Seismic analysis of submarine slopes: retrogressive and three-dimensional effects

Azizian, Alireza (2004) Seismic analysis of submarine slopes: retrogressive and three-dimensional effects. Doctoral (PhD) thesis, Memorial University of Newfoundland.

[English] PDF - Accepted Version Available under License - The author retains copyright ownership and moral rights in this thesis. Neither the thesis nor substantial extracts from it may be printed or otherwise reproduced without the author's permission. Download (35MB)

Abstract

Failures of submarine slopes, caused by earthquakes, rapid sedimentation, storm waves, etc., have resulted in significant damage near- and off-shore, in many areas of the world. High costs of off-shore projects, such as oil exploration projects, necessitate using accurate methods for assessing slope resistance and possible extent of slope failures due to rare events such as earthquakes that may lead to considerable devastations. -- The stability analyses of submarine slopes, to this date, are mostly based on the classical methods of slope stability analysis such as the limit equilibrium method. While appropriate for static slope stability analysis, those methods have some limitations when used in seismic analysis of saturated soil slopes involving soil liquefaction. This study aims at filling some gaps in the current approach by using a state-of-the-art method for effective stress, seismic analysis of submarine slopes. The method proposed here implements a fully coupled, dynamic, finite element approach and a multi-yield surface plasticity model for simulating non-linear soil behaviour under dynamic loads. -- According to the geological and geophysical investigations of past submarine failures, an important phenomenon observed in such events is the significant retrogression of failure, initiated as a slope failure and extending back to a long distance in a nearly flat seabed. Accurate prediction of the extent of retrogression is of crucial importance when assessing the safety of seabed facilities. In addition, seabed images showing crescent shaped escarpments of failures indicate significant three-dimensional (3D) characteristics of such failures. Most slope stability analysis methods, and in particular those for dynamic analysis, are based on the two-dimensional, plane strain simplifying assumption. Assessment of 3D effects in seismic slope stability analysis is therefore essential for obtaining relatively more accurate numerical results. Moreover, geotechnical investigation in submarine environment is much more costly than on land. Geotechnical data regarding submarine soils are rather scarce and insufficient for stability analyses where vulnerability to liquefaction is of great importance. Obtaining best estimates of soil properties from such scarce data based on some statistical methods is of great importance for numerical predictions. -- The objectives of this research are aimed at addressing the needs of geotechnical practice: (1) to provide a procedure for analyzing seismically induced retrogressive slope failures and to use this procedure for explaining the mechanisms and identifying the main factors affecting the extent of those slope failures; (2) to assess the three-dimensional effects in seismic analysis of submarine slopes, in order to provide geotechnical practitioners with a reliable tool for extrapolating the results of manageable 2D seismic analyses to real 3D configurations; and (3) to design a procedure for constitutive model parameter calibration based on liquefaction strength analysis, using limited amount of experimental data and accounting for uncertainties in soil properties. -- To the author's knowledge, the two aspects of slope stability analysis addressed here, namely, simulation of retrogressive slope failures and 3D seismic analysis of saturated soil slopes, have not been investigated in a consistent manner so far. -- By modelling the retrogressive failures, the study highlights the importance of accounting for the potential of retrogression in regions that are seemingly safe but can be affected by such phenomenon. Risk assessment of infrastructures (e.g. pipelines) located on such seemingly safe zones should include estimation of retrogression distance. This is similar to accounting for the potential hazard of debris run-out for infrastructures located below the potentially unstable slopes. In this part of the study, a new method is introduced for simulating successive failures due to loss of support. For the various configurations of seabed slopes analyzed here for assessing the effects of gentle seafloor slope and presence of a layer with low permeability, it was found that the final linear extent of retrogressive failures are 5 to 20 times larger than those of the initial failure, which is usually the only stage of failure accounted for in practice. -- Three-dimensional effects are assessed by comparing results of two- and three-dimensional analyses, in terms of predicted displacements, shear strains, and excess pore water pressure ratios. Limits of applicability of the 2D, plane strain analysis assumptions are quantitatively assessed. Some regression models are also presented that express ratios of 3D to 2D predictions as a function of slope width/height ratio and earthquake peak acceleration. The results of the present dynamic, fully coupled, non-linear analyses are also compared with those of static slope stability analyses. The comparison indicates that the trend of decrease in the ratio of 3D/2D response as a function of slope width/height ratio is very similar for both approaches. However, the applicability limit of the 2D assumption is found to be slightly lower in dynamic analysis (width/height ratio of about 3 - 5, with larger values corresponding to larger seismic accelerations) than in static analysis (width/height ratio of about 5) for the same level of tolerance (15%). Moreover, for B/H > 6- 7, the differences between 3D analysis predictions on the symmetry plane of the slope and 2D analysis predictions are found to be insignificant. -- Soil constitutive model parameters used in the analyses are obtained and calibrated for two types of sand, namely Nevada and Fraser River sands in a loose state, using available information from the literature as well as results of some recently performed laboratory soil tests. -- Response Surface Methodology is used in several parts of this study for the efficient identification of the most important parameters (or factors) that affect analysis results (or responses). It is used for soil parameter calibration where some specific information regarding soil behaviour is not available, yet a set of parameters can be estimated that can re-produce the observed behaviour of soil as indicated by liquefaction strength analysis. This methodology is also used for identifying the significant factors, and then obtaining regression models, to quantify the 3D effects.

Actions (login required)

View Item