Full-scale Test of Uplift Resistance of Trenched Pipes

International Journal of Offshore and Polar Engineering (ISSN 1053-5381) Copyright by The International Society of Offshore and Polar Engineers Vol. 23, No. 4, December 2013, pp. 298 306 http://www.isope.org/publications Full-scale Test of Uplift Resistance of Trenched Pipes Gudmund Eiksund 1 Department of Civil and Transport Engineering, NTNU, Trondheim, Norway Hans Langø GeoPartner Marin AS, Trondheim, Norway Even Øiseth 1 Rambøll, Trondheim, Norway Trenching is a frequently used method for protection of offshore pipelines against dropped objects and fishery activities. To provide additional protection the trenches are usually back-filled. The backfill material may be soft, partly remoulded clay or sand present at the seabed or sand mix and crushed rock installed after pipe lay. Offshore pipelines may experience large axial forces due to temperature expansion. As a part of the pipeline design, the buckling failure modes must be analysed. In the buckling analyses, an important parameter is the upheaval resistance from the backfill material. To verify the design of a trenched offshore pipeline in the Norwegian Trench, a full-scale test was performed by SINTEF and NTNU. The test program included measurement of uplift resistance in soft clay, sand mix and crushed rock. This paper presents the test setup, the backfill materials and the measured uplift resistances. The measured resistances are compared with theoretical models for uplift resistance for different backfill materials. Recommendations are given regarding the uplift resistance pipes in trenches backfilled with sand mix or crushed rock. INTRODUCTION Pipelines used for transportation of hot content will experience thermal expansion at start-up of production. In many cases the pipelines are buried in trenches for protection or temperature insulation. Due to sliding friction between pipe and soil, thermal expansion will result in increased axial force. If the lateral support from the soil is insufficient, the axial force increase may result in global buckling. The buckling may appear downwards, laterally or upwards depending on the soil support and the pipeline curvature. According to DNV-F110, uncertain or unknown soil conditions are a common cause of construction delays and cost escalations for marine pipeline projects. Soil investigations to determine the soil conditions both for intact and backfilled soil are required to perform pipeline buckling analysis. In addition to the uncertainties in the soil conditions, there are uncertainties in the soil response at known soil conditions. This paper presents methods for estimation of the uplift resistance in different backfill materials and results from a full-scale test of the uplift resistance for pipes in a trench backfilled with soft clay, sand mix and crushed rock. The uplift resistances discussed in this paper include only the resistances originated from shear stress changes in the soil when lifting the pipe. The weight and buoyancy of the pipe in soil or water are not included. resistance can then be expressed as: R = N c s u D (1) N c = bearing capacity factor s u = undrained shear strength at the centre of the pipe D = pipe diameter Closed form solutions for the lateral resistance for piles with circular cross-section were derived by Randolph and Houlsby (1984). They found N c = 9 14 and 11.94 for a smooth and rough pipe surface, respectively. These results are also applicable for the uplift resistance of pipes at large burial depth if full tension is assumed between pipe and soil. At shallow burial depth the uplift resistance may be reduced due to a failure mechanism reaching the surface. The burial depth can be described by the embedment ratio (H/D), as defined in Fig. 1. Martin and White (2012) presented results from detailed finite element analysis of the vertical resistance of pipes in cohesive soil. The results were presented as a function of the pipe embedment, interface properties between soil and pipe and the weight- UPLIFT RESISTANCE IN COHESIVE SOILS At large burial depth the uplift resistance R will be limited by local failure around the pipe. In cohesive material the local 1 Employed at SINTEF when the field tests were performed. Received January 31, 2013; revised manuscript received by the editors April 20, 2013. The original version was submitted directly to the Journal. KEY WORDS: Pipelines, trench, uplift resistance. Fig. 1 Embedment definition and failure modes

International Journal of Offshore and Polar Engineering, Vol. 23, No. 4, December 2013, pp. 298 306 299 As the pipe approaches the soil surface, the failure modes change from a local failure around the pipe to a global failure mode reaching the surface, as shown in Fig. 1. The global uplift resistance for shallow embedment can be approximated, as presented in Eq. 4, the resistance is assumed to be the sum of the reversed bearing capacity on the lower part of the pipe and the wedged failure above the pipe. R = 1 2 9 14 + r i 2 83 s u D + 2s u H H/D < 9 14 + r i 2 83 /4 (4) Fig. 2 Uplift resistance as function of embedment with tension between soil and pipe to-strength ratio D/s u. The uplift resistance with buoyancy and pipe weight excluded is, however, independent on the weight-tostrength ratio for the case with full tension between pipe and soil. Only results for weightless soil are therefore included in this discussion. For a smooth pipe (r i = 0) with full tension, Martin and White (2012) found that the bearing capacity factor increases linearly from 4.5 for H/D = 0 to 9.14 for H/D = 1 5. For a rough pile, the bearing capacity factor increased linearly from 5.7 for H/D = 0 to 11.94 for H/D = 2 2. The results are included in Fig. 2. For H/D > 2 2 these results coincide with the closed form solution proposed by Randolph and Houlsby (1984). The results by Martin and White (2012) can be approximated with the bilinear expression presented in Eq. 2. 9 14 1 ( 1 + H/D ) 1 + 0 306 r 2 1 4 + 0 6r i s u D i R = H/D < 1 4 + 0 6 r i (2) 9 14 1 + 0 306 r i s u D H/D 1 4 + 0 6 r i The resistance according to Eq. 4 for r i = 0 and r i = 1 is included in Fig. 2. The resistance found by Martin and White (2012) exceeds the resistance according to Eq. 4, indicating that the simplified wedge failure mode does not capture the entire soil resistance. To investigate the transition in failure mode from local failure at deep embedment to a failure mode reaching the surface at shallow embedment, the pipe uplift case was analysed by the authors using the finite element method (FEM). The analysis was performed using the software PLAXIS 2D 2011. The calculated bearing capacity factors for the embedment ratios H/D = 2 and 3 and interface roughness r i = 1 and H/D = 1 5 and 2 for r i = 0 02 are shown in Fig. 2. Due to numerical limitations, r i = 0 could not be analysed. The FEM analysis is done with weightless soil and pipe. The failure modes presented as incremental shear strains are shown in Fig. 3. The result from the FEM analysis corresponds well with the results reported by Martin and White (2012) and confirms that the wedge failure mode assumed in Eq. 4 underpredicts the resistance at intermediate embedment depths. The local failure mode at H/D = 3 shows only slightly higher resistance than the global failure modes reaching the surface at H/D = 2. For the case r i = 1, the local failure mode involves an area reaching about 1D to each side of the pipe and 1 2 D above and below the pipe while the global failure mode involves an area about 10D wide and from 2D below the pipe up to the surface. A practical implication of these findings is that the uplift resistance for a pipe in a trench with soft backfill material will be governed by the strength of the backfill close to the pipe. Installation of stronger material on top of soft clay backfill may change the failure mode from global to local but not provide a significant In DNV-RP-F110, the expression presented in Eq. 3 is proposed for the undrained uplift bearing capacity factor as a function of embedment and pipe-soil interface roughness (r i. The expression consists of a constant term 2 and a term approaching 2 83 1 + r i. The constant term can be interpreted as the reversed bearing capacity acting on the lower half of the pipe while the other term provides a gradual transition to the local failure mode for deep embedment. [ N c = 2 1 + 1 ( atan H ) ] 1 + r 3 D i H 4 5 (3) D The expression in Eq. 3 corresponds with the factors found by Randolph and Houlsby (1984) for H/D > 4 5. In Fig. 2, Eq. 3 is compared to the results from Martin and White (2012). Equation 3 shows higher resistance at shallow embedment, particularly for smooth pile surface (r i = 0), and does also require larger embedment to reach full resistance. One reason for the higher resistance for the (r i = 0) case is that the term for the resistance below the pipe in Eq. 3 is independent of the surface roughness. Fig. 3 Failure modes for different embedment ratios illustrated by contour plots of the incremental shear strain

300 Full-scale Test of Uplift Resistance of Trenched Pipes increase in uplift resistance unless the pipe is in direct contact with stronger material. The uplift resistance in case of full tension between pipe and soil is independent of the soil unit weight. With no tension between pipe and soil, the resistance will be lower at shallow embedment and also dependent of the ratio /s u is the submerged unit weight of the soil. In case of no tension, the underside of the pipe will be in contact with the soil only below embedment depths the overburden stress causes reversed failure. Assuming a failure mode according to strip loading, the depth full contact appears can be shown to be H > 2 + s u /. The no-tension condition will be relevant for cases the trench is backfilled with partly remoulded clay and free water voids between lumps of clay. The calculated uplift resistance according to Martin and White (2012) for the submerged soil unit weight = s u /D and no tension is included in Fig. 4. With this unit weight, a gap between the underside of the pipe and soil may open at the embedment H/D < 2 +. As seen in this figure, the uplift resistance in case of no tension approaches the full tension cases (N c = 9 14 and 11.94) at embedment H/D = 5, corresponding well with the depth for reversed failure. DNV-RP-F110 recommends the expression included in Eq. 5 for calculation of the uplift resistance for a shallow embedded pipe the failure extends to the surface and with no tension between pipe and soil. { ( )} R = HD D 2 + 2s 8 u H (5) The resistance according to this expression is included in Fig. 4. The resistance consists of the sum of the weight of a soil wedge above the pipe and the shear strength along the sides of the soil wedge. The effect of pipe roughness is not included. The expression corresponds well with the results found by Martin and White (2012) for a smooth pipe with no tension between soil and pipe at H/D < 2 5 but exceeds the results for the rough pipe case at H/D > 3 5. UPLIFT RESISTANCE IN COHESIONLESS SOILS The uplift resistance in cohesionless soils is usually calculated using wedge uplift models the resistance is divided into two contributions: the weight of the soil above the pipe and the shear force acting on the slip surface along the soil wedge. The slip surface can be either vertical or inclined. The inclined slip surface model has been experimentally shown to be closer to the real deformation mechanism by White et al. (2001). In this model, the inclination of the wedge is equal to the dilatancy angle. However, the closed form solution for the inclined model is algebraically equivalent to the vertical slip model. For back calculation of uplift tests presented in this paper, only the vertical slip model is considered. In the vertical slip surface model, the shear force on the soil wedge will be a function of the internal soil friction and the lateral stress. This uplift resistance model presented in Eq. 6 is discussed by Trautmann et al. (1985), Schaminée et al. (1990), DNV-RP- F110 and White et al. (2008). { ( )} R = H D D 2 + K tan H 2 (6) 8 K = lateral earth pressure coefficient accounting for the increase in vertical stress during uplift = angle of internal friction. The earth pressure coefficient and the internal friction may be combined in an uplift factor f up = K tan. The main uncertainty in using this resistance model is the lateral earth pressure coefficient. The lateral pressure will depend on the degree of compaction or relative density I D, the dilation angle and the shape of the trench. Trautmann et al. (1985) presented results from an experimental study of the uplift force-displacement response of buried pipes. In this paper recommendations are given for peak and residual uplift resistance as a function of sand density and pipe embedment. For very loose sand, a nearly constant value was found for H/D > 4 indicating a local failure mode. Schaminée et al. (1990) presented results from full-scale tests on axial and uplift resistance for pipes buried in crushed rock, dense and loose sand, and remoulded clay. The uplift tests in sand and crushed rock showed results similar to Trautmann et al. (1985). Based on the test results, Schaminée et al. (1990) recommend uplift factors f up = 0 6, 0.4 and 0.25 for crushed rock, sand and very loose sand, respectively. DNV-RP-F110 recommends K = 1 sin for loose sand while a lateral pressure, according to passive earth pressure included in Eq. 7, is recommended for medium and dense sand. K = ( 1 + tan 2 tan 1 + r) 2 (7) Fig. 4 Uplift resistance as function of embedment with no tension between soil and pipe and = s u /D r = a roughness parameter with value close to 1. A limit equilibrium solution for the uplift resistance of pipes and plate anchors in sand is proposed by White et al. (2008). This solution includes three independent parameters: the peak friction angle crit, the dilation angle and the effective unit weight of the soil. By applying a correlation between the dilation angle, the relative density I D and grain crushing strength proposed by Bolton (1986), the solution is recast in terms of crit and I D. Since crit and grain crushing strength show minimal variation for typical sands used in offshore applications, the main material parameter required for design is I D. The solutions proposed by White (2008) do not, however, cover crushed rock as backfill material since I D is a poor measure for the properties for this material.

International Journal of Offshore and Polar Engineering, Vol. 23, No. 4, December 2013, pp. 298 306 301 Uplift factor f up Test results / model Soil condition H/D =1 5 H/D =4 H/D =8 Trautmann Dense sand 0.57 0.57 0.78 (max force) Medium sand 0.31 0.60 0.54 Loose sand 0.37 0.27 0.16 Schaminée Crushed rock 0.60 Sand 0.40 Very loose sand 0.26 White I D =0 7 1.14 0.97 0.91 D =0 25 m I D =0 5 0.99 0.72 0.64 =10 kn/m 3 I D =0 3 0.71 0.49 0.44 I D =0 1 0.42 0.32 0.29 DNV =44, r = 0 85 0.94 1 RP F110 =34, r = 0 94 0.62 1 =32, K O =1 sin 0.29 1 Uplift factors interpreted from the tests presented in this paper Table 1 Uplift resistance factors for different models A comparison of the reported test results by Trautmann et al. (1985), Schaminée et al. (1990), the theoretical solution proposed by White et al. (2008), and the DNV recommendation is presented in Table 1 in terms of the uplift factor f up. The results reported by Trautmann et al. (1985) and the model proposed by White et al. (2008) show relatively large variation in uplift factor for different embedment ratios. This large variation may be partly explained by the failure modes being poorly represented by the resistance model in Eq. 6. In particular, this may be the case for the Trautmann tests for dense sand compaction in layers was conducted to achieve the desired density. Table 1 does also include friction angles and roughness for the DNV model (Eqs. 6 and 7 combined). The parameters for the DNV model are adjusted to give uplift factors corresponding to the resistances measured in the full-scale test presented later in this paper. In addition to the uplift resistance, the force-displacement response is an important input to upheaval buckling evaluations for buried pipelines. In DNV-RP-F110, a tri-linear curve for the upheaval resistance in cohesionless soils is proposed fully mobilized resistance is reached at a displacement ( f ranging from 0.5 1.0% of the backfill cover height. Thusyanthan et al. (2010) presented results from six different uplift tests and found poor correlation between cover height and f. Thusyanthan et al. (2010) proposed a new relation the displacement at peak resistance is given as a function of pipe diameter and embedment ratio, as shown in Eq. 8. The main purpose of the full-scale tests was to investigate the uplift resistance of a pipe in a trench backfilled with soft clay, sand or crushed rock. The effect of sand and crushed rock backfill on soft clay was of particular interest. This was an important issue for pipeline projects the buckling evaluations depend on the lateral resistance from the soft clay backfill. Based on the test results, recommendations were given regarding possible improvements in the uplift resistance using sand or crushed rock. The pipe uplift tests were performed by NTNU and SINTEF at a test site in Spongdal close to Trondheim. TRENCH GEOMETRY At the test site 10 trench sections were excavated. Each section was 1.6 m deep, 0.7 1.0 m wide and 2.5 m long. A cross-section of the trenches, including the pipe, is shown in Fig. 5. PIPE SECTIONS Pipe sections with 2000-mm length, 273-mm diameter and closed ends were manufactured for the uplift tests. The pipes were coated with epoxy to get a surface roughness close to a typical pipeline used in offshore projects. The pipe sections were filled with water to reduce the buoyancy. The weight of a pipe section, including water, was 2.80 kn, corresponding to a unit weight of 24.4 kn/m 3. The pipe sections were positioned 100 mm above the trench bottom, as shown in Fig. 5. INSTRUMENTATION The uplift force and displacement was measured using two sets of load cells and deformation sensors, as shown on Fig. 6. The load cells were mounted on steel rods between the pipe and the pulling rig while two deformation sensors were mounted between a separate steel beam and the steel rods on the pipe. The deformation reference beam was connected to the ground through two soil anchors. The measurements were recorded using a computer. LOAD EQUIPMENT A Geotech 1500 geotechnical drill rig was used to pull the pipe upwards at a velocity of 10 mm per minute. The pulling capacity for the drill rig was 100 kn. f = D 0 02 e 1 2 H /D (8) f = displacement at maximum resistance H = height of backfill above the pipe H = H D/2. FULL-SCALE TESTS The uplift resistance is important for the evaluation of pipe upheaval buckling. As shown both by theoretical models and numerical analysis discussed above, the uplift resistance is strongly dependent on the pipe embedment. However, the embedment is a rather uncertain value for real pipelines. To improve the protection and uplift resistance of pipelines, additional backfill of sand mix or crushed rock is frequently installed. Fig. 5 Idealised trench geometry

302 Full-scale Test of Uplift Resistance of Trenched Pipes Grain Dry Total unit Fraction density density weight Porosity Material mm g/cm 3 g/cm 3 kn/m 3 n Sand mix 0 32 2.72 1.74 20.30 0.36 Crushed rock 25 75 2.70 1.44 18.55 0.47 Table 2 Properties for the granular backfill materials Fig. 6 Test set-up and instrumentation BACKFILL MATERIALS Clay Clay from the Onsøy test site was used in the test to represent backfill of soft marine clay. The clay at the Onsøy test site has been investigated in several research programmes. According to Lunne et al. (2006), the natural water content ranges from 61 to 65%, the sensitivity from 3 to 7 and the plasticity index from 34 to 41%. The clay used in this test was collected at 2 to 3-m depth below the dry crust and brought by truck to the test site. At the test site the clay was stored in a steel container. The index properties measured on the collected material were: Liquid limit: w l = 49 2% Plasticity limit: w p = 25 9% Plasticity: I p = 23 3% Grain density: s = 2 78 g/cm 3 Unit weight: = 17 3 kn/m 3 Water content: w = 49% The water content and liquid limit were lower than the values for the Onsøy clay reported by Lunne et al. (2006). The reason for this difference is believed to be caused partly by the transportation and storage of the clay and partly by local variations at Onsøy. To achieve properties similar to soft clay backfill in an offshore trench, the clay was remoulded in a concrete mixer while adding water. The amount of water added corresponded to about a 1% increase in water content. To secure repeatability and similar conditions for each batch of clay prepared in the concrete mixer, a fall cone test (60 g and 60 was used to monitor the achieved remoulded shear strength. The shear strength was interpreted according to the procedure presented in NS 8015 11.2 mm cone penetration corresponds to an undrained shear strength of 2 kpa. Alternative correlations between cone penetration and undrained shear strength do, however, exist. Hansbo (1957) proposed separate relations between the cone penetration and shear strength for remoulded and undisturbed clay. The Hansbo relation for a 60 g and 60 cone penetrating 11.2 mm in remoulded clay results in a shear strength of 1.4 kpa. The actual shear strength of the clay in the test trench was therefore uncertain. The uplift tests were performed 1 3 days after mixing of the clay, not allowing significant strength increase due to consolidation. The development in undrained shear strength with time due Fig. 7 Grading curves for sand and crushed rock to thixotropic effects was tested using a falling cone apparatus. A minor increase in shear strength was found 2 3 days after mixing. Sand mix and crushed rock Sand mix and crushed rock from a quarry at Sløvåg close to Bergen was used as backfill above the clay. The material properties are listed in Table 2 while the grading curves are shown in Fig 7. The trench was filled with water before installing sand or rock. The density was measured by placing a steel bucket in the trench before installing backfill material. The bucket volumes were 12.4 litres and 49.5 litres for the sand mix and crushed rock density measurements, respectively. TEST PROGRAM A total of eight uplift tests with different layering and backfill masses above and around the pipe were performed. In all the tests, the top of the pipe (TOP) was placed 1200 mm below terrain level, corresponding to an H/D ratio of 4.9. The first four tests included combinations of sand and remoulded clay while the sand was replaced by crushed rock in the last four tests. The combinations of different backfill materials are summarized in Table 3. Height of Backfill Height of sand/ clay above top Test no material rock backfill (mm) of pipe (mm) 1 Sand mix 1200 0 2 Sand mix 1000 200 3 Sand mix 800 400 4 Sand mix 1200 No clay 5 Crushed rock 1200 No clay 6 Crushed rock 1000 0 7 Crushed rock 800 200 8 Crushed rock 1200 400 Table 3 Backfill material in the different tests

International Journal of Offshore and Polar Engineering, Vol. 23, No. 4, December 2013, pp. 298 306 303 Fig. 8 Geometry tests 1 and 6 In tests 1 and 6, clay was filled from the bottom level of the trench and up to the top level of the pipe. Above the clay there was 1200 mm of sand (test 1) or crushed rock (test 6) (see Fig. 8). In tests 2 and 7, clay was filled from the bottom level of the trench and up to 200 mm above the top level of the pipe (i.e. 1000 mm below new terrain level in the trench). Above the clay there was 1000 mm of sand (test 2) or crushed rock (test 7) (see Fig. 9). In tests 3 and 8, clay was filled from the bottom level of the trench and up to 400 mm above the top level of the pipe (i.e. 800 mm below new terrain in the trench). Above the clay there was 1000 mm of sand (test 3) or crushed rock (test 8) (see Fig. 10). In tests 4 and 5, only sand or crushed rock was filled in the trench (see Fig. 11). TEST RESULTS The uplift resistance R was interpreted as the force contributing to shear stress changes in the soil. The total weight of the pipe (in air) is subtracted while the buoyancy of the pipe in soil is added to the measured force, as shown in Eq. 9: R = R m W tot + soil V pipe (9) R = uplift resistance R m = measured force Fig. 10 Geometry tests 3 and 8 W tot = total weight of the pipe section in air soil = total soil unit weight V pipe = volume of the pipe section When evaluating the uplift resistance, the buoyancy in soil must be included in clay, sand mix and crushed rock. During the uplift tests, the pipe section was partly surrounded by materials with slightly different unit weight. For simplicity, the total unit weight of clay (17.3 kn/m 3 was used for all tests and uplift distances, giving buoyancy of 1.01 kn/m for the pipe in soil. The uplift resistance in clay is evaluated using the theory for local failure around a pipe, as presented in Eq. 1: R = N s u D. In sand and rock, the uplift resistance is evaluated relative to the weight of a vertical wedge above the pipe and friction along the sides of the wedge. The weight of the soil wedge is adjusted for the observed surface upheaval by adding the term H: R = { H + H D D 2 ( 8 )} + f up H 2 (10) = submerged unit weight of the soil f up = uplift resistance factor H = distance from the centre of the pipe to the soil surface H = measured surface upheaval Surface upheaval The width and the heave of the deformed zone at the surface of the trench were measured during each test. For most of the tests Fig. 9 Geometry tests 2 and 7 Fig. 11 Geometry tests 4 and 5

304 Full-scale Test of Uplift Resistance of Trenched Pipes Fig. 12 Deformed zone for test 1 with sand, the maximum width at the surface upheaval zone was approximately 600 mm. The maximum heave was approximately 120 mm. Figure 12 shows a sketch with different positions of the pipe and corresponding widths of the deformed zone. In the tests with crushed rock, the maximum width at the surface ranged from 600 to 800 mm and the maximum heave was approximately 100 mm. In the back calculations, the surface upheaval is assumed to be proportional to the pipe lifting distance reaching 100 mm at 1200 mm lifting distance. Resistance in sand mix and clay The test results show good repeatability for pulling through the clay and the sand layers (see Fig. 13). A maximum uplift resistance of 14.3 kn/m was measured in Test 4 at about 50-mm pulled distance with only sand in the trench. The maximum force was only slightly lower in Test 1 with clay up to the pipe top, while the maximum force was reached at approximately 250-mm pulling distance. The resistance according to Eq. 10 using f up = 0 62 fits well with the measured resistance in the sand mix. As shown in Table 1, the friction factor corresponds with DNV approach (Eqs. 6 and 7) using a friction angle of 34 and a roughness of r = 0 94. The pulling force through clay is about 3.5 4 kn/m for both tests 2 and 3. When the top of the pipe is pulled into the sand layer, the pulling force is considerably increased. No significant increase in resistance is found prior to the first contact between the pipe and the sand layer. This shows that the local failure mode Fig. 14 Tests 5-8: recorded load per meter against vertical deformation for the tests in clay and rock the clay is flowing around the pile is, to a very limited extent, influenced by the presence of a sand mix layer above. As shown in Fig. 3, the failure zone reaches only 1 D above the pipe 2 for the case r i = 1 while, for the case r i = 0 02, the failure zone only reaches 0 13D above the pipe. When the bottom of the pipe is leaving the clay layer, there is a drop in the pulling force, indicating the loss of suction forces from the clay on the underside of the pipe. This shows that that tension between pipe and clay below the pipe was present in the test. Resistance in crushed rock and clay The tests using crushed rock did also show good repeatability for pulling through the clay and crushed rock backfill. The curves plotted in Fig. 14 are very similar to the ones found for tests 1 4. This shows that the same mechanisms occur whether the fill masses above the clay are sand mix or crushed rock. Maximum pulling force is registered for test 5 with only crushed rock in the trench. A maximum resistance of 17 kn/m was found after 60- mm pulling distance. The maximum force is only slightly lower for test 6 with rock from the level of the pipe top. The response is somewhat softer with a maximum resistance after approximately 100 mm. The resistance according to Eq. 10 using f up = 0 94 fits well with the measured resistance in crushed rock. As shown in Table 1, this friction factor corresponds with DNV approach (Eqs. 6 and 7) using a friction angle of 44 and a roughness of r = 0 85. Distance at Peak Resistance As can be seen from Fig. 13 and Fig. 14, the pulled distance at peak resistance for tests 4 and 5 are 50 and 60 mm, respectively. This result corresponds well with Eq. 9 giving f = 49 mm for a cover height of 1200 mm and pipe diameter 273 mm. The corresponding range (0.5% 1.0% of pipe diameter) recommended in DNV-RP-F110 is 14 27 mm. This result supports the findings by Thusyanthan et al. (2010) regarding displacement at peak resistance. Failure modes in sand mix and crushed rock Fig. 13 Tests 1-4: recorded load per meter against vertical deformation for the tests in clay and sand mix The uplift resistance for tests 4 and 5 was back-calculated by the authors with the finite FEM program PLAXIS 2D using the friction angles corresponding to the back-calculated f up values. A low stiffness was used since the backfill materials were uncompacted. In the analysis, the maximum interface friction was

International Journal of Offshore and Polar Engineering, Vol. 23, No. 4, December 2013, pp. 298 306 305 Measured Calculated Backfill Friction Dilatancy resistance resistance Test material angle angle kn/m kn/m 4 Sand mix 34 0 14 3 14 2 5 Crushed rock 44 14 17 16 5 Table 4 Uplift resistance in sand mix and crushed rock FEM Fig. 15 Failure modes (incremental deviatoric strains) for tests 4 and 5 back-calculated using PLAXIS set to 50% of the shear strength of the backfill material. The horizontal stress coefficient for the backfill was assumed to be K 0 = 1 sin while the surrounding clay was modelled as linear elastic material with E = 10 MPa and = 0 4. The dilatancy angles were chosen relative to the friction angles according to the recommendations given by Bolton (1986). Cohesion c = 0 2 kpa, E = 1 MPa and = 0 4 were used for both sand mix and rock. The material parameters and the calculated resistances are summarized in Table 4. The calculated failure modes are shown in Fig. 15 illustrated by incremental deviatoric strains. The calculated resistances correspond well with both the measured resistances and the resistance back-calculated using Eq. 9. The failure modes found in the FEM analysis correspond to the inclined wedge mode discussed by White et al. (2001) the inclination is determined by the dilatancy angle. The inclination is close to vertical for the sand mix while the crushed rock shows an inclination close to the dilatancy angle of 14. The analysis does also show that the trench geometry was sufficiently wide not to influence the failure mode. The relatively narrow trench may, however, cause some horizontal confinement resulting in higher uplift resistance. Resistance in clay As discussed above, different correlations between undrained shear strength and drop cone penetration exist. The measured uplift resistance in clay of 4 kn/m may be used to evaluate the undrained shear strength of the clay. The test arrangement is similar to a T-bar test, and a bearing capacity factor similar to the T-bar factor N T was expected. According to Lunne et al. (2005), the T-bar factor for undisturbed Onsøy clay ranges from 11 to 13.4 with 11.9 as the average value. However, the clay used in the test presented in this paper was fully remoulded and therefore without sensitivity. According to Hansbo (1957), the ratio between interpreted shear strength for 11.2 mm cone penetration (60 g and 60 for remoulded and intact clay is 0.86. If we apply this ratio on the T-bar factor, the corresponding T-Bar factor for remoulded Onsøy clay becomes 11 9 0 86 = 10 2. This value corresponds well with the recommendations, according to plasticity theory, of 10.5 by Randolph and Houlsby (1984). Back-calculation of the undrained shear strength using a T-Bar factor 10.5 and a measured uplift resistance of 4 kn/m gives: s u = R/N T /D = 4 kn/m/10 5 /0 273 m = 1 4 kpa This fits well with the Hansbo fall-cone relation for remoulded clay 11.2 mm penetration corresponds to s u = 1 4 kpa. Some characteristic differences in the uplift resistance in clay between tests 2 and 3 (sand) and tests 7 and 8 (rock) were measured. In tests 2 and 3 (Fig. 13), the maximum resistance of 4 kn/m was reached after 15-mm pulled distance followed by a gradual reduction to 3.5 kn/m. In test 7, 3.8 kn/m was reached after about 15-mm pulled distance followed by a gradual increase to 4.3 kn/m after 80-mm pulled distance. The corresponding values for test 8 were 3.5 kn/m and 3.9 kn/m. The reason for this difference is believed to be caused by the time period between clay mixing and uplift testing. Tests 2 and 3 were performed two and three days after clay mixing while tests 7 and 8 were performed one day after clay mixing. The additional time before testing for tests 2 and 3 did therefore allow for development of thixotropic increase in strength and thus some sensitivity in the clay, explaining the peak in resistance at 15 mm displacement. CONCLUSIONS The uplift tests presented in this paper show that the short-term initial uplift resistance for a pipe in a trench backfilled with soft clay will not be significantly increased by installing additional backfill of sand or crushed rock. A significant increase can only be expected if the pipe is in contact with the sand or crushed rock layer. Strength increase due to consolidation may, however, provide some increase in the uplift resistance. However, the uplift resistance shortly after backfilling is frequently the most relevant design condition. The measured uplift resistance in backfilled remoulded clay is found to correspond to a local failure model with a bearing capacity factor N = 10 5. The undrained shear strength of soft clay backfill is, however, not easy to measure accurately. The failure modes during uplift of a pipeline are similar to the failure modes for the T-Bar test. In order to predict reliable resistances in soft remoulded clay, the T-bar test is recommended as test method. The measured displacements at peak resistance in sand mix and crushed rock backfill are found to be 2 3 times larger than the recommendations given in DNV-RP-F110 and in good correspondence with an equation proposed by Thusyanthan et al. (2010). Back-calculation of the measured resistance in sand mix and crushed rock shows that the vertical wedge model, the friction along the wedge sides is calculated using an uplift resistance factor f up, is applicable for the estimation of the uplift resistance. Through back-calculation, the uplift test resistance factors f p = 0 62 and 0.94 with corresponding internal friction of 34 and 44 were found for sand mix and crushed rock, respectively. FEM analysis of the uplift test resistances, applying the same friction angles in a Mohr-Coloumb model, shows good correspondence with the measured peak resistances. REFERENCES Bolton, MD (1986). The Strength and Dilatancy of Sands, Géotechnique, Vol 26, No 1, pp 65 78. DNV-RP-F110, Global Buckling of Submarine Pipelines, Det Norske Veritas, Oct 2007.

306 Full-scale Test of Uplift Resistance of Trenched Pipes Hansbo, S (1957). A New Approach to the Determination of the Shear Strength of Clay by the Fall-Cone Test, Proc Roy Swedish Geotech Inst, Vol 14. Lunne, T, Randolph, MF, Chung, SF, Andersen, KH, and Sjursen, M (2005). Comparison of Cone and T-Bar Factors in Two Onshore and One Offshore Clay Sediments, Proc Int Symp Front Offshore Geotech, IS-FOG 2005, pp 981 989. Lunne, T, Berre, T, Andersen, KH, Strandvik, S, and Sjursen, M (2006). Effects of Sample Disturbance and Consolidation Procedures on Measured Shear Strength of Soft Marine Norwegian Clays, Can Geotech J, Vol 43, No 7, pp 726 750. Martin, CM, and White, DJ (2012). Limit Analysis of the Undrained Bearing Capacity of Offshore Pipelines, Géotechnique, Vol 62, No 9, pp 847 863. Randolph, MF, and Houlsby, GT (1984). The Limiting Pressure on a Circular Pile Loaded Laterally in Cohesive Soil, Géotechnique, Vol 34, No 4, pp 613 623. Schaminée, PEL, Zorn, NF, and Schotmann, GJM (1990). Soil Response for Pipeline Upheaval Buckling Analyses: Full-Scale Laboratory Tests and Modelling, Proc 22nd Offshore Tech Conf, Houston, OTC Paper No 6486, pp 563 572. Standard Norge (1988). Geotechnical Testing, Laboratory Methods: Determination of Undrained Shear Strength by Fall-Cone Testing, NS 8015, Standard Norge, Lysaker, Norway. Trautmann, CH, O Rourke, TD, and Kulhawy, FH (1985). Uplift Force-displacement Response of Buried Pipe, J Geotech Eng, Vol 111, No 9, pp 1061 1076. Thusyanthan, NI, Mesmar, S, Wang, J, and Haigh, SK (2010). Uplift Resistance of Buried Pipelines and DNV-RP-F110 Guidelines, Proc Offshore Pipeline Tech Conf, Amsterdam. White, DJ, Barefoot, AJ, and Bolton, MD (2001). Centrifuge Modelling of Upheaval Buckling in Sand, Int J Phys Model Geomech, Vol 2, pp 19 28. White, DJ, Cheuk, CY, and Bolton, MD (2008). The Uplift Resistance of Pipes and Plate Anchors Buried in Sand, Géotechnique, Vol 58, No 10, pp 771 779. Proceedings of the 10th (2013) ISOPE Ocean Mining & Gas Hydrates Symposium Szczecin, Poland, September 22 26, 2013 DEEP-OCEAN MINERALS AND PROCESSING, EXPLORATION AND ENVIRONMENT, DEEP-OCEAN MINING SYSTEMS AND TECHNOLOGY (Mining Systems, Ship, Pipe, Nodule Lift, Buffer, Link, Oceanfloor Miner, and Miner Control) GAS HYDRATES (Fundamentals, Properties, Geotechnical and Geochemical Characteristics, Development) The Proceedings (ISBN 978-1-880653-92-0; ISSN 1946-0066): $100 (ISOPE Member: $80) in a single volume (CD-ROM) is available from www.isope.org or www.deepoceanmining.org, ISOPE, P.O. Box 189, Cupertino, California 95015-0189, USA (Fax +1-650-254-2038; orders@isope.org)