Marine Geology 213 (2004) 457 – 480 www.elsevier.

com/locate/margeo

Slope stability assessment of the Helland Hansen area offshore the mid-Norwegian margin
D. Leynauda,*, J. Mienerta, F. Nadimb
b a Department of Geology, University of Tromsø, NO-9037 Tromsø, Norway International Centre for Geohazards/Norwegian Geotechnical Institute, NO-0806, Oslo, Norway

Accepted 30 September 2004

Abstract Numerous large-scale submarine slides affected the continental slope offshore mid-Norway during the Holocene. One of them, the Storegga slide, is located south of the Vbring Plateau, stretching from the North Sea Fan to the Helland Hansen arch. With a volume of 3200 km3 and an influenced area of 95,000 km2, the Storegga slide, which occurred about 8250 years before present (BP), represents one of the largest submarine slides worldwide. The slope stability assessment concentrates on the northern sidewall (Helland Hansen area) where gas hydrates are suspected from the presence of a bottom-simulating reflector (BSR) on reflection seismic data and where few slope stability investigations have been performed compared to the Ormen Lange area. The limit equilibrium and finite element methods (FEM) were used for the evaluation of static and dynamic (seismic) stabilities of the dipping seabed. To account for the uncertainties in the soil parameters, a probabilistic approach was applied by coupling the limit equilibrium model with the first- and second-order reliability methods (FORM and SORM) and the Monte Carlo simulation. The finite element simulation for the seismic loading indicates that a strong earthquake (N6.5 Ms) could be a potential trigger for slope failure but only down to 30 m subsurface. Obviously, one needs a significant pore pressure build-up and loosening of the sediments to explain the slide on such gentle slopes. More critical preconditions to failure in the deeper sediments should exist to explain the thickness of the sliding slab, knowing that the northern sidewall is approximately 150 m high. Cyclic loading due to a series of earthquakes could explain the slide, affecting the shearing resistance in a marine clay unit (weak layer) by excess pore pressure generation (drained conditions). The degradation of undrained shear strength with cyclic loading plays a major role in the instability process. D 2004 Elsevier B.V. All rights reserved.
Keywords: Storegga slide; risk assessment; slope failure; limit equilibrium; finite element; first- and second-order reliability methods

1. Introduction
* Corresponding author. Tel.: +33 298 22 42 57; fax: +33 298 22 45 70. E-mail addresses: dleynaud@ifremer.fr, leynaud@ibg.uit.no (D. Leynaud). 0025-3227/$ - see front matter D 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.margeo.2004.10.019

Submarine landslides are frequent both on passive and active continental margins, especially on the continental slope (e.g., Mienert et al., 2002). This is

458 D. Leynaud et al. / Marine Geology 213 (2004) 457–480 Fig. 1. (a) Location of the Storegga slide offshore Norway (Stor). Tr&nadjupet and Bjbrnbyrenna slides areas are shown (grey areas). (b) Location of boreholes 6405/2, 6404/5, long sample MD99-2288, and seismic line NH9651-202 on the enlargement.

0 28. and excess pore pressure generation are among the most obvious triggers of submarine failures. Rapid deposition of sediment can often play a destabilizing role in this area. The geotechnical parameters come from borehole 6404/5-GB1 (966 m water depth).. 2003). 3.0 43.D. Ip.8 18. S t .4 3. Cyclic loading from earthquakes or waves.558 to 1.9 17. 2001). The headwall is 290 km long and is located at a water depth of 200–300 m (Haflidason et al.3 43. where the effects of gravity on the downslope sediments are increased. Basic concepts 3. Methodology 3. The location of the two major Holocene slides (Storegga and Tr&nadjupet slides) on the mid-Norwegian margin suggests a relation with the deeper structures and processes on the margin (Berndt et al. 1).3 17.4 18.2 1.1 4. reducing the effective strength of the sediments (Bryn et al. total unit weight. sensitivity. the aim is also to verify the quality of the results (safety factor and failure probability) to ensure the reliability of the different approaches and the similarity of the final estimates.2 1. not all failure mechanisms are fully understood.0 St 3. is between 1. 2003). EOB is 309 m below seabed Unit 1 2 3 4 5 6 Depth (m) 0–30 30–65 65–123 123–152 152–270 270–EOB Clay Very soft to soft Medium to stiff Very stiff Very stiff to hard Hard becoming very hard Hard Clay content (%) 42 43 41 35 43 34 c tot (kN/m3) 16. The probabilistic approach is thus applied to observe the effect of uncertainty of each parameter on the likelihood of failure and the failure probability during a specific time period.178 and 1. gas hydrate dissociation. OCR. The slide occurred during a multiphase event approximately 8250 years ago (Haflidason et al. Six soil units (from very soft to very hard clays) were defined from borehole 6404/5 down to a maximum depth of 309 m (Table 1). and possible existing excess pore pressure. 2.3 32.1.6 4. Topographic difference defining the Vbring high (inferred to play a stabilizing role on the margin) have been increased by differential subsidence and uplift of the margin (Bryn et al.9 2.7–1.328 (0.4 26.4 29.148 eastward) (Haflidason et al. Excess pore pressure within the marine clays due to rapid loading of glacial deposits is considered as one of the main destabilization factors. w. The slides are defined as translational from the slip planes and the thickness of removed sediment (compared to the lateral extension).1.0 OCR 1. As the modelling uses some simplifications. water content. and long piston corer sample MD99-2288 southwest of Helland Hansen (NGI. Storegga slide area: geological and geotechnical settings The Storegga slide complex is located to the south of the Vbring Plateau (Fig. 1998 and 2000) (Fig. plasticity index.4 38.. the weight of the solid (sediment) particles up to the seafloor. Soil shear strength A soil element below the seafloor is subjected to a total stress that depends on the weight of the water above the element.1. 6405/ 2 (532 m water depth). 2002). 2001).5 47. and INO3) correspond to marine clay units (Bryn et al.7 27. The effective stress affecting the sediment matrix corresponds to the total stress minus Table 1 Summary of soil conditions and the basic recommended soil parameters for borehole 6404/5-GB1/1A.2 1. southwest of the working area. 2003).6 w (%) 71.2 33. 2002).0 17.6 Ip (%) 35. 1). . Leynaud et al. The slip planes (corresponding to the seismic horizons TNS. The calculated slope angle inside the failed area.. Inasmuch as many parameters are involved in causing seabed instability. / Marine Geology 213 (2004) 457–480 459 the steeper part of the margin. TNR... undrained shear strength u (direct simple shear).3 3.. g.2 s DSS u (kPa) 3–80 80–130 130–260 260–325 325–580 580–700 c tot.2 1. over consolidation ratio..2 1. s DSS.

e. infinite slope analysis). Consequently. Infinite slope analysis (adapted from Nash. In absence of excess pore pressure. c is the total unit weight of the soil. Leynaud et al. Undrained loading conditions mean that the loading process is (relatively) rapid enough so that excess pore pressure does not have time to dissipate through the porous medium. / Marine Geology 213 (2004) 457–480 the effect of water pressure (Terzaghi and Peck. one considers the equilibrium of forces ..1 means 10% of g). As the undrained case is considered (u=0). log-spiral. the factor of safety F is defined as the shear strength divided by the mobilized shear stress. 1987) s su ð6Þ F¼ ¼ s cVzsinbcosb By including a uniform horizontal seismic acceleration one gets su F¼ ð7Þ ð cVzsinbcosbÞ þ ð KX czcos2 bÞ where K X is the maximum horizontal component of seismic acceleration in g (0. . (Nash. the equation becomes s ¼ su ¼ constant ð2Þ Inasmuch as the slope is considered infinite. Slope stability evaluation by limit equilibrium method In the limit equilibrium method. and uV is the internal friction angle. 1967). i. where cV is the cohesion of the soil. one can use the effective (submerged) unit weight of the soil (total unit weight minus the unit weight of water) to evaluate effective stresses. . circular. the soil volume above the assumed slip surface is divided into vertical slices. rV is the effective n normal stress. uV=0 ). For the undrained case (present study. the failure criterion becomes s=s u. ð4Þ ð5Þ where W is the slice weight (W=czb).).460 D. 2. The factor of safety (FOS) is defined as P P resisting forces resisting moments F¼ P % P loading forces loading moments available shear strength ð3Þ ¼ mobilized shear strength The simplest case occurs when a slab of soil is assumed to slide on a plane parallel to the ground surface (Fig. interslice forces are parallel to the slope and cancel each other out ( Q L=Q R) so one gets: P ¼ W cosb ¼ rl so W r ¼ cos2 b ðr ¼ vertical stressÞ b T ¼ W sinb ¼ sl so W s ¼ sinbcosb ðs ¼ mobilized shear stressÞ b where s u is the undrained shear strength. This approach is appropriate to model the Storegga slide using a back-analysis (translational slide from the observed slip planes). and cV is the effective unit weight of the soil. Failure occurs when the mobilized shear strength required for equilibrium exceeds the maximum shear strength available. etc.1. concave. As mentioned earlier. 3.2. For concave failure surfaces. one analyses the forces (or stresses) acting on an assumed failure surface (plane. The shear strength of the sediments typically follows the Mohr–Coulomb model where the shearing resistance s per unit of area is related to the effective normal stress acting on the soil at a specific depth as in the following equation: s ¼ cV þ rn ðuVÞ Vtan ð1Þ Fig. 1987). 2.

and slopes. The physical behaviour is modelled for the elements defining the structure.3. h is the depth of the base of the soil column. and the connection between the elements approximates the behaviour of the soil continuum. The ground acceleration giving a FOS of unity in the pseudostatic analysis is called the cut-off acceleration.D.2. and the permanent displacements accumulate during an earthquake in a stepwise fashion.1. some key questions regarding the stability of a clayey slope under a strong earthquake are the following: ! Will the soil mass experience essentially elastic deformations during the earthquake. the pseudostatic FOS is believed not to be a good measure of the seismic performance of the slope during an earthquake event. how large will be the permanent displacements? Will the earthquake loading trigger creep of the slope sediments. Leynaud et al.2. / Marine Geology 213 (2004) 457–480 461 applied to each slice to estimate the stability along the failure surface (method of slices) (Nash. for example. the inertial force caused by ground acceleration is applied as an effective static load equal to the mass of soil times the peak or the effective acceleration. 3. 3. For these geotechnical structures.3. Dynamic approach with finite elements In the dynamic approach. For a clay-rich submarine slope. Newmark (1965) presented the sliding block model for computing the seismically induced permanent displacements in dams. Newmark (1965) pointed out that a pseudostatic FOS having less than unity during the maximum-earthquake-induced load does not necessarily indicate a slope failure with large movement of soil masses. The pseudostatic analyses provide the factor of safety against slope instability as a function of the peak earthquake acceleration. In the latter. However. It is not physically possible to transmit accelerations higher than the cut off value to the soil mass because the shear strength is fully mobilized along a slip surface. embankments. The maximum earthquake loads only last a fraction of a second. smax. The maximum shear stress on a soil element (deformable body behaviour) subjected to a ground surface acceleration a max is. 3. If the slope is subjected to an earthquake with peak acceleration higher than the cut-off value. Finite element method (FEM) The concept of the FEM is to model an object with simple blocks or small elements. This procedure has been widely used for modelling strongly nonlinear geotechnical problems by using the elasto-plastic analyses (Griffiths and Lane. the earthquake-induced sliding occurs only in one direction. the consequence of full mobilization of soil shear strength along a slip surface is a limited permanent deformation. or will it completely loose its strength and develop into a slide? The latter situation would be the case if.2. leading to large shear displacements after the earthquake? Will the soil keep its shear strength after being exposed to the high cyclic shear stresses induced by the earthquake. Pseudostatic and dynamic undrained slope stability 3. ! ! 3. The seismic performance of the geotechnical structure under consideration is deemed acceptable if the earthquake-induced permanent deformation is less than the cut-off value to the soil mass. Hence. Pseudostatic approach The evaluation of slope stability under earthquake loading is commonly based on pseudostatic analyses. and. strains and stresses can thus be estimated at any location. liquefaction takes place. time) normalized . 1999). permanent deformations along the slip surface will take place.2. 1987). as long as the soil does not completely lose its shear strength. Earthquake-induced shear stress A simplified procedure for estimating the earthquake-induced stresses has been proposed by Seed and Idriss (1971).2. or will nonreversible plastic deformations occur? If plastic deformations do occur. and r d is a stress reduction coefficient with a value less than 1 (Fig. the seismic-accelerationinduced shear stress (total stress approach) is modelled through the FEM using a representative accelerogram (horizontal acceleration vs. 3).def ¼ ch amax rd g ð8Þ in which c is the unit weight of the soil.1.

2.4. In the absence of a seismic event record (related to large-magnitude seismic events) providing PGA for each seismic area. 3.2. In the Eurocode-8 regulations and for conventional buildings. 3. As seen previously. a 10. Liquefaction of sand Liquefaction occurs in sand when the vertical effective stress is cancelled by the pore pressure in excess. Leynaud et al.462 D.2. 3. This corresponds to a 475-year return period. deformations would be acceptably small for a FOS greater than 1. resulting in a liquid behaviour.000year return period in our case. one has to consider a more or less complete loss of strength in the soil to move the soil mass. Hynes-Griffin et al.000-year return period was used.or 10.3. 1971). PGA=0.02 s). The earthquake records used to model the seismic loading in the study area is the Friuli Tarcento earthquake (Ms=6.1. Sediment-loosening potential To initiate the slide on a very gentle slope (18 slope angle for the Storegga slide case study).08 g) with a duration of 10 s (sampling=0. The failure (FOS lower than 1.0) is not sufficient to explain the slide on such regional scale and slope angle (gravity forces insignificant on 18 slope angle) if permanent deformation is not large enough. ENV.0 (for dams). the sediment has no internal shear strength. 2000). providing a much easier way to remove the soil mass in the upper tens of meters of the sub-seabed. (1984) stated that using a seismic coefficient equal to one-half the PGA. For back-calculation.35 g) with a duration of 10 s (sampling=0. Different soil conditions could explain such liquefaction. 1998-1-1) to get an event providing the expected seismic accelerations (PGA). a PGA value with 90% probability of no exceedance during 50 years is required. A common way is to consider a pseudostatic acceleration related to the PGA. one will use onehalf of PGA to represent the pseudostatic seismic acceleration. PGA is commonly defined with a probability of no exceedance during a certain period of time. to the peak ground acceleration (PGA) value expected in the area for a specific return period (475. PGA=0. Determination of maximum shear stress (from Seed and Idriss. / Marine Geology 213 (2004) 457–480 Fig.02 s) and the El Centro earthquake (Ms=7. this could trigger some liquefaction. The . Subsequently. a high excess pore pressure (95%) could be expected in case a strong earthquake occurs. the use of PGA (lasting a fraction of second) overestimates the lower average effective acceleration acting on the soil during a longer period and leads to conservative results.3. Selection of a max The peak ground acceleration (PGA) represents the maximum acceleration experienced by a small particle of the soil during the course of the earthquake motion. Depending on clay content and liquid limit (Andrews and Martin. So. 3. In this case.

Mienert et al. Fig. Bunz and Mienert. 4. 4).L CRRL ¼ ¼ scyc CSR ð9Þ CRR and CSR are related to a representative number of cycles corresponding to the seismic event considered.D. Dissociation ¨ of gas hydrates can increase the excess pore pressure of the gas phase up to a critical value (percent of vertical effective stress) and thus may trigger an isotropic loss of strength in the soil. Extra sensitive clay Sensitivity is defined as the ratio of the peak shear strength to the residual strength of a material. / Marine Geology 213 (2004) 457–480 463 FOS against liquefaction can then be estimated from the cyclic stress approach developed by Seed and Idriss (1971): FSL ¼ cyclic shear stress required to cause liquefaction equivalent cyclic shear stress induced by earthquake scyc. one can observe another type of liquefaction which corresponds to the complete destruction of the delicate particles packing.3. 2004. Seismic profile NH9651-202 with borehole 6404/5-GB1/1A and sidewall locations. 4. Description of the probabilistic approach Contrary to the deterministic approach which uses one single constant value (mean value in general) for each parameter describing the soil behaviour. . 1998.. Melting of gas hydrates The presence of gas-hydrate-bearing sediments in the study area is inferred from the observation of a well-defined bottom-simulating reflection (BSR) on seismic profiles (Stoll and Bryan.3. 1983. Fig.2. In some circumstances. 1979) from the NE flank of the slide scar (Bugge. when the sensitivity of the clay is very high (sensitivity higher than 30). 3. the probabilistic approach considers their variability and define them using a mean and a standard deviation value.3. Leynaud et al. 3.

often referred to as Monte Carlo or Stochastic simulation. and U is the standard normal distribution function.Q The design point is found by optimisation techniques. Leynaud et al. A limit state function g(x) or performance function is defined such that g(X)z0 when the domain is stable and g(X)b0 when the domain has failed.U 2) of independent Gaussian variables with zero mean and unit standard deviation using Rosenblatt’s (1952) transformation. . and modelling uncertainty). respectively) are transformation of arbitrary random uncertainty vectors into independent. Example of Monte Carlo simulation outputs. and probability distribution of the performance function is derived from the calculated values (Fig. First. The FORM approximation (Fig. Considering numerous trials. and the hyperplane g(U)=0. Let X represents a vector of random variables (such as soil properties.2. Fig. This is the most likely bfailureQ point and is referred to as the bdesign point. the expected mean value. 6) is done in two steps: (1) The vector of basic random variables X=(X 1. 5). 5. This limit state function can be represented by a function defining the FOS less unity.Q ai is the direction cosine of random variable Ui (transform of X i). The vector of the direction cosines of the random variables (ai) is called the vector of sensitivity factors. n is the number of basic random variables X. the performance function defining the boundary between safe and unsafe (failure) domains. the probability of failure P f is: ! n X Pf % Pð gðU Þb0Þ ¼ P ai U i À bb0 ¼ UðÀ bÞ i¼1 ð10Þ where P(. . / Marine Geology 213 (2004) 457–480 4. b is the distance between the origin. load effects.and second-order reliability methods The basic concepts of first. Random variable values are generated according to their probability distribution. (2) In the transformed space. The subroutines devel- oped by Gollwitzer (1999) were used for the FORM and SORM approximation.X 2) is transformed into a vector U=(U 1. standard normal vectors and approximation of the failure boundary in a certain point of this area.and second-order reliability methods (FORM and SORM. .) reads bthe probability that. geometry parameters. Monte Carlo simulation The expected mean value and standard deviation of the performance function is estimated using a simulation method. standard deviation. 4. and the performance function is calculated for each set generated. The (transformed) limit state function is linearised at the point of maximum probability density. typically several thousands. and the distance b is called the reliability index.1.464 D.

2. Typically. one corrects the pseudostatic acceleration value to get a similar earthquake-induced effect according to the following relationship: corrected PSA ¼ average total unit weight  PSA average effective unit weight ð12Þ 5. Slope/W (limit equilibrium) where PSA is the pseudostatic seismic acceleration. 1977) was used in our evaluation. Basic principles for FORM method for two random variables. The expected value of e is 1. Quake/W (Geo-slope. Leynaud et al.0 for an unbiased analysis method. which sum is equal to unity. to the user to solve the interslice force indetermination. A large and complete set of methods of slices is available This product is a geotechnical finite element software for the dynamic analysis of earth structures . As the earthquake acceleration affects the total soil mass (including water). F acceptable is unity. / Marine Geology 213 (2004) 457–480 465 Fig. for conventional static loading. The performance function for the slope stability was defined as: g ¼ eF À Facceptable ð11Þ where F is the computed safety factor. Softwares 5. quantifies in i a relative manner the contribution of the uncertainty in each random variable X i to the total uncertainty. The general limit equilibrium method (Fredlund and Krahn. 2001) uses the limit equilibrium theory to compute the FOS of earth slopes. The random variable e is introduced to account for the modelling uncertainty. 5.D. Once the geometry is defined. 6.1. 2001) Slope/W (Geo-slope. This is done to avoid computation problems related to the water mass. The square of the direction cosines or sensitivity factors (a2). one applies for each layer an average effective unit weight and undrained shear strength representative of the soil profile (static). F acceptable is the acceptable minimum safety factor.

000 165.000 290.000 56. 7. damping ratio has been fixed to 2%.466 D. Ln: sediment layers with different mechanical properties.and second-order reliability methods (FORM and SORM). 1999). This implies that permanent movement of the sliding mass occurs during specific time period when the FOS falls below 1. Fig. The first improvement concerns the geometry of the two-wedge model. / Marine Geology 213 (2004) 457–480 Table 2 Elastic (linear) parameters used for the dynamic approach (Quake/W) Depth (m) 0–20 20–30 30–40 40–70 70–90 90–110 110–140 140–360 E modulus (kPa) 9000 15. The angles a and b are defined on Fig. 1992). W 1 and W 2 are the submerged weights of wedge 1 and wedge 2. this model has been improved (Nadim et al. Sigma/W (Geo-slope. 1965). 2003). X 1 and X 2 correspond to the forces (respectively on wedges 1 and 2) due to the seismic acceleration. A closed-form solution for FOS can be obtained by considering the equilibrium of the two wedges and assuming the same safety factor F on all slip planes: F¼ max max max S1 cosðb À aÞ þ S2 sinðb À aÞ þ S3 þ Pcosa W1 sinbcosðb À aÞ þ W2 sina þ X1 cosbcosðb À aÞ þ ðX2 À PÞcosa Poisson’s ratio corresponding to constant volume (undrained conditions). 7.3.4.5) considering undrained conditions (no volume change during loading). Recently. S 2 . According to the variation of damping ratio of fine-grained soil with cyclic shear strain amplitude and plasticity index (Kramer. and thus one has to consider a specific model to define the FOS. No graphic interface is provided. Li. 7). Total stresses computed with finite elements (static and dynamic) can be used in Slope/W to get a more accurate evaluation. 2001) The Sigma/W software for finite element stress and deformation analysis was used to estimate the FOS using the shear strength reduction technique (Matsui and San. Modified two-wedge model used with STRUREL software (Nadim et al. We considered a linear elastic model (Table 2) with a Young’s modulus E U estimated from the following relationship (Duncan and Buchignani.000 126. The deformation analysis of the soil subjected to earthquake shaking uses a Newmark-type procedure (Newmark. Fig. and S 3 and are the maximum shear resistance along the sliding planes shown on Fig. 5.0 (when the average acceleration becomes larger than the yield acceleration). L1. STRUREL (limit equilibrium method—two-wedge model) Reliability analyses were performed using the STRUREL software (Gollwitzer.5 5.. This analysis is based on first. 7). 1.. The second one is where a m=600 (NGI.000 91. Two-dimensional elements with a quadrangular mesh were used (40Â40 and 20Â60 m or more depending on the model. . 2003).000 33. The Poisson’s ratio m is constant (m=0. Leynaud et al. L2. one considers the possibility to have one failure coming out within the slope (this corresponding to a negative a parameter. In this new model. and DSS s u is the undrained shear strength in direct simple shear. 1995). under earthquake loading and was used for our dynamic modelling. 1976): EU ¼ am sDSS u ð13Þ ð14Þ max max max where S 1 . and P is the passive lateral resistance at the toe of the slope.000 Poisson’s ratio (v) 0. A two-wedge slope stability model was developed to simulate a sliding block (wedge 2) with a collapsing mechanism (wedge 1) (Fig. 1998) (empirical parameter).

which are corrected for overconsolidation ratio (OCR) and effective vertical stress according to the Stress History and Normalized Soil Engineering Properties (SHANSEP) relationship.2. 8. Input parameters for modelling 6. 1997) provides the main horizon locations in term of depth (m) and two-way time (ms). pV 0 is the vertical effective stress assuming hydrostatic Às Á conditions. First. The northern sidewall slope angle is found to be around 58 from both the seismic profile and the highresolution bathymetric map (Fig. one can estimate the horizon location along the seismic line. has been used in this work (Ladd and Foott. . This approach allows defining more realistic s u values from overconsolidation ratio (OCR) and effective stress conditions. one obtains a geometrical model for the ð16Þ where s u is the reference undrained shear strength. Thus. OCR is the overconsolidation Fig. The s u values have been reestimated near the sidewall. the following correction is applied to the s u values for computing the shear resistance: sub ¼ sDSS ð1 þ ðj À 1Þsin2bÞ u ð15Þ soil in the vicinity of the sidewall (Fig. The geotechnical report (NGI. One can also estimate the s u reduction related to excess pore pressure generation. 1 and 4). Interpreted horizons from seismic line NH9651-202. 9). Undrained shear strength—SHANSEP model The Stress History and Normalized Soil Engineering Properties (SHANSEP) model. several horizons have been defined to fit the layers reported in the borehole 6404-5/GB1. This model has been used with the limit equilibrium method (twowedge and circular failure models). One define the undrained shear strength in each layer by the following SHANSEP-type equation: su ¼ aðp0 À DuÞOCRm V   p0 þ Dp À Du m V ¼ aðp0 À DuÞ V p0 À Du V where b is the angle of the failure plane with the horizon. 6. The program works in two steps. and then (second step) it calculates the probability of failure around these values (with reliability index). The latest improvement concerns the strength anisotropy which is taken into account C DSS through a global anisotropy factor j=S u /S u (defined for the entire soil profile). Leynaud et al. a ¼ rVu NC = ratio of undrained shear ac strength to vertical effective stress for normally consolidated clay.1. according to the SHANSEP approach. The average preslide slope angle (for the back-analysis) was considered to be 18. 6. 1974). Assuming a laterally constant P wave velocity distribution in each layer. Thus. / Marine Geology 213 (2004) 457–480 467 related to the undrained shear strength values. which includes the effects of overconsolidation and vertical effective stress. Soil geometry Using the NH9651_202 seismic profile (Figs.D. it looks for the critical geometry (criticals length a and angle b providing the lowest safety factor). 8 and Table 3).

and Du is the excess pore pressure (assumed to be due to incomplete consolidation). m is a dimensionless c 0 exponent (typically between 0. ratio=PV/( pV ÀDu). / Marine Geology 213 (2004) 457–480 Horizons TWT (ms) 1311–1330 1330–1350 1350–1372 1372–1395 1395–1468 1468–1506 1506–1528 1528–1575 1575–1606 1606–1651 Estimated Vp (m/s) 1579 1500 1636 1478 1589 1526 1636 1574 1742 1600 Estimated depth at the sidewall (m) 0–? ?–16 16–26 26–33 33–61 61–77 77–85 85–93 . . Norsk Hydro). DSS A comparison of SHANSEP modelling. .3. pV is the maximum past consolc 0 c idation stress=pV +DpÀDu. Weichselian Late Saalian Saalian Saalian Saalian Saalian Seabed–BNA0 BNA0–BNA1 BNA1–BNA2 BNA2–BNA3 BNA3–BNA4 BNA4–BNA5 BNA5–BNA6 BNA6–TNC1 TNC1–TNC TNC–BSR Model estimated at the sidewall. 6.and 10. Leynaud et al.468 Table 3 Summary of depths for seismic units Unit Depth (m) 0–15 15–30 30–48 48–65 65–123 123–152 152–170 170–207 207–234 234–270 Period D.90). –285 HH0 HH1 HH2 HH3 HH4 HH5 HH6 HH7 HH8 HH9 Holocene Middle Weichselian M. Eemian/E. Peak ground acceleration (PGA) in the Helland Hansen area PGA values considered for 475. 10.W. These correspond to the maximum value Fig. 9. Table 3). .65 and 0. This confirms a good estimate for the undrained shear strength profile at the sidewall location concerning the first 20 m depth.1–? ?–29 29–46 46–58 58–122 122–153 153–171 171–189 .. Slope angle around the seismic line area... The long sediment core MD99-2288 (NGI.W. M. ?–134 Estimated s DSS u at the sidewall (kPa) 2. s u C (direct simple shear) and s u (CAUC triaxal test) profiles is shown on Fig.25. 2000) located near the sidewall provides some undrained DSS shear strength in the direct simple shear mode (s u ) values ranging from 35 to 37 kPa at 18–19 m depth (s u estimated around 31–33 kPa from SHANSEP relationship using a=0.. Dp is the removed over0 burden stress=pVÀpV +Du. SURFER software and high-resolution bathymetric data (200Â200 m. .000-year return periods in the Helland Hansen area are shown in Table 4.

5 second). Using the seismic activity parameters from NORSAR (1998) and a public software (SEISRISKIII.5 The Friuli seismic event contains mainly high frequencies (periods lower than 0. Ms=6. / Marine Geology 213 (2004) 457–480 469 Fig.35 Pseudostatic acceleration ( g) 0. Ms=5. Ms=5.3/6.1. 1895.175 it is relevant to consider accelerograms corresponding to large. Historical seismic activity (for magnitudes above 4.0 (65.3). 6.0 (65.4) on the Norwegian continental margin (from NORSAR.000-year return period offshore Norway (NORSAR.4. 1987).0).4.5). Ms=5. 1998) with outlines of the major slides. It follows that the maximum PGA value offshore Norway occurs in the vicinity of the Storegga slide headwall.44).1/6.05 0.as well as medium-magnitude events in our study.and 10.68/2. May 6.10 0.D. 1976. Therefore.2/6.0/6. Leynaud et al. Largest seismic events in the vicinity of the Storegga slide: 1866. Ms=5. SHANSEP modelling compared to s DSS and s C values. 1988. the frequency content of the ground motion can be quite different because highmagnitude events contain more energy in low frequencies and have longer duration. Bender and Perkins. Earthquake records It is well known that the same value of peak ground acceleration (PGA) can be due to different combinations of magnitude and distance.0 (64. 1998) 90% probability of no exceedance during (years) 50 1000 Return period (years) 475 (present-day) 10. 1913. Ms=5.3 (65. The peak Fig. 6. 10. Fig. we made a simplified seismic hazard map for the area in the vicinity of the Storegga slide area (Fig.7 (Lat/Lon: 65. with 90% probability of no exceedance (NORSAR. . 12) to get an outline of the PGA potential spatial distribution in this area.2/6.0). 1969. 1998). In spite of the same PGA.000 (backcalculation) PGA ( g) 0.5). Friuli earthquake.3 (63. Ms=5. u u offshore Norway. 1958. 11. 11 shows the historical seismic activity on the Norwegian continental margin. Table 4 Maximum peak ground acceleration with 90% probability of no exceedance for 475.

Leynaud et al. Simplified seismic hazard map in the vicinity of the Storegga slide area for a 10. 14).4.3. GMT software (Wessel and Smith. 13. and a maximum spectral acceleration of 0. El Centro (Imperial Valley) earthquake. and a maximum spectral acceleration of 0. Site effect (Friuli earthquake) Using AMPLE2000 software (Pestana and Nadim.25 s (Fig. The isotropic linear elastic model has been used with the initial shear Fig. 1988).80 g is reached for 0. . Ms=7. 15). May 18. Ambraseys’ attenuation relationship (1995. 2000).5 Friuli earthquake (damping=5%). ground acceleration (PGA) is 0.6-s periods (Fig. 13). 1940.25.25 and 1.1 g on the bedrock (Fig. 6. Grid smoothed using a Gauss filter/200 km. / Marine Geology 213 (2004) 457–480 Fig.082 g.2 The El Centro seismic event contains low and high frequencies (periods between 0. The PGA is equal to 0.35 g.4.470 D.and 0. Acceleration time history (A) and spectral acceleration for the Ms 6. Seismic zonation from NORSAR.2.0 s). 50th percentile curve). 6. one have estimated the 1-D seismic response of the site assuming a Friuli-type earthquake normalized to 0. 12.30 g is reached for 0.000-year return period (PGA value with 90% probability of no exceedance during 1000 years).

Excess pore pressure generation The earthquake-induced excess pore pressure can play a major role in the decrease of shearing resistance and thus in the sliding mechanism. m=0. 1997). Earlier site response studies performed for the Helland Hansen area indicate that this reduction could be a factor of two or more (NGI. 2000) has been used to simulate the 1-D site soil response and excess pore pressure generation under seismic loading. . It should be noted that soil response effects will filter out the high frequencies and reduce the seabed accelerations significantly with respect to the bedrock accelerations given in Table 4. large Once the site effect of the submarine soil (accelerogram on soft soil equivalent seafloor) has been estimated (elastic behaviour).35. AMPLE2000 software (Pestana and Nadim. Fig. 7. Acceleration time history (A) and spectral acceleration for the Ms 7. slenderness parameter. The simple direct simple shear (DSS) model for lightly consolidated soils (Pestana and Biscontin.D.5% damping. 1998): Gmax ½ zŠ ¼ 50 þ 1620*z with Gmax ¼ shear force=unit area ðkPaÞ and angular deformation z ¼ depthðmeterÞ ð17Þ The dynamic approach has been performed using these seismic events to observe the effects of the magnitude (frequency content and spectral accelerations) on the shear stress and excess pore pressure generation. 2000) is considered to estimate the amount of excess pore pressure developing with cyclic loading.2 El Centro earthquake (damping=5%). b=0. 1. 15. The simple DSS model parameters for the Helland Hansen area were provided within AMPLE2000 software package (failure ratio. Friuli earthquake. 14. we use it as input seismic event in the finite element software (Quake/ W) to model a more realistic dynamic approach with an accelerogram which models the seismic response of the soft soil. / Marine Geology 213 (2004) 457–480 471 Fig. stiffness at small strains G max defined as follows (NGI.5. Accelerogram and Spectral acceleration estimated at the seafloor level (red line) using an elastic model (AMPLE2000 software) compared to the accelerogram recorded on the bedrock (black line). Leynaud et al.

10 g (slope angle: 58) PGA=0. Table 6 Earthquake-induced excess pore pressure ratio estimated using AMPLE2000 for 0.23bamax) from Seed et al. One observes a large difference for the excess pore pressure generation (Tables 5 and 6). Two earthquake records have been used to model the seismic loading: the Ms 6. suggesting a very stable situation. Slope stability assessment of the Helland Hansen area 8. strain obliquity angle. 0. G p=10. with the layer thicknesses and the undrained shear strength profile estimated at the headwall.5 m 0.1.472 D. (2001). depending on the frequency content of the seismic acceleration.35 g (slope angle: 18) 7. considering rd mean plus one standard deviation.5% 82.35 g (slope angle: 18) 7. It was used considering the uncertainty of the pseudostatic approach (using the half of the PGA). .175-g pseudostatic acceleration depending on depth and shear stress reduction factor curves. expected peak acceleration). one gets a FOS related to the depth (Fig. Excess pore pressure values. are shown in Tables 5 and 6. PGA=0. It is noticed that the excess pore pressure generation is higher with a 0. Infinite slope analysis The infinite slope analysis.0% 9.5 m 3% 12% 82.5magnitude seismic event.5 m 0.35 g PGA high-frequency event (Table 5).5 m 0. Back-calculation of the slide 8.000-year return period) in percent of the initial vertical effective stress Depth PGA=0. a max. the curves show that displacements are still acceptably small.175-g pseudostatic seismic acceleration (half of the PGA value) and the nonlinear shear mass participation factor (Mwz6.0 for the static case (Table 8).5 (k=0.25% 3.5 m 8% 20% 42. backbone curve parameter.5 Friuli earthquake (high-frequency content) and the Ms 7. Considering a 0.22% 3.1 El Centro earthquake (low frequency content). k=30). Ms 7.5 m 5% 15% 62.35 g PGA (10.8.2 El Centro earthquake (5% damping). As the undrained shear strength remains constant with excess pore pressure generation (after complete consolidation).10 g PGA (475-year return period) and 0.5% damping).35 g PGA (10.10 g (slope angle: 58) PGA=0.2% 8.5 Friuli Tarcento earthquake (1.1. W=28. Leynaud et al. Furthermore. normalized with respect to the vertical effective stress.35% 5. Makdisi and Seed (1978) proposed some displacements computed by Newmark’s method as a function of the acceleration ratio ((k)/(a max)) (k. Safety factor (infinite slope analysis) with 0. For an acceleration ratio around 0.5 m 12% 40% 22.30% 4.2% 22. / Marine Geology 213 (2004) 457–480 Table 5 Earthquake-induced excess pore pressure ratio estimated using AMPLE2000 for 0. 16).1.5 m 0.175. one has to consider the drained case to find out whether or not this latter is more critical than the former one (whether drained shear strength is lower than undrained shear strength).5 m 1. seismic coefficient reducing the safety factor to unity.5 m 3% 12% Fig.000-year return period) in percent of the initial vertical effective stress Depth PGA=0.10 g PGA (475-year return period) and 0. Considering that the equivalent depth to bedrock is much higher in the Ms 6.35 g) and for a 7. The excess pore pressure generation estimated with the El Centro earthquake (low frequencies) will be used in the present-day slope stability assessment to consider a low-frequency generation. provides an FOS around 14.1 g PGA low-frequency seismic event (Table 6) than with a 0. 16.2% 62.2% 42. One have to notice that these curves correspond to embankments with a height ranging from 15 to 75 m. The FOS is above unity using the mean value of rd but remains lower than unity down to approximately 50 m depth. h=25.

this approach underestimates displacements and thus overestimates the stability in our area (FOS). one observe that failure occurs for s cyc/ DSS s u =0. the Makdisi and Seed’s approach does not consider any large displacement in the shallow part of the seafloor.995. same depth and cycles number) at 10.5 m depth: s u=18 kPa. r n is the effective normal stress.1 g. considering 10 cycles for the maximum seismic event (Ms=7. we use the following relationship: s ¼ c V þ rV tanðuV Þ n ð18Þ Fig.35 g at 10. permanent displacements calculated on the failure surface when the critical acceleration is reached (FOS lower than 1. 17.13 with a max=0. 10. / Marine Geology 213 (2004) 457–480 473 Helland Hansen area. at 80% excess pore pressure). Drained conditions As the earthquake-induced excess pore pressure generation affects drained shear strength. one gets a cyclic shear stress value for the failure. depth for drained (0%.5 m depth). CSR ¼ 0:65 amax rvo rd g rV vo where c is the cohesion of the soil (7 to 10 kPa. 18). It is estimated that an excess pore pressure generation exceeding 50% of the vertical effective stress is necessary to reach more critical conditions in the area compared to undrained case. decrease to CSR=0.47 with a max=0.0 earthquake) (Fig. To do that.0) remain small between 20 and 30 m depth. 17). Slope/W returns a critical dynamic FOS around 0.9 (down to 20–30 m depth) (Table 8). 2000). Thus. borehole 6405/2). Using c and / parameters (NGI. Previous results show that excess pore pressure generation modelling remains below 50% even with the more critical seismic event considered (Tables 5 and 6). V and u is the internal friction angle (28 to 29. 25%.5 m depth. one shows that 80% excess pore pressure can be reached down to 30 m depth with a 0. one needs to check the stability using the drained conditions to test the more critical case which could lead to failure. it is not possible to evaluate the safety factor against liquefaction (corresponding to 95% excess pore pressure) but only a safety factor against 80% excess pore pressure generation or failure.35 g PGA seismic acceleration generated by an ElCentro-type earthquake. With the present values.35 g PGA acceleration and a 10-cycle seismic event (Ms 7. borehole 6405/2). r vo =57 kPa V scyc ðCRRÞfield ¼ V ¼ 0:9ðCRRÞSS rvo ðCRRÞSS ¼ 0:31 so ðCRRÞfield ¼ 0:28 FS80% EPP ¼ FScyclic failure ¼ ¼ 0:6 DSS Assuming the same cyclic ratio [(s cyc/s u )=0.1.0). 10 cycles. Slope/W–Quake/W softwares The finite element approach (Quake/W) for the dynamic loading provides a stress state in the soil with a 0. However. Cyclic stress approach Using the cyclic DSS test performed on the borehole 6405/2-GB1 (Unit 2.1.1. 8. 8. 8.995 (failure occurring before any liquefaction CSR=0.3. Leynaud et al.D. The ðCRRÞfield 0:28 ¼ 0:47 CSR .5 m depth] at a greater depth (similar clay content down to 270 m depth: 34–43%).2. and 50% excess pore pressure) and undrained conditions. Using a computed timevarying shear stress.4.5 kPa. Shear strength profiles vs. we find that the drained shear strength is still less critical than the undrained one (Fig. 10.

pressure. the factor of safety against 80% excess pore pressure (or failure) remains below unity down to 80 m depth.. and shear wave velocity (mean value).23ba max) (Seed et al.6. 2004a). 2001). / Marine Geology 213 (2004) 457–480 Fig. 10. Sensitive clay No high sensitive clay has been observed in the boreholes 6404/5-GB1/1A [sensitivity=4. (2001) has been used for this estimate. 19) and the sensitivity at this depth (sensitivity=4. . Gas hydrate dissociation Sultan et al. suggesting that the liquefaction potential could be much higher at this location and depth (Sultan et al. 8. including the shear stress reduction factor with depth. As the uncertainty concerning the stress reduction factor is very large at such depth (100 m depth). horizontal PGA. an uncertainty in the spatial variability of the sensitivity has to be considered. with approximately a 30-kPa excess pore pressure generation (4% of the vertical effective stress) at 100 m depth below seafloor. here. Assuming that gas hydrates dissociation results in a reduction of the undrained shear strength corresponding to the drop from the peak to the residual strength values.6 or lower..1. Leynaud et al.1. 2004b). However. a high sand content has been found at 20 m depth (borehole MD99-2288) with plasticity index Ip=20 and liquid limit WL=35. assuming a constant cyclic behaviour of the soil for the first 100 m depth.000 years BP (from Sultan et al. 0. depth for different peak ground seismic accelerations (a max). Excess pore pressure generation estimated from gas hydrates dissociation. EOB is 124 m depth below seabed). 8. the shear strength should drop to a residual value which is around 44 kPa.0 or lower. although failure is very unlikely to occur. Considering that the main hydrates dissociation took place between 80 and 120 m depth below seafloor (Fig. borehole 6404/5. The limit equilibrium method (Slope/W) combined to the finite element stress model (using Quake/W software) computes only small permanent displace- Fig. 18.474 D. The FOS is then estimated using the infinite slope analysis (Table 7). end of Boring (EOB) is 309 m below seabed] or 6405/ 2-GB1 (sensitivity=4.6.8. taking into account the magnitude. Safety factor against 80% excess pore pressure (and failure) vs.. (2004b) proposed a model to explain some retrogressive failures from gas hydrates dissociation at the top of the gas hydrate stability zone (GHSZ) related to the postglacial temperature. Using a reduction factor corresponding to the mean value plus one standard deviation (M wz6. In addition. reduction factor proposed by Seed et al.5. the FOS could be obviously drastically reduced. 19. Tables 1 and 3). one cannot efficiently propose any FOS with a high accuracy. and gas solubility changes.

2. 9).2. and a sketch shows the results in Fig.0) down to 30 m depth Infinite slope Failure very unlikely at 100 m depth +0.45 0. A summary of present-day FOS is shown on Fig. a SHANSEP=0.35 g PGA seismic acceleration (El-Centrotype earthquake) considering residual shear strength at 100 m depth from gas hydrate dissociation (s u reduced to 44 kPa).2. The previous modelling results are summarized in Table 8.0) at 100 m depth . and the s u profile estimated at the sidewall. Figs. Leynaud et al.0 s for both cases). 20. One have to notice that the critical failure surface does not correspond to the time of the PGA (2. Present-day slope stability The slope stability assessment is conducted for the north sidewall where the steepest slope observed on the seismic line NH9651-202 is around 58. 0.25) and 1. particularly for the shear strength distribution in the soil around the sidewall (s u deduced from OCR distribution). this corresponds to a more realistic modelling. the limit equilibrium (Slope/W) provides a critical pseudostatic FOS around 1. PSA=0. 4).11-g corrected pseudostatic seismic acceleration (Eq.D.21) (Table 11).0 (El-Centro-type earthquake) (Table 10).1 1. The El-Centro-type seismic event remains the most critical one regarding the dynamic FOS. the soil geometry (thicknesses).7 using the limit equilibrium method (Slope/W) and 4. 21 and 22 show the most critical failure surfaces (with critical FOS) depending on the length of the failure surface (ElCentro-type earthquake. a SHANSEP=0.175 g PGA Method/ software Factor of safety Infinite slope Higher than 10 Infinite slope Failure (FSb1.9 (circular failure).35 g PGA) Cyclic stress approach (0.9 for the shear strength reduction method (Sigma/W). The highest slope angle (around 58) observed for the sidewall (Fig.21) similar to the dynamic one (FOS=2. Fig. and gives a FOS around 2.0) down to 30 m depth Gas hydrate dissociation (s u=44 kPa at 100 m depth) +0. ments (1.25). 100 m depth) Factor of safety (pseudostatic acc=0.0 E–03 m) on a failure plane at 100 m depth with a 0. 8. The modified two-wedge model gives a lower FOS in the range 1. 8.37–1.1 s for the El Centro accelerogram) but later on (after 5.9) (Table 9).175 g) and the residual shear strength at 100 m depth (44 kPa) corresponding to gas hydrate dissociation for different values of the shear strength reduction factor with depth (100 m) Shear strength reduction factor (r d. including the reduction factor accounting for the shear stress reduction with depth. (12).0) down to 50 m depth Slope/W+Quake/W Failure (FSb1.35 g PGA) Seed and Idriss (1971) Failure (FSb1. The twowedge model allows us to get a pseudostatic approach.05 0.1. Norsk Hydro. The slope angle in the vicinity of the seismic profile has been checked on using a high-resolution bathymetric map (200Â200 m.0875 g.63–1.05 g). 23.175 g Â0. The FE dynamic modelling using Quake/W software (stress distribution with time) and different accelerograms (0. / Marine Geology 213 (2004) 457–480 Table 7 Factor of safety estimated using the infinite slope analysis (pseudostatic approach. Static case The Geoslope products provide similar FOS: 4. Table 8 Factors of safety for 18 slope angle (horizontal component for earthquake loading) Case Static Pseudostatic (0.1 (Friuli-type earthquake) to 3.175 g slope angle: 18) 0 0.05 means that equivalent acceleration for estimating induced shear stress at 100 m depth is 0. 3.67 (a SHANSEP=0.1 g PGA) combined to the limit equilibrium method (Slope/W software) gives some FOS ranging from 4. 9) confirms the value obtained from the seismic line (Fig.4 (a SHANSEP=0.35 g PGA Slope/W+Quake/W No failure (FOSN1.3.1 475 The two-wedge model (new geometry+OCR correction+strength anisotropy) gives a lower FOS (3.2.21) (Table 9).95 r d=0. 8.175 g PGA) Dynamic approach (0. Pseudostatic and dynamic cases Using a 0.26 (a SHANSEP=0.05=0.

This illustrates that strength degradation. The input acceleration below the critical one can trigger permanent deformation after few cycles because of the significant reduction in undrained shear strength.21). Friuli (0.1 g) 3. the effect of strength degradation with cyclic loading (Cascone et al. The degradation index which represents the ratio of the undrained shear strength after cyclic loading to the initial one prior to cyclic loading depends not only on the number of cycles but also the cyclic stress ratio. If the average acceleration is larger than the yield acceleration (leading to a FOS below 1. the sliding depends on the permanent deformation occurring on a failure surface.0 (5.9 4.2 g) 3.21) FOS g FOS g 4. and cycles frequency. 9. For both cases. and the overconsolidation ratio.7 – STRUREL (new geometry) 3.05 a S=0.1 g) 4.40 – 1. / Marine Geology 213 (2004) 457–480 Fig. Zhou and Gong (2001) propose a model for strain degradation of saturated clay under cyclic loading with parameters such as OCR.6 s) Static FOS Pseudostatic (acc=0.67 – 1.63 plane.8 s) El Centro (0. cyclic stress ratio. Using the Monte Carlo simulation (using Slope/W software) and FORM/SORM methods (STRUREL.4 (6.37 .2 1. 20. one gets some probability of failure for the present-day sidewall (Table 12). the cumulative permanent displacement would be insufficient to move the soil mass on the sliding Table 9 Factors of safety for present-day profile Software Slope/W STRUREL Sigma/w (SSRM) 4.1 Friuli+Site Effect (0.9 1. resulting in a situation much more prone to sliding after the failure occurs..05 a S=0. They show that Table 10 Factors of safety for present-day profile (dynamic approach) Earthquake type Dynamic FOS (Slope/W+ QUAKE/W) a SHANSEP=0.25) Pseudostatic (acc=0.0) during few time steps. Discussion For the back-analysis and considering the low slope angle.476 D. Leynaud et al. 1998) was not considered. the type of soil (plasticity index Ip).9 1. as expected. Maximum failure depth modelling for different approaches (failure but no large permanent deformation expected from finite element approach). This would be much easier with the present-day profile which exhibits a larger slope angle and consequently an increased downslope gravity component. two-wedge model) for the pseudostatic case. Table 13 shows failure probabilities (FORM/SORM with two-wedge model) using the most critical undrained shear strength profile (a SHANSEP=0. will affect the development of slope displacement.25.

175 g) using the infinite slope analysis explains a failure down to (2) Fig. Vertical exaggeration=2:1. 1995). Obviously.7 Halten Terrace earthquake). Leynaud et al. higher PGA. Vertical exageration=2:1. 0.0 at 5. Additionally. 0. one needs some other factors to increase the thickness of the sliding sediments (Hydrates.35 g PGA) with one of the largest historical events (the 09. FS=3.1 g PGA). one can expect higher PGA values in the vicinity of a magnitude Ms 7. one could also explain a failure down to 30 m depth which is far from the 110 m height of the sidewall. 21. Critical failure surface obtained with the dynamic approach (El Centro earthquake. 10.6 s. low frequencies have a higher effect on degradation. Present-day slope stability assessment. This confirms a high potential of seismic activity in terms of large events (NORSAR. static case) shows that the continental slope (18 slope angle) is very stable if subjected only to gravity loading. investigations should focus on degradation of marine sediment with cyclic loading. Present-day slope stability assessment. using one cyclic DSS test. . However. .D.43 g PGA for a 10-km depth event).0 seismic event (0. The highest expected PGA offshore the midNorwegian margin is in the vicinity of the Storegga slide area (0.0. Vucetic and Dobry (1988) show that low plasticity clays are more prone to a higher degradation for a given cyclic strain.03. Finally. but a spatial variability has to be taken into account on a such slide scale.9 at 5.1 g PGA). showing that monotonic strength at the end of an earthquake shaking is related to cyclic strain amplitudes. For more accurate slope stability evaluations in this area. Kramer (1995) notes that cyclic strength of a cohesive soil element depends on the combination of average shear stress and cyclic shear stress. FS=3. .1866 Ms 5.35 g PGA). From the Ambraseys attenuation relationship (Ambraseys. etc. 1998).). Conclusions Back-calculation of the slide: (1) A high FOS (FOSN10. / Marine Geology 213 (2004) 457–480 477 Fig. one needs to assume a failure process along with liquefaction (complete loss of strength in the sediment) to explain the mass movement on a gentle slope (18 slope angle). 22.3 s. Longest critical failure surface (fully specified) obtained with the dynamic approach (El Centro earthquake. This is not in agreement with the nature of the soil (no extrasensitive clay and no liquefaction potential from cyclic DSS test for 0. The site effect (time history acceleration on the seafloor or soft soil) could be high enough to initiate some failures at a greater depth. The pseudostatic approach (0.

23.4 E–09% (FORM) 1. The earthquake-induced shear stress could initiate some failures with large displacements (meter) in the first 20 m depth during the first 10 s of an ElCentro-type earthquake. Leynaud et al.21.0) with 0. Including the effect of gas hydrate dissociation (Sultan et al.0 between 0 and 50 m depth).05 g)+r d 0.4% – FORM/SORM (SHANSEP+ new geometry) b=188/a=À337 m 3. .75% 1.1 g PGA) Quake/W+Slope/W 2.26 (3) 50 m depth (FOSb1.2 E–16 0.1 E–09% (SORM) Monte Carlo Table 11 Factors of safety for present-day profile (dynamic approach) Approach Softwares FOS a SHANSEP=0.. infinite slope analysis).0 s) Pseudostatic (0. 2004b).35 g PGA seismic event.23% – Present-day slope stability.05 g) Pseudostatic (0.2 E–15 0.05 g)+(r d ) Two-wedge model (modified) 2.10 with ((s u)/(sensitivity)) at 100 m depth) and reach a critical FOS (FOS below 1. Furthermore. / Marine Geology 213 (2004) 457–480 Table 12 Failure probabilities with FORM and SORM and Monte Carlo simulation methods Method FORM (SHANSEP) 418/a=0 1. Dynamic+El Centro (0. but with only small displacements (decimeter scale).1 to 100 m depth.30 (6. The finite element stress distribution model a critical state at 30 m depth (failure) with a 0.25. Summary of safety factors vs. a SHANSEP=0.175 g pseudostatic acceleration (r d=0.21 a SHANSEP=0. the uncertainty related to the shear Fig. One has to consider the residual shear strength value (assuming remoulding of sediment) to reduce drastically the initial static safety factor (static FOS=3. a 30-kPa excess pore Static Pseudostatic (0. However.478 D. methods for present-day evaluation. pressure generation (4% of the vertical effective stress) is not sufficient to destabilize the slope. the Storegga slide area is located exactly on the Eastern Jan Mayen Fractured Zone (EJMFZ).

05 g pseudostatic acceleration) models no failure (FOS above 1.21).R. S.9).3 E–07% (FORM) 1. Offshore Site Investigation and Geotechnics. P..8 E–07% (SORM) 479 Pseudostatic (0. SEISRISK III: a computer program for seismic hazard estimation.. Berg. C. D.. pp. D.. Earthq. Bungum for information regarding the fine seismic zonation for Norway. R.. 2003. Lien. Soc. Dr. Geol. K. Solheim. One candidate is associated with gas hydrate melting... No failure is expected from the finite element modelling with 0. The pseudostatic approach combined to the shear stress reduction factor (modified two-wedge model) gives a FOS similar to the dynamic one (FOS=2.21.C. Leynaud et al.1 g PGA (FOS=3. An insightful review of Drs. This confirms that the PGA parameter is not sufficient to perform a realistic pseudostatic or dynamic slope stability assessment and that one needs to consider the frequency content of the seismic acceleration. Solheim. Struct. The prediction of earthquake peak ground acceleration in Europe. H. 24.. T. Mienert. C. Rise. Perkins. J. Bender. but the site effect (soil response to seismic loading) could be much more critical than the expected one. This last parameter is mainly relevant to define the critical slope stability area at a local or regional scale. B. One can state that the slope is stable even with a 0. The finite element approach (shear stress distribution with time) does not model any critical state (failure) at 100 m depth with a similar configuration. The Storegga geomodel and its use in slide risk evaluation: geological and geotechnical site investigation in the Storegga slide area. 1772... 0312. Acknowledgements This work is a contribution to the COSTA project. 1995. N.3.26). Lindholm and Dr. H. Dynam. stress reduction factor value at 100 m depth is very high and does not allow one to draw any conclusion.21 to fit the s u DSS profile. New Zealand. N.. Auckland. A. Seismic volcanostratigraphy of the Norwegian margin: constraints on tectono-magmatic breakup processes. Eng. 413 – 426. L. 2001. G.A.D.M. Paper vol. Rasmussen. Rise. The dynamic FOS drops to 2.1). The Storegga slide complex: repeated large scale sliding in response to climatic cyclicity. 1987. Ottesen. Sultan and R. D. (2) . and could explain weak layers development in deeper sediments. El-Centro-type earthquake).8 E–07%. Lien. 219 – 232. but the effect of excess pore pressure in the marine clays due to the rapid loading by the overlying glacigenic sediments has to be considered as well. Kluwer Academic Publishers.. Haflidason. References Andrews. Bull. The Friuli Tarcento earthquake (high-frequency content) has less effect on the shear stress (FOS=4.. Sultan for fruitful discussions and data concerning gas hydrates dissociation modelling.F. L. Present-day slope stability: (1) Both softwares provide a similar static FOS which means that the slope is stable (FOS above 3. 158. / Marine Geology 213 (2004) 457–480 Table 13 Failure probabilities with FORM and SORM methods Method FORM/SORM (SHANSEP+new geometry) b=188/a=À337 m 2.. Martin. In: Locat. 2002. Submarine Mass Movements and their Consequences.05 g)+r d Present-day slope stability. with the critical SHANSEP shear strength profile estimated using a S=0. 2000. Bryn. Forsberg. pp. F. J. J. Geol.0) even with the critical SHANSEP modelling (a S=0.... Berndt. Proceedings 12th World Conference on Earthquake Engineering. confirming that the present-day slope is stable in this area. Ottesen.S. 467 – 490. P. one needs deeper and more critical preconditions to failure to fully explain the slide (100 m high sidewall).. considering parameter uncertainties. Criteria for liquefaction of silty soils.. Berg. a SHANSEP=0.. Bryn.1-g PGA seismic event occurring in this area. N. (Eds. Although the triggering of the Storegga slide is most likely caused by a large earthquake (postglacial crustal uplift).. U. 215 – 222. We would like to thank Norsk Hydro for providing a complete set of data in the Storegga slide area. Planke. K. Surv.. The pseudostatic approach (0...0. The corresponding failure probability (SORM method) is around 1.). Popescu significantly improved the manuscript. A. Ambraseys. C.. R. Tsikalas.N. D.

N. Sejrup. Lane. Berndt. Tamotsu.S. John Wiley and Sons. F. H. B. Springer Verlag. Posewang. Vorren... J. 729 pp. 2002. ´ SPL-2. R.H. Norwegian council for Building Standardization (NBR). Nonlinear site response analysis of submerged slopes.P. G. 45 pp. J.480 D. Mbre and Vbring soil investigation 1997.L. France. pp. 1967.www. San. 2001. SLOPE/W. M... Sejrup. Peck. QUAKE/W. Weaver.. Earthquake Hazard and Submarine Slides—Effects of local soil response on earthquake loads. J. Vorren. Department of Civil Engineering. 45 – 53. Long Sample MD99 2288.. K. Publ. Cochonat.. J. Buchignani. J. A comparative review of limit equilibrium methods of stability analysis. 763 – 786. Schlqter. Canada. General rules—Seismic actions and general requirements for structures..B.P. 139 – 160. 14 February. Seed... 2001. Mary E. F. 1984. NGI report 963014-2.. Baumann. Paper vol. A simplified model describing the cyclic behaviour of lightly overconsolidated clays in simple shear.O. 213. Fredlund. Development of a Seismic Zonation for Norway.). vol. Pestana.. A.. Mienert. B04102.. 1977. J. J. Motta. Canals. 1983. UCB/GT/ 2000-03. 11th European Conference on Earthquake Engineering.... C. Remarks on multivariate transformation. Software documentation. Geophysical signature of gas hydrates along the North-Eastern Atlantic Margins: possible hydrate-bound margin instabilities and possible transfer of methane from oceanosphere to atmosphere. 653 pp. Ormen Lange. University of California. OTC . S. P. Zhou.. Vucetic. 1988. STRUREL. with special emphasis on the Storegga area. Haflidason. Geotechn.. Ann. Report No. 1998. 2004a. Degradation of marine clays under cyclic loading. Billet.G. 972521-1. Statistics 23 (3)..ce. 1994. 8–12 April. Bryn. P.). Prentice Hall.. Math.. 2001.A. R. I.. P. Geol. Geophys. Dept.. Acoustic imaging of gas hydrate and free gas at the Storegga slide. S. 849 – 867. Geol.. In: Anderson..B. Dennielou. Texas. Lien. IKU Pet. US Army Corps of Engineers. 340 pp. Norwegian Geotechnical Instiute.. 291 – 321. M. Jeanjean. Duncan. Laberg. Newmark.. In: Wefer.. 2000. 470 – 472. S. Soc.. H.F.. Engng. May 2003. Journal of the Soil Mechanics and Foundations division. 59 – 70. Miscellaneous paper GL-84-13. Strain degradation of saturated clay under cyclic loading. Maugeri.. www. 2nd ed. University of California. Can. I and II. Bugge.html. Geotechn. T. J.). technical reference and cookbook.M. 1979. 179 – 193. XI European Union of Geoscientists meeting. Simplified procedure for evaluating soil liquefaction potential.. Urgeles. Report N. 387 – 403. pp. A. London Spec.. 2001.. Effects of earthquakes on dams and embankments. Geotechnique 49 (3). D.. Smith. K. 1999.. of Civil and Environmental Eng’g.berkeley. Griffiths. Arley G. Rationalizing the seismic coefficient method. Geo-Slope International Calgary. Ocean Margin Systems. 94 pp. M. Long. 1999.B. (Eds. Mienert. Gas Hydrates: Relevance to World Margin Stability and Climate Change.S. ENV1998-1-1...I. Soil Foundations 32 (1). Krahn. Slope stability analysis by finite elements. Simplified procedure for estimating dam and embankment earthquake-induced deformations. Cattaneo. Hynes-Griffin. Dep. D. Oslo.html. R.. Mienert.. T8.edu/Programs/ Geoengineering/faculty/Pestana/seismic_triggering. vol.. Proc. Eurocode 8—Part 1-1—Design provisions for earthquake resistance of structures. Sultan... 1965. P. J. pp. 133 – 149. W.. 429 – 439. Physical properties of sediments containing gas hydrates. Sultan. Am. Comparison of slope stability methods of analysis. 110. M.. Berkeley.. Mienert. van Veering. M. (Eds. C.. X. Strasbourg. (Eds. A.. 1978. Haflidason. Pestana. 11 – 75. 1997. H. Stoll. 1988. B.. F. Ladd. 1987. Triggering Mechanisms of Slope Instability Processes and Sediment Failures On Continental Margins: A Geotechnical Approach. Houston. Soc. and Biscontin. J. J. Mienert. Haflidason. 17 pp. 379 – 401. J. 1992. (Eds.. D.. 275 – 292. M.. pp. / Marine Geology 213 (2004) 457–480 paper 15203. Cochonat. The generic mapping tools. M. H. Mienert. 2003.. 97 (SM9). Geotechnical Engng Report No.. R. Hebbeln. 109. D. N. 38. H. 51 pp. Laberg. Nadim. 1952. 2000. New design procedure for stability of soft clays. European Margin Sediment Dynamics. Slope Stability. J. Res.P. D. Leynaud et al. 1 – 12. New York. 1976. 84. Terzaghi. G. E. 1 – 8. 137. P. Finite element slope stability analysis by shear strength reduction technique. Mar. Geo-slope. Cascone. M. 1998. P.B.. Berkeley. University of California. Norway. Gravdal. vol. ASCE 104 (GT7). J. T. Geol.edu/Programs/ Geoengineering/faculty/Pestana/seismic_triggering. Eng. Gollwitzer. Res. 14. Proceedings of ASCE. R. Mar. Res. Of Civil and Environ. Seed.T. R.S.). 1995.C.. J. P06-031-1. Probabilistic slope stability analyses of the Sigsbee Escarpment..G. T. Idriss. C. 2003.M. Proceedings. D. Bryan. P. 67 pp. Norway. School of Ocean and Earth Sciences and Technology.. 1971. Submarine landslides on continental margins.ce. Kramer. 1998. Wilson. pp. The Northern Storegga slide escarpment—morphology and features.. Can... 2004. T. E. Berkeley. NORSAR.D. Norwegian Geotechnical Institute. J.M... Seismic response of clay slopes.. et al. Bqnz. In: Henriet. 114 (2). 152 pp.. Trincardi. G.. and Nadim. Civil Eng. Geophys. Publ... A structural reliability analysis program-system.M.A. University of California. H. 208 – 212. An Engineering Manual for Settlement Studies. Submarine slides on the Norwegian continental margin. 2000. J. 1249 – 1273. J. Oslo. Rosenblatt. J. pp. 1629 – 1634..M. Paris. User’s guide.B. Wessel. Springer. Krunic.. Makdisi.. Jbrgensen. Ka-Ching... Soil Mechanics in Engineering Practice. Nash. GT7. Eng. J.N. 2004b. H.. Offshore Technology Conference ’03.. Richards. 162 p. Berkeley. J.. Geotech. In: Mienert.P. Norwegian Geotechnical Institute. Effect of Gas Hydrate Melting On Seafloor Slope Stability. Matsui. 992527-1. Recent advances in soil liquefaction engineering and seismic site response evaluation.V. Geotechnical Engineering report No UCBGT 2000–04. Foott. Dobry. University of Hawai. R. 1974. J.L. Div..berkeley. 1998. F. Gong. Seed.... Geotechnical Earthquake Engineering. and Franklin. Foucher. Geotechnique 15 (2). 213. 618 and 458 pp. The Storegga slide: chronology and flow mechanism Paper presented at the. Geotech. John Wiley and Sons. D. J.

Sign up to vote on this title
UsefulNot useful