Suction caisson extraction resistance in Gulf of Guinea clay

Transcription

1 Frontiers in Offshore Geotechnics III Meyer (Ed.) 2015 Taylor & Francis Group, London, ISBN: Suction caisson extraction resistance in Gulf of Guinea clay C. Gaudin, M.F. Randolph, S. Gourvenec, C.D. O Loughlin & D.J. White Centre for Offshore Foundation Systems, The University of Western Australia, Australia J.-L. Colliat TOTAL E&P, Pau, France ABSTRACT: This paper presents the results and interpretation of a series of centrifuge tests performed to model the extraction of stiffened and unstiffened suction caissons in Angola clay. Parameters investigated during the testing programme included (i) the setup time after installation, which varied between 0 and 360 days, and (ii) the extraction rate, which spanned over three orders of magnitude. The interpretation revealed fundamental aspects of the caisson behaviour notably associated with the installation soil kinematics, and its influence during the subsequent extraction on both the internal and external friction, and their evolution with setup time. The effect of setup time can be reasonably approximated using conventional cavity expansion theory followed by radial consolidation. The effect of the extraction rate can be assessed using a total stress analysis but accounting for both strain rate effects and consolidation occurring while mobilising the peak extraction resistance. Recommended friction factors and bearing capacity factors are provided accounting for the parameters investigated. 1 INTRODUCTION The capacity of suction caissons is governed by different mechanisms associated with soil softening/ hardening at the wall interface and soil kinematics around the tip and stiffeners. These mechanisms are relatively well understood (see notably Andersen et al. 2005), but depend strongly on the caisson geometry and the mechanical characteristics of the soil. Establishing geotechnical design parameters therefore remains challenging and may require modelling exercises to account for specific features of the caisson geometry or soil conditions encountered at the site. In 2006, Total E&P initiated a centrifuge testing programme undertaken by IFSTTAR (formerly LCPC) to assist in the design of suction caissons used in an oil development in the Gulf of Guinea, offshore Angola. The programme was prompted by the specific features of the Gulf of Guinea seabed material, which notably exhibits a high plasticity index (from 70 to 150), a yield stress ratio greater than 1 (associated with post-sedimentation structural and bonding effects), and a high sensitivity (from 3 to 6). A comprehensive review of Gulf of Guinea sediment characteristics is presented in Colliat et al. (2010a). The centrifuge test programme focused on the installation of suction caissons with and without internal stiffeners, and the subsequent sealed extraction capacity, as a function of loading rate and postinstallation consolidation time (Colliat et al. 2010b). Results provided insights into the suction caisson performance, but also raised issues, notably related to the very low inferred interface friction (when compared to the soil remoulded strength) and a limited dependence of the caisson capacity on the setup time. To investigate further and potentially settle these issues, a second set of centrifuge tests, featuring a nearly identical testing programme was performed at UWA. That programme is the subject of this paper. Gaudin et al. (2014) presented the first part of the testing outcomes, focusing on the installation results and on the effect of the internal ring stiffeners on the penetration resistance. It was demonstrated that very limited soil flow occurred around the ring stiffeners as water became trapped between the stiffeners. This results in essentially zero inner skin friction, so the penetration resistance could be accurately predicted assuming a nominal internal friction, τ in = 1 kpa above the first stiffener, an external friction factor, α ex = τ ex /s u of 0.3, a caisson skirt tip bearing factor of 7.5 and a bearing factor on the first stiffener (only) increasing from 5 to 15 with penetration depth to account for the additional work to extrude the soil plug as the caisson penetrates. This result was on contrast with IFSTTAR results where a gap between the stiffeners and the wall allowed water to freely move along the wall so inner skin friction could be generated. The paper presents the second part of the testing outcomes, focusing on the extraction capacity as a function of ring stiffeners, setup time and pullout rate. It highlights the importance of strain rate effects and strength recovery. 251

2 2 CENTRIFUGE MODELLING The centrifuge soil sample characteristics, suction caisson models and testing procedure have been presented extensively in Gaudin et al. (2014) and are covered here in brief. Table 1. Suction caisson prototype dimensions. 2.1 Soil sample characteristics Gulf of Guinea clay was mixed with fresh water to reconstitute normally consolidated samples with similar water content profile to that in-situ. Shear strength profiles, inferred from in-flight T-bar penetrometer tests performed at normalised velocity v/d of 0.25, have a strength at mudline of s um = 1.3 kpa and strength gradients k ranging from 1 kpa/m to 1.4 kpa/m during the course of tests conducted in each sample. The latter value aligns with lower bound shear strength profiles measured in-situ. Caisson capacities have been scaled to reflect an average strength gradient of 1.2 kpa/m. Piezocone dissipation tests indicated a coefficient of consolidation c v of about 3 m 2 /year at σ v = 55 kpa. However, significant other differences were observed between the centrifuge samples and in-situ soil. These relate notably to the shear strength ratio, s u /σ vo, and the soil sensitivity, S t, which were both significantly lower for the centrifuge samples ( 0.28 against 0.45 and 1.7 against 3 6, respectively). Laboratory tests showed that the differences were not due to different salinity of the pore fluid (the centrifuge sample being reconstituted with added fresh water), but from structural and bonding effects, present in-situ, which could not be recreated by the sample reconstitution process. An important feature of Gulf of Guinea clay is its relatively high strength rate dependency. The increase in shear strength per log cycle increase in shearing rate range from 16 to 19%, based on CU triaxial tests performed over six orders of magnitude of strain rates (Colliat et al. 2010a). 2.2 Suction caisson models Two suction caisson models were manufactured, with overall geometry summarised in Table 1 in prototype dimensions (at a scaling ratio of 1:200). The models represent an 8 m diameter and 24 m long prototype suction caisson, used to anchor riser towers in Site C of block B17, offshore Angola (Colliat et al., 2007). The unstiffened caisson model, labelled USC, featured a flush skirt, 0.08 m thick, slightly thicker than the m used in-situ (due to machining constraints). No surface treatment was applied on the skirt, resulting in a relatively smooth interface. The stiffened caisson model, labelled SC, featured 5 horizontal ring stiffeners 0.58 m wide (measured in the horizontal plane, as compared to m in-situ) and 0.2 m thick (measured in the vertical plane). The first ring stiffener is located 4 m from the base of the caisson, with the subsequent stiffeners spaced 4 m apart. The model caissons were machined from a full block of duraluminium. There were no gaps between the stiffeners and the caisson inner surface where water could travel freely up the inner skirt. This was demonstrated in Gaudin et al. (2014) to influence significantly the soil kinematics around the stiffeners, resulting in very little soil flow round the stiffeners. 2.3 Testing programme and procedure The testing programme included 11 tests featuring vented and sealed extraction, at normalised pullout rates of vd/c v of 0.42 and 168.2, and after consolidation prototype periods of 0, 30 and 366 days. Four of these cases were performed using the unstiffened caisson (USC) and eight used the stiffened caisson (SC). Note that vd/c v = corresponds to fully undrained conditions and that the difference in extraction rates translates into a difference in undrained strength of about 44% (based on 17% increase in strength per log cycle increase in rate). For tests at vd/c v = 168.2, the interpretation uses the strength inferred from T-bar tests without rate correction. The test names reflect the sample the test was performed in (S1 or S2), the type of caisson (stiffened SC or unstiffened USC), the test number, the extraction mode (s for sealed or v for vented) and the extraction rate (q for vd/c v = or s for 0.42). Following penetration (either jacked or by selfweight followed by suction), the suction caisson was maintained in place ( z = 0) for the required consolidation period, before being pulled out under displacement control at the targeted velocity v. The purpose of the vented extraction tests was to provide direct insight into changes in external and internal friction (since no plug uplift takes place) with setup time and pullout rate. The vented tests enable the post-consolidation changes in friction factor to be isolated, at least for the combined internal and external friction (separate assessment can be made assuming an internal to external friction ratio). 252

3 Table 2. Summary of extraction results. Peak Consol. Extract. extract. period rate resist. Proto. T v Extract. vd/c v Tests (d) ( ) mode ( ) F e (MN) S1-USC4vq V S2-USC4vs 0 0 V S1-SC5vq V S2-SC5vs V S2-SC7vs V S2-USC3sq S S1-SC1sq S S2-SC1sq S S1-SC2ss S S2-SC2ss S S2-SC6ss S RESULTS 3.1 Summary of results Table 2 presents the peak extraction resistance for each test. The time factor T v characterising the level of consolidation for each test is introduced and defined as c v t/d 2 eq, where t is the consolidation time and D eq the caisson equivalent diameter (Randolph 2003). Although there are some anomalies, the following trends are noted: Effect of sealing. Sealed extraction of the stiffened caissons evidently exhibits higher capacity than vented extraction due to the base resistance of the soil plug. Effect of stiffeners on internal friction. For vented extraction, unstiffened caissons have a higher extraction resistance than stiffened caissons. This indicates that the stiffened caissons have lower inner friction, even after recovery post-installation. Effect of stiffeners on external friction and plug uplift resistance. By contrast, for sealed extraction, stiffened caissons exhibit higher capacity than unstiffened caissons, indicating higher external friction (taking plug uplift resistance as unaffected). Observed effect of extraction rate. The capacity of both vented and sealed caissons is reduced by the 500-fold reduction in extraction rate in the slow tests. The magnitude of this effect depends on the presence of stiffeners and the consolidation period. Capacity is reduced by about for the vented stiffened caissons (comparing S1-SC5vq with S2- SC5vs and S2-SC7vs) and with the low end of this range observed at shorter consolidation times. The effect of rate is only 1.2 for vented unstiffened caissons, although this comparison is made for different consolidation times (comparing (S1-USC4vq and S2-USC4vs). The sealed stiffened caissons exhibit a reduction in capacity with decreasing extraction rate of 1.05 after 30 days of consolidation (see tests S1-SC1sq and S1-SC2ss) and of about 1.65 after about 366 days consolidation (see tests S2-SC1sq and S2-SC2ss). Theoretical effect of extraction rate. Direct allowance for strain rate effects via the typical soil element response (Colliat et al. 2010) suggests a ratio of about 1.44 between the two different extraction rates (at identical consolidation time). Evidently, for both the vented and sealed extraction, other factors influence the caisson capacity. Effect of consolidation time. For vd/c v = 168.2, there is an evident increase of stiffened caisson extraction resistance with consolidation time of about 1.4 from 27.2 days of consolidation (test S1-SC1sq) to 357 (test S2-SC1sq). No effect of consolidation time is observed for vd/c v = Back calculation of design parameters Vented rapid extraction. Design parameters are first calculated for vd/c v = and 32 days of consolidation. According to the consolidation solution of Teh and Houlsby (1991), about 30% of full primary consolidation is expected after 32 days for the conditions considered in this study. The predicted extraction resistance for vented test S1-USC4vq is calculated as the sum of the internal and external friction, and reverse end bearing at the skirt tip for which a bearing factor of N c,tip = 7.5 is adopted. The experimental result is best fitted by assuming external and internal friction factors of α ex = 0.87 and α in = 0.53, respectively, i.e. 2.3 times the value deduced from penetration (see Gaudin et al. 2014). The ratio α ex /α in was taken as 1.63 (although this figure may not be strictly appropriate for the present tests), following Jeanjean et al. (2006) who observed lower consolidation inside the caisson from centrifuge model tests featuring double-wall caisson. The exercise is repeated for the stiffened caisson, assuming a nominal internal friction above the first stiffener of 1 kpa (resulting from no soil flow round the stiffeners during penetration). This yields an external friction factor of Predicted and measured extraction resistances are compared in Figure 1. Vented slow extraction. For tests at vd/c v = 0.42, there is uncertainty about the exact drainage conditions, and hence what intrinsic differences in frictional capacity might be expected between rapid (undrained) and slow (quasi-drained) extraction. The difference in capacity observed for the unstiffened caissons is relatively small. There is also evidence (essentially for piles), that drained and undrained shaft capacities show little differences, with similar changes in normal effective stresses. The soil caisson system is not a constant stress system, but more closely resembles a constant volume system. Accordingly, drained shearing may reduce the normal effective stress due to volumetric collapse of the sheared material, just as undrained shearing reduces the normal effective stress due to generation of excess pore pressure. In practice, the drained capacity might be estimated as a function of the effective stress σ h and the drained friction angle 253

4 Figure 1. Vented unstiffened and stiffened caissons predicted and measured extraction resistances for vd/c v = δ r. However, following the above consideration and acknowledging the difficulty in predicting both the change in effective stresses and the interface friction angle, it is more convenient to use the same friction factor approach to predict the slow capacity, noting that αs u = σ h tanδ r. Another advantage of this approach is the ability to account directly for strain rate effects. Figure 2 presents the predicted extraction resistance for both the stiffened and unstiffened caisson. The extraction resistance has been factored down by 44% to account for a reduction in strength of 44% due to the difference in strain rates. In contrast to test S1-USC4vq, identical internal and external friction factors are assumed for test S2-USC4vs since extraction was performed immediately after penetration. For the unstiffened caisson, a friction factor of 0.86 provides the best fit. This value is higher than the average value of 0.7 obtained for the faster test S1-USC4vq (Fig. 1, combining internal and external friction factors), which experienced a higher (initial) consolidation time (32 days against 0). There is however an evident time effect at vd/c v = 0.42, due to the long period required for the peak extraction resistance to be mobilised. For test S2-UCS4vs, although extraction began after 32 days of consolidation, the peak extraction resistance was only mobilised after 509 days, potentially resulting in about 78% of consolidation (against the estimated 30% for the fast 32-day test, S1-USC4vq). However, if consolidation is assumed to take place, the assumption of identical external and internal friction factors may no longer be valid.assuming the same distribution, as for test S1-USC4vq, would result in external and internal friction factors of α ex = 1.08 and α in = 0.65 (after 78% consolidation) compared to the values of α ex = 0.87 and α in = 0.53 (after 30% of consolidation).this is broadly consistent. Figure 2. Vented unstiffened and stiffened caissons, predicted and measured extraction resistances for vd/c v = 0.42, accounting for strain rate effects. Sealed extraction. The capacity of the sealed suction caisson is subsequently predicted assuming passive suction is mobilised (as demonstrated by the pore pressures of about kpa at caisson base, not presented in the paper). The net undrained uplift capacity of the caisson (i.e. excluding the caisson submerged weight) is expressed as the sum of a reverse end bearing component at the caisson base and the external friction along the caisson skirt, noting that the overburden component at the tip level is cancelled by the weight of the soil plug. The data were fitted using a value of 5 for the plug bearing capacity N c,plug. This is lower than the range of suggested (Andersen et al. 2005), but reflects the limited displacements (averaging around 0.1D) required to reach peak capacity, essentially accounting for the fact that full mobilisation of the reverse end bearing component does not coincide with the peak of the overall extraction resistance. For 32 days of consolidation, back calculated values of the external friction factor α ex are 0.4 and 0.6 for the unstiffened and stiffened caissons, respectively. The value of 0.6 for the stiffened caisson is consistent with the value of 0.54 obtained for unstiffened vented extraction. For the unstiffened caisson a lower value of α ex of 0.4 is required, assuming a plug bearing factor identical to the stiffened caisson (an assumption justified by pore pressure measurements at the base, not presented in the paper). This value is consistent with the values inferred for the stiffened caisson, assuming that slower consolidation takes place for the latter (as discussed later in the paper). However, it contrasts with the value of α ex = 0.87 obtained for the vented extraction. A possible explanation relates to the test 254

5 Table 3. Design parameters after 30 days of consolidation. Table 4. Measured and back-calculated friction factors. Caisson Extract. N c,tip N c,plug α ex α in Unstiff. Vented 7.5 N/A Sealed N/A N/A Stiff. Vented 7.5 N/A 0.54 (1 kpa) Sealed N/A N/A procedure for the vented caisson, whereas penetration was undertaken beyond touchdown, pushing the plug within the soil and generating higher excess pore pressures around the caisson that would take longer to dissipate. For extraction at vd/c v = 0.42, α ex = 1.27, 1.01 and 1.28 is calculated for tests S1-SC2ss, S2-SC2ss and S2-SC6ss, respectively (accounting for strain rate effects). These values are relatively high, but consistent with the value of 1.17 obtained after 366 days for stiffened caisson tests at vd/c v = The calculated friction factors are summarised in Table 3 for tests at vd/c v = and 30 days of consolidation. 4 DISCUSSION 4.1 Effect of internal ring stiffeners Internal ring stiffeners have an effect on the suction caisson capacity. Comparison of tests S2-USC3sq and S2-SC1sq indicates that the stiffeners increased the undrained vertical sealed capacity by a ratio of Analysis of both the penetration (see Gaudin et al. 2014) and extraction stages has demonstrated that the internal friction is reduced to nearly zero. The water entrapped between the ring stiffeners prevents full flow around the stiffeners, limiting the contribution of the internal friction to the total penetration resistance to less than 5%. There is no regain of internal friction with consolidation (demonstrated from the vented extraction results). However, this does not affect the undrained uplift capacity of the sealed stiffened caisson, which was measured to be 64% higher than for the unstiffened caisson, after 366 days of consolidation, indicating a higher external friction for stiffened caissons. This is most likely related to the soil flow during penetration. Due to the presence of the stiffeners, more soil is expected to flow outside the caisson, generating (after consolidation) higher effective stresses along the caisson external walls and, consequently, higher friction (although a longer time scale for consolidation). For the vented extraction (i.e. without contribution of the plug uplift), the higher soil displacement and external friction is overshadowed by the negligible internal friction, resulting in the unstiffened caisson exhibiting a higher vented undrained uplift capacity by a factor of Effect of consolidation time and pullout rate Interpretation of piezocone tests performed in the centrifuge samples yielded a t 50 time of about 95 days for Cons. time (d) T v = c v t/d 2 eq U = u/ u i (%) 0% 30% 73% 78% 84% 89% α ex CET Stiff Sea Stiff Ven. Unstif. 0.4 Sea. Unstif Ven. the caissons, and t 90 of more than 1500 days. Accordingly, significant setup is expected between 30 and 366 days, the two consolidation periods used during testing. Indeed, the capacity of sealed stiffened caissons extracted at vd/c v = increases by a ratio of 1.41 after 366 days of consolidation. This translates into an increase in external friction factor from 0.60 to The latter value is slightly higher than the value predicted using conventional cylindrical cavity expansion theory (CET) for open piles (Randolph 2003). The prediction is presented in Table 4 for the full consolidation range, assuming a residual interface friction angle δ r of 21 (obtained from direct shear tests performed at 0.05 mm/min), a rigidity index G/s u of 100 relevant for Gulf of Guinea clay (Low et al. 2010), and a remoulded strength immediately after penetration of 0.3 (consistent with the friction factor of 0.3 back calculated during penetration). The cavity expansion analysis captures reasonably well the increase in external friction factor with consolidation time for the stiffened caisson, and can be used confidently for design, provided that the soil sensitivity used is corrected to reflect the estimated friction factor at zero consolidation. As discussed, the difference in extraction rate results in a difference in shear strength of about 44%, which does not totally account for the measured differences in capacity. During slow extraction, the peak resistance is reached after a significant time, comparable with consolidation times. As a result, excess pore pressure present at the beginning of the process will dissipate further, adding consolidation to what was already achieved during the post-installation consolidation phase. The superposition of these two aspects (consolidation and strain rate) explains the increase in difference in uplift capacity between tests performed at vd/c v = and vd/c v = Slow extraction tests experienced a much greater degree of consolidation by the time failure occurred, resulting in a larger setup and higher friction, which balances the reduction in strength due to strain rate. This is illustrated symbolically in Fig. 3, where a constant difference in capacity is observed provided that the consolidation time is corrected to reflect the time to reach peak capacity. The point is further demonstrated in Table 4, where friction 255

6 Figure 3. rates. Combined effects of consolidation time and strain ACKNOWLEDGMENTS This work forms part of the activities of the Centre for Offshore Foundation Systems (COFS), currently supported as a node of the Australian Research Council Centre of Excellence for Geotechnical Science and Engineering. The centrifuge programme and subsequent interpretation were funded by Total E&P, which support is gratefully acknowledged. Centrifuge tests were performed with the assistance of Manuel Palacios, Dave Jones and John Breen. REFERENCES Figure 4. Evolution of external and internal friction factors with time factor. factors calculated for test at vd/c v = 0.42, accounting for rate effects and corrected for additional consolidation, are compared with cavity expansion solutions. The agreement is convincing. 5 CONCLUDING REMARKS Figure 4 presents the set of recommended friction factors gathered in Table 4 as a function of time factor, T v. These values are believed appropriate for design of suction caissons in Block 17 in the Gulf of Guinea, offshore Angola, acknowledging that the absolute values of α back-calculated from the model tests need to be reduced due to strain rate effects, depending on the actual penetration or extraction rates of the caisson. The model caisson tests induce very high shear strain rates along the shaft, which can be estimated as 20 to 40 times the ratio v/d (Einav and Randolph 2006). This gives an average strain rate of about 30%/s for tests at vd/c v = 168.2, which is 4 orders of magnitude higher than typical laboratory strain rates. This would require α to be factored by 0.6 for design, reducing the value at 900 days from an average 1.14 to 0.69, a value closer to the range of reported by Colliat & Colliard (2010) and suggested by Andersen & Jostad (2002) for full consolidation. Andersen K.H. & Jostad H.P Shear strength along outside wall of suction anchors in clay after installation. Proc. of the 12th International Offshore and Polar Engineering Conf., Kitakyushu. Japan, Andersen K.H., Murff J.D., Randolph M.F., Clukey E.C., Erbrich C.T., Jostad H.P., Hansen B., Aubeny C., Sharma P. & Supachawarote C Suction anchors for deepwater applications. 1st Int. Symp. Frontiers in Offshore Geotechnics. Gourvenec & Cassidy (Eds) Colliat J.-L., Colliard D Set-up of suction piles in deepwater Gulf of Guinea clays. Proc. of the 2nd International Symposium on Frontiers in Offshore Geotechnics. Gourvenec & White (Eds) Colliat J.-L., Dendani H. & Schroeder, K Installation of suction piles at deepwater sites in Angola. Proc. 6th Int. Offshore Site Investigation and Geotechnics Conf.: Confronting New Challenges and Sharing Knowledge, London, England Colliat J.-L., Dendani H., PuechA. & Nauroy J.-F. 2010a. Gulf of Guinea deepwater sediments: Geotechnical properties, design issues and installation experiences. 2nd Int. Symp. Frontiers in offshore Geotechnics, Gourvenec & White (Eds) Colliat J.-L., Dendani H., Jostad H.P., Andersen K.H., Thorel L., Garnier, J. & Rault, G. 2010b. Centrifuge testing of suction piles in deepwater Nigeria clay Effect of stiffeners and setup time. 2nd Int. Symp. Frontiers in offshore Geotechnics, Gourvenec & White (Eds) Einav I. & Randolph M.F Effect of strain rate on mobilised strength and thickness of curved shear bands. Géotechnique. 56(7): Gaudin C., O Loughlin C.D., Hossain M.S., Randolph M.F. & Colliat J.-L Installation of suction caissons in Gulf of Guinea clay. Proc. of the 8th International Conf. on Physical Modelling in Geotechnics, Perth, Australia. 1: Jeanjean P., Znidarcic D., Phillips R., Ko H. Y., Pfister S. & Schroeder K Centrifuge testing on suction anchors: double-wall, stiff clays, and layered soil profile. Offshore Technology Conf., Houston. OTC Low H.E., Lunne T., Andersen K.H., Sjursen M.A., Li X. & Randolph M.F Estimation of intact and remoulded undrained shear strength from penetration tests in soft clays. Géotechnique. 60(11): Randolph M.F Science and empiricism in pile foundation design. Géotechnique. 53(10): Teh C.I. & Houlsby G. T An analytical study of the cone penetration test in clay. Géotechnique. 41(1):