Centrifuge model study on spudcan footprint interaction 3

In the commentaries SNAME, 2002b, a general guideline on theacceptable minimum distance of one spudcan diameter measuredfrom the edge of the bearing area to the edge of footprint isrecom

2.2 Guideline - SNAME 2002

SNAME (2002a) considered the footprint problem in two operationalsequences as follows:

i) Use an identical jackup design (same footing geometries and leg

spacing) and locate it in exactly the same position as the previousrig

ii) However, it is unlikely that two jack-up units have an identical

design as the structures of most units are often custom-made andthe deployment of units is subject to availability In this situation,SNAME (2002a) suggests to carefully position the jackup on a newheading, or/and with one footing located over a footprint and theothers in virgin soil, to alleviate the potential for spudcan sliding

In the commentaries (SNAME, 2002b), a general guideline on theacceptable minimum distance of one spudcan diameter (measuredfrom the edge of the bearing area to the edge of footprint) isrecommended However, footprint features due to various factors(such as soil type, penetration depth and time) are not clearlydefined in the guideline When and where the spudcan-footprintinteraction becomes problematic to new rig installation still remainunidentified

If it is not possible to avoid spudcan-footprint interaction, the guidelinesuggests infilling the footprints with imported materials.The material selectionshould recognize the potential for material removal by scour, and differences

in material stiffness No further detail is given in the guideline The suitabilityand effectiveness of infilling to ease the footprint problem in various soilconditions are still questionable

2.3 Spudcan-footprint interaction

2.3.1 A single leg

When a spudcan penetrates into the seabed, there are two force componentsthat are equal in magnitude but acting in opposite direction, so that the wholesystem is in equilibrium These two force components are the soil bearingresistance and the preload from the structure Hence, the interaction between anew spudcan installation and old footprint is not merely dependent on soilconditions but also the manner the structure interacts with it If the jackup leg

is fully restrained where no lateral movement is permitted, the spudcan wouldtend to penetrate vertically and large lateral loads will develop On the otherhand, if the jackup leg is fully flexible where no lateral movement is resisted,the spudcan will tend to slide into the adjacent footprint (Stewart & Finnie,2001) The leg penetration responses with fully restrained and fully flexible atthe leg are illustrated schematically in Fig 2.1 In reality, the jackup leg-hullconnections are neither fully rigid nor fully flexible and that makes theproblem complex The interrelation between structural response andgeotechnical response with these two conditions can be further illustrated byexamining the predicted horizontal load-displacement response as shown inFig 2.2 (Stewart & Finnie, 2001) The intersection point of the geotechnicaland the purely elastic structural response predicts the potential horizontal loadthat develops in the jackup leg and the corresponding horizontal displacement

at the spudcan This implies that the load imposed on the spudcan is a function

of horizontal displacement and vice-versa

Dean & Serra (2004) examined this problem differently Conceptually,they considered the forces needed to install a spudcan vertically over a

boundary between a region of stronger soil (less disturbed soil) and a region ofweaker soil (disturbed soil), see Fig 2.3 The force needed to push aside thestronger soil is larger than what is needed for the weaker soil Depending onvarious factors, a net reaction force inclined to vertical will then be obtained.This force has to be provided by the jackup structure for the spudcan to movevertically or else it will move in an inclined direction and slip towards theweaker soil They termed the phenomenon as a ‘weak-seeking’ response Onthe other hand, if an open cylinder is pushed vertically, a larger reaction forcefrom the stronger soil, producing a resisting moment that must be provided bythe jack-up Else, the footing would rotate towards the stronger soil Theytermed this as ‘strong-seeking’ behaviour.

The probable spudcan-footprint interaction described above is solelypurely based on postulations without being experimentally proven Animportant aspect that was not considered in the postulations is that the soilfailure mechanism which may alter the foundation behaviour Both studiesassumed the uneven soil bearing resistance dominates the interaction, whichmay not be the case for footprint problem As far as the author is concerned,there is very little study on the soil failure mechanism reported in the publicdomain

2.3.2 System behaviour

Most of the footprint studies were conducted based on a single leg In reality, arig typically consists of 3-leg The legs pass through openings in a barge hullwhere its deck serves as the platform for drilling equipment and othermachinery The basic function of leg-hull connection is to allow forces totransit between the legs and the hull Each connection commonly consists of a

pair of upper and lower guides and a jacking system and/or fixation system(Fig 1.5) Hence, the spudcan-footprint interaction is not only governed byhow a single leg interacts with the footprint, but also how the whole rigresponse to the induced forces and/or displacements The interaction involvingthe entire rig is termed as system behaviour When the system behaviour isaccounted for, the footprint does not just interact with a single jack-up leg butinstead with the whole jack-up The structural stiffness is not only dependent

on the leg above the spudcan and its connection to the hull, but also thestiffness of the hull, the hull connections to the other legs and their stiffness,lastly foundation stiffness of the other spudcans (Jardine et al., 2001)

Randolph et al (2005) postulated that if the movement of the jack-upstructure as a system is discounted, the spudcan loads and displacementmeasured are potentially significantly lower than those if the system behaviourwas accounted for They recommended full three-legged jack-up experimentalstudies However, to-date, no simulation either numerically or experimentally

on the system behaviour is reported in the public domain However, owing tothe complexity of the problem due to many parameters are involved, itbecomes extremely difficult to study the influence of each parameteraccurately if a full jack-up is modelled

2.3.3 Experimental modelling

2.3.3.1 Effect of offset distance

Stewart & Finnie (2001) conducted a series of centrifuge model tests toexamine the effect of offset distance of spudcan installations adjacent tofootprints The offset distance is the distance from the spudcan centre to thefootprint centre The tests were performed using the 1.2 m diameter drum

centrifuge at University of Western Australia Over-consolidated kaolin claywith an undrained shear strength, suof 12 kPa at the mudline to 53 kPa at adepth of 20 m was used The tests were conducted at 200g with a 60 mmdiameter model spudcan (12 m in prototype) The jack-up leg was rigidly fixed

to the vertical actuator A footprint was created by performing spudcanpenetration and extraction The spudcan was then re-penetrated at offsetdistance 0, 0.25, 0.5, 0.75, 1.0, 1.5 and 2.0 times footing diameter, D Theyfound that the vertical load was reduced to between 40% and 70% of the loadobtained during initial penetration for various offset distances (Fig 2.4) Animportant finding was that considerable large horizontal forces developedduring re-penetration at the offset distance from 0.5D to 1.0D and a peak of1.3 MN horizontal force developed at the offset distance of 0.75D (Figs 2.5(a)

& (b))

Cassidy et al (2009) reported a similar qualitative trend of thebehaviour due to the offset distance for spudcan re-penetration The modelspudcan used followed the design of jack-up rig Mod V with a prototypediameter of 18.2 m The tests were conducted in 250g in clay of lightly over-consolidated profile The undrained shear strength profile measured using theT-bar was approximated to be 7.5 kPa at the mudline with 2 kPa/m increasewith depth.Fig 2.6 shows the induced maximum horizontal H and moment Mloads on the spudcan They found that the effect of footprint is minimal foroffset distance of greater than one and a half spudcan diameters Among fouroffset distances tested (D/4, D/2, D, 3D/.2), the induced H and M at an offsetdistance of D/2 was found to be the greatest

The above studies had identified a range of critical distance (whererelatively high horizontal force and moment are likely to occur) that should beavoided to minimize the risk of sliding into footpritnt As the same spudcanwas used to create the footprint and to perform the spudcan re-penetration, theeffect of varying spudcan sizes was not addressed in both studies.

2.3.3.2 Effect of leg stiffness

It has been suggested that the leg flexural rigidity would influence theinteraction between a spudcan and a footprint (Stewart and Finnie, 2001; Foo

et al 2003; Dean and Serra, 2004) Fig 2.7 shows a comparison of theequivalent stiffness EI/L3 of four different jack-up rigs over the leg length L

by Foo et al (2003) They pointed out that the equivalent stiffness differencebecomes less distinct when a jack-up operates in deeper water with longer leglength It is apparent that there is no typical leg stiffness and the equivalent legstiffness is dependent on the length Foo et al (2003) postulated that theinfinitely stiff leg would take the most bending moment as compared to onethat is fully flexible

Cassidy et al (2009) reported a set of drum centrifuge test resultsinvestigating the effect of the leg stiffness on spudcan-footprint interaction.The leg stiffness of two jack-ups, the 116C and the Mod V, was modelled inthe centrifuge Two sets of legs were manufactured for each jack-up type, asshown in Fig 2.8 The model legs used have prototype stiffnesses rangingfrom 1.69 × 1011 Nm2 to 2.75 × 1012 Nm2 They found no distinguishabledifference in the induced horizontal force, H during spudcan-footprintinteraction for the 4 different leg stiffnesses with a similar experimental set-up(rigid connection), see Fig 2.9 They attributed this surprising result to the

experimental set-up With a fixed connection between the top of the loadingleg and the centrifuge actuator, the displacement and rotational movementswere restrained and subsequently reduced the contribution of the leg stiffness

to the overall system feasibility However, this finding may only be applicable

to the range of leg stiffnesses and forces that were tested Beyond thesestiffness and loading ranges, the penetration response may be different

2.3.3.3 Effect of preload

Three test results of different preloads for the initial penetration were reported

by Cassidy et al (2009), see Fig 2.10 The preloads level were 40 MN, 60

MN and 80 MN, corresponding to preload pressure of 154 kPa, 231 kPa and

308 kPa respectively for the Mod V spudcan (prototype footing area of 259.7

m2) The peak horizontal force for reinstallation at offset distance 0.5D fromfootprint created by preload of 60 MN and 80 MN were very similar, but wassignificantly less for the 40 MN preload They accounted this as a reflection ofthe change in footprint shape for different preloads as a deeper crater wascreated in a higher preload case On the other hand, higher moment wasinduced on the spudcan for the higher preload case As the undisturbed soilhad a linearly increase shear strength profile, the variation in strength of thesoil beneath the footprint and the differential in bearing pressures alsoincreased with depth This may attribute to higher moment was induced in thehigher preload case

2.3.3.4 Effect of seabed irregularities

Teh et al (2006) performed a series of 1g tests to investigate the effects ofsloping seabed (30o inclined to the horizontal) and footprint on loads

developed in jack-up leg on sand and clay Two types of foundation namelyspudcan and skirted foundation were used Similar to Stewart & Finnie (2001),the rig flexibility was not accounted for in the model They observed a muchhigher horizontal force for the installation on the footprint than on the slope Inthe study on the spudcan and skirted foundation behaviour interaction withfootprints in clay, they found that the maximum moment that developed onboth spudcan and skirted foundation was comparable, but the moment profilevaried with penetration depth (Fig 2.11) The horizontal loads on these twodifferent foundations acted in the opposite direction This implies that theoverall moment developed at the lower guide of a jack-up might be different.The probable mechanism that caused this behaviour was not discussed in thepaper.

2.3.4 Numerical modelling

2.3.4.1 Limit equilibrium method

Stewart & Finnie (2001) modelled the spudcan-footprint interaction using asimple analytical technique: slope stability coupled with bearing capacitysolutions Footprint was modelled as a steep-sided crater with uniform shearstrength Differential earth pressures due to non-uniformly embedment wereestimated using the equation presented by Rowe & Davis (1982) Loadsdeveloped at the spudcan due to the collapse of crater wall were estimatedassuming limiting equilibrium on a simplified prescribed failure planeextending from the crater toe up to the spudcan edge They claimed that theresults are loosely consistent with what were measured in the centrifuge tests.The authors assumed the induced loads on the spudcan were due to thecollapse of the crater wall This may not be the case as depending on the soil

types, the footprint crater may more likely be a sloping ground rather than asteep-sided wall In addition, the authors ignored the effect of the non-uniformity of the shear strength profile of a footprint Hence, the validity ofthis simulation approach is in question.

2.3.4.2 Finite element simulation

Jardine et al (2001) conducted numerical study on jackup penetration intoinfilled footprint craters on layered cohesive material Two infill types namelymedium dense sand and dense gravel were used An idealized structural modelfor plane strain model was adopted where the spudcan was considered as aninfinitely long prism, shown in Fig 2.12 The footprint was modelled asvertical sides with firm base (Fig 2.13(a)) This is not a realistic feature assloping upper sides are more common as observed in the field Separate finiteelement runs with a mesh representing a different stage of penetration wereperformed (Fig 2.13(b)) The initial shear strength was downgraded by 7%and used as sufor the modeled footprint Three sets of leg stiffness were alsostudied They found that the vertical capacity was significantly reduced by20%, and the full capacity could only be achieved with sustaining relativelylarge lateral leg movement, leg forces and leg bending moment The resultalso revealed that the developed leg forces and bending moment were afunction of leg stiffness where the leg forces increase systematically with legstiffness They pointed out the difference in stiffness between the infilledmaterial and the subsoil would also greatly affect the effectiveness of thismitigation method Jardine et al (2002) presented an updated version of theanalysis in which the finite element mesh employed was further refined Withthe improved analysis, it was found that the composite foundation capacity

was reduced and significant lateral and rotational footing displacements wasresulted, with large lateral forces and bending moments building-up on thejack-up's leg.

A follow-up study was reported by Grammatikopoulou et al (2007).Using a similar analysis method as Jardine et al (2002), the authors examinedpotential infilling solutions for a planned jack-up rig installation in NorthEverest that was eccentrically located over a 3 m-deep craters formed by theearlier rig They considered two solutions: i) infilling the craters with granularmaterial, and ii) combined crater infilling with capping gravel loadingplatforms Examples of geometry for options i) and ii) are shown in Figs.2.14(a) and (b) The gravel platform was assumed to have a maximum sizethat could be accommodated in the site with 10 m wide, 5 m high and a steepslope of 1:1 The gravel infill properties were assumed to have effective unitweight of 10 kN/m3, effective angle of shearing resistance of 40o and angle ofdilation of 30o The coefficient of earth pressure Ko was assumed to be 0.5 forthe soil below the spudcan base, and assumed to be 2 on either side of thespudcan where passive failure is expected It should be noted that the trueinfill properties would depend on the material acquired and the placementprocedure Despite the problem of installing a spudcan over an infilled crater

or gravel loading platform which was essentially a three dimensional interaction problem, it was modelled as a plane-strain problem The su profilewhich was reduced by 7% was used for the analysis of the future rig Whenthe spudcan penetrated the infilled crater (Option 1), the soil movementpatterns were remarkably asymmetrical This led to development of intolerablehorizontal forces and moments The soil movement reverted to a more

soil-symmetrical pattern at lower penetration depth, but at this stage, the jack-uplegs would have been damaged For Option 2, the existence of the loadingplatform had stiffened the initial response that brought forward the onset oflarge asymmetric movements This may likely lead to structure damage wellbefore achieving the required bearing pressure For both options, the computedinduced forces and moments were considerably larger than the structuralcapacity of the rig used before reaching the desired rig preload pressure Thenumerical analyses indicated that neither infilling options were able tomitigate the problem.

Based on the numerical analyses reported by Jardine et al (2002) andGrammatikopolou et al (2007), the crater infilling solutions to footprintproblems may not be advisable when considering clay foundations (Jardine,2009)

2.3.5 Other preventive/mitigation measures

2.3.5.1 Stomping

Jardine et al (2001) briefly discussed, stomping as one of the practical options

to ease spudcan-footprint interaction problem, also reported in ResearchReport 289 (MSL, 2004) Stomping is a designed soil displacementprogramme using the spudcan to perform a series of emplacements initiallyfurther from the footprint centre and finally at desired position (Fig 2.15) Theauthors pointed out three disadvantages for stomping as it involved significantrig-time; could only practice in mild weather; and lastly, the operation waslimited by the clear distance between the fixed platform and the jack-up.Neither experimental nor numerical studies on the stomping method are found

in the public domain

2.3.5.2 Rack Phase Difference (RPD) monitoring

RPD is simply the difference in elevations between the rack teeth of the chords

of any one leg (Fig 2.16) It is used to measure the inclination of the legrelative to the hull and hence may be used to estimate the leg loads Foo et al.(2003) suggested using RPD monitoring during jacking to prevent the legsbending beyond their limits and to track and control spudcan sliding bymonitoring RPD

 Upslope-down slope repeated sliding and final position

 Towing the rig with spudcan on-bottom “ploughing” into final position

 Modify the seabed either by excavation or softening including SwissCheessing

Each mitigation method has its own functionality in treating the footprintproblem They may not be suitable for all soil conditions As such, study onthe footprint characteristics in various soil conditions is essential to identifythe predominant factors in the spudcan-footprint interaction The findings arepotentially useful in selecting a suitable mitigation method for easing thefootprint problem for a particular soil condition

2.4 Spudcan foundation behaviour

Soil flow mechanisms during spudcan penetration and extraction play a keyrole in determining a footprint features including the extent of soil disturbance,seabed profile and probable stress state of the soil beneath the footprint Thedefinition of spudcan footprint can be found inSection 1.3

2.4.1 Spudcan penetration

Hossain et al (2004) observed the top view of full spudcan centrifuge test onnormally consolidated clay (Figs 2.17(a) – (c)) and found a soil distortionzone of 1.67D (D is the spudcan diameter) consisting of soil heaving uparound the spudcan edge when the spudcan edge touched the soil surface Thelateral distortion zone remained the same after the spudcan was fullyembedded In addition to this, a finite element analysis with smooth soil-spudcan interface using H-adaptive RITSS method (details of finite elementmodelling technique can be found in Hu and Randolph, 1998) was alsopresented in the paper Deep failure mechanism, where the soil flow wasentirely local after a penetration d/D = 0.75 where d is penetration depth(Figs 2.18(a) – (c)) The lateral distortion was confined within a zone of 1.6D,which agreed well with the centrifuge test observations (1.67D)

Purwana (2007) conducted a detail study on the soil flow mechanismfor spudcan penetration and extraction in normally consolidated clay A half-spudcan modeling using particle image velocimetry (PIV) technique coupledwith close-range photogrammetry were used to measure the soil movement

He categorized the spudcan penetration process into five transitional stages atdifferent depth ratio, d/D = 0.1, 0.35, 0.5, 0.75, 1.0 and 1.35 At d/D = 0.1, thefull spudcan base touched the soil surface and the side wall of the spudcan was

fully embedded A major displacement field extended about 0.8D below thespudcan base and from the spudcan center outward At d/D = 0.35, an opencavity formed and remained stable at this stage At d/D = 0.5, soil at the side

of the spudcan started moving upward and inward and flow back on top of thespudcan At d/D = 0.75, a stable cavity was formed above the spudcan and atd/D = 1.0, a fully localized soil backflow around the spudcan edge wasobserved Other important findings are the major soil displacement beneath thespudcan consistently extended to 0.8D deep and soil at the spudcan base wasbeing compressed throughout the spudcan penetration

Hossain et al (2004) and Purwana (2007) both found the lateraldistortion at soil surface was in the range of 1.5 to 1.6D and deep failuremechanism was initiated after d/D = 0.75

2.4.2 Spudcan extraction

Purwana (2007) also examined the soil flow mechanism during spudcanextraction He found that at 2.0 m uplift spudcan displacement, the breakoutfailure occurred where the spudcan separated with the soil below it (Fig 2.19).After the breakout displacement, the soil flowed from the spudcan top to thespudcan base The subsequent extraction can be characterized as uplifting asoil column above the spudcan coupled with massive downward movement ofthe adjacent overburden soil The lateral extent of downward soil movementextended from the spudcan edge to about 1D at the soil surface Thisphenomenon remained until the spudcan was fully extracted Shortly after thespudcan was fully extracted from the soil, an immediate collapse of the cavity,where the adjacent near-surface soil flowed into the cavity occurred and a

depression was formed on the model ground These observations wereconsistent with the findings by Xie (2009).

2.5 Shear strength profile

Siciliano et al (1990) investigated the change in the shear strength in consolidated clay due to remoulding effect caused by spudcan penetration Thespudcan was installed and extracted at 100g but the measurements ofundrained shear strength profile of footprint were conducted at 1g They foundthat the variation in undrained shear strength profiles was a function of thedistance from the spudcan centre The result (Fig 2.20) revealed that the soilprofile at one diameter and greater from the spudcan centre were essentiallythe same Therefore, the authors suggested no remoulding of the soil beyondthat distance The accuracy of the results may be discounted as the authorsfailed to consider the effect of stress relief and the delay in measurements afterthe centrifuge spinning down from 100g to 1g

over-Stewart (2005) conducted centrifuge test to investigate the remouldedshear strength profile of a footprint The strength measurements were done in-flight using T-bar penetrometer The results were compared with Siciliano et

al (1990), seeFig 2.21 In general, the strength reduced to about 40% to 80%

of the intact strength at the center of the footprint Only a little reduction instrength was observed at a distance of about 1.5 to 2D from the centre of thefootprint The results also revealed that a zone of relatively intense remouldingwas confined within 0.75D from the footprint centre The soil was lessremoulded beyond 0.75D and extended to a radial distance of about 1.5D to2D

The effect of soil reconsolidation was not considered in both studies.The soil underwent reconsolidation after a footprint was formed (Gan et al.,

2007 and Leung et al., 2007) The soil condition beneath a footprint gainedstrength with time as the excess pore pressure generated by the spudcanactivity dissipated A detailed study on the soil properties change with timewill be discussed inChapter 5

2.6 Shear strength measurement devices

Cone penetration tests (CPTs) are commonly used in the offshore soilinvestigation to estimate the soil strength profiles Randolph et al (2005)reported that the field measurement of sucan be improved by using ‘full-flow’penetrometers, particularly the T-bar and ball penetrometers The advantage ofusing full-flow penetrometers is to minimise the correction for overburdenstress.Fig 2.22 shows the alternative full-flow penetrometers with cone

T-bar penetrometer was first introduced by the University of WesternAustralia for shear strength profiling of soil sample in centrifuge model tests(Stewart & Randolph, 1991 & 1994) In recent years, T-bar has beenimplemented for offshore site investigations (Randolph et al., 1998 and Lunne

et al., 2005) Plastic solution for the flow around a cylinder (Randolph &Houlsby, 1984) and sphere (Randolph et al., 2000) provided the basis forobtaining shear strength directly from the measured penetration resistance.The bearing capacity factors for T-bar, NT-bar predicted based on the lowerbound and upper bound solutions against the material roughness is shown inFig 2.23 It was noted that the NT-bar values fell within a range of 9.4 to 10 forfully smooth material and converged to 11.8 for fully rough material The soil

shear strength can be deduced directly from the measured penetrationresistance, qT-bar and the T-bar factor, NT-barusing the equation below:

The lower bound and upper bound solution for ball penetrometer based

on Tresca and Von Mises soil models are also studied and shown in Fig 2.24.The Nball values vary from 10.98 to 11.8 for perfectly smooth interface and15.1 to 15.5 for perfectly rough interface based on Tresca model Theplasticity solutions suggested that the penetration resistance using the ball wasroughly 25% higher than those acquired by the T-bar However, theexperimental comparison between the T-bar and the ball conducted in theUniversity of Western Australia proved otherwise The resistance measured bythe T-bar and the ball showed no consistent difference for materials rangingfrom kaolin to calcareous silt, with probably a difference of up to 5% in thepenetration bearing pressure, as presented in Fig 2.25 (Watson et al., 1998;Newson et al., 1999 and Watson, 1999) Similar finding was made by Chung

et al (2006) where the T-bar and ball penetrometer demonstrated a verysimilar resistance profile (Fig 2.26) The discrepancy between experimentfindings and theory deserves further research Chung et al (2006) also foundthat T-bars with aspect ratios (length/diameter) from 4 to 10 showed verysimilar resistances (Fig 2.27) Thus, they suggested no significant effect onthe T-bar resistance for the aspect ratio within the range of 4 to 10