Doctor of Philosophy
Department of Civil and Mining Engineering
Bertoldi, Carlo Ernest, Computer simulation of progressive failure in soil slopes, Doctor of Philosophy thesis, Department of Civil and Mining Engineering, University of Wollongong, 1988. http://ro.uow.edu.au/theses/1232
This thesis is concerned with the simulation of progressive failure within soil slopes considering, as a basis, the widely accepted concept of limit equilibrium. In particular, the investigations reported here concern the influence of strain-softening on the overall safety factor, the identification of local failures and the propagation of failure within a slope. Several methods of analysis were developed and successfully implemented. A number of case histories were analysed using these methods and the influence of progressive failure evaluated. Some of these methods take the initial stress field into consideration.
Theories of progressive failure are often based on the well established mechanism of strain-softening associated with soils. The extent of strain-softening at different locations within a slope or along a slip surface is generally unknown. Complete strain-softening (a residual factor of one) along a slip surface must occur after overall failure of a slope and. if no overstress has occurred anywhere and relative deformations have been small. no strain-softening would have occurred (a residual factor of zero). These are the limiting cases and thus the overall residual factor may have a value between zero and unity. More importantly, the local residual factor (as distinct from the overall residual factor) may vary from point to point along a slip surface.
A method of simulation was developed to study the factor of safety of a slope considering any arbitrary distribution of the local residual factor along the potential slip surface. Typical distributions represent failure initiating from the crest of a slope or the toe of a slope or from somewhere in the interior. The effect of the type of shear strength distributions on the factor of safety was examined. Moreover, a relationship between the average shear strength and the factor of safety was established. Criteria for acceptability of rigorous (Morgenstern and Price type) methods of slices were highlighted.
A number of methods for simulating local failure and its propagation were developed. In these methods the excess shear stress resulting from strain-softening of failed segments of the slip surface must be redistributed to other segments. Once this redistribution occurs, more segments may fail and then further redistribution occurs. This process of progressive failure continues until no more failures occur. At that stage the factor of safety is the lowest one associated with progressive failure. In one of the methods an assumption is made on the manner of this redistribution e.g. uniform distribution over the segments or linear distribution with its maximum near the last failed segment. In other methods no such assumption is made and the new 'failed' segments are identified during successive limit equilibrium calculations, one for the stage corresponding to no failures and the other for the stage with initial local failures. Considering all the failed segments, a new analysis is made and, comparing this with the previous one, further local failures may be identified. In this way, the iterative technique enables the simulation of the progressive failure process and the associated redistribution of shear stress occurs automatically.
It is well known that an initial stress field in a slope may have a significant influence on its stability. Therefore, it was considered appropriate to develop methods of analysis of progressive failure which took a given initial stress field into consideration. Two different approaches were considered for the development of these methods. The first approach is to consider the initial stress field in the identification of local failures and then to follow up with methods similar to those mentioned in the previous paragraph. The second approach is to simulate progressive failure as a transformation from the initial stress field to the stress field associated with limit equilibrium. In the latter case two different types of analysis are required for two parts of a sloping mass at each stage of progressive failure. As the failure progression process continues, the relative size of these masses changes. The interaction between the two masses is taken into consideration by including the limiting earth pressures as extreme cases of possible interaction.
The successful implementation of all the methods is demonstrated in relation to a number of case histories. The influence of progressive failure on the stability is influenced not only by the soil shear strength and brittleness but also by slope and slip surface geometry. The initial stress field may have a significant influence on the extent of progressive failure.