from “punch-through”, an
Foundations
Spudcan in-situ measurement concept is suggested in this paper to control
Soil-structure interaction the installation process of spudcan foundations. First, three-
Bearing capacity dimensional finite element studies using a Coupled Eulerian–
Coupled Eulerian–Lagrangian Lagrangian method are carried out to simulate the penetration
In-situ measurement process. The numerical results have been validated with existing
analytical solutions and centrifuge model test data. Furthermore,
parametric studies are carried out to quantify the influences of the
sand thickness and shear strength of the clay on the bearing
capacity of spudcans. Based on the numerical studies an idea for
the development of an in-situ measurement concept is suggested
to control the spudcan penetration process in-situ.
Ó 2011 Elsevier Ltd. All rights reserved.

1. Introduction

Jack-up rigs are a type of mobile platforms which are disembarked on the seabed and used for site
exploration as well as oil well drilling. Sometimes the soil conditions are difficult when a thin sand
layer overlies a weaker clay stratum. The bearing capacity of the soil may be overestimated, and rapid

* Corresponding author.
E-mail address: g.qiu@tuhh.de (G. Qiu).

0951-8339/$ – see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.marstruc.2011.06.005
 G. Qiu, S. Henke / Marine Structures 24 (2011) 528–550 529

uncontrolled penetration (“punch-through”) can occur. This may lead to severe damages or even loss of
the jack-up rig. Therefore, it is necessary to predict the bearing capacity of the spudcan footing
accurately.
In Meyerhof [1,2] or Hanna and Meyerhof [3] an analytical method and design charts are introduced
to predict the bearing capacity of a footing on sand overlying clay. Further investigations using
centrifuge tests are presented in Craig and Chua [4], Teh et al. [5,6] and Lee [7]. Most of the centrifuge
tests focused more on medium and dense sand, where dramatic punch-through can occur, than on
loose sand. Through a serial of centrifuge tests Teh et al. [6] revealed that the ratio of upper sand layer
thickness over spudcan diameter and the ratio of bearing resistance between the upper sand and
underlying clay affect the development of the spudcan bearing resistance and the existing design
methods fail to predict the spudcan qnom-d profile in sand overlying clay. The original qnom-d profile is
simplified with three key characteristic bearing resistances [6]: q0 (spudcan bearing resistance at
d ¼ 0), qpeak (peak spudcan bearing resistance) and qint (spudcan bearing resistance at the sand–clay
interface). Two new design frameworks for estimating the characteristic bearing resistances have been
developed by Teh et al. [8] and Lee et al. [9]. Nowadays, the use of numerical methods may be a helpful
tool to predict the bearing capacity of spudcan foundations [10–13]. First numerical calculations using
a Coupled Eulerian–Lagrangian approach (CEL) to model spudcan penetration can be found in Tho et al.
[14]. Tho et al. [15] investigated spudcan penetration into different soil profiles using the CEL method.
Applications of the CEL method for the analysis of spudcan penetration into sand over clay were carried
out by Qiu et al. [16] and Henke and Qiu [17], in which a hypoplastic constitutive model was used to
describe the sand.
In the present paper the CEL method is used to simulate the spudcan penetration into uniform clay
and sand or into layered soil with a sand layer overlying weaker clay layer respectively. The numerical
results are validated in comparison with analytical solutions and centrifuge tests [4,5]. This is done to
show that the numerical simulations are suited to capture the behavior of spudcan penetration into
uniform or layered seabed.

2. Numerical model

A three-dimensional finite element model using the Coupled Eulerian–Lagrangian method is pre-
sented to investigate the installation process of spudcans. The usage of classic FE-codes often leads to
contact problems and distortion of the mesh. To deal with these problems the finite element calcu-
lations are carried out using the Coupled Eulerian–Lagrangian method (CEL) implemented in the
commercial program Abaqus [18]. For general geotechnical problems, a Lagrangian mesh is used to
discretize structures, whereas an Eulerian mesh is used to discretize the subsoil. The interface between
structure and subsoil can be represented using the boundary of the Lagrangian domain. The Eulerian
mesh, which represents the soil that may experience large deformations, is able to overcome problems
like mesh and element distortions in finite element simulations.
Several benchmark calculations in Qiu et al. [19] reveal that the CEL approach is well suited to solve
numerical problems involving large deformations which cannot be solved satisfactorily using the
classical finite element method.

2.1. Geometry and mesh

The soil is modeled as an Eulerian domain. 3D Eulerian elements with reduced integration (element
typ: EC3D8R), which are the only available Eulerian elements in Abaqus, are used to discretize the soil.
Thus, the axisymmetric boundary value problem must be simulated in a 3D model. Due to the
symmetry, only one fourth of the whole model is considered in the three-dimensional analysis. Two
numerical models are generated to investigate the penetration process of a spudcan into both uniform
(see Fig. 1(a)) and stratified deposits (see Fig. 1(b)). The geometry of the spudcans, which are used in
centrifuge tests from Craig and Chua [4] (Fig. 1(c)) and Teh et al. [5] (Fig. 1(d)), is shown in Fig. 1. In this
paper, the spudcan is modeled as discrete rigid body. The penetration of a spudcan into the soil is
simulated displacement controlled with a constant penetration velocity. The depth of penetration d is
defined as zero after the cone completely penetrated into the soil (see Fig. 1(c) and (d)).
530 G. Qiu, S. Henke / Marine Structures 24 (2011) 528–550

a model a
b model b
c d
void 14 void 4

1.6
sand 7

penetration penetration
depth d depth d
0.75
sand or clay clay
167° 2 0.54
46 154°
46 0.15
0.71
0.54
0.6 0.5
1.46 0.86
D = 14
z
x
D=6
y
28 28

Fig. 1. (a) Numerical model for spudcan penetration into uniform sand or clay bed; (b) numerical model for spudcan penetration
into a sand layer overlying uniform clay; spudcan geometry used in (c) Craig and Chua [4] and (d) Teh et al. [5]; unit [m].

2.2. Influence of the mesh density and penetration velocity

Two sets of preliminary calculations for the penetration of a spudcan into uniform clay (su ¼ 39 kPa)
are carried out to study the dependency of the solution on the mesh density and penetration velocity.
Three meshes with different coarseness are used (see Fig. 2). The Eulerian domain in mesh A is meshed
with 10,918 elements; whereas mesh B consists of 48,360 elements and mesh C of 259,308 elements.
The pressure-displacement curves for different degrees of coarseness of the FE-mesh are shown in
Fig. 3(a). It can be seen, that the FE-solutions converge with decreasing element size. The solution using
mesh B coincides with the solution using mesh C. Thus, mesh B is selected for the following simulations
to consider both accuracy and efficiency of the simulation.
The penetration of the spudcan into the soil is simulated displacement controlled with a constant
penetration velocity. Rate independent constitutive models are used to simulate the soil (see Section
2.4). After a set of preliminary calculations with penetration velocities of 0.25 m/s, 0.5 m/s and 1 m/s
Fig. 3(b) reveals, that the penetration velocity does not influence the results significantly. Regarding the
computational time a penetration velocity of 0.5 m/s is selected for the following simulations.

Fig. 2. FE-meshes for the preliminary study regarding mesh dependency.

 G. Qiu, S. Henke / Marine Structures 24 (2011) 528–550 531

a -0.5
0
0.5

Penetration d [m]
1
1.5
2
2.5
3
mesh A
3.5 mesh B
mesh C
4
0 100 200 300 400 500
Bearing pressure q [kPa]

b -0.5
0
0.5
Penetration d [m]

1
1.5
2
2.5
3
v = 0.25 m/s
3.5 v = 0.5 m/s
v = 1 m/s
4
0 100 200 300 400 500
Bearing pressure q [kPa]
Fig. 3. Pressure-displacement curves for penetration of a spudcan into uniform clay subject to (a) the coarseness of the FE-mesh; (b)
the penetration velocity.

2.3. Contact formulation

The contact between the soil material in the Eulerian domain and the spudcan meshed with
Lagrangian elements is described using the “general contact” algorithm, which enforces the use of the
penalty contact method. This contact method works without discrete contact elements. The “general
contact” algorithm incorporating the finite-sliding formulation, which allows arbitrary motion of the
surface, is well suited to simulate highly non-linear processes involving large deformations [18].
There is no general consensus regarding the friction coefficient between spudcan and soil in
numerical simulations. Various authors modeled the soil-spudcan interface as either fully smooth or
fully rough (see Table 1). The cone roughness factor a is defined as:

a ¼ au =su (1)
for the clay–spudcan interaction [20], where au is the maximum shear stress that can be mobilized
at the cone surface and su is the local value of the undrained shear strength. The cone roughness factor
between sand and spudcan [21] can be calculated with
532 G. Qiu, S. Henke / Marine Structures 24 (2011) 528–550

Table 1
Summary of the friction coefficient between spudcan and soils used in different numerical analysis.

Authors Soil-spudcan interface Numerical code

Liu et al. [10] Smooth or rough AFENA
Hossain and Randolph [11] Smooth or rough AFENA
Liu and Hu [12] Smooth or rough AFENA RITSS
Tho et al. [14] Smooth Abaqus/Explicit CEL
Yu et al. [13] Rough AFENA

a ¼ tand=tan4 (2)

where d is the interface friction angle between the sand and spudcan and f is the friction angle of the
sand. A cone roughness factor of a ¼ 0.5 is suggested by InSafeJIP [22] for soil-spudcan interaction.
SNAME [23] suggests a ¼ 0.4 for clay–spudcan interaction and a ¼ 0.5 for sand-spudcan interaction.
White et al. [24] used a centrifuge to assess the sand-spudcan behavior and suggested a roughness
factor of a ¼ 0.6 for the sand-spudcan interface.
Two sets of preliminary studies for penetrating a spudcan into uniform clay (su ¼ 39 kPa) and sand
(4 ¼ 38 , j ¼ 8 ) are carried out in this paper to study the dependency of the soil-spudcan interface. The
soil-spudcan interface is modeled as either fully smooth or fully rough, in additional calculations a fric-
tion coefficient a ¼ 0.5 is also taken into account. The Nc0 values from Houlsby and Martin [20] for a ¼ 0,
0.5 and 1 with a cone angle b ¼ 167 (for 2Rr/sum ¼ 5 and h/2R ¼ 0.25, intermediate values determined by
linear interpolation) are Nc0 ¼ 7.247, 7.887 and 8.303. The difference between smooth and rough interface
is about 13%. As shown in Fig. 4(a), the influence of the roughness on the bearing capacity of a spudcan on
clay is very small. The difference between fully smooth and fully rough is only 5% after 4 m penetration
from the numerical simulations, which is smaller than the variation of the Nc0 values.
The Ng values from Cassidy and Houlsby [21] for a ¼ 0, 0.5 and 1 with a cone angle b ¼ 167 and
friction angle of sand 4 ¼ 38 (intermediate values determined by linear interpolation) are Ng ¼ 36.8,
84.3 and 94.3. The scattering of Ng value between a ¼ 0.5 and 1 is about 12%. Fig. 4(b) shows, that the
bearing capacity of the spudcan on sand is increasing with an increasing roughness between sand and
spudcan. With a friction coefficient a greater than 0.5 this increase becomes less distinct. The difference
between a ¼ 0.5 and fully rough is about 5% after 2 m penetration. Therefore, the friction coefficient
a ¼ 0.5 between spudcan and soil is adopted in this paper.

2.4. Constitutive models

The clay is modeled as an elasto-plastic material obeying the TRESCA failure criterion. The
friction angle 4 and the dilation angle j are set to 0 . The clay is assumed to be undrained and the
Poisson’s ratio n is taken as 0.49 (sufficiently high to give minimum volumetric strains, while
maintaining numerical stability) [25,13]. A constant stiffness ratio of Eclay/su ¼ 500 is adopted for all
calculations [25,13], where su is the undrained shear strength and Eclay is the Young’s modulus of
the clay.
The loose sand is modeled as an elasto-plastic material obeying the MOHR-COULOMB failure criterion.
The dilation angle j is set to 0 for loose sand, whereas the dilation angle is set to j ¼ 430 for dense
sand [26]. If not otherwise specified, the Young’s modulus of dense and loose sand is set to Esand, d ¼ 50,
000 kPa and Esand, l ¼ 10, 000 kPa in this paper.
To recalculate the centrifuge test T2 from Teh et al. [5], the dense sand layer is modeled using the
hypoplastic constitutive law. The theory of hypoplasticity was developed at the University of Karlsruhe
in Germany, in particular by Kolymbas [27] and Gudehus [28]. The calculations in this paper are based
on the version of von Wolffersdorff [29] with the extended concept of intergranular strain [30]. The
hypoplastic model is implemented in the form of a user subroutine VUMAT in Abaqus/Explicit by Kelm
[31] to model the non-linear and an elastic behavior of dry granular soils. Typical soil characteristics
like dilatancy, contractancy, different stiffnesses for loading and unloading as well as the dependency
of the stiffness on pressure and void ratio can be simulated. The hypoplastic parameters for the silica
 G. Qiu, S. Henke / Marine Structures 24 (2011) 528–550 533

a -0.5
0
0.5

Penetration d [m]
1
1.5
2
2.5
3
smooth
3.5 = 0.5
rough
4
0 100 200 300 400 500
Bearing pressure q [kPa]

b smooth
-0.5
= 0.5
rough
0
Penetration d [m]

0.5

1.5

2
0 1 2 3 4 5
Bearing pressure q [MPa]
Fig. 4. Pressure-displacement curves for penetration of a spudcan into (a) uniform clay and (b) uniform sand subject to friction
coefficient between spudcan and soil.

sand used at the University of Western Australia (UWA) can be found in Bienen et al. [32] and are
shown in Table 2.

3. Validation of the numerical models by comparison with centrifuge test results

In this section centrifuge tests carried out by Craig and Chua [4] and Teh et al. [5] are recalculated
using the CEL method. The numerical results are compared to the test data in Figs. 5,6,8. The simu-
lations for validation are summarized in Table 3.

3.1. Penetration into uniform clay

A set of tests of penetration into uniform clay with three different undrained strengths su ¼ 39 kPa,
63 kPa and 87 kPa were carried out by Craig and Chua [4]. In all tests no free surface water was taken
into account. The unit weight of the clay is set to gc ¼ 20 kN/m3 in the numerical simulations. The
534 G. Qiu, S. Henke / Marine Structures 24 (2011) 528–550

Table 2
Hypoplastic parameters for the silica sand according to Bienen et al. [32].

Parameter Pescription Unit Value


4c Critical state friction angle [] 30 (34)a
hs Granular hardness [MPa] 1354
n Exponent [-] 0.34
ed0 Minimum void ratio [-] 0.49
ei0 Critical void ratio [-] 0.76
ec0 Maximum void ratio [-] 0.86
a Exponent [-] 0.18
b exponent [-] 1.3
mR Stiffness ratio at a change of direction of 180 [-] 3.1
mT Stiffness ratio at a change of direction of 90 [-] 5.2
R Maximum value of intergranular strain [-] 0.0001
bR Exponent [-] 0.6
c Exponent [-] 5.7
a
34 is adopted in this paper.

loading response of the spudcan penetrating the clay was predicted by Craig and Chua [4] using
a theory based on conventional shallow foundation analyses after full penetration (d > 2 m) of the
spudcan:

q ¼ 7:97su þ gc d (3)
where gc is the unit weight of the clay and d the depth of penetration. The predicted pressures are
depicted in Fig. 5 together with the centrifuge test results of Craig and Chua [4] and the numerical

Fig. 5. Bearing pressure response of a spudcan penetrating into uniform clay; comparison of FE, analytical and centrifuge test results.
 G. Qiu, S. Henke / Marine Structures 24 (2011) 528–550 535

2
Penetration d [m]

FEM case S1
8 Craig and Chua
prediction

0 2 4 6 8 10
Bearing pressure q [MPa]

Fig. 6. Bearing pressure response of a spudcan penetrating into uniform sand; comparison of FE results, analytical predictions and
centrifuge test results.

solutions. The calculated bearing pressures increase with increasing depth of penetration. This matches
well with the centrifuge tests. The FE-solutions are smaller compared to the test results, especially in
the first 2 m of penetration. After 4 m of penetration, the numerical results match well the test results.

3.2. Penetration into uniform sand

Regarding the centrifuge tests of penetration into uniform sand the model was flooded with water
[4]. The water was above the sand surface such that the sand was saturated and dense. The indicated
mobilized friction angle 40 was supposed to be between 37.5 and 38.5 . The numerical simulation is

Fig. 7. Velocity field after 8.8 m of spudcan penetration into (a) uniform clay and (b) uniform sand.
536 G. Qiu, S. Henke / Marine Structures 24 (2011) 528–550

a b
0 0

1 1

2 2

Penetration d [m]
Penetration d [m]

3 3

4 4

5 5

6 6
FEM case SC1
7 Craig and Chua 7 Esand=5MPa
FEM (Yu et al., 2009) Esand=10MPa
8 prediction 8 Esand=100MPa

0 0.1 0.2 0.3 0.4 0.5 0.6 0 0.1 0.2 0.3 0.4 0.5 0.6
Bearing pressure q [MPa] Bearing pressure q [MPa]

Fig. 8. Bearing pressure response of a spudcan penetrating into loose sand overlaying clay; (a) comparison of numerical solution
with the centrifuge results; (b) investigation of the influence of the stiffness of sand on the numerical results.

carried out using a friction angle 40 ¼ 38 and an effective unit weight of g0 s ¼ 11 kN/m3. Craig and Chua
[4] predicted the bearing capacity q using Eq. (4)

q ¼ 0:3Dg0s Ng þg0s dNq (4)

where D is the diameter of the spudcan. The two factors Ng and Nq are depending on the friction angle
of the sand.
The numerical solution is compared to the centrifuge test and predicted results after Craig and Chua
[4] in Fig. 6. Generally, the FE-solution matches well the centrifuge test results. The resistance deter-
mined from the FE-simulation is slightly smaller than the test results at the beginning and is slightly
greater after 2 m penetration. After the cone has completely penetrated into the sand (d ¼ 0 m) the
bearing capacity of the FE-solution increases linearly. This trend matches well with the analytical
solution.
The bearing pressures observed in the centrifuge tests are higher compared to the numerical results.

This is due to the higher friction angle (37.5 –38.5 ) of the sand used in the centrifuge.
The velocity field after 8.8 m penetration into uniform soil is shown in Fig. 7 to visualize the failure
mechanisms depending on the soil, in which the spudcan is penetrating. From Fig. 7(a) it can be

Table 3
Summary of numerical simulations for validation with centrifuge tests.

Case Numerical Spudcan Clay su [kPa] Sand

model (Fig. 1) geometry (Fig. 1)  
4[] j[] H [m]
C1a (a) (c) 39 –
C2a (a) (c) 63 –
C3a (a) (c) 87 –
S1a (a) (c) – 38 8
SC1a (b) (c) 30 32 0 7
SC2b (b) (d) 10 34 4 5
SC3b (b) (d) 10 see Table 2 5
a
Recalculation of the centrifuge tests from Craig and Chua [4].
b
Recalculation of the centrifuge test T2 from Teh et al. [5].
 G. Qiu, S. Henke / Marine Structures 24 (2011) 528–550 537

seen, that the cavity, which is punched into uniform clay, remains open. This matches very well with
observations after Craig and Chua [4] in the absence of free water. The clay around the spudcan edge
begins to flow back onto the exposed top of the spudcan. The failure mechanism is known as “flow
failure”. Similar flow mechanisms are observed by Hossain et al. [25] in centrifuge tests as well as
numerical analyses. By contrast, “wall failure” can be observed in Fig. 7(b). After penetration into the
sand the cavity inside the uniform sand becomes unstable. The sand at the surfaces falls down into the
cavity, such that the wall collapses and refills the cavity above the spudcan.

3.3. Penetration of spudcan into loose sand overlying clay

A series of centrifuge tests were carried out by Craig and Chua [4] in which saturated sand overlaid
uniformly consolidated clay. The effective unit weight of sand g0 s ¼ 11 kN/m3 and clay g0 c ¼ 10 kN/m3
are adopted for numerical simulation and analytical prediction. In the test T27 the sand was placed as
loose as possible. The angle of friction was adopted as 32 and the shear strength of the clay bed is
chosen to be 30 kPa. This test is simulated in this paper.
A new simplified conceptual model for predicting the penetration profile of a foundation (flat based
foundation or spudcan with conical based inclined at 13 to the horizontal) in sand overlying clay has
been developed by Lee [7] and Lee et al. [9]. The penetration profile can be constructed by joining the
values of the peak resistance qpeak, the post–peak resistance qpost–peak and the resistance profile in the
underlying clay. Same as Lee [7], the spudcan used by Craig and Chua [4] (with conical based inclined at
6.5 to the horizontal) is approximately regarded as a flat based foundation by predicting the pene-
tration profile. The relative density of ID ¼ 0.24, the parameter in Bolton’s equation Q ¼ 10 and the
trapped sand thickness of 0.9H (H is the sand thickness) are adopted in the prediction.
The numerical result in this study is compared to the centrifuge result [4], numerical result after Yu
et al. [13] and prediction calculated using the method after Lee [7] in Fig. 8(a). The FE result shows
similar progress as the centrifuge test result. The peak value from the numerical simulation is 8% higher
than the qpeak from analytical prediction. The bearing capacity out of the FE-simulation in this study is
about 24% higher compared to the centrifuge test result, whereas the FE result from Yu et al. [13] is 15%
higher than the centrifuge test result. The difference may be caused by the mobilized friction angle of
the loose sand, the friction coefficient between loose sand and spudcan and the unit weight of the soil.
The mobilized friction angle depends on the stress level, the friction angle of the loose sand in the
centrifuge test may be lower than 32 . In addition, the friction coefficient of the interface between
loose sand and spudcan in the centrifuge test may be smaller than m ¼ 0.5. Furthermore, the unit
weight of the sand and the clay were not explicitly specified in Craig and Chua [4]. The same unit
weights of sand and clay used in numerical simulation, are used in the analytical prediction.
In addition, two simulations with Young’s modulus of loose sand Esand, l ¼ 5 and 100 kPa are carried
out to investigate the influence of the stiffness of the sand on the bearing capacity of a spudcan. It can
be seen from Fig. 8(b), that the stiffness of the sand has no significant influence on the bearing capacity
of a spudcan after about 4 m penetration. The reduction in stiffness only lowers the inclination of the
reaction pressure.
The soil velocity fields at different penetration depths d ¼ 0.3, 3.8 and 8.3 m are shown in Fig. 9. Just
after the cone of spudcan is penetrated into the sand (d ¼ 0.3 m), soil movements can be observed in
both layers. The dilatancy in loose sand is not obvious and can be ignored, hence the dilatancy angle is
set to 0 in the simulation. Therefore, the angle of the shear band with the vertical is equal to 0 . A sand
wedge with cylindrical form is formed between the spudcan base and the sand/clay interface. The
shape differs from truncated conical shape predicted by Meyerhof [1]. The cylindrical block with an
area equivalent to the spudcan area is pushed into the clay. After 8.8 m penetration of the spudcan the
cylindrical block has a height of 7.1 m (1.01H), which matches well with the observation of Craig and
Chua [4] (0.98H).

3.4. Penetration of spudcan into dense sand overlying clay

A series of centrifuge tests were carried out by Teh et al. [5] to investigate the failure mechanism of
a penetrating spudcan through dense sand overlying clay. The failure modes at different spudcan
538 G. Qiu, S. Henke / Marine Structures 24 (2011) 528–550

Fig. 9. Velocity fields at different penetration depths for penetrating a spudcan in loose sand (sand thickness H ¼ 7 m) overlying clay
(case SC1).

penetration depths are illustrated with particle image velocimetry (PIV) analyses and close-range
photogrammetry method [33]. The test T2, in which the thickness of dense sand layer is H ¼ 5 m, is
recalculated using the CEL method (case SC2 and SC3 in this paper). In the case SC2 the dense sand is
modeled as an elasto-plastic material obeying the MOHR-COULOMB failure criterion. According to
O’Loughlin and Lehane [34] the friction angle 40 is set to 34 and the dilatancy angle is therefore
considered to be j ¼ 4 . In the case SC3 the dense sand is modeled using a hypoplastic constitutive law.
The hypoplastic parameters are shown in Table 2. In order to distinct a significant peak value in phase B
(see Fig. 10) the critical state friction angle is set to 34 in this study. The dense sand in the test T2 from
Teh et al. [5] has a relative density index of ID ¼ 0.85 and effective unit weight g0 ¼ 10.8 kN/m3, which
are adopted in the numerical simulation. The underlying clay, which is modeled using the TRESCA model,
has a normally consolidated strength ratio su/s0 v0 ¼ 0.185.
In Fig. 10 the numerical results are compared to the results from centrifuge test T2. The bearing
resistance begins to increase only after the spigot of the spudcan has fully penetrated into the sand
surface (phase A) [5]. Subsequently the bearing capacity increases rapidly in both numerical and
centrifuge results until the peak bearing capacity qpeak is reached (phase B). The bearing resistance
continues to increase in case SC2. In contrast, in case SC3 a marked reduction of the bearing resistance
(phase C) and a second peak with a smaller bearing capacity (phase D) can be observed. This is in good
correlation with the centrifuge test results T2. The subsequent penetration leads to a reduced bearing
resistance (phase E) in the centrifuge test, whereas the numerical result increases more slowly in
case SC3.
In Figs. 11 and 12 the velocity fields from the numerical simulations (case SC2 and SC3) are
compared with the results from centrifuge test T2 [5].

 In phase A, the soil movement is limited to the sand layer, which is observed in all three cases.
 The three soil flow patterns in phase B are also very similar. A movement both in sand and in clay
can be observed. A sand wedge is formed between the spudcan base and the sand–clay interface.
The deformation mechanism in case SC2 shows close relation to the projected area method [35].
The angle of project to vertical is about q ¼ tan1 ð1=2Þ. The subsequent penetration (phase C to F,
case SC2) does not lead to further changes of the failure mechanism in case SC2 and the angle of
shear band keeps constant by q ¼ tan1 ð1=2Þ. The deformation mechanism in case SC3 is same as
the centrifuge test observation (test T2) in phase B: a truncated cone mechanism [1] is established.
The sand within the wedge under the spudcan moves downwards, whereas the sand located
outside the sand wedge moves radially upwards [5].
 G. Qiu, S. Henke / Marine Structures 24 (2011) 528–550 539

Fig. 10. Bearing pressure response of a spudcan penetrating into dense sand overlaying clay; comparison of numerical solution with
centrifuge results after Teh et al. [5].

 In phase C, a radial and upwards motion of the sand outside the sand wedge cannot be observed.
The soil movements resemble the projected area method [35] in all three cases. The inclination of
the shear band to the vertical is reduced in case SC3 and test T2 compared to phase B, which leads
to a reduction of the bearing resistance of the spudcan. It is believed that the shear strength of
dense sand is reduced to its residual value at this state [3,5,6].
 In phase D, a rigid sand plug with sidewall tapered inwards is observed in test T2, whereas the
inwards inclination of the shear band in case SC3 is not so distinct. The inclination of the shear
band keeps constant q ¼ 0 during further penetration (phase E and F) in case SC3, which looks
similar to the observation of the penetration of a spudcan into loose sand overlying clay (section
3.3).
 In the deep penetration state (phase F) the sand plug has a cylindrical form in case SC3, which
matches well with the centrifuge test T2. However the sand plug in case SC2 has a half spherical
form. The heights of numerically received sand plugs are 4.2 m (case SC2) and 4.4 m (case SC3)
respectively, which are smaller compared to a height of 5 m obtained from the centrifuge test T2.
At this state, the clay around the sand plug begins to flow back onto the top of the spudcan.

4. Parametric studies

Using model b (Fig. 1) parametric studies are carried out to determine the influence of the friction
angle of the sand and the shear strength of the clay on the bearing capacity of the spudcan (see Fig. 1(c))
on loose sand. In addition, the thickness of the sand layer H is varied between 3, 5, 7, 9 and 11 m. All
simulations of the parametric study are summarized in Table 4.
Fig. 13 shows the loading responses of the numerical simulations (case G1–G5). It can be seen that
the bearing resistance of a spudcan depends on the thickness of the sand layer. The thicker the sand
540 G. Qiu, S. Henke / Marine Structures 24 (2011) 528–550

Fig. 11. Velocity field of a spudcan penetrating into dense sand overlaying clay; comparison of numerical solution with the
centrifuge results after Teh et al. [5] in phases A to C.
 G. Qiu, S. Henke / Marine Structures 24 (2011) 528–550 541

Fig. 12. Velocity field of a spudcan penetrating into dense sand overlaying clay; comparison of numerical solution with the
centrifuge results after Teh et al. [5] in phases D to E.
542 G. Qiu, S. Henke / Marine Structures 24 (2011) 528–550

Table 4
Summary of the simulations carried out throughout the numerical parametric studies.

Case Clay su [kPa] Sand

 
40 [ ] j0 [ ] H [m]
G1 30 27.5 0 3, 5, 7, 9, 11
G2 30 32 0 3, 5, 7, 9, 11
G3 30 37.5 0 3, 5, 7, 9, 11
G4 50 32 0 3, 5, 7, 9, 11
G5 100 32 0 3, 5, 7, 9, 11

layer, the higher the bearing capacity of the spudcan. Furthermore, the pitch of the bearing pressure
response increases with increasing bearing capacity. The pitch of the curve h ¼ DqD/Dd at d/D ¼ 0 and
the peak value of the bearing capacity qpeak are defined in Fig. 14.

4.1. Influence of the friction angle of sand 40

In general, a punch-though occurs in loose sand overlying clay only if a weaker clay layer is
underlying a thick sand layer. From the parametric studies, it can be seen that a punch-though is not
liable to occur in the case of a thin sand layer (H/D < 0.5) (see Fig. 13). It matches well with numerical
results from Yu et al. [13].
The peak value qpeak and the pitch h as function of the normalized sand thickness H/D are show in
Fig. 15. In the numerical study, the friction angle of the sand is varied (40 ¼ 27.5 , 32 and 37.5 ). Fig. 15
reveals that the bearing capacity of the spudcan increases with increasing friction angle. Through in the
present case (su ¼ 30 kPa), the friction angle of the sand only has small influence on the bearing
capacity of the spudcan. The friction angle 40 increases from 27.5 up to 37.5 , the peak bearing capacity
qpeak increases only up to 20%.
The dimensionless bearing capacity value (Ng) for conical footing on sand has been calculated by
Cassidy and Houlsby [21]. An increase of about twelve times of the Ng value [21] (with b ¼ 180 , a ¼ 0.5)
is detected due to an increase of friction angle 40 from 27.5 up to 37.5 . However, an increase of 22%
(H ¼ 7 m) and 32% (H ¼ 11 m) is established for comparison the peak bearing capacity of the case G1
(40 ¼ 27.5 ) with G3 (40 ¼ 37.5 ) back-calculated by using the analytical method after Lee [7]. The
footing pressure and the weight of the sand plug will be resisted by the frictional resistance along the
sides of sand plug and the bearing capacity of the underlying clay [7] when qpeak occurs. The friction
angle of sand only influences the friction angle along the sides of the sand plug and the dispersion
angle of the sand plug. Fig. 9 shows, that a cylindric sand plug is pushed into the underlying clay. The
dispersion angle of the sand plug keeps unchanged by 0 for spudcan penetration in loose sand
overlying clay. Therefore, the friction angle of sand has only limited influence on the bearing capacity of
a spudcan penetration in loose sand overlying clay compared to pure sand.

4.2. Influence of the shear strength of the clay su

The peak value qpeak and the pitch h as function of the normalized sand thickness H/D are shown in
Fig. 16. The shear strength of the clay is varied (su ¼ 30, 50 and 100 kPa). The shear strength has
significant impact on the bearing capacity of the spudcan qpeak and the pitch h. The higher the shear
strength of the underlying clay, the greater the bearing capacity of the spudcan and the pitch h.

4.3. Correlation between dcrit and H

Teh et al. [5] defined the distance between the lowest elevation of the spudcan’s widest cross-
sectional area at qpeak and the original sand–clay interface as the effective thickness Heff, which can
be calculated with

Heff ¼ H  dcrit (5)

 G. Qiu, S. Henke / Marine Structures 24 (2011) 528–550 543

a b
0 0

0.2 0.2

0.4 0.4
d/D [-]

d/D [-]
0.6 0.6

H=3m H=3m
0.8 5m 0.8 5m
7m 7m
9m 9m
11 m 11 m
1 1
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Bearing pressure q [MPa] Bearing pressure q [MPa]

c d
0 0

0.2 0.2

0.4 0.4
d/D [-]
d/D [-]

0.6 0.6

H=3m H=3m
0.8 5m 0.8 5m
7m 7m
9m 9m
11 m 11 m
1 1
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0 0.2 0.4 0.6 0.8 1
Bearing pressure q [MPa] Bearing pressure q [MPa]

e
0

0.2

0.4
d/D [-]

0.6

H=3m
0.8 5m
7m
9m
11 m
1
0 0.2 0.4 0.6 0.8 1 1.2 1.4
Bearing pressure q [MPa]

Fig. 13. Bearing pressure response of a spudcan penetration into loose sand overlying clay (a) case G1; (b) case G2; (c) case G3; (d)
case G4; (e) case G5.
544 G. Qiu, S. Henke / Marine Structures 24 (2011) 528–550

-0.2

-0.1

0.1
d/D [- ]

0.2

0.3

0.4

0.5
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Bearing pressure q [MPa]

Fig. 14. Definition of the pitch of bearing capacity h ¼ DqD/Dd, the peak bearing pressure qpeak and its depth dcrit.

in which dcrit is the depth of qpeak (see Fig. 14). The empirical correlation for Heff ¼ 0.88H is suggested
by Teh et al. [5] for dense sand overlying clay to suggest a means to estimate the depth of qpeak. Based
on the numerical results, a best-fit value of the constant dcrit ¼ 0.66H with a sum of squares of residuals
RSS ¼ 0.12 is derived for the loose sand overlying clay investigated in this study (see Fig. 17(a)). This
gives Heff as 0.33H. An effective thickness Heff ¼ 0.67–0.82H was calculated by Yu et al. [13] for loose
sand overlying clay. These values are smaller than 0.88H reported by Teh et al. [5] for dense sand
overlying clay. This leads to the conclusion that the depth of qpeak is deeper in the case of loose sand
compared to dense sand overlying clay. This is due to the fact that the maximum shear strength in the
dense sand is mobilized after smaller deformation, whereas the shear strength increases relatively
slow in the loose sand and can be mobilized after larger deformations (see Fig. 17(b)).

5. Recommendation to control the penetration process

5.1. Concept including a CPT

In Quah et al. [36], an in-situ monitoring concept for spudcan penetration in layered soils is
proposed using a cone penetrometer. A similar concept is described in this section.
Before the spudcan penetration, a cone penetration test is usually carried out. Out of this CPT test
the parameters sand layer thickness H, friction angle of the sand 40 and the undrained shear strength su
can be determined. With respect to these results the numerical results in Fig. 15 for example can be
adapted to the in-situ soil conditions to estimate the maximum bearing capacity of the spudcan.
It has to be remarked that the results described in this paper are only valid for the investigated boundary
conditions. In a real case the numerical study has to be carried out using the in-situ soil conditions.
Furthermore, the numerical simulations are not fully verified. Therefore, it is necessary to include safety
factors into the calculations which have to be determined with respect to the complexity of the problem.

5.2. New idea for the concept without CPT

From Figs. 15,16 it can be seen, that the peak bearing pressure qpeak and the pitch of bearing pressure
h increase with increasing thickness of the sand layers. In addition, the pitch of the penetration curve
 G. Qiu, S. Henke / Marine Structures 24 (2011) 528–550 545

a 1
case G1:
case G2:
0.8 case G3:

qpeak [MPa]
0.6

0.4

0.2

0
0.4 0.5 0.6 0.7 0.8 0.9
H/D [-]

b 13
case G1:
12 case G2:
case G3:
11
10
[MPa]

9
8
7
6
5
4
0.4 0.5 0.6 0.7 0.8 0.9
H/D [-]
Fig. 15. (a) Peak bearing pressure qpeak and (b) pitch of bearing pressure h as a function of the normalized sand thickness (case G1,
G2, G3).

increases with increasing bearing capacity. The relationship between peak bearing pressure qpeak and
pitch h is shown in Fig. 18. All five cases lie close together, forming a straight line. The straight line can
be expressed as:

qpeak ¼ f h (6)

with a best-fit value of the constant f ¼ 0.083 using the sum of squares. The residual is RSS ¼ 0.22. It has
to be noted that the empirical constant f ¼ 0.083 is only valid for the spudcan with a cone angle of 167
(as shown in Fig. 1(c)) and a Young’s modulus of the loose sand of Esand ¼ 10, 000 kPa.
According to the numerical results shown above, the relationship between peak bearing pressure
qpeak and the pitch of bearing pressure h can be expressed as shown in Eq. (6). The stiffness of the loose
sand in-situ should be known which is obviously the hardest task regarding this monitoring concept.
The empirical constant f of a specific spudcan under specific ground conditions can be determined out
546 G. Qiu, S. Henke / Marine Structures 24 (2011) 528–550

a 1.8 case G2: su = 30 kPa

case G4: su = 50 kPa
1.6 case G5: su = 100 kPa
1.4

qpeak [MPa]
1.2
1
0.8
0.6
0.4
0.2
0
0.4 0.5 0.6 0.7 0.8 0.9
H/D [-]

b 18
case G2: su = 30 kPa
case G4: su = 50 kPa
16 case G5: su = 100 kPa

14
[MPa]

12

10

0.4 0.5 0.6 0.7 0.8 0.9

H/D [-]
Fig. 16. (a) Peak bearing pressure qpeak and (b) pitch of bearing pressure h as a function of the normalized sand thickness (case G2,
G4, G5).

of numerical simulations or centrifuge tests before the jack-up platform is put into operation. The
reaction force of one leg F and its penetration d into the seabed must be measured in-situ. To obtain
a reliable result a continuous measurement has to be employed throughout the penetration process, as
discrete measurements at two or three depths are not sufficient. The measured pitch h at the instance
when the cone completely penetrates into the seabed d ¼ 0 m, can be calculated as

DFD
h¼ (7)
DdA
where A is the area of the spudcan. With the determined value h, the maximum bearing capacity can be
calculated from Eq. (6) and compared to the in-situ measured pressure qcurrent ¼ F/A. Subsequently it
has to be decided, whether the spudcan is penetrated further or not. A scheme of the new concept is
shown in Fig. 19.
It has to be point out that this concept will only work, if the necessary data like bearing pressure and
penetration depth of the spudcan is monitored in detail throughout the whole penetration process.
 G. Qiu, S. Henke / Marine Structures 24 (2011) 528–550 547

a 1 b
case G1
case G2
case G3
0.8 case G4
case G5
Fitted line
0.6
dcrit /D [-]

0.4

0.2

0
0 0.2 0.4 0.6 0.8 1
H/D [-]

Fig. 17. (a) Correlation between Heff/D and H/D; (b) mobilization of shear resistance in dense and loose sand.

Therefore, the present concept represents an idea of the authors to further improve the monitoring of
spudcan penetration processes especially with respect to difficult soil conditions. Thus, the offshore
industry should be encouraged to move toward such ideas to significantly reduce the danger of failure
and to gain deeper insight into the mechanisms in the subsoil during the penetration process of
spudcan footings.
Concluding, it has to be pointed out, that the monitoring concept has to validated using in-situ
measurements to be judge the accuracy and reliability of the concept. Though, with regard to

case G1
1.4 case G2
case G3
1.2 case G4
case G5
Fitted line
1
qpeak [MPa]

0.8

0.6

0.4

0.2

0
0 2 4 6 8 10 12 14
[MPa]

Fig. 18. Relationship between peak bearing pressure qpeak and pitch h.
548 G. Qiu, S. Henke / Marine Structures 24 (2011) 528–550

Fig. 19. An idea for a new concept to control the penetration process of jack-up platform.

the validation of the numerical models in this study and the validation of similar models in
[32], it can be seen that the numerical simulations provide a very accurate tool to predict the
load-bearing behavior of spudcans throughout penetration if the necessary information of the soil
is available.

6. Conclusions

In this paper, 3D FE-analyses using the CEL method are carried out to simulate the installation
process of spudcan foundations into loose sand overlying weak clay.
In a first numerical study, centrifuge tests of spudcan penetration into a uniform clay or sand bed are
recalculated. The recalculated bearing pressures using CEL method are in good agreement with the
results from centrifuge tests. The holes formed due to the penetration of the spudcan into the two
different soils were also observed in centrifuge tests. Two different failure mechanisms (“flow failure”
and “wall failure”) are observed depending on penetration into clay or sand respectively.
In further calculations the penetration of spudcans into a layered soil with a loose sand layer
overlying a weaker clay is investigated. The comparison with centrifuge test result and analytical
prediction show a good correlation. Therefore, it can be stated that the presented numerical model is
well suited for limit state analysis of spudcan penetration in loose sand overlying weak clay.
The penetration of a spudcan into dense sand overlying clay is simulated. Both Mohr-Coulomb
model and hypoplastic model are used to simulate the dense sand and the numerical results are
compared with test results from Teh et al. [5]. The hypoplastic modeling shows great potential to
improve the prediction of the bearing capacity of a spudcan penetration in dense sand overlying clay
using the CEL method. The numerical results could be further improved by using the visco-hypoplastic
model [37] to simulate the underlying clay.
 G. Qiu, S. Henke / Marine Structures 24 (2011) 528–550 549

In further research work, a numerical parametric study is carried out. The bearing capacity increases
with increasing sand layer thickness H, sand friction angle 40 and clay shear strength su. A critical depth
dcrit ¼ 0.66H is obtained from the numerical results. The peak value qpeak can be expressed by the linear
equation

qpeak ¼ f h

The empirical constant f is only dependent on the stiffness of the loose sand in-site and the geometry
of the spudcan. The properties of clay and the sand thickness have no remarkable influence on f. Based on
the relationship between qpeak and h, an idea for a possible in-situ measurement concept is suggested to
control the installation of spudcan foundations founded on loose sand overlying weak clay.

Acknowledgment

The present work has been funded by the German Research Foundation (DFG) in the framework of
the research training group GRK 1096 “Ports for Container Ships of Future Generations”. The authors
thank the DFG for funding this work. Furthermore, the authors appreciate the academic use of the
commercial program Abaqus.

References

[1] Meyerhof GG. Ultimate bearing capacity of footings on sand overlying clay. Canadian Geotechnical Journal 1974;11(2):
223–9.
[2] Meyerhof GG. Bearing capacity of anisotropic cohesionless soils. Canadian Geotecnical Journal 1978;15(4):592–5.
[3] Hanna AM, Meyerhof GG. Design charts for ultimate bearing capacity of foundations on sand overlying soft clay. Canadian
Geotechnical Journal 1980;17:300–3.
[4] Craig WH, Chua K. Deep penetration of spud-can foundation on sand and clay. Gèotechnique 1990;40(4):541–56.
[5] Teh KL, Cassidy MJ, Leung CF, Chow YK, Randolph MF, Quah C. Revealing the bearing capacity mechanisms of a penetrating
spudcan through sand overlying clay. Gèotechnique 2008;58(10):793–804.
[6] Teh KL, Leung CF, Chow YK, Cassidy MJ. Centrifuge model study of spudcan penetration in sand overlying clay. Gèo-
technique 2010;60(11):825–42.
[7] Lee KK. Investigation of potential spudcan punch-through failure on sand overlying clay soils, PhD Thesis: University of
Western Australia, Centre for Offshore Foundation Systems; 2009.
[8] Teh KL, Leung CF, Chow YK, Handidjaja P. Prediction of punch-through for spudcan penetration in sand overlying clay. In:
Offshore Technology Conference, OTC20060; 2009.
[9] Lee KK, Randolph MF, Cassidy MJ. New simplified conceptual model for spudcan foundations on sand overlying clay solis,
2009, Offshore Technology Conference, OTC20012.
[10] Liu J, Hu Y, Kong X. Deep penetration of spudcan foundation into double layered soils. China Ocean Engineering 2005;
19(2):309–24.
[11] Hossain MS, Randolph MF. New mechanism-based design approach for spudcan foundations on single lager clay. Journal
of Geotechnical and Geoenvironmental Engineering 2009;135(9):1264–74.
[12] Liu J, Hu Y. Numerical simulation of the installation and extraction process of spudcan foundations in clay. In: Proceedings
of the ASME 28th International Conference on Ocean, Offshore and Arctic Engineering. Hawaii, USA, no. OMAE2009–
79222; 2009.
[13] Yu L, Hu Y, Liu J. Spudcan penetration in loose sand over uniform clay. In: Proceedings of the ASME 28th International
Conference on Ocean, Offshore and Arctic Engineering, Hawaii, USA, no. OMAE2009–79214; 2009.
[14] Tho KK, Leung CF, Chow YK, Swaddiwudhipong S. Application of Eulerian finite element technique for analysis of spudcan
and pipeline penetration into the seabed. In: 12th International Jack-up Conference, London; 2009.
[15] Tho KK, Leung CF, Chow YK, Swaddiwudhipong S. Eulerian finite element technique for analysis of jack-up spudcan
penetration. International Journal of Geomechanics; 2010. doi:10.1061/(ASCE)GM.1943-5622.0000111.
[16] Qiu G, Henke S, Grabe J. 3D FE analysis of the installation process of spudcan foundations. In: 2nd International
Symposium on Frontiers in Offshore Geotechnics (ISFOG), Perth WA; 2010. pp. 685–690.
[17] Henke S, Qiu G. Zum Absetzvorgang von Offshore-Hubplattformen. geotechnik 2010;33(3):284–92.
[18] Dassault Systèmes. ABAQUS, Version 6.8EF documentation; 2008.
[19] Qiu G, Henke S, Grabe J. Application of a coupled Eulerian-Lagrangian approach on geomechanical problems involving
large deformations. Computers and Geotechnics 2011;38(1):30–9.
[20] Houlsby GT, Martin CM. Undrained bearing capacity factor for conical footing on clay. Gèotechnique 2003;53(5):513–20.
[21] Cassidy MJ, Houlsby GT. Vertical bearing capacity factors for conical footings on sand. Gèotechnique 2002;52(9):687–92.
[22] Osborne JJ, Teh KL, Houlsby GT, Cassidy MJ, Bienen B, Leung CF. Improved guidelines for the prediction of geotechnical
performance of spudcan foundations during installation and removal of jack-up units; 2010.
[23] SNAME. Guidelines for site specific assessment of mobile jack-up units T&R Bulletin 5-5 and 5-5A, panel OC-7 site
assessment of jack-up rigs. Society of Naval Architects and Marine Engineers; 2008.
[24] White DJ, Teh KL, Leung CF, Chow YK. A comparison of the bearing capacity of flat and conical circular foundations on
sand. Gèotechnique 2008;58(10):781–92.
550 G. Qiu, S. Henke / Marine Structures 24 (2011) 528–550

[25] Hossain MS, Hu Y, Randolph MF, White D. Limiting cavity depth for spudcan foundations penetrating clay. Gèotechnique
2005;55(9):679–90.
[26] Brinkgreve RBJ. Plaxis, 2D-Version 8, Material models manual; 2002.
[27] Kolymbas D. Computer-aided design of constitutive laws. International Journal for Numerical and Analytical Methods in
Geomechanics 1991;15:593–604.
[28] Gudehus G. A comprehensive constitutive equation for granular materials. Soils and Foundations 1996;36(11):1–12.
[29] von Wolffersdorff P-A. A hypoplastic relation for granular material with a predefined limit state surface. Mechanics of
Cohesive-frictional Materials 1996;1:251–71.
[30] Niemunis A, Herle I. Hypoplastic model for cohesionless soils with elastic strain range. Mechanics of Cohesive-frictional
Materials 1997;2(4):279–99.
[31] Kelm M. Numerische Simulation der Verdichtung rolliger Böden mittels Vibrationswalzen, PhD Thesis, Veröffentlichungen
des Instituts für Geotechnik und Baubetrieb der TU Hamburg-Harburg. Hamburg, 6; 2004.
[32] Bienen B, Henke S, Pucker T. Numerical study of the bearing behaviour of circular footings penetrating into sand. In:
Proceedings of the 13th International Conference of the International Association for Computer Methods and Advances in
Geomechanics (IACMAG 13), Melbourne, Australia; 2011.
[33] White DJ, Take WA. Soil deformation measurement using particle image velocimetry (piv) and photogrammetry. Gèo-
technique 2003;53(7):619–31.
[34] O’Loughlin CD, Lehane BM. Measure and prediction of deformation patterns beneath strip footings in sand, research
report; 2003.
[35] Kraft LM, Helfrich SC. Bearing capacity of shallow footing sand over clay. Canadian Geotechnical Journal 1982;20:182–5.
[36] Quah M, Foo K, Purwana O, Keizer L, Randolph M, Quah C. An integrated in-situ soil testing device for jack-up rigs. In: 2nd
Jack-Up Asia Conference and Exhibition, Singapore; 2008.
[37] Qiu G, Grabe J. Explicit modeling of cone and strip footing penetration under drained and undrained conditions using
a visco-hypoplastic model. geotechnik 2011. doi:10.1002/gete.201100004.