You are on page 1of 15

OTC 23184-PP Design of Gravel Banks - A Way to Avoid Jack-Up Spudcan Punch-Through Type of Failure

Lindita Kellezi, GEO-Danish Geotechnical Institute; Henrik Stadsgaard, Maersk Drilling

Copyright 2012, Offshore Technology Conference This paper was prepared for presentation at the Offshore Technology Conference held in Houston, Texas, USA, 30 April3 May 2012. 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 have not been reviewed by the Offshore Technology Conference and are subject to correction by the author(s). The material does not necessarily reflect any position of the Offshore Technology Conference, its officers, or members. Electronic reproduction, distribution, or storage of any part of this paper 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 OTC copyright.

Abstract For the installation of a three-legged jack-up rig at an oil field / drilling location in the North Sea conventional and finite element spudcan penetration analyses and engineering assessments are carried out. As the location, based on the preliminary assessments, indicated critical soil conditions with regard to rig installation, an improvement / increase in the seabed bearing capacity was aimed by designing / constructing gravel banks on the seabed, one at each spudcan location, respectively, as a way to avoid jack-up foundation punch through type of failure. The soil data at the location included three boreholes with sampling and laboratory testing combined with piezocone penetration tests (PCPTs) one at each spudcan centre, respectively. Furthermore, a site survey was available covering the area of interest. Based on all the available soil data design lower / upper bound soil profiles applicable to penetration analyses at each spudcan location for virgin seabed conditions were assessed first. Those profiles and the results from the different, conventional and finite element analyses carried out together with a validation study are presented. Based on them the decision for modifying the seabed by constructing gravel banks was taken. The gravel banks were designed based on the Plaxis finite element modelling applying lower bound soil parameters calculating final heights of 4.0 m, 2.5 m, 2.0 m for Port, Starboard and Bow legs, respectively. The top diameter and the slope of the banks are chosen to be the same, 50 m and 1:2.5, respectively. Except the small strain (plastic) and updated mesh (UM) analyses an attempt was made to adjust the conventional solution for multilayered soil conditions in comparison to finite element results. For the new conditions, the modified seabed by construction of the gravel banks, sufficient foundation bearing capacities and no risks for punch-through / rapid penetrations were calculated. The spudcan-gravel bank-soil interaction systems able to support the maximum preloads for each leg determined based on the environmental load analyses. The gravel banks were constructed and short after that the jack-up rig was successfully installed measuring shallow spudcan penetrations as predicted. Introduction Prediction of the jack-up spudcan penetrations is an important issue in the process of rig installation. Unexpected sudden and rapid penetrations can be of major risk for the stability and equilibrium of the jack-up structures. Classical conventional solutions (Hansen 1970), which are applications of bearing capacity equations for homogeneous soil conditions and modified procedures for layered soil profiles (Jacobsen et al., 1977), (Hanna & Meyerhof 1980) are normally used for the spudcan penetration prediction. However, for layered soil condition the modified conventional procedures are not always sufficiently accurate and realistic comparable to field observations. As analytical procedures have several limitations, alternative analyses based on numerical modeling are investigated and presented through the years. Applications of those methods mainly based on the finite element analyses in comparison to analytical solutions for the conventional, non-skirted spudcans are among others elaborated in (Kellezi & Strmann 2003), (Kellezi et al., 2005), (Kellezi & Kudsk 2009). A case history related to a jack-up installation in the North Sea is considered in the following. As part of the assessment, due to multilayered / critical soil conditions conventional and finite element analyses of the spudcan penetrations for virgin seabed

OTC 23184-PP

are carried out. As the results of the analyses indicated risk for punch-through / rapid penetration for the calculated maximum preloads at each leg, an improvement / increase in the seabed bearing capacity was aimed by designing gravel banks, one at each spudcan location, respectively. The concept of the application of the gravel banks as a way to avoid punch-through risk by locally increase the thickness of the existing seabed sand layer is previously investigated and applied in other locations for a jack-up rig with skirted spudcans (Kellezi et al., 2007), (Kellezi et al., 2008). The necessity of the gravel banks at the current location for the installation of the jack-up with conventional spudcans, the way they are designed and their benefit in fulfilling the required safeties are elaborated in the following sections. Spudcan Dimensions and Loads The spudcans of the current rig have the equivalent radius of 8.9 m and the full contact area of 249 m2. Distance from spudcan base (full contact) to spudcan tip is 1.6 m. The tip has a minimum radius of 0.75 m and a maximum radius of 1.55 m. The central distance between the spudcans is 62 m. The spudcan geometry is given in Figure 1. The initial load is about 6824 tons / leg and the maximum preload about 15700 tons / leg, giving a maximum spudcan pressure equal to 619 kN / m2.

Figure 1: Spudcan geometry

Jack-up Location The position / heading of the jack-up is north-west and such that the spudcan centres nearly coincide with the location of the available boreholes / PCPTs carried out during a site investigation prior to the rig installation. The water depth at the site is about 71.5 m Lowest Astronomical Tide (LAT). The seabed at the jack-up location is essentially flat deepening gently to the southeast. The mapped seabed is dominantly composed on fine to very fine sand. Seabed features do not imply any concerns regarding the rig installation. Seismic Survey and Shallow Soils A site survey has been carried out at the location. The survey with pinger and sparker covered a wider area encompassing several jack-up locations. Shallow lithology at the location comprises a thin veneer of Holocene sand underlined by Base Forth formation (medium dense to dense sand). Below this lie deposits of Coal Pit Formation comprising stiff to very stiff clays, silty sands and interlaminated clays and silty sands. Geotechnical Investigation The geotechnical investigation at the location included three boreholes (F1, F2, F3) with alternating sampling and PCPTs at respective spudcan positions. The boreholes show thin veneer of loose to medium dense sand propagating from the seabed to (0.3 0.5) m in depth, underlined by medium dense to dense sands to depths of (4.1 6.2) m. Below the sand, firm to stiff clay deposits are found to the end of the boreholes, with an exception at borehole / PCPT F3, where a layer of medium dense sand is found at (11.5 19.2) m below seabed. Thin laminate / pockets of clay / silt / sand have been found at all the three boreholes / PCPTs. The soil stratification in the boreholes reflects the shallow soils prediction given by the seismic survey,

OTC 23184-PP

although, the Base Forth formation seems to be somewhat deeper. The PCPTs show cone resistance qc of up to 20 MPa in surficial sands and to about 30 MPa in the deeper sand deposit encountered at borehole PCPT F3. In clay layers, the cone resistance qc varies, and excluding the peaks caused by lamination or other inclusions, it is generally in the interval of (0 - 5) MPa. Strength tests on clay are carried out on the extracted samples. The undrained shear strength tests are carried out using pocket penetrometer, torvane, and unconsolidated undrained (UU) triaxial tests. The laboratory tests show good match with the PCPT results for the adopted cone factor Nkt = (15 20). Design Lower / Upper Bound Soil Profiles, Virgin Seabed Based on all the soil data, design lower / upper bound soil profiles for virgin seabed conditions are assessed applicable to penetration analyses at each spudcan location. Some of the interpreted lower bound undrained shear strength profiles are given in Figures (2-4) where a lower bound (blue line) and an alternative lower bound (red line) are shown.

Alternative Lower Bound Lower Bound

Figure 2: Undrained shear strength and friction angle at the Borehole / PCPT (Port Leg)

The upper / lower bound undrained shear strength cu, for the clay layers has been determined from the results of the laboratory tests (mainly UU triaxial tests) and net cone resistance, qnet, using correlation factor Nkt = (1520). This would fit with an average equivalent isotropic shear strength, being the average of the active, direct shear and passive shear strength. The way

OTC 23184-PP

how the designed undrained shear strength profile follows the variation of qnet / Nkt depends also on the experience of the design engineer. As the current analyses are based on characteristic soil parameters (no material coefficients applied, just as for a common spudcan penetration prediction following (SNAME 2002), a conservatively selected lower bound profile (blue lines in Figures (2-4) are chosen for the current assessments. The angles of internal friction for the seabed sand layers are derived from the results of the PCPTs as peak values and critical state values. Critical state friction angles equal to 30 are applied as lower bound strength for the sand layers in the current analyses as shown in Figures (2-4).

Alternative Lower Bound Lower Bound

Figure 3: Undrained shear strength at the Borehole / PCPT (Starboard Leg)

Relative density of the sand, Dr, is interpreted first from the PCPT data. The peak angle of internal friction is then based on the relative density. The values of the angle of internal friction utilized for the following analyses are re-assessed by reducing the peak strengths or deriving the critical state angles as a difference of the peak and dilatancy values. This is because the angle of frictions depends on the volumetric changes related to the stress level. On high stress levels, dense sands dilate less than on low stress levels. However for loose sands the dilation is smaller, reducing to zero for high stress levels. The friction angles measured in the laboratory testing are the peak angles obtained while the sand is free to contract / dilate. However, to account for the fact that in order to mobilize failure into the clay, large strains within the sand may result leading

OTC 23184-PP

to a drop in the friction angle. Therefore, the highest values that are applied in the current bearing capacity computations are however the critical state friction angles, i.e. the angles that correspond to the failure associated with no volumetric changes. The difference between the peak angles, and the critical state angles, represent the angles of dilatancy or a fraction (0.8) of the angles of dilatancy - depending on the type of the test, i.e. for triaxial or plane strain respectively. This results with roughly (2 5) difference between the two test types that are commonly taken for assessments of lower and upper bound failure strengths.

Alternative Lower Bound Lower Bound

Figure 4: Undrained shear strength at the Borehole / PCPT (Bow Leg)

Design Lower / Upper Bound Soil Profiles, Seabed Modified by Gravel Banks Based on the bearing capacity analyses for virgin seabed conditions, which are discussed in the following, gravel banks were found necessary to be constructed at the spudcan locations in order to increase the lower bound capacity to punch through / rapid penetrations. The gravel banks are designed based on the lower bound soil parameters and applying conventional solutions for infinite extended banks and finite element analyses for a prechosen geometry of the banks. After several analyses gravel banks with heights 4.0 m, 2.5 m, 2.0 m, were designed for Port, Starboard and Bow legs, respectively. The top diameter and the slope were chosen to be the same, 50 m and 1:2.5, respectively.

OTC 23184-PP

Considering the above heights and assuming infinite extension of the banks in the horizontal plane, the design lower / upper bound soil profiles at each borehole / PCPT or spudcan location, after the construction of the gravel banks are modified compared to the virgin seabed as shown in Tables (1-3). The upper bound parameters are only used to get an impression of the expected maximum capacities at the leg locations. For the gravel / rockfill material an angle of internal friction of 40 was chosen based on a conservative interpretation of the data for the stress dependency of the angle of friction for such material. The actual gravel material was expected to be loosely compacted, but relatively uniformly graded material. However, it consists of high strength particles meaning that the particle interlocking is not easy destroyed by crushing. Hence, the chosen strength was considered realistic. The deformation module for the gravel was chosen E=35000 kPa. Soil Type Depth of Layer (m) Unit Weight ' (kN/m3) Undrained Shear Strength cu (kPa) Angle of Internal Friction ()

GRAVEL 0.0 4.0 9.5 40 / 45 SAND, medium dense to dense 4.0 8.1 9.5 30 / 35 CLAY, firm to stiff 8.1 17.5 8.5 (80-60) / (105-80) CLAY, firm to stiff 17.5 28.5 8.5 80 / 105 CLAY, firm to stiff 28.5 33.0 8.5 (125-80) / (165-105) CLAY, firm to stiff 33.0 34.0 8.5 (125 / (165) CLAY, firm to stiff 34.0 38.5. 8.5 (80-60) / (105-80) CLAY, stiff to hard 38.5 44.0 8.5 (60-210) / (80-280) Table 1 Design lower / upper bound soil profile, Borehole / PCPT F1, gravel bank height=4.0 m, (Port Leg)

Soil Type

Depth of Layer (m)

Unit Weight ' (kN/m3)

Undrained Shear Strength cu (kPa)

Angle of Internal Friction ()

GRAVEL 0.0 2.5 9.5 40 / 45 SAND, medium dense to dense 2.5 8.7 9.5 30 / 35 CLAY, firm to stiff 8.7 9.0 8.5 110 / 145 CLAY, firm to stiff 9.0 18.0 8.5 (80-45) / (105-60) CLAY, firm to stiff 18.0 31.5 8.5 (70-75) / (90-100) CLAY, firm to stiff 31.5 32.8 8.5 (75-120) / (100-160) Table 2 Design lower / upper bound soil profile, Borehole / PCPT F2, gravel bank height=2.5 m, (Starboard Leg)

Soil Type

Depth of Layer (m)

Unit Weight ' (kN/m3)

Undrained Shear Strength cu (kPa)

Angle of Internal Friction ()

GRAVEL 0.0 2.0 9.5 40 / 45 SAND, medium dense to dense 2.0 6.5 9.5 30 / 35 CLAY, firm to stiff 6.5 13.5 8.5 (80-70) / (105-90) SAND, medium dense to dense 13.5 21.2 9.5 30 / 35 CLAY, firm to stiff 21.2 25.0 8.5 100 / 130 CLAY, firm to stiff 25.0 28.0 8.5 75 / 100 CLAY, firm to stiff 28.0 40.9 8.5 105 / 140 Table 3: Design lower / upper bound soil profile, Borehole / PCPT F3, gravel bank height=2.0 m, (Bow Leg)

Conventional Analyses for Virgin and Modified Seabed Conventional or classical bearing capacity analyses are often based on limit equilibrium methods. Limit equilibrium is a state of equilibrium corresponding to a failure criterion that defines the strength of the soil. Failure is usually defined as a state when the strength is fully mobilized along the entire failure surface. The penetration analyses follow the guidelines given in (SNAME 2002). The calculations are based on design soil parameters with partial coefficients, m = 1.0. In the calculations the loads are considered as purely static. To define footing penetration versus load, assessment of the static bearing capacity of the spudcan at various depths is carried out. The bearing capacity is calculated based on (Hansen 1970) theory and in-house program developed from the experience with spudcan penetration predictions. The spudcan is simplified to a circular footing having a flat bottom. The effect of the actual spudcan shape is taken into account. Squeezing of the clay layer underlying the sand during the footing penetration is considered and implemented. Soil backflow is included in the analyses. The punch through mechanism was analysed based on the load spreading factor (SF) according to (SNAME 2002), (Jacobsen et al., 1977) and authors experience and by the punching mechanism or the method described by (Hanna & Meyerhof 1980) utilizing the coefficients given from Figure 5 derived from the same authors.

OTC 23184-PP

During spudcan penetration through the sand layers a sand plug is assumed to develop and lead the spudcan penetration through clay. This is confirmed by the finite element analyses. The conventional analysis utilized the thickness of the plug estimated with respect to finite element results. Punching failure is investigated by varying load spread factor (SF), which depends on the ratio between the spudcan diameter and the sand thickness, the average strength of the sand layers etc. SF adopted for the analysis has been considered also in relation to the results of subsequent finite element analyses. For the modified seabed conditions the soil profile is constructed by adding a layer of gravel over the seabed with the corresponding height depending on the spudcan location. In the analyses the seabed level is assumed at the top of the banks.

Figure 5 Coefficients of punching shear resistance under vertical load

Finite Element Analyses for Virgin and Modified Seabed Finite element modeling of the spudcan penetration is carried out with (Plaxis 2008) finite element software. Due to the multilayered soil profile finite element modelling of the large spudcan penetrations is complex. When using finite element methods both the deformations and the failure criterion are defined through a material model. In this way deformation corresponding to a certain level of soil strength utilization can be chosen. Hence, small deformations or large deformations theories can apply. Considering the program limitations, the following assumptions / simplifications are made in building the finite element model in Plaxis: Due to symmetrical vertical loading / penetration, axisymmetric modelling of the spudcan-soil interaction system is carried out taking the three-dimensional effect into account. Sand layers are modelled in drained condition using Mohr-Coulomb constitutive model. Clay layers are modelled in undrained conditions using Mohr Coulomb or Tresca soil model, assuming ideal elastic-perfectly plastic soil behaviour. Isotropic shear strength and tension cut-off is applied. The dilatancy angle was taken as =-30. Soil softening and rate effects can be taken into account in advanced soil models, however, not available in the finite element software. The initial geostatic stresses are calculated for virgin soil first based on the K0-procedure. In line with the conventional analyses, finite element analyses utilize effective unit weights for the seabed soil layers. Hence, the water table is placed at the bottom boundary of the model. Mesh has been generated using cubic (15-noded) triangular finite elements, which results with lower bearing capacity than when using quadratic (6-noded) triangular elements. The spudcan is modelled as a weightless rigid body. Interface elements are used between the structure and the soil with a strength reduction factor of 0.7.

OTC 23184-PP

The deformation parameters (E module) for sand layers are calculated based on the PCPT data, and for the clay layers on the undrained shear strength using the correlation E = 200cu. These assessments of the soil stiffnesss are supported from the laboratory testing. However, for the current assessment the strength parameters play the main role. The analysis starts with spudcan in-place with full base contact with the seabed or initial penetration of 1.6 m. Vertical displacement is applied at the full base of the spudcan and the reaction force is computed and recorded. Soil backflow generally is not / cannot be modelled in the Plaxis finite element analyses. Therefore, corrections of the bearing capacity are made when plotting the Plaxis curves at the depth where backflow is expected to occur. In the current analyses, the finite element results are considered only for the depths where the backflow / inflow are not expected to affect the results. For the modelling of the spudcan-gravel bank-soil interaction, the gravel bank is constructed first giving changes in the initial stress conditions and excess pore pressure at the underlying clay layers. The dissipation of the excess pore pressure in time is conservatively not considered in the analyses.

Figure 6: Incremental shear strain (failure pattern) at 3 m spudcan penetration (Port Leg) Left small strain (Plastic) analysis; Right large deformation (UM) analysis

Figure 7: Shear strain at 3 m spudcan penetration (Starboard Leg) Left small strain (Plastic) analysis; Right large deformation (UM) analysis

OTC 23184-PP

Figure 8: Shear strain at 3 m spudcan penetration (Bow Leg) Left small strain (Plastic) analysis; Right large deformation (UM) analysis

At first, mesh sensitivity analyses have been carried out and the optimal mesh with respect to element size (inversely proportional to the computational time) and obtained accuracy has been chosen for the final analyses. Computations, using both small strains (Plastic) and large deformation or updated mesh techniques are performed. Some results for the modified seabed (including the construction of the gravel banks) are given in Figures (6-8). Plaxis Small Strains and Updated Mesh Analyses Some elaborations on the Plaxis small strains plastic analysis and plastic updated mesh analyses are given in this section. Generally in finite element analysis, the influence of the geometry change of the mesh on the equilibrium conditions is neglected. This is usually a good approximation when the deformations are relatively small as is the case for most engineering structures. However, there are circumstances under which it is necessary to take this influence into account. Typical applications where updated mesh analyses may be necessary include the analysis of reinforced soil structures, the analysis of large offshore footing collapse problems and the study of problems where soils are soft and large deformations occur. When large deformation theory is included in the finite element program, some special features are considered like: Additional terms in the structure stiffness matrix are included to model the effects of large structural distortions on the finite element equations. A procedure to model correctly the stress changes that occur when finite material rotations happen is implemented. This particular feature of large displacement theory is usually dealt with by adopting a definition of stress rate that includes rotation rate terms. The finite element mesh is updated as the calculation proceeds. This is done automatically within Plaxis when the updated mesh option is selected. These calculation procedures are in fact based on an approach known as an Updated Lagrangian formulation.

It is not possible to give simple guidelines that may be used to indicate when an updated mesh analysis is necessary and when a conventional analysis is sufficient. If the geometry changes are large (on a real scale!) then significant importance of geometric effects might be suspected. In this case the calculations are repeated using the updated mesh option. Validation Study Conventional and Finite Element Plaxis Analyses Several validations of the Plaxis analyses applicable to offshore foundations are previously performed by the authors. However, due to the differences found in the calculated spudcan bearing capacities based on the conventional (load spreading factor or punching mechanism) and Plaxis analyses for the current multi-layered virgin seabed, additional validations were carried out. In this validation study the accuracy of the conventional analysis is at first investigated assuming a theoretical profile (Test

10

OTC 23184-PP

Site) consisting of a simply two-layer system: sand over clay. The virgin soil profile interpreted from the borehole / PCPT F1 is simplified, (however, keeping most of the features), by excluding the strength reduction with depth for the clay layer underlying the sand as shown in Figure 2 (blue line). The results for the Test Site with sand friction angles of 35 and 40 and clay with constant cu = 80 kPa are given in Figure 9 left. It is observed that conventional analyses utilizing SF 1:4 & 1:3 fit well with Plaxis updated mesh (UM) results at depth of 2.6 m (spudcan tip to base, plus spudcan height over the maximum area), while Plaxis small strain (plastic) analyses give unrealistically low capacities. The same Test Site but with friction angle of 30 for the seabed sand layer is carried out afterwards. The conventional analysis utilizing SF 1:5 overestimate the peak bearing capacity in comparison to Plaxis updated mesh (UM) analyses, and even more Plaxis small strain (plastic) analyses. The results of the method using load spreading factor (SF) are comparable with the Plaxis updated mesh (UM) results only for SF reduced to 1:12 as shown in Figure 9 right. This is supported by the study of (Jacobsen et al., 1977). The above analyses for the Test Site are conducted for varying sand friction angles and keeping the underlying clay strength constant. As a result, it is concluded that when the strength of the top sand layer is not very large in comparison to the strength of the underlying clay layer, the conventional methods utilizing SF or punching mechanism may not comply well with the finite element results giving lower bearing capacity or giving more a risk for rapid penetration than punch through. Conclusions of the validation are as in the following:

Plaxis updated mesh (UM) analyses result with larger bearing capacities than Plaxis small strain (plastic) analysis. Application of the Plaxis updated mesh (UM) analysis on the spudcan penetration is more or less validated. Plaxis small strain (plastic) analysis is evaluated to be conservative and non-complying with the conventional methods. However, for clay soils where softening behaviour is indicated from the laboratory tests (reduction of the undrained shear strength with increasing strain), small strain analyses can be used in absence of a softening soil model in Plaxis. When the relative difference in the soil strengths for sand over clay type of soil profile is small, conventional method applied for strong over soft layer type of failure, needs to be updated by reducing the SF with respect to the results of Plaxis updated mesh (UM) analysis or alternatively re-assessing the coefficients given in Figure 5.

Figure 9: Conventional and FE results of spudcan penetration on the Test Site Left medium to very dense sand; Right loose to medium dense sand

OTC 23184-PP

11

Results of the Conventional and FE Analyses for Virgin and Modified Seabed The results for the conventional and finite element analyses, small strain (plastic) and updated mesh (UM) for virgin and modified seabed by the construction of the gravel banks are summarized in Figures (10-12) for the borehole / PCPT F1 (Port Leg), F2 (Starboard Leg) and F3 (Bow Leg, respectively. The application of SF = 1:12 is implemented in the lower bound conventional analyses for all three soil profiles at the three spudcan locations, virgin seabed. If the lower bound friction angle for the sand layer is chosen larger (corresponding in this case to the upper bound value) conventional method results with SF (1:5-1:4) and Plaxis updated mesh (UM) analyses would compare relatively well.

Figure 10: Conventional and FE results of spudcan penetration on virgin and modified (by gravel banks) seabed (Port Leg)

The application of SF 1:12 seem to work well for the current lower bound virgin seabed conditions. However, as the SF value depends on the geometry of the soil profile, the relative strength of the layers, the strength variation within the layers etc., such conventional analyses need to be confirmed by alternative methods such as Plaxis updated mesh (UM) method. For the modified seabed, assuming the construction of the gravel banks as designed with height of 4.0 m, 2.5 m, 2.0 m for Port, Starboard and Bow legs, respectively and the top diameter and the slope of the banks chosen to be the same, 50 m and

12

OTC 23184-PP

1:2.5, respectively, beside the small strain (plastic) and updated mesh (UM) analyses an attempt is made to derive the conventional solution, which compare well with the finite element results. These solutions are given in Figures (10-12). At the borehole / PCPT F2 where the strength decrease with depth is more pronounced as noticed from the blue line in Figure 3, there is more discrepancy between those methods. The finite element results in the form of incremental shear strain or failure patterns are shown in Figures (6-8) for both types of the finite element analyses. Conclusions and Recommendations The engineering assessment for the installation of a jack-up rig, whos spudcan geometry is given in Figure 1, at a drilling location in the North Sea is carried out. Based on the available soil data design lower / upper bound soil profiles for virgin seabed conditions, applicable to penetration analyses at each spudcan location are assessed first. Those soil profiles are presented in Figures (2-4) and the results of the different, conventional and finite element, small strain (plastic) and updated mesh (UM) analyses are given in Figures (10-12) for each spudcan location.

Figure 11: Conventional and FE results of spudcan penetration on virgin and modified (by gravel banks) seabed (Starboard Leg)

OTC 23184-PP

13

Due to the observed discrepancies between conventional and Plaxis analyses results for the virgin seabed design soil profiles (lower bound), a validation study was performed assuming a Test Site consisting of two-layer system, sand over clay. The analyses for the Test Site are conducted for varying sand friction angles and keeping the underlying clay strength constant.

Figure 12: Conventional and FE results of spudcan penetration on virgin and modified (by gravel banks) seabed (Bow Leg)

The results are given in Figure 9. Based on them it is concluded that when the strength of the top sand layer is not very large in comparison to the strength of the underlying clay layer, the conventional methods utilizing load spreading factor (SF) or punching mechanism applying the coefficients given from the graphs in Figure 5 may not comply well with the finite element results, which give lower bearing capacity or more a risk for rapid penetration than for punch through. Both Plaxis analyses, small strain (plastic) and plastic updated mesh (UM) are investigated. The results with Plaxis UM seem to compare reasonably well with the conventional methods, while Plaxis small strain (plastic) analyses result with smaller capacities. Based on the validation study the correction for SF from 1:5 to 1:12 is implemented in the lower bound conventional analyses for all three virgin soil profiles at the locations as given in Figures (10-12). If upper bound strength is applied to the sand, which based on the calculation of peak and critical state friction angles could be the case, conventional

14

OTC 23184-PP

method results with SF (1.5-1:4) and Plaxis UM compare relatively well, while plastic analyses give reduced capacities as can be observed in Figure 9. The application of SF 1:12 seem to work well for the current lower bound conditions. However, as the SF value depends on the geometry of the soil profile, the relative strength of the layers, the strength variation within the layers etc., such conventional analyses need to be confirmed by alternative methods such as Plaxis UM method or by applying other finite element software.. For the modified seabed by the construction of the gravel banks as designed based on the Plaxis finite element modelling having heights of 4.0 m, 2.5 m, 2.0 m for Port, Starboard and Bow legs, respectively and the top diameter and the slope of the banks the same, 50 m and 1:2.5, respectively the results in the form of the failure pattern / figure are given in Figures (6-8) and in the form of bearing capacity or penetration curves in Figures (10-12). The differences in the bearing capacity given from the small strain (plastic) analyses and plastic updated mesh (UM) analyses can be explained with the differences in the shape and size of the failure patterns given in Figures (6-8). In these figures it is easy to see how the radial extent of the gravel bank influences on the shape of the failure mechanism. The gravel bank has a high capacity and will push the failure line out radially in the underlying clay, rather than going through the gravel. At a certain distance the failure mechanism will find the lowest bearing capacity by going through the bank, which means a larger radius of the bank, has no advantages. The necessary extensions of the gravel banks are determined by comparing the results with infinite extended banks. Except the small strain (plastic) and updated mesh (UM) analyses an attempt was made to adjust also in this case conventional solution for multilayered soil conditions in comparison to finite element results. At the borehole / PCPT F2 where the shear strength decrease with depth is stronger as indicated in Figure 3, there is more discrepancy between those methods. For the modified seabed by the gravel banks as designed no risk for punch through / rapid penetration was expected at the calculated preload levels. From the results given in Figures (10-12) sufficient bearing capacities at each leg was interpreted in comparizon to the preloads to be applied. Independent spudcan-soil interaction and after that spudcan-gravel bank-soil interactions analyses are carried out for the current project. This means that structure-foundation interaction has been neglected being on the safe side. Such interaction was taken into account from (Kellezi et al., 2007) and (Kellezi et al., 2008) for the world largest platform installed at another location
offshore Norway.

Construction of the Gravel Banks, Jack-Up Rig Installing and Field Observations The gravel banks were built-up / constructed as shown in the plan view drawing given in Figure 13.

Figure 13 Plan drawing of the gravel bank cones as constructed at the location

OTC 23184-PP

15

Figure 14 Bathymetry of the seabed after the construction of the gravel banks

The measured spudcan penetrations during the rig installation corresponded approximately to the depth of the spudcan full base contact (1.6 m) with the top of the gravel bank. Those values plotted in Figures (10-12) fit quite well with the lower bound conventional penetration curves given also in the same figures. Regarding Plaxis analyses one of the programs limitations is soil-spudcan interface modeling. Therefore, the finite element analyses start with the spudcan already in full contact with gravel bank meaning 1.6 m initial spudcan tip penetration. In addition to that, some penetration is required to mobilize the gravel / sand strength while forcing the spudcan to penetrate down. Hence, Plaxis analyses (plastic) or (UM) are used only to estimate the peak lower bound bearing capacity at rapid penetration / punch through. To simulate the gradual spudcan penetration starting with the tip contact at the top of the gravel bank finite element modeling with Abaqus as in (Kellezi & Stromann 2003) and (Kellezi & Kudsk 2009) can be chosen, however much more time and cost demanding. In the conventional analyses the gradual penetration of the spudcan from the tip contact with the gravel bank, to the full base contact are calculated. Hence, these analyses are considered more accurate with this respect. For this reason, they are included in Figures (10-12) together with the Plaxis curves. Based on the above, when considering the general uncertainties in the water depth measurements, and as a result, tolerances in the calculated / recorded penetrations, the integrated conventional / finite element analyses presented in this paper seem to well predict the rig installation and confirm that the gravel banks are a way to avoid spudcan punch-through type of failure. References
Hansen J.B. (1970), A Revised and Extended Formula for Bearing Capacity. Bulletin No. 28. The Danish Geotechnical Institute. Jacobsen M. Christensen V. K. & Sorensen S.C. (1977), Gennemlokning af tynde sandleg Penetration of thin sand layers. Vag-och vattenbyggaren 8-9 1977. Hanna A. M. & Meyerhof G. G. (1980), Design charts for ultimate bearing capacity of foundations on sand overlying clay. Canadian Getechnical Journal, Vol. 17, page 300-3003. Kellezi L. & Stromann H. (2003), FEM Analysis of Jack-up Spudcan Penetration for Multi-Layered Critical Soil Conditions. Proceeding of BGA International Conference on Foundations, ICOF2003, Dundee, Scotland, page 410-420. Kellezi K., Kudsk G. & Hansen B. P. (2005), FE Modelling of Spudcan Pipeline Interaction. Proceedings, Frontiers in Offshore Geotechnics, ISFOG 2005, September, Perth, Australia, page 551 557. Kellezi L. Kudsk G. & Hofstede H., (2007), Seabed Instability and 3D FE Jack-up Soil-Structure Interaction Analysis 14th European Conf. on Soil Mech. & Geotech. Eng. ECSMGE 2007, September, Madrid, Spain, Proc. Volume 5 page 247 - 252. Kellezi L. Kudsk G. & Hofstede H., (2008), Jack-up Rig Foundation Design Applying 3D FE Structure-Soil Interaction Modelling 2nd BGA International Conf. on Foundations ICOF 2008, June, Dundee, Scotland, Proc. Volume 1 page 937-948. SNAME, Technical & Research Bulletin 5-5A. (2008), Guidelines for Site Specific Assessment of Mobile Jack-Up Units. Plaxis 2D Version 9.02 (2008), Finite Element Code for Soil and Rock Analysis. Delft University of Technology & Plaxis bv. The Netherlands. Kellezi L. & Kudsk G. (2009), Jack-up Foundation, FE Modelling of Punch Through for Sand over Clay 12 International Conference on Jack-up Platform. Sept. London UK, Proceeding page 1-12.

You might also like