Marine Structures 23 (2010) 187–208

Contents lists available at ScienceDirect

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

Assessing appropriate stiffness levels for spudcan

foundations on dense sand
Mark J. Cassidy a, *, George Vlahos b, Mathew Hodder a
a
Centre for Offshore Foundation Systems, The University of Western Australia, 35 Stirling Highway, Crawley, WA 6009, Australia
b
Vibropile (Aust) Pty Ltd, Melbourne, Australia

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

Article history: Before a jack-up can operate at a given location, a site-specific

Received 28 June 2009 assessment of its ability to withstand a design storm during oper-
Received in revised form 2 March 2010 ation must be performed. During this assessment, the complex
Accepted 25 March 2010
state of stress and strain under a spudcan is usually simplified to
a value of foundation stiffness that is integrated as a boundary
Keywords:
condition into the structural analysis. Soil stiffness is a critical
Jack-up
parameter affecting the foundation and structural load distribution
Spudcan foundations
Stiffness, footings/foundations and displacements, and the jack-up natural period and dynamic
Plasticity response. The level of spudcan stiffness is an area of intense interest
Model tests and debate. This paper assesses appropriate stiffness levels for
Soil–structure interaction numerical simulation. Utilising results from a detailed “pushover”
Sand experiment of a three-legged model jack-up on dense sand, the
paper compares the experimental pushover loads and displace-
ments on the hull and spudcans to numerical simulations using
different assumptions of spudcan stiffness. These include pinned
and encastré footings, linear springs and a force-resultant model
based on displacement-hardening plasticity theory. Constant
stiffness levels are shown to be inadequate in simulating the
experimental pushover test. The non-linear degradation of stiffness
associated with the latter force-resultant model is critical.
Ó 2010 Elsevier Ltd. All rights reserved.

1. Introduction

Most jack-up rigs use inverted conical footings known as spudcans as their foundations. Within the
offshore industry, there is considerable interest in determining the level of moment restraint they

* Corresponding author. Tel.: þ61 8 6488 3732; fax: þ61 8 6488 1044.
E-mail address: mcassidy@civil.uwa.edu.au (M.J. Cassidy).

0951-8339/$ – see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.marstruc.2010.03.003
188 M.J. Cassidy et al. / Marine Structures 23 (2010) 187–208

provide, as this level of fixity is often a vital component in the site-specific assessment of a jack-up unit.
This is because inclusion of rotational restraint tends to reduce critical assessment criteria such as
member stresses at the leg/hull connection and the lateral hull deflection. If fixity is assumed within
current site-assessment guidelines SNAME [1] initial rotational stiffness levels have been derived from
recommendations based on monitored offshore data [1–3]. During a numerical analysis these initial
values degrade as the combined loading on the foundation approach the defined failure envelope [1].
The level of spudcan fixity, as well as the overall pushover capacity of a jack-up, is explored in this
paper. A 1:250 scale model jack-up is used to experimentally measure the spudcan and hull load and
displacements during a pushover event on dense silica sand. The experiment reported here furthers
the extensive testing of Vlahos [4] and Vlahos et al. [5,6] which explored the pushover behaviour of
jack-ups on over-consolidated clay. It was performed at 1-g on the laboratory floor as part of a series of
six experiments that studied the effect of different spudcan configurations on both the degree of fixity
and the overall pushover capacity of a model jack-up sitting on dense sand (the programme included
a conical spudcan with spigot, a flat plate, two skirted caissons with different skirt length to diameter
ratios and a spudcan with three prongs [7]). Fig. 1 shows the dimensions of the spudcan used in the
experiment discussed in this paper.
The experiment is used to back calculate stiffness levels of the spudcan and jack-up system.
However, it also provides relevant data to evaluate the performance of numerical analysis techniques.
Therefore, the pushover experiment has been retrospectively simulated using a plane-frame structural
finite element program. Numerical assumptions for the spudcan response have included (i) pinned and
encastré footings, where infinite stiffness is assumed in the vertical and horizontal direction and zero
and infinite in the rotational degree of freedom respectively, (ii) linear springs with stiffness levels as
assumed in the SNAME document [1], and (iii) by using a plasticity modelling approach (see for
example Refs. [8–11]).

2. Model jack-up experiments

2.1. Model jack-up

The model rig used in the experiment was a 1:250 scaled version of jack-ups used offshore. The
prototype leg length was assumed at 150 m, with the remainder of the parameters being average
values of five large jack-ups designed by the Houston-based company Friede & Goldman Ltd (see
Vlahos et al. [5] for more details). As shown in Figs. 1 and 2 and detailed in Table 1, the leg length of the
model jack-up was 600 mm and the spudcan diameter 72 mm. Variations between the true 1:250 scale

Load Reference Point

16mm L.R.P.
13°
1.5mm
Undisturbed Soil Surface
°
13° 130
76°
Reference
Position
w
Ø72mm M

H
u
Displaced
Position

Fig. 1. Spudcan used in experiments, including the sign convention for spudcan loads (V, M, H) and displacements (w, q, u) at the
load reference point (LRP) assumed.
 M.J. Cassidy et al. / Marine Structures 23 (2010) 187–208 189

Vertical Actuator

Hull
Potentiometer
& Tilt Sensor
Vertical Actuator Arm Holder
Dead Weights Horizontal Actuator

Footing
Potentiometer Load Cell
& Tilt Sensor
Holder
1 leg upwave
2 legs downwave Reaction Frame

Testing Tank SandSample

Clay sample

Fig. 2. Scaled model jack-up used in experiments (in pushover mode).

model and the as built model exist, as detailed in Table 1. The largest dissimilarity is in the scaling of the
axial rigidity (EA), as it was not possible to consistently scale both flexural (EI) and axial rigidity. As the
overall structural behaviour of a jack-up unit is dominated by bending of the legs, priority was given to
flexural similarity. This is reflected in Table 1, though more detail of the model scaling assumptions for
the experiments is provided in Vlahos et al. [5].
The model jack-up was strain-gauged so that Vertical (V), Moment (M) and Horizontal (H) loads at
the spudcan and leg-hull connection could be measured. LVDTs and tilt meters were used to measure
the spudcan and hull displacements and rotations (w, u and q respectively). These can be seen attached
to one of the footings in Fig. 3. Shown in Fig. 2 is the 600 mm diameter tub that the model jack-up was
installed into and then tested for its pushover capacity. The tub was filled with superfine silica sand,
with key properties provided in Table 2. The samples were prepared by raining silica sand into the tub
from a 125 cm drop height, at 80 mm/s hopper speed and with an opening in the hopper of 1.16 mm.
This ensured a relative density of around 87%.

Table 1
Properties of the model jack-up unit (after Vlahos et al. [5]).

Model characteristics Prototype (average case) 1:250 Scaled model Built model
Leg length: full and partiala 150 m 600 mm 600 mm
Aft leg separationb 51 m 204 mm 216 mm
Aft and forward leg separationc 45 m 180 mm 187 mm
Breadth of hull 78 m 312 mm 324 mm
Length of hull 70 m 280 mm 280 mm
Depth of hull 10 m 40 mm 40 mm
Second moment of area of leg (I) 7.2 m4 1843.2 mm4 1892 mm4
Cross-sectional area of leg (A) 1 m2 16 mm2 73.8 mm2
Spudcan diameter 18 m 72 mm 72 mm
Leg flexibility (EI/L) 9.27 GNm 593 Nm 608 Nm
a
Lower guide (bottom of the hull, see Fig. 3) to spudcan Load Reference Point.
b
Measured from centre to centre.
c
Measured from centreline of legs.
190 M.J. Cassidy et al. / Marine Structures 23 (2010) 187–208

Fig. 3. Horizontal and vertical LVDTs and tilt meter attached to one leg.

Full details of the apparatus will not be provided here, however, they can be found, along with
a detailed description of an experimental test of a spudcan on over-consolidated clay in Vlahos et al. [5].
Further, full explanation of the calibration factors for all of the load cells and the methodology used to
determine the footing and hull displacements from the LVDT and tilt meter measurements are given in
Vlahos [4].

2.2. Testing procedure

The aim of the physical experiment was to investigate the overall behaviour of the model rig, but
specifically the load sharing between the three independent legs and the stiffness (load–displacement
behaviour) of the entire system. By using a vertical actuator, weights and a horizontal actuator, the
model rig is subjected to a combination of vertical and horizontal loads at the hull level (simulating
a prototype preloading and then environmental loading situation). The tests also determine failure of
the jack-up system, though noting that this is limited to spudcan foundation failure (e.g. sliding) or
system instability (e.g. jack-up overturning). Plasticity within the structural members was not
considered.
Imitation of the operational conditions of offshore jack-ups was required for the experimental
results to be interpreted for appropriate fixity levels. Jack-ups are usually towed to site floating on the
hull with the legs elevated out of the water. On location, the legs are lowered to the sea-bed, where
they continue to be jacked until adequate bearing capacity exists for the hull to climb out of the water.
The foundations are then preloaded by pumping sea-water into ballast tanks in the hull. This ‘proof
tests’ the foundations by exposing them to a larger vertical load than would be expected during service.
The ballast tanks are emptied before operations on the jack-up begin.

Table 2
Superfine silica sand properties (after Cheong [28]).

Property Value Units

Specific gravity, Gs 2.65
Average particle size, d50 0.190 mm
Particle sizes, d10, d20 0.099, 0.135 mm, mm
Minimum dry density, rmin 1516.7 kg/m3
Maximum dry density, rmax 1829.6 kg/m3
Minimum voids ratio, emin 0.448
Maximum voids ratio, emin 0.747
Critical state friction angle, 40 cv 34.9 

Peak friction angle, 40 peak 39.4 

 M.J. Cassidy et al. / Marine Structures 23 (2010) 187–208 191

Table 3
Summary of pushover test.

Experimental Numerical

Vpreload (N) Vself-weight (N) Relative density, Id Vpreload V0 Vself-weight Vself-weight

Total (N) Each foot (N) Foot 3 (N) Total, foot 1 & 2 (N)
450 280 0.87 450 150 80 200

This preloading procedure was followed in the experiment with the self-weight of the jack-up
targeted at roughly 60% of the total preload. Experimental preloading was achieved using a combina-
tion of a vertical actuator and dead-weights. Initially, the model jack-up was lowered whist hanging
using a vertical actuator. After the jack-up penetrated to its own self-weight, further preloading was
performed by pushing an attached vertical loading arm using the same vertical actuator (in this test
this was to a level of 450 N). The actuator was then stopped and detached reducing the load to simply
the jack-up “self-weight” (this was then made to be 280 N by use of additional weights, Fig. 2). The
preload ratio was therefore w0.6, with measured values given in Table 3. The normalised vertical stress
levels range between s/gR z 40 and 60. These values are more consistent with similar laboratory floor
tests of single shallow circular foundations by Gottardi et al. [12], Byrne and Houlsby [13] and Bienen
et al. [14], than the lower values of w7 found offshore [15]. Testing at prototype stress levels is possible
in a geotechnical centrifuge, as discussed for the spudcan application on sands by Cassidy [15].
After the preloading process, vertical and horizontal LVDTs and tilt meters were connected to all
footings and to the hull. A horizontal actuator and load cell were also connected to the hull so that the
pushover test could commence (see Vlahos [4], Vlahos et al. [5,6] or Cassidy et al. [7] for more details of
the experimental set-up). The orientation of the horizontal actuator and jack-up was such that when
pushed there was one leg “upwave” and two legs “downwave”. The two downwave legs were named
Foot 1 and Foot 2 and the single upwave leg Foot 3, as shown in Fig. 4.
In summary, the testing procedure was as follows:

1) Sample Preparation: A sample of dry dense silica sand was prepared in a 600 mm diameter tub.
2) Preloading Process: The load cells were initialised with the jack-up in a hanging position and the
jack-up then lowered using a vertical actuator. After the jack-up penetrated to its own self-weight,
further preloading was performed by pushing the jack-up with a vertical actuator (see Fig. 2). The
level of preload was recorded and is given in Table 3. Once the pre-determined preload level was

Fig. 4. Footing and leg names.

192 M.J. Cassidy et al. / Marine Structures 23 (2010) 187–208

Fig. 5. Connection of load cell to hull (for horizontal pushover).

reached the actuator was stopped and the vertical load reduced. The final jack-up self-weight
consisted of the weight of the model jack-up itself and some additional weights added to the
centroid position of the hull (Figs. 2, 6 and 7). The final “self-weight” is given in Table 3.
3) Preparation for pushover: Vertical and horizontal LVDTs and tilt meters were connected to all
footings and to the hull (Fig. 3). The horizontal actuator and load cell were connected to the hull at
the upwave position (Figs. 2 and 5).
4) Pushover: The jack-up was pushed horizontally at the hull using an actuator attached horizontally
on a frame attached to the sand tub (Fig. 6) and a pin jointed arm with load cell (Fig. 5). All spudcan
and hull load and displacements were measured during this stage. The test was stopped when one
of the LVDTs had extended out of range (Fig. 7 shows the jack-up after the pushover test was
completed).
5) Final check: The zeros were checked again with the jack-up hanging in the original position.

With all of the loads and displacements measured, the load and displacement paths followed in
each leg and at the hull were recorded.

3. Experimental results

The aim of the physical experiment was to investigate the overall load–displacement behaviour of
the model rig, but specifically the (i) load sharing between the three independent legs, (ii) pushover
capacity, and (iii) level of fixity in the footings. Although preload loads and displacements were
measured only the pushover results will be presented in this paper.
Fig. 8 shows the horizontal movement of the hull against the total pushover load (also equal to the
combined base shear of the footings). A peak horizontal pushover load was reached at 24.7 N before
a softening behaviour occurred. Figs. 9–11 detail the loads and displacements recorded at the spudcans,
and these figures show that the peak load coincided with sliding and uplift of the single upwave leg.
This uplift is shown in the photograph of Fig. 12. It is also shown in Fig. 9 that the horizontal load
carrying capacity of the footings is roughly the same until the initiation of sliding in the upwave leg (at
w20 N applied horizontal load). At this stage the horizontal load in the upwave leg drops off until it has
been lifted off the soil and reaches a state of no vertical load as well as no horizontal or moment load.
The applied load has also peaked and Fig. 11 shows the downwave legs plunging into the soil, with
significant movements in all degrees of freedom. The lifting of the upwave spudcan clear of the sand
 M.J. Cassidy et al. / Marine Structures 23 (2010) 187–208 193

Fig. 6. Horizontal actuator pushing over jack-up (during test).

Fig. 7. Jack-up after pushover test is completed.

194 M.J. Cassidy et al. / Marine Structures 23 (2010) 187–208

30

Appl ied horiz ontal force, H ap p

25

(al so total base shear) (N)

20

15

10 L2 L3

5 L1 Happ

0
0 5 10 15 20 25
u @ Hull Centre (mm)

Fig. 8. Horizontal displacement during pushover of model jack-up.

surface is considered to be the failure point of the jack-up system. However, further measurements
were made as the model jack-up pivoted about the two downwave legs.
Fig. 10 shows the combined load paths followed by each footing during the pushover event. As the
applied horizontal load is increased the upwave footing experiences a loss of vertical load, whilst the
downwave footings gain vertical load at half the rate (maintaining vertical equilibrium). As also shown
in Fig. 10 all of the footings initially develop similar moment reactions under the applied horizontal
load. However, with applied load the moment carrying capacity decreases and this behaviour of
moment shedding worsens throughout the test. A greater reliance on the change in vertical load
between the upwave and downwave legs to accommodate the jack-up overturning moment occurs,
and this is reflected in the increasingly non-linear vertical load paths of Fig. 9.

3.1. Degree of fixity

The assumption of rotational fixity on spudcan footings has been used by operators and designers in
an attempt to improve site-specific assessments of jack-ups. This is because “conservative” practice has
been using the pinned footing assumption, exacerbating the moment at the lower guide of the hull. The
use of some rotational fixity reduces stresses at the top of the legs, usually one of the key failure
criteria.1 The experimental data obtained for the model jack-up pushover on dense silica sand is used to
review the amount of fixity available.
Wong et al. [16] proposed that the degree of secant rotational fixity be defined as

Mfooting
f ¼
: (1)
Hfooting L=2

This fixity (f) is the ratio of moment at the spudcan (Mfooting) to the moment assuming an encastré
footing (calculated as the horizontal force at the spudcan) (Hfooting) multiplied by the distance to the
inflexion point, which for the encastré footing is halfway along the leg length (L/2). Using this defi-
nition, a value of 1 represents the fully fixed condition (encastré assuming infinite rotational restraint)
and a value of 0 indicates pinned conditions (no rotational restraint). Fig. 13(a) shows the degradation
of fixity as the pushover proceeds. For this case the fixity reduces in an approximately linear fashion
during the pushover. Fig. 13(a) shows that there is still fixity in the downwave footings even after
failure of the system.

1
It should be noted that the pinned assumption is not always conservative and can lead to a misunderstanding about the rigs
behaviour during extreme events (see Cassidy et al. [20], for instance). This can be due to incorrect modelling of the jack-ups
natural period, but also because of the assumptions of infinite vertical and horizontal stiffness of the spudcan.
 M.J. Cassidy et al. / Marine Structures 23 (2010) 187–208 195

160

140

120

100
Leg 1
V (N) 80
Leg 2
60
Leg 3
40

20

-20
0 5 10 15 20 25 30
Applied horizontal load (also total base shear) (N)

1400

1200 Leg 1
Leg 2
1000
Leg 3
800
M (Nm m )

600

400

200

-200
0 5 10 15 20 25 30
Applied horizontal load (also total base shear) (N)

14

12 Leg 1
Leg 2
10
Leg 3
8
H (N)

-2
0 5 10 15 20 25 30
Applied horizontal load (also total base shear) (N)

Fig. 9. Spudcan loads measured against applied pushover force.

The overall secant lateral stiffness of the model jack-up (horizontal applied load divided by hori-
zontal hull displacement) is shown in Fig. 13(b). Later in this paper, these derived experimental stiff-
ness values will be compared to the numerically applied values.

4. Numerical simulations

4.1. Back analysis using JAKUP

Retrospective numerical simulations of the physical experiment have been performed using the
displacement-hardening plasticity model, Model C [17], and the structural analysis program JAKUP
196 M.J. Cassidy et al. / Marine Structures 23 (2010) 187–208

14

12 L2 L3
10
L1 Happ
8
H (N)
6

2 Leg 1
Leg 2
0
Leg 3
-2
-20 0 20 40 60 80 100 120 140 160
V (N)

1400

1200
L2 L3
1000
L1 Happ
800
M (Nm m )

600

400

200 Leg 1
Leg 2
0
Leg 3
-200
-20 0 20 40 60 80 100 120 140 160
V (N)

14

12 L2 L3
10
L1 Happ
8
H (N)

2 Leg 1
Leg 2
0 Leg 3
-2
-200 0 200 400 600 800 1000 1200 1400
M (Nmm)

Fig. 10. Experimentally measured spudcan loads.

[18–21]. They have also been compared to the more traditional approaches of pinned, fixed and linear
spring foundation assumptions. The JAKUP program is described in detail in the theses by Thompson
[18] and Cassidy [19]. For this simulation the following features of the program were used:

 The program uses the finite element method modelled by time-stepping.

 The structure is modelled as a 2-dimensional “bar stool” model, with the legs and hull represented
by elastic finite elements.
 M.J. Cassidy et al. / Marine Structures 23 (2010) 187–208 197

2
plunging of
downwave legs
1

w (m m )
-1

-2
L2 Leg 1 uplift of
L3
upwave
-3 Leg 2
L1 Happ fotings
Leg 3
-4
0 5 10 15 20 25 30
Applied horizontal load (also total base shear) (N)

2.5

Leg 1
2 Leg 2
Leg 3
1.5
θ (deg)

0.5

0
0 5 10 15 20 25 30
Applied horizontal load (also total base shear) (N)

3.5 Leg 1
Leg 2
3
Leg 3
2.5
u (m m )

1.5

1
0.5
initial sliding
0

-0.5
0 5 10 15 20 25 30
Applied horizontal load (also total base shear) (N)

Fig. 11. Measured spudcan displacements.

 The foundation can be modelled as pinned, encastré, linear elastic springs, or using one of two
advanced displacement-hardening plasticity models in terms of force-resultants on the founda-
tion. In this case only the model for sand, known as “Model C” is used. It is described below and
details are also given in Table 4.

The structural properties reflected the physical experiments, with the details given in Fig. 14. The
single leg properties were doubled for the 2 downwave leg case and the hull modelled as a single
member of the same properties (although the stiffness was increased twenty-fold to simulate a stiff
198 M.J. Cassidy et al. / Marine Structures 23 (2010) 187–208

Fig. 12. Spudcans after pushover test. (a) View of single upwave leg (Foot 3). (b) View of all legs (upwave to right of picture).

hull body). A horizontal load was applied at structural node 3 (on the upwave section of the hull,
Fig. 14) and was increased until failure in the foundation models occurred.
The following were the foundation assumptions investigated (with the sophistication of the model
increasing down the list):

 Pinned footings (infinite vertical and horizontal stiffness but zero rotational stiffness)
 Encastré (infinite vertical, horizontal and rotational stiffness)
 Linear springs (using fixity levels derived from SNAME [1])
 A force-resultant approach using “Model C” for spudcan foundations in sand.

4.2. Force-resultant model for spudcans on sand – Model C

One of JAKUP’s major advantages is the implementation of “Model B” and “Model C” – displacement-
hardening plasticity models for spudcans on clay and sand respectively. Model C, which is used for
 M.J. Cassidy et al. / Marine Structures 23 (2010) 187–208 199

Fixed 1
0.9
a Leg 1
0.8 Leg 2
0.7 Leg 3

Fixity = M/(HL/2)
0.6
0.5
0.4
0.3 L2 L3
0.2
L1 Happ
0.1
0
Pinned 0 5 10 15 20 25 30
Applied horizontal load (also total base shear) (N)

Level of fixity developed in spudcans

80
Secant hori z ontal stiffnesss = Hap p / u

70 b Experiment
Fixed
60 Pinned
50
(N/ m m )

40

30

20

10

0
0 5 10 15 20 25
u @ hull centre (mm)

Secant horizontal stiffness for hull

Fig. 13. Derived spudcan fixity and jack-up system stiffness from experiments. (a) Level of fixity developed in spudcans. (b) Secant
horizontal stiffness for hull.

the analyses here, is based on a series of experiments described in Gottardi et al. [12]. In displacement-
hardening plasticity theory the response of the foundations is expressed purely in terms of force-
resultants; and is expressed in a way that makes it directly applicable in numerical analysis of structures.
Model C has four major components:

(1) An empirical expression for the yield surface in three dimensional vertical, moment and horizontal
loading space (V, M/2R, H) (note: moment is normalised by the radius of the spudcan, R). Once the
yield surface is established, any changes of load within this surface will result only in elastic
deformation. Plastic deformation can result, however, when the load state reaches the surface.
(2) An empirical displacement-hardening expression to define the variation of the size of the yield
surface with the plastic component of vertical displacement. Though the shape of the yield surface
is assumed constant, it expands with vertical plastic penetration (as the footing is pushed further
into the soil). Yield surface expansion is shown in Fig. 15 where the size of the surface is deter-
mined by the point on the surface at maximum V values (defined as V0).
(3) A model for elastic load–displacement behaviour within the yield surface. Finite element work has
shown that cross coupling exists between the horizontal and rotational footing displacements [22–
24], with a linear elastic incremental force–displacement relationship of the form

0 1 2 30 1
dV Kv 0 0 dw
@ dM A ¼ 4 0 Km Kc 5@ dq A (2)
dH 0 Kc Kh du
200 M.J. Cassidy et al. / Marine Structures 23 (2010) 187–208

Table 4
Parameters for Model C used in JAKUP retrospective simulations.

Constant Type Explanation Value used Values [17] Notes

R Footing Footing radius 0.036 m (Various) For spudcan dimensions
dimension see Fig. 1.
g Soil Unit weight of soil 17.1 kN/m3 10 kN/m3 Cheong [28]

g Elasticity Shear modulus factor 500 or 4000 4000 Eqn. (4).1 in Cassidy [19].
Also see Table 5.
k1 Elastic stiffness factor (vertical) 2.65 2.65 As in Bell [22]. These non-
k2 Elastic stiffness factor (moment) 0.46 0.46 dimensional factors
k3 Elastic stiffness factor (horizontal) 2.3 2.3 combine with an
k4 Elastic stiffness factor (horizontal/ 0.14 0.14 assumed shear modulus
moment coupling) to form the elasticity
matrix Eq. (2).

h0 Yield surface Dimension of yield surface 0.116 0.116

(horizontal)
m0 Dimension of yield surface 0.086 0.086
(moment)
a Eccentricity of yield surface 0.2 0.2
b1 Curvature factor for yield surface 0.90 0.90
(low stress)
b2 Curvature factor for yield surface 0.99 0.99
(high stress)

b3 Flow rule Curvature factor for plastic 0.55 0.55

potential (low stress)
b4 Curvature factor for plastic 0.65 0.65
potential (high stress)
ah Association factor (horizontal) 1.0–2.5 1.0–2.5 ah0 ¼ 1.0 and ahN ¼ 2.5
am Association factor (moment) 1.0–2.15 1.0–2.15 am0 ¼ 1.0 and amN ¼ 2.15
k0 Rate of change in association 0.125 0.125
factors

f Hardening law Initial plastic stiffness factor
Ng Bearing capacity factor (peak)
dp Dimensionless plastic penetration
at peak
0.007625 0.144 Based on single footing
110 150–300 tests performed in this
0.0861 0.0316 sand. Bearing capacity
values change during
embedment, but derived
using 4 ¼ 38 ; fp ¼ 0.8;
a ¼ 1.0 and factors of
Cassidy and Houlsby [31].

1. Only parameter values are given here. Equations as set out in Houlsby and Cassidy [17].

where Kv, Km, Kh and Kc represent stiffness values and w, q and u are the vertical, rotational and
horizontal displacements respectively. These values are defined in Tables 4 and 5.

(4) A suitable flow rule to allow prediction of the ratios between plastic footing displacements during
yield.

For a detailed description of Model C reference can be made to Houlsby and Cassidy [17], Cassidy
[19] and Cassidy et al. [25,26]. The assumptions made in the numerical simulations are detailed in
both Table 4 and Fig. 15. The Model C parameters assumed are also given in Table 4. These represent,
apart from the hardening law, values described by Houlsby and Cassidy [17], where reference to the
full set of equations can also be made. The hardening law parameters represent the best fit to the
spudcan and sand used in the experiment. They were derived from a complementary series of
vertical load tests of a single spudcan of equivalent size and shape in sand at the same relative
density. Further details of the experimental and numerical results are available in the data report of
Cassidy et al. [7].
 M.J. Cassidy et al. / Marine Structures 23 (2010) 187–208 201

Applied
216 mm
Applied preload and
preload and self-weight
self-weight Applied
Horizontal
Node 3
Load a node

For single leg:

200mm
E = 193 GPa
I = 1892 mm4
A = 73.8 mm2
As = 4.92 mm2
G = 80 GPa
Double leg
For hull:

I = 37840 mm4
400mm As = 24.6 mm2

For spud-cans:
Single leg
R = 36 mm (see Fig. 1 for
spudcan dimensions)

Fig. 14. Structural dimensions and properties used in retrospective numerical simulation.

A six-degree of freedom version of Model C, known as the ISIS model, has also been published by
Bienen et al. [14]. It is incorporated in the three dimensional structural analysis program SOS3D, with
example non-symmetric jack-up analyses described in Bienen and Cassidy [27].

4.3. Numerical simulation procedures

The same installation and pushover “procedure” as was performed in the physical experiments was
followed in the numerical simulations.
While using the force-resultant Model C, the total preload was divided evenly between the three
footings and this expanded the yield surface to an apex value of V0 on each footing. The vertical load
was then reduced to the self-weight of the jack-up before the pushover commenced. At this stage, the
numerical load state was roughly within the “centre” of the expanded yield surface and with applied
horizontal load the foundations initially displayed elastic behaviour. This numerical set-up procedure
is depicted in Fig. 16. The vertical load on each footing was measured during the experimental process
and occasionally slight tilting occurred. This resulted in uneven vertical load distribution and this has
been accounted for in the numerical simulations by applying the measured vertical “self-weight” loads
on each Model C footing. The level on each footing is detailed in Table 3.
For the other foundation assumptions, the level of preload is irrelevant. Pinned and encastré have
infinite vertical stiffness and therefore no vertical settlement, and the vertical movement in the linear
springs case is only related to the self-weight at the start of the pushover. None of these account for the
history of stress under the spudcan due to the preloading process.
202 M.J. Cassidy et al. / Marine Structures 23 (2010) 187–208

V0
(describes apex of
yield surface)

Vertical
plastic
displacement

Vertical load
H penetration
Expansion
M/2R of surface
curve

Yield surface in
(V, M/ 2R, H) load space V

Fig. 15. Theoretical basis of Model C.

4.4. Results of the numerical simulation

The load and displacement paths of the experiment and retrospective simulations are shown in Figs.
17–19. Hull displacements and lower leg guide moments have also been compared against the
pushover loads, and are reported in Cassidy et al. [7]. In Fig. 17, two Model C simulations are reported
with the only difference being the assumption of the elastic shear modulus of the sand (detailed in
Table 5). The first assumes G ¼ 44.7 MPa (derived from a value of g ¼ 4000 and the methodology
detailed in [19]) and is a value based on the scaled model experimental data of Gottardi et al. [12] and
Cheong [28]. However, a value of g ¼ 500 was shown to be a more reasonable fit to the data of
monitored offshore jack-ups [1,3]. The linear springs represent elastic stiffness values taken from the
‘Guidelines for the Site Specific Assessment of Mobile Jack-Up Units’ published by SNAME [1]. The
comparison of the elastic stiffness for the Model C analyses and the SNAME values are given in Table 5.
Only the initial Model C case (g ¼ 4000) is shown in the latter Figs. 18 and 19.

Table 5
Rotational stiffness assumed under footing.

Footings assumption G (MPa) Km, Eq. (2) (kNm/rad) Comments

Pinned 0
Linear springs 11.4 1.96 g ¼ 230(0.9 þ DR/500) and
G/pa ¼ g(V/Apa)0.5 from SNAME [1]
Model C 5.6 Variable (0.96 elastic value) Elastic values uses, g ¼ 500 (as derived from
Noble Denton [2] and Cassidy et al. [3], but utilising
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
G=pa ¼ g 2Rg=pa [19])
Model C 44.7 Variable (7.67 elastic value) g ¼ 4000 (as derived from tests of Gottardi et al. [12]
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
and G=pa ¼ g 2Rg=pa [19])
Encastré Infinite
 M.J. Cassidy et al. / Marine Structures 23 (2010) 187–208 203

H Preload = V0 preload / Vself-weight

M/2R

V self-weight

Initial yield surface in

(V, M/2R, H) load space V0 preload

Fig. 16. Initial yield surface as set-up by numerical preloading.

Fig. 17 shows the horizontal hull displacements calculated for the applied loads for all of the
simulations. Initially the Model C (g ¼ 4000) analysis shows linear behaviour until the load condi-
tions touch the yield surface. This occurred initially in the downwave footings at an applied load of
only 7.7 N. At this point this produced a reduction in the stiffness of the foundations as a coupled

a 50
Experiment
45
Appl i ed hori z ontal force, H ap p

Model C g=500
(al so total base shear) (N)

40
Model C g=4000
35 Pinned
30 Linear springs - SNAME
25 Fixed
20
15
10
5
0
0 5 10 15 20 25
u @ Hull Centre (mm)

Complete test

b 30
Appli ed hori z ontal force, H ap p

25
(al so total base shear) (N)

20

15 Experiment
Model C g=500
10 Model C g=4000
Pinned
5 Linear springs - SNAME
Fixed
0
0 0.5 1 1.5 2 2.5 3 3.5 4
u @ Hull Centre (mm)

Zoom in on initial push

Fig. 17. Comparison of hull displacement in retrospective numerical simulations and experimental pushover results. (a) Complete
test. (b) Zoom in on initial push.
204 M.J. Cassidy et al. / Marine Structures 23 (2010) 187–208

16

14 Model C
L2 L3 Linear Springs - SNAME
12
Experiments
10 L1 Happ

H (N) 8

2
0

-2
-20 0 20 40 60 80 100 120 140 160
V (N)

1400
Model C
1200
Linear Springs - SNAME
1000 Experiment

800
M (Nm m )

600

400

200

-200
-20 0 20 40 60 80 100 120 140 160
V (N)

16

14 L2 L3

12 L1 Happ
10

8
H (N)

Model C Leg 1&2

6
Model C Leg 3
4 Linear Springs - SNAME
2 Exp. Leg 1
Exp. Leg 2
0
Exp. Leg 3
-2
-200 0 200 400 600 800 1000 1200 1400
M (Nmm)

Fig. 18. Comparison of numerical and experimental loads on spudcans.

elastic–plastic stiffness is evaluated. The increased hull displacement due to this effect is seen clearly
in Fig. 17, as well as a change in the footing load and displacement paths in Figs. 18 and 19 (the sharp
transition in the loads is clearly defined in Fig. 18). The upwave footing started to yield later, at an
applied load of 11.2 N.
When using the force-resultant Model C at the footings the numerical simulations all “stopped” due
to the failure of the upwave footing (Foot 3). This was numerically predicted in the JAKUP simulation
and came about because the spudcan load state in Foot 3 reduced to a state of zero and no tensile
capacity was possible. For the elasticity assumption of Cassidy [19] (g ¼ 4000) this occurred at 31.1 N,
a value slightly higher than physically recorded. Fig. 17 shows that at this position the prediction of hull
 M.J. Cassidy et al. / Marine Structures 23 (2010) 187–208 205

4.5
4
3.5
3
L2 L3
2.5
u (m m )
2 L1 Happ
1.5 Model C Leg 1&2

1 Model C Leg 3
Exp. Leg 1
0.5
Exp. Leg 2
0 Exp. Leg 3
-0.5
-4 -3 -2 -1 0 1 2
w (mm)

2.5
Model C Leg 1&2
Model C Leg 3
2
Exp. Leg 1
Exp. Leg 2
1.5 Exp. Leg 3
θ (deg)

0.5

0
-4 -3 -2 -1 0 1 2
w (mm)

4.5
Model C Leg 1&2
4
Model C Leg 3
3.5 Exp. Leg 1
3 Exp. Leg 2

2.5 Exp. Leg 3

u (m m )

2 initial sliding
1.5
1
0.5
0
-0.5
0 0.5 1 1.5 2 2.5
θ (deg)

Fig. 19. Comparison of numerical and experimental displacement of spudcans.

movements were also reasonable (though the ability to predict the recorded post peak rotations were
not possible).
The failure mode predicted was the same as observed in the pushover experiment, with the upwave
spudcan lifting clear of the sand surface. By assuming the upwave spudcan is a free-node with zero
stiffness, further numerical analysis using a displacement based or even a Riks method approach is
possible. However, this was not considered in this paper as the jack-up system had failed and the
pushover capacity had already peaked.
206 M.J. Cassidy et al. / Marine Structures 23 (2010) 187–208

80

Secant horiz ontal stiffnesss = Hap p / u

Experiment
70
Model C g=500
60 Model C g=4000
Fixed
50

(N/m m )
Linear springs - SNAME
40 Pinned
30

20

10

0
0 0.5 1 1.5 2 2.5 3 3.5 4
u @ hull centre (mm)

Fig. 20. Comparison of numerical and experimental secant horizontal stiffness values.

The loads and displacements of the footings calculated during the retrospective simulations of the
pushover are shown in Figs. 18 and 19 respectively. Model C retrospectively simulates the experiments
well with prediction of the transition to a plastic type behaviour at the yield surface shown in both the
experiments and the numerical predictions. The displacement-hardening behaviour also seems to
follow the measured load and displacement paths. The ability to follow separate load paths in different
footings (as shown in Fig. 18) is one of the advantages of using force-resultant models based on
displacement-hardening plasticity. Rig failures involving, for example, shallow sliding of a windward
spudcan (as was the case here) and/or ‘plunging’ of a leeward spudcan (for instance in clay soils, as was
recorded by Vlahos [4]) cannot be simulated in a realistic manner when the foundations are modelled
as pin joints or linear springs. For elastic foundations, say for pinned, encastré and linear spring
foundations, the deformation analysis continues indefinitely (for an elastic structure). This was shown
in Figs. 18 and 19 to be unreasonable.
No tensile capacity was possible in the experiments due to the use of dry sand. This is also the
assumption of Model C with the yield surface starting at the apex of V ¼ H ¼ M ¼ 0 (see Figs. 15 and 16).
It can be seen that for both the experiments and the numerical simulations the loads on the upwave
spudcan reached that apex. The numerical yield surface was also reducing in overall size due to a heave
in the leg. By finally reaching the apex of the surface “failure” occurred in the numerical analysis.
Another pleasing prediction in the Model C analysis was the point and magnitude of the horizontal
upwave leg sliding, as shown in Fig. 19.
The overall secant lateral stiffness values experimentally derived (previously shown in Fig. 13(b))
are compared to numerical values in Fig. 20. For the pinned footings, encastré and linear springs the
response is linear and the stiffness constant. This is shown by the horizontal lines in Fig. 20. These
simplified assumptions are again shown to be inappropriate. In the force-resultant Model C simula-
tions, only the initial response is linear and after both upwave and downwave spudcans have yielded,
the response becomes non-linear and follows a similar degradation of stiffness to that observed in the
experiment. The lateral stiffness for these cases eventually lies below the pinned condition. This is
possible because the numerical simulations account for finite elastic spudcan stiffness in all three
degrees of freedom. It was also the case for the experimental results, highlighting that the pinned
condition is not the limiting case (and degradation in horizontal and vertical stiffness is also required).

5. Conclusion

In this paper the results from a monotonic pushover experiment of a model jack-up on dense sand
and retrospective numerical simulations have been presented. Failure of the jack-ups occurred due to
sliding and then uplift of the upwave footing. The load combination tracked to zero for this footing.
Though the jack-up continued to rotate and slide after failure of the upwave footing, only a reduced
horizontal pushover load could be sustained.
 M.J. Cassidy et al. / Marine Structures 23 (2010) 187–208 207

Rotational fixity of all of the footings was derived from the experimental results. Although initial fixity
levels were close to fully fixed, and for small loads, the levels recommended by SNAME were appropriate
it was clearly demonstrated the degradation of fixity to zero during the course of the pushover. This non-
linearity could only be simulated by the macroelement force-resultant approach, with the Model C
displacement-hardening plasticity model shown to simulate this type of monotonic behaviour
adequately, including the ultimate horizontal load capacity and sliding failure of the upwave leg.
All of the model experiments were conducted at unit gravity. Further verification of the results is
required at stress levels found in the prototype conditions offshore. Some progress has been made,
with similar pushover experiments in the geotechnical centrifuge at UWA reported in Bienen [29] and
Bienen et al. [30]. However, detailed measurement of spudcan loads and displacements was not
achieved, but is still required. The assumptions detailed in the published literature on Model C [17,26]
should also be considered in its application.

Acknowledgements

The research presented here was conducted when the second author was a doctoral student at The
University of Western Australia. The support of Dr Matthew Quah of the Offshore Technology Devel-
opment of Keppel FELS for this research is appreciated. The contribution of Dr Chris Martin of Oxford
University and former University of Western Australia undergraduate student Mr Michael Prior to the
initial development of the model jack-up experiment are gratefully acknowledged.

References

[1] SNAME T&R 5-5A. Site specific assessment of mobile jack-up units. Research Bulletin 5-5A. New Jersey: Society of Naval
Architects and Marine Engineers; 2002.
[2] Noble Denton Europe & Oxford University. The calibration of SNAME spudcan footing equations with field data, Report No
L19073/NDE/mjrh, Rev 3; 2005.
[3] Cassidy MJ, Houlsby GT, Hoyle M, Marcom M. Determining appropriate stiffness levels for spudcan foundations using jack-
up case records. Oslo, Norway: Proc. 21st Int. Conf. on Offshore Mechanics and Arctic Engineering (OMAE); 2002.
OMAE2002-28085.
[4] Vlahos G. Physical and numerical modelling of a three-legged jack-up structure on clay soil. PhD thesis, The University of
Western Australia; 2004.
[5] Vlahos G, Martin CM, Prior MS, Cassidy MJ. Development of a model jack-up unit for the study of soil–structure interaction
on clay. International Journal of Physical Modelling in Geotechnics 2005;5(2):31–48.
[6] Vlahos G, Cassidy MJ, Martin CM. Experimental investigation of the system behaviour of a model three-legged jack-up on
clay. Applied Ocean Research 2008;30(4):323–37.
[7] Cassidy MJ, Vlahos G, Hodder M. Push-over experiments of a model jack-up unit in dense silica sand. Geo:03291. Centre for
Offshore Foundation Systems, The University of Western Australia; 2003.
[8] Roscoe KH, Schofield AN. The stability of short pier foundations in sand. British Welding Journal 1956;August:343–54.
[9] Schotman GJM. The effects of displacements on the stability of jackup spudcan foundations. Houston: Proc. 21st Offshore
Technology Conf.; 1989. OTC 6026.
[10] Martin CM. Physical and numerical modelling of offshore foundations under combined loads. DPhil thesis, University of
Oxford; 1994.
[11] Williams MS, Thompson RSG, Houlsby GT. Non-linear dynamic analysis of offshore jack-up units. Computers and Struc-
tures 1998;69(2):171–80.
[12] Gottardi G, Houlsby GT, Butterfield R. The plastic response of circular footings on sand under general planar loading.
Géotechnique 1997;49(4):453–70.
[13] Byrne BW, Houlsby GT. Observations of footing behaviour on loose carbonate sands. Géotechnique 2001;51(5):463–6.
[14] Bienen B, Byrne BW, Houlsby GT, Cassidy MJ. Investigating six degree of freedom loading of shallow foundations on sand.
Géotechnique 2006;56(6):367–79.
[15] Cassidy MJ. Experimental observations of the combined loading behaviour of circular footings on loose silica sand.
Géotechnique 2007;57(4):397–401.
[16] Wong PC, Chao JC, Murff JD, Dean ETR, James RG, Schofield AN, Tsukamoto YK. Jackup rig foundation modelling II. Houston,
Texas: Proc. 25th Offshore Technology Conference; 1993. OTC 7303.
[17] Houlsby GT, Cassidy MJ. A plasticity model for the behaviour of footings on sand under combined loading. Géotechnique
2002;52(2):117–29.
[18] Thompson RSG. Development of non-linear numerical models appropriate for the analysis of jack-up units. DPhil thesis,
University of Oxford; 1996.
[19] Cassidy MJ. Non-linear analysis of jack-up structures subjected to random waves. DPhil thesis, The University of Oxford;
1999.
[20] Cassidy MJ, Eatock Taylor R, Houlsby GT. Analysis of jack-up units using a Constrained NewWave methodology. Applied
Ocean Research 2001;23:221–34.
208 M.J. Cassidy et al. / Marine Structures 23 (2010) 187–208

[21] Cassidy MJ, Taylor PH, Eatock Taylor R, Houlsby GT. Evaluation of long-term extreme response statistics of jack-up plat-
forms. Ocean Engineering 2002;29(13):1603–31.
[22] Bell RW. The analysis of offshore foundations subjected to combined loading, MSc thesis, Oxford University; 1991.
[23] Ngo Tran CL. The analysis of offshore foundations subjected to combined loading. DPhil thesis, University of Oxford; 1996.
[24] Doherty JP, Deeks AJ. Elastic response of circular footings embedded in a non-homogeneous half-space. Géotechnique
2003;53(8):703–14.
[25] Cassidy MJ, Martin CM, Houlsby GT. Development and application of force resultant models describing jack-up foundation
behaviour. Marine Structures 2004;17(3–4):165–93.
[26] Cassidy MJ, Byrne BW, Houlsby GT. Modelling the behaviour of circular footings under combined loading on loose
carbonate sand. Géotechnique 2002;52(10):705–12.
[27] Bienen B, Cassidy MJ. Advances in the three-dimensional fluid-structure–soil interaction analysis of offshore jack-up
structures. Marine Structures 2006;19(2–3):110–40.
[28] Cheong J. Physical testing of jack-up footings on sand subjected to torsion. Honours thesis, The University of Western
Australia; 2002.
[29] Bienen B. Three-dimensional physical and numerical modelling of jack-up structures on sand. PhD thesis, The University
of Western Australia; 2008.
[30] Bienen B, Cassidy MJ, Gaudin C. Physical modelling of the push-over capacity of a jack-up structure on sand in
a geotechnical centrifuge. Canadian Geotechnical Journal 2009;46(2):190–207.
[31] Cassidy MJ, Houlsby GT. Vertical bearing capacity factors for conical footings on sand. Géotechnique 2002;52(9):687–92.