OTC 18923
Design and Operation of an Extractable Caisson Foundation for Offshore Structures
M. Hesar, Subsea 7 (London UK)
Copyright 2007, Offshore Technology Conference
This paper was prepared for presentation at the 2007 Offshore Technology Conference held in
Houston, Texas, U.S.A., 30 April–3 May 2007.
This paper was selected for presentation by an OTC Program Committee following review of
information contained in an abstract submitted by the author(s). Contents of the paper, as
presented, have not been reviewed by the Offshore Technology Conference and are subject to
correction by the author(s). The material, as presented, does not necessarily reflect any
position of the Offshore Technology Conference, its officers, or members. Papers presented at
OTC are subject to publication review by Sponsor Society Committees of the Offshore
Technology Conference. Electronic reproduction, distribution, or storage of any part of this
paper for commercial purposes without the written consent of the Offshore Technology
Conference is prohibited. Permission to reproduce in print is restricted to an abstract of not
more than 300 words; illustrations may not be copied. The abstract must contain conspicuous
acknowledgment of where and by whom the paper was presented. Write Librarian, OTC, P.O.
Box 833836, Richardson, TX 75083-3836, U.S.A., fax 01-972-952-9435.
Abstract
This Paper describes the development and first successful
application of a caisson foundation providing a reliable and
easily extractable foundation solution for offshore structures.
The specific structure described is a two piece template called
Pin Pile Installation Frames (PPIF, 400te) and Locator Arm
(LA, 275te). Pin piles are typically used as alternatives to
mudmats for temporary support of jacket structures in deep
water soft clay sites, where the required mudmat area would
be prohibitively large. The PPIF and LA were successfully
used in May 2006 to install eight 96” diameter, 100m
penetration pin piles for two jackets in BP’s Gunashli field in
175m water depth in the Caspian Sea. The piles were installed
well within the ±150mm lateral and vertical positional
accuracy and 0.1° inclination tolerance.
The Paper illustrates how a diffuser system utilising a
geotextile product routinely used in onshore construction
practice, in conjunction with a passive suction system was
developed and used to provide a simple yet effective means of
avoiding the soil adhesion problem. 1:20 scale 1g physical
model tests and laboratory interface friction tests on the
special paint system for skirts are discussed. The philosophy
developed for sizing the capacity of passive suction system
and the back-up active injection system is explained. Also
described are the nonlinear geotechnical FE analyses, which
made it possible to justify minimizing the size of caisson
foundations beyond Code factor of safety to ensure extraction
within 5 minutes, whilst maintaining the required system
reliability.
Introduction
Azerbaijan International Operating Company (AIOC) is
developing the ACG field in the Azerbaijan sectopr of the
Caspian Sea. Phase 3 of the project involves installation of
two platforms in the Deep Water Gunashli (DWG) field, the
Drilling/Utilities/Quarters (DUQ) platform and the
Production/Compression/Water Injection/Utilities platform
(PCWU), see Fig. 1.
During installation the platforms will be placed on four
pin-piles, pre-installed one under each corner leg. These pin-
piles replace the conventional mudmats that were used to
support the Phase 1 and 2 platforms in the project for on
bottom stability. The reason for this change in Phase 3 was a
combination of factors including, soft soil conditions and
excessive wind resistance on the large sized mudmats during
towing on barge to site.
The eight pin piles were installed by deploying two pieces
of temporary structures, the Pin pile Installation Frame (PPIF)
and Locator Arm (LA), see Fig. 2. Both these structures were
supported on circular caisson foundations (referred to as
mudmats for simplicity in the remainder of this paper). At
DUQ platform location the PPIF was docked onto the pre-
drilling template and placed on the seabed first. The LA,
which incorporates the pin pile driving sleeve, was then
connected to each of the four faces of the PPIF structure to
locate the position of the corner pin piles in turn, see Fig. 1.
This paper presents the results and supporting geotechnical
calculations for the mudmat foundations of PPIF and LA
structures. The nominal water depth at the site is 175m.
The submerged weights of PPIF and LA are 200.6 tonnes
and 141.8 tonnes, respectively. Penetration of the skirted
mudmats into seabed will take place under the self weight of
each structure. Extraction of the mudmat foundations from
seabed were to utilize a passive suction (PS) system with an
active water injection system (AWIS) forming a backup
system. In the PS system the mudmats will be subjected to an
uplifting force (crane lift) and water will be sucked into the
cylindrical cavity through suction ports as a result of negative
pressures developing there in response to the applied load. In
the AWIS system water is injected into the cylindrical cavity
below the mudmat base to piston the mudmat out of the
seabed. Both systems comprise valved penetrations on
individual mudmats. To avoid the well known adhesion
problem between the mudmat base and seabed clay and to
ease the breakout of mudmats during lifting off from the
seabed all mudmats are provided with a special layer of
Terram geotextile underneath the base plate.
Review of Previous Applications of Suction
Foundations
The effectiveness of caisson foundations for permanent
support of offshore structures has been well established. In
recent years caissons have been used to provide foundation
support to a variety of offshore structures, ranging from
concrete gravity bases1 and steel jackets2,3 to subsea facilities4
and pipelines; in both clay and sand seabed conditions.
However, in order to avoid the notoriously protracted
2 OTC 18923
“breakout” phenomenon during extraction5, it is desirable to
have a foundation system that can be extracted from the
seabed swiftly at the end of normal service life of the facility
or end of operation6. Examples include all types of fixed
structures supported on shallow foundations, mobile offshore
production units used for exploitation of marginal fields,
jackup rigs, or heavy structures deployed repetitively on the
seabed for some subsea operation.
Present Design
The foundation design comprises four skirted and one un-
skirted mudmats for the PPIF and 2 skirted and one un-skirted
mudmats for the LA, Fig. 2. All mudmats are 6.2m in diameter
and depth of skirts is 0.75m. The un-skirted mudmat of the
PPIF is termed the “5th” mudmat (towards centre of PPIF, not
shown in Fig. 2) and that of the LA is termed the “Apex”
mudmat. The skirts were painted with a special friction
reducing paint system to minimise adhesion between the skirts
The underbase of all mudmats was lined with a geotextile
product (Terram) which acts as a diffuser to distribute and
equalise the water pressures. Both active injection and passive
suction systems for extracting the mudmats were designed.
Separate suction ports were provided for the active and
passive systems. A programme of 1:20 scaled model tests
together with a set of interface friction tests was carried out by
Fugro Limited in the UK. These tests revealed that the Terram
diffuser is very effective in eliminating the underbase
adhesion during the removal operations, both with the active
water injection and passive suction schemes. The skirt
adhesion -value obtained from the interface tests between
clay and painted skirt surfaces is around 0.37.
Skirted mudmats have 4 number 160mm diameter passive
suction holes, except for the two Row-1 mudmats of the PPIF,
which have 2 additional 14” holes. These served to improve
the breakout time from seabed in the final lifting of the PPIF
at the end of pin-piling operations. The removal systems were
designed to release the skirted mudmats from seabed in less
than 30 minutes. With the contingency margins adopted the
actual breakout times turned out to be between 5 to 10
minutes.
Site and Soil Conditions
DUQ and PCWU Jackets are located at the Deep Water
Gunashli site. The seabed is gently sloping, with a level
difference across diagonal legs of each jacket of about 1m.
Apart from a deep borehole site investigation campaign by
Fugro in 2004, aimed at the design of platform main pile
foundation, a shallow site investigation was performed by
Andrew Survey, also in 2004. This survey comprised
piezocone penetration testing (PCPT) to 10m, piston/gravity
cores capable of penetrating up to 6m depth and box cores.
The aim of this SI was to provide a better definition of the
shallow soil conditions both in terms of strength parameters
and lateral variability. During this site investigation a
relatively large volume of soil was recovered from the site by
utilising piston core tubes to be used in a medium sized model
soil bed for scaled model tests discussed later. A programme
of laboratory soil testing was conducted following the field
work on the piston samples recovered.
Soil Conditions
The surface layer of gravely shelly sand was encountered in
all of the core locations and the majority of CPT locations. It
comprised a very thin, 1 to 10cm, locally up to 20cm thick
layer of shelly fine to medium sand and fine to medium
gravel.
Beneath the sand layer a relatively uniform calcareous clay
layer was encountered down to at least 10m. The clay has a
shear strength of 8kPa at the surface, increasing uniformly to
35kPa at 10m. Laminations of organic matter and pockets of
silty fine sand occur throughout the deposit investigated.
Design Soil Parameters
Fig. 3 shows the combined plots of undrained shear strength
profiles. The results of laboratory tests are also shown
superimposed on these plots. The characteristic strength
profile revealed from these plots is as follows:
Su = 8 + 2.7 z (Su units in kPa)
The presence or otherwise, nature, depth, and degree of
variability of the top gravely sand layer are of special
importance for the present mudmats. All the core samples
were split and photographed prior to logging. A typical
photograph of the cores is shown in Fig. 4. Permeability
values of the surficial granular layer are in the range 2.25x10-
8 and 4.0x10-4m/s. If it could be guaranteed that this surficial
layer would be present at the time of pin-piling operations,
then there would be no need for the geotextile diffuser layer
lining the underbase of the mudmats. However, prior to pin
pile installation operations at least one well was to be drilled at
DWG and it was likely that significant drill cuttings would be
present on the seabed spread over the site. For these reasons
the high permeability benefit from the insitu granular layer
could not be relied upon and hence design of mudmats
assumed cohesive seabed conditions.
The clay sensitivity with depth obtained from a number of
different tests and instruments is variable, with an average
value of 3.0 over the deopth of interest. Atterberg Limits
within the depth of interest is reasonably uniform and
indicates an average plasticity Index of around 27 in the upper
3.5m.
Combined Bearing-Sliding Capacity
The combined bearing-sliding-moment capacity of shallow
foundations typically falls on the surface of an egg-shaped 3D
envfelope7. In the absence of moment loading, the envelopes
degenerate to a 2D cross-section of the 3D version. Both
conventional bearing capacity envelope methods8 and
nonlinear finite element analysis techniques were used in the
design of mudmats. A load and resistance factor design
methodology was adopted in accordance with API LRFD9.
Factored loads generated from structural analysis of the PPIF
and LA structures were used. In the FE analyses the
characteristic soil strength was derated by the appropriate
material resistance factor. One of the most important aspects
of the design of these mudmats is the consideration of the
“breakout” phenomenon. The mudmats need to be kept to a
small size in order to reduce weight and facilitate easy release
from the seabed, while at the same time be able to carry the
design loads without failure or excessive settlement.
OTC 18923
Foundation Loads
The maximum load on the 5th mudmat occurs when the PPIF
was being installed over the DUQ drilling template. At this
stage the weight of the PPIF was supported on the tops of the
two template support piles, together with the 5th mudmat
forming a three-point support. The factored load on the 5th
mudmat for this condition was 1472kN. Similarly each time
the LA was being set-up its weight was carried on a three-
point support, the two claws at the PPIF end and the Apex
mudmat at the pile sleeve end. The load on the Apex mudmat
for this load condition was 1100kN.
Reaction loads on skirted mudmat supports were produced
from the SESAM structural model. The worst scenario loading
on the skirted mudmats occured on one of the aft mudmats of
the LA when the pile was being stabbed into the sleeve cone
(and inadvertently landed on the edge of the cone). For this
load case 25% of the pile weight was applied statically onto
the pile cone top rim in the structural analysis. The horizontal
and vertical mudmat reaction components for this scenario
were 123kN and 1874kN, respectively.
Conventional V-H Stability Envelope Results
Bearing capacity analyses were performed in accordance with
DnV8 recommendations for linearly increasing soil shear
strength with depth. Ultimate and allowable bearing capacity
envelopes (V-H diagrams) were produced for skirted and un-
skirted mudmats. V-H diagrams for both skirted and un-
skirted mudmats assumed a ‘rough’ base condition.
Fig. 5 shows the envelopes for un-skirted PPIF 5th and LA
Apex mudmats together with the maximum load conditions. It
can be seen that for the mudmat diameter of 6.2m the
conventional analysis methods predict a vertical bearing
capacity with an available factor of safety (FOS) in the order
1.21, less than the API LRFD required value of 1.5.
Similar envelopes for the skirted mudmats showed that for the
latter case even the ultimate capacity of the mudmats would be
exceeded.
The above results suggest that in the conventional sense
the mudmat area is not sufficient. The conventional bearing
capacity formulae make a number of simplifying assumptions,
chiefly among them is the assumption of a uniformly loaded
area, i.e. a flexible foundation with uniform pressure with no
soil containment of skirts. This shortcoming, together with the
restrictions imposed by seabed lifting requirements
(desirability of smaller mudmats to make breakout from the
seabed easier) motivated the use of non-linear FE analysis.
The results obtained from FE analyses are discussed below.
Finite Element Model and Results
A number of finite element models were used to analyse the
bearing capacity, settlement and stiffness of the mudmat
foundations. Fig. 6 shows a typical deformed shape contour
plots of the early 3D FE model used to derive foundation
lateral, vertical and combined capacities and stiffness for
skirted and non-skirted mudmats. Due to symmetry, in the
subsequent analyses only one half of the full domain was
analysed. In the absence of any soil test results which might
have provided parameters for an advanced soil constitutive
model, the soil was represented with the Mohr-Coulomb
constitutive model, calibrated for undrained clay response.
3
The undrained internal friction angle, u was set equal to zero
and a perfectly plastic post-yield behaviour was selected.
Therefore, the yield and failure surfaces were coincident
implying an associated flow rule. The Poisson’s ratio was set
equal to 0.49 to simulate no volume change conditions for
undrained clay.
The steel structure of the mudmat was modelled with shell
elements. Fig. 7 shows the steel structure of a typical skirted
mudmat. Photographs taken at the site during fabrication are
presented in Fig. 8a and 8b. Distribution of contact soil
pressures against skirts as well as bending moments and shear
forces were obtained from this model. For the un-skirted
mudmats the skirt elements were simply removed from the
model, leaving a flat base.
Combined vertical and horizontal loads were applied to the
centre node of mudmat at the top level of web stiffeners and
ramped up proportionally to twice the maximum design values
in order to define the full load-displacement response of the
mudmat.
The mudmat skirts were coated with friction reducing
paint to make the extraction of mudmats easier, and the
laboratory interface tests (discussed later) showed that the
prevailing -factor with the specified paint system was of the
order 0.37. In order to account for contact and desired -factor
along skirt interface, a 3D FE model was developed which
allowed for large-sliding frictional contact between the outer
face of skirts and soil. A finer mesh grading in the vicinity of
the mudmat foundation was used together with a lining of
special infinite elements outside of this core mesh. In this way
a relatively fine mesh of well proportioned and shaped
elements captured the elasto-plastic behaviour of the soil and
contacting surfaces with good accuracy, while not
compromising the contribution of the rest of the soil domain.
A typical output showing contours of vertical displacement
at the end of the analysis run with the latter model is shown in
Fig. 9. The classical “displacement bulb” shapes can be seen,
and they are continuous across the boundaries between the
finite and infinite element meshes.
A set of three load-displacement curves obtained from the
above model are shown in Fig. x10. The top curve shows the
results with an upper bound estimate of undrained Young’s
modulus profile varying between 200*Su at seabed to 500*Su
at 16m below mudline. The -factor used on skirt interface
was 0.5. For comparison the analysis was repeated with the
lower bound estimate of Young’s modulus profile varying
between 100*Su at seabed to 400*Su at 16m below mudline.
The results shown in Fig. x10 suggest that a minimum
vertical capacity of at least 2000kN is obtained. At
approximately 2000kN load the curves begin to flatten and
this load level was treated as ultimate load. Load levels
beyond 2000kN conservatively ignored, since element
distortions may mean the results are unreliable beyond this
point. Therefore according to the Load and Resistance Factor
Design philosophy an additional margin of safety of at least
6% exists.
Foundation Settlement
During setting up of the PPIF a substantial proportion of the
structure weight is carried by the 5th mudmat for a short
period until the skirted mudmats were jacked down. Once the
skirted mudmats was embedded the 5th mudmat was extracted
4 OTC 18923

and the PPIF structure will be levelled. Similarly for the LA DAYS AFTER INSTALLATION 2 4 7 14 30 60
the Apex mudmat supported the sleeve end until the aft
mudmats were lower to the seabed and jacked in to take up Consolidation Settlement (mm)
their share of the load. For these conditions the immediate
elastic settlement of the un-skirted mudmats was of
importance. (Upper Bound) Consolidation
During operation the skirted mudmats were expected to Settlement
10 15 20 23 25 30
undergo some degree of consolidation settlement.
Consolidation is a very slow process in clays due to their soft (Lowe Bound) Consolidation
skeleton and very low permeability. These time-related Settlement
consolidation settlements were calculated using an FE model 0.5 1 1.2 3 4 8
described below.

Immediate Settlements From Elastic Theory Foundation Stiffness

Elastic settlements of the 5th and Apex mudmats during setting Foundation stiffness was calculated from both the 3D finite
down on the seabed were estimated using the conventional element model and conventional methods provided in
elastic theory hand calculation10. For the high estimate Reference 8. In Fig. x13 the results of horizontal load against
stiffness estimate Ir values selected were 100 at mudline and horizontal displacement are compared for a mudmat with and
500 at 12.8m depth. Fig. x11 shows the profiles of predicted without skirts. The effect of skirts is to double the horizontal
elastic settlement of the PPIF 5th mudmat with depth below sliding capacity as well as increasing the lateral stiffness by
mudline for the low estimate and best estimate rigidity index about 40%.
values under the loads expected. In general the initial stiffnesses from FE analysis are
higher than DnV results. The results from both sets of analyses
Consolidationn Settlements using Coupled FE are summarized below in Table-2:
Analysis
Simulation of time dependent consolidation effects are best Table-2 Mudmat foundation stiffness values
performed using the FE methods, since closed form solutions
are only available for specific conditions such as 1-D cases ABAQUS ABAQUS DNV
and cannot model arbitrary shaped foundations and boundary (INITIAL) ( AT 50%ULTIMATE (REF. 8)
conditions. An axisymmetric FE model was used for these LOAD)
coupled stress-pore pressure analyses. In this model the
(kN/m)
mudmat was represented with axisymmetric shell elements.
Free drainage was allowed to take place from beneath the Horizontal 39,000 29,000 34,000
mudmat base plate as well. This simulates the presence of the Vertical 70,000 60,000 51,000
diffuser with the passive suction valves being left open.
Consolidation was simulated for a hypothetically long time of
33 years in order to obtain the theoretical final settlement as a
Jack-to-Mudmat Connection Design
reference value. Upper and lower bound estimates of soil
When the LA mudmats approach the sloping seabed with an
stiffness and permeability coefficients were used to obtain a
attitude relative to the seabed, the skirt penetration is
range of predicted consolidation settlement. Elastic parameters
essentially axially symmetric. However once the base plate
were the same as those used for elastic settlement calculations.
starts to contact the seabed the eccentric loading causes a
Permeability coefficients were obtained from the results of a
bending moment about the jack connection point.
few oedometer tests that were performed on specimens from
the top layers as part of the deep borehole SI.
Skirt Frictional Resistance
Typical consolidation settlement-time curves for the PPIF
The penetration and extraction resistance of mudmat skirts
5th mudmat with the high and low soil stiffness estimates are
were assessed following the well established NGI
shown in Fig. x12. The results of consolidation settlement
methodology11. In this method the unit skirt friction, f, is
prediction for the PPIF skirted mudmats under still water loads
calculated as:
after a selection of time periods relevant to the installation
activities are listed in Table-1 below. The static load for this
case is 1232kN. Consolidation settlement values have been f = Aw . DSS . SuDSS
rounded up to the nearest 5mm for the upper bound case. The
lower bound settlement estimates are negligible. Where:

Aw = Skirt Wall Surface Area in contact with soil

Table-1 Consolidation settlements after a selection
of time periods following installation
DSS = -factor for skirt-soil interface
SuDSS = Undrained shear strength in Direct Simple Shear
(DSS) test
In the absence of any DSS test data, a typical correlation
factor between triaxial compression and DSS strengths of 0.8
OTC 18923 5

was used. The -factors for extraction were selected using the where, GMF is the Gross Mudmat uplift Force. It is the
relation: EMF that causes a differential suction head to develop across
the Terram layer and hence suck the water into the mudmat
 = Ct (1/ St) chamber allowing it to be released. In Fig. x16 curves have
been plotted for different size of total passive openings in the
Where: mudmat. All mudmats have four passive holes of 160mm
Ct = Clay thixotropy factor diameter each (equivalent to one 12” diameter opening),
St = Clay sensitivity except the Row-1 mudmats of PPIF which have an additional
two 12” holes (i.e. total of 3x12” opening size, the brown
A thixotropy factor of 1.2 was selected following the curve).
recommendations in Ref. 15 for estimated time lapses between As an example, to extract the Row-1 mudmats in a time of
setting up and lifting off of the PPIF of the order one month 1.1 minutes, an EMF of 500kN would be required. Therefore
(exaggerated to account for a breakdown scenario). The - for total skirt frictional resistance of 10te (at the start of
factors calculated were 0.33 for penetration and 0.4 for extraction) and mudmat self-weight of 10te, the gross upward
extraction. force required to extract the Row-1 mudmats would be

Sizing Passive Suction System Capacity 500kN = GMF -100kN – 100kN,

Basic data from the standard Terram data sheet was used in
developing the Passive Suction system. A spreadsheet giving a Gross Mudmat uplift Force of 700kN. The Green
program was used to perform the calculations. The essential (uppermost curve) in Fig. x15 depicts the force-time
steps in the calculation process are summarized below: relationship for all the other skirted mudmats. For these
mudmats if the same constant force of 670kN was to be
applied, they would be completely extracted in 3.3 minutes
Base area A
Sizing Active Injection System Capacity
The criterion for designing the size of active injection system
was to limit the required time to less than the practical time of
d 30 minutes. The design pump capacity was quoted at
120m3/hr. The total chamber volume of mudmat to fill is 22.5
m3. Hence it was expected that the pumps should be capable
of extracting the mudmats out in 11 minutes. In practice
1. Uplift Force is F (assumed constant)
piping would occur in the soil shortly before the skirts were
2. Suction Pressure, P = F/A
completely out, especially when the skirt tips rose to the level
3. Suction Head, h = P/9.81
of granular layer on the seabed. If this was to happen water
4. Suction-Flux, q = qTerram . h/hTerram
injection would be continued, while the structure was lifted
(Units: m3/s per m2 of Terram)
out by the crane.
5. Volume Increment, V = A ( dn+1 –dn)
The final design of the lifting system proved tom be
6. Upward Displacement Increment, d governed by the response characteristics of the derrick barge
7. Time to Lift by d (and fill volume V), in the design sea state. This entailed incorporation of a
t = V/(q x A valves) Cranemaster damper element into the lifting line in order to
8. Extraction Time, T = t . V/V damp out the “shock” forces. In the actual field operations the
9. Extraction Speed, S = d/T passive system performed perfectly adequate and the active
system was not needed. The active injection system was only
used as an additional aid under the unskirted mudmats during
Passive + Active Injection Case, t = V/[(q . Avalve) + Qactive]
lifting off of each structure clear of the seabed to give a
precautionary protection to the Cranemaster damper element.
The latter Passive + Active Injection Case was for the final
lift off of the PPIF from seabed at the end of operations, when
Model Tests on Extraction Resistance
only the two Row-1 mudmats were embedded. Water was to
Due to the novel nature of the concept used to overcome the
be actively injected while the passive suction holes were open
mudmat sticktion problem a series of 1:20 laboratory scaled
to allow the mudmats to be released from seabed easier.
model tests were carried out by Fugro Ltd in the UK. In
For typical skirted mudmats extracted by retracting the
addition, a series of interface friction tests were also
jacks Fig. 15 shows the relationship between Effective
performed by Fugro Ltd to assess the range of likely values of
Mudmat Forces (EMF) and time required to completely
remove the skirts from soil. The EMF is the net uplift force -factor that might develop between the painted surfaces of
which is required to cause the passive flow into the expanding skirts and clay. These interaface tests are described and
chamber under the mudmat as it moves upwards: discussed later.
The prime objective of these tests was to verify that the
EMF = GMF – (mudmat self-weight) – (skirt friction) Terram-lined base plate separated from the seabed as the
design intended. Both Passive Suction and Active Water
Injection Systems were tested. The following means of
6 OTC 18923
mudmat release mechanism from seabed were simulated in the
model tests:
 Base Plate fitted with Terram lining and provided
with active or passive holes
 Base Plate with no lining and provided with passive
holes
Three possible states of seabed condition were considered
in the test programme:
1. Insitu conditions, where the seabed is covered by a
nominal layer of granular material, simulated by a
3mm thick layer of sand
2. Seabed being covered by an impermeable layer of
drill cuttings, simulated by a 2mm thick layer of
bentonite slurry.
3. Bare clay, corresponding to the condition of a seabed
being jetted to remove drill cutting materials and in
the process removing the granular layer as well.
In all, four separate models were milled from single pieces
of Aluminium alloy (dural) all to 1:20 geometric scale. The
model skirts were 38mm deep and 0.75mm thickens. In order
to keep the correct scaling, painting was avoided. Instead, the
model skirts were polished to a smooth surface, except for one
case where rough skirts were simulated. The smoothing was
intended to simulate the friction reducing effect of paint on
skirts. Three pore pressure transducers were used to monitor
the underbase cavity water pressure. A selection of
photographs taken during the model testing programme is
shown in Fig. 16.
The tests proved that the presence of Terram ensures a
clear and easy separation of the mudmat from soil in both the
active and passive cases and in all three seabed soil conditions
tested.
Discussion of Passive Tests
In nine of the passive tests the models were pulled at a rate of
8.33 mm/sec corresponding to the design prototype extraction
time of 30 minutes. In all of these tests the mudmat was
recovered successfully.
After the design of testing apparatus and start of the
programme it was decided to do an additional test. This was
to simulate snatch loading of the crane-hook lifting the PPIF
structure off the seabed on completion of the pin-pile
installation operations. In this case the weight would be
supported on the two Row-1 mudmats of PPIF, fully
embedded in the seabed together with the un-skirted 5th
mudmat. seven attempts were made on this test to pull the
mudmat model off the soil bed at the faster rate of 25mm/sec
corresponding to prototype extraction time of 10 minutes. The
outcome of this test is quite significant, even though the
apparatus was not designed for it and equipment failures were
caused.
The results showed a strong dependence of the mudmat
resistance on uplift speed. Higher under-base suctions
recorded in this test resulted in loads that were up to three
times higher than the other passive tests. Indeed the attempted
model scale pulling speed of 25mm/sec was based on
limitations of the testing equipment (loading motor and frame,
etc.). Even this pulling rate could not be sustained by the
loading motor and after an extraction of 6-7mm the rate
reduced. The maximum underbase suction recorded in test 12
was 14kPa, compared to 5.5kPa in the other passive tests. This
observed dependence of pulling force on uplift displacement is
reproduced by the analytical predictions presented later in the
paper.
Discussion of Active Tests
In two of the tests the active injection system with the most
onerous conditions of structure self weight (20kg and 55 kg
surcharge) sitting on the mudmat with bare clay seabed was
simulated. Even in these tests the diffuser element (Terram)
allowed the injected water to be distributed under the base
plate and the mudmat was lifted off the seabed. In all the
active injection tests piping occurred generally prior to
complete extraction of the skirts. However, in these tests
piping occurred earlier than active tests without surcharge.
The piping was severest in test with 55kg of surcharge.
Soil Skirt Interface Model Tests
The skirt surfaces on both inside and outside of the mudmats
were painted with friction reducing paint. A two-coat paint
system was used to give a low friction finished surface
condition. A number of interface tests were performed by
Fugro Ltd. on three different surface conditions:
1. The smooth Aluminium with three different
consolidation times: 23, 89 & 963 minutes.
2. Paint system 1- only first coat applied on a steel
specimen
3. Paint system 2– both coats applied on a steel specimen
Surface condition of the smooth Aluminium alloy (Dural)
was as used in the 1:20 scaled model tests discussed above.
In all, five tests were performed (3 on smooth aluminium
and 2 on painted steel specimens. The tests were performed in
a manner to simulate the conditions that would prevail in the
field and also in the model tests, in terms of set-up time,
shearing rate (8mm) and soil specimen interface orientation.
The soil specimens used were all clay from depths ranging
0.75m -1.65m depth.
The following conclusion was reported: The peak average
interface friction ratio ( -factor) for Aluminium alloy as used
in the mudmat model tests is 0.37. The peak average interface
friction ratio for both Paint systems 1 and 2 is 0.35.
These -values agree well with the calculations based on
DSS strength, described above; between 0.33 for penetration
and 0.4 for extraction.
Conclusions
Scaled model tests and FE analyses have shown that the
Terram diffuser is very effective in eliminating the underbase
adhesion during the removal of mudmats, both with the active
water injection and passive suction schemes. However, one
test simulating an extraction time of 10 minutes (prototype),
i.e. three times faster than the design value, showed that the
force required is almost three times the value for extraction
time of 30 minutes (prototype). Therefore, any “snatch”
loading should be avoided during the field operations.
OTC 18923 7

Laboratory interface tests show that the selected paint
system reduces skirt friction effectively. The skirt adhesion -
value obtained from the interface tests was around 0.37.
Maximum in place loading occurs during pile stabbing into
the sleeve cone. The size of mudmats has been kept to a
minimum in order to minimise problems during removal.
Although the bearing capacity of mudmats has been verified to
be adequate using non-linear FE models, care should be taken
during pile stabbing operations to avoid this load scenario and
potentially disturbing the levelling of the LA.
Maximum consolidation settlement of the longest loaded
PPIF mudmats is predicted to be of the order 30mm over a
period of 1 month. The maximum immediate elastic settlement
of the PPIF 5th mudmat is estimated to be of the order 100mm.
It is recommended that the settlement of mudmats be
monitored during field operations to assist in any necessary re-
levelling of the PPIF structure.
Acknowledgement
The author is grateful to BP Exploration UK for permission to
publish this work.
5. “On the Extrication of large objects from the ocean bottom (the
breakout phenomenon)” Foda, M.A., J. Fluid Mech., Vol. 117,
p. 211-2331.
6. Hesar, M. “Optimal Inclined Bearing Capacity of Shallow
Foundations”, Proc. 16th ISOPE Conf., Paper JSC-416, San
Francisco, 2006.
7. Bransby, M.F. & Martin, C.M. (1999). “Elasto-plastic
modelling of bucket foundations” Proc. 7th Int. Symp. on
Numerical Models in Geomechanics, Graz, pp 425-430.
8. Det Norsk Veritas, Classification Notes No. 30.4
“FOUNDATIONS”, February, 1992
9. API RP2A-LRFD, 'Recommended Practice for Planning,
Designing and Constructing Fixed Offshore Platforms - Load
and Resistance Factor Design', American Petroleum Institute,
First Edition, July 1, 1993
10. Janbu, N., Bjerrum, L. and Kjaernsli, B. (1956): Norwegian
Geotechnical Institute Publication No. 16
11. Norwegian Geotechnical Institute, Joint Industry Project
“Skirted Foundations and Anchors in Clay: Set-up Effects
outside Skirt Wall”, Report No. 524071-2, 11 May 1999

References
1. Tjelta, T.I., Aas, P.M., Hermstad, S. and Andenaes, E. “The
skirt piled Gullfaks C installation”, OTC 6473, Houston 1990.
2. Tjelta, T.I., “Geotechnical aspects of Bucket Foundations
replacing piles for the Europipe 16/11-E jacket”, OTC 7379,
Houston 1994.
3. Kolk, H.J. and Kay, S. “North Nemba Flare Bucket
Foundations”, OTC 13057, Houston 2001
4. Fisher, R. “Gravity Base Design for Subsea Structures”, Proc.
Int. Conf. Offshore Site Investigation and Geotechnics –
Diversity and Sustainability, London, November 2002.
8 OTC 18923

Figure 1 Bathymetric plot with PPIF and LA positions on DWG

sites

Figure 2 Conceptual Schematics of PPIF and LA

OTC 18923 9

Fig. 3 Undrained shear strength profile Fig. 4 Photograph of a typical Core sample from seabed
10 OTC 18923

2000

1800

1600

1400

1200
Vertical Load (kN)

1000

800

600

400

200
Fig. 7 Mudmat structure modeled with shell elements

0
0 50 100 150 200 250 300
Horizontal Load (kN)

Ultimate Rough Surf ace Factored Rough Surf ace Factored Design Loads

Fig. 5 Stability envelopes for un-skirted PPIF 5th and LA Apex

Mudmats

Fig. 8 (a) PPIF structure under construction

Fig. 6 Typical Deformed Shape plot showing Contours of Vertical

Displacement (Combined Lateral-Vertical Loading)

Fig. 8 (b) Skirted mudmat during fabrication in Baku

OTC 18923 11

ACG Phase 3 DWG - Pin Pile Installation System Mudmat

Immediate Elastic Settlement of 5th PPIF Mudmat

M udmat Elastic Settlement (mm)

0 20 40 60 80 100 120
0

10

Depth Below Mudline (m)

15

Fig. 9 Fig. 4.9 Contours of Vertical Displacement, Combined

Lateral-Vertical Loading ABAQUS model with 3D Infinite
Elements, magnification factor = 10.
20

25

30

Lower Bound Settlement Upper Bound Settlement

Fig. 11 Immediate seabed settlement profiles under 5th mudmat

from Elastic Theory

ACG Phase 3 DWG - Pin Pile Installation System Mudmats

Consolidation Settlement of Typical PPIF Mudmat

0.00 0.00 0.10 10.00 1000.00 100000.00

-60

-70

-80
Settlement (mm)

Fig. 10 Vertical Load versus Vertical Displacement response from -90

model shown in Fig. 4.9

-100

-110

-120
Time (Days)

Low Estimate High Estimate

Fig. 12 Consolidation settlement time history of skirted mudmats

12 OTC 18923

AIOC PROJECT - PHASE 3

DWG Pin-Pile Instalation Frame Mudmats
Horizontal Load-Displacement Response

700

600

500
Lateral Load (kN)

400

300

200
Fig. 16 (a) Laboratory scale model for the active injection system
testing

100

0
0.000 0.010 0.020 0.030 0.040 0.050
Lateral Displacem ent (m)

LA-6m LA-6m-sk75

Fig 13 Lateral load-displacement response with and without skirts

AIOC - ACG PHASE III - Deep Water Gunashli

PPIF Row-1 Skirted mudmat extraction Constant Force
Extraction Times extrapolated from vol. increments)

900

800 6x12inch valves

"5x12inch valves"

700 4x12inch valves

Effective Mudmat Uplift Force (kN)

"3x12inch valves"
600
"2x12inch valves"

4x6inch valves
500
Fig. 16 (b) Laboratory scale model for the passive suction system
testing
400

300

200

100

0
0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0
Extraction Time (Minutes)

Fig. 15 Effective mudmat uplift force v. Extraction time for

different passive suction hole opening sizes
OTC 18923 13

Fig. 16 (c) Scale model in the test tank