PROBABILISTIC STABILITY ANALYSIS OF SLOPES

Abstract

The slope stability analysis is usually carried out by the deterministic approach rather than a probabilistic approach. Conventional methods of analysis and design of slope employs the values of soil parameters as average values emanating from multiple tests conducted for this purpose. However, due to inherent variability in soils, the soil properties can not be considered as deterministic quantities. The probabilistic approaches provide more rational results in terms of factor of safety and reliability indices. In the present work, the analysis of slopes has been carried out considering the soil parameters as a random variable using Monte Carlo simulation. Bishop's simplified method was chosen as the deterministic model. This model employs cohesion, angle of internal friction and pore water pressure for the calculation of factor of safety. A computer program has been written in C++ to accomplish the iterations at various steps. The random numbers honoring lognormal distribution of given mean and standard deviation were generated using polar method. These numbers worked as an input to the deterministic model and probability distribution function of factor of safety was obtained. A detailed parametric study was conducted to the effect of number of• simulations on probability distribution function of factor of safety. Both the slopes. (Lodalen slope and two layer slope) were repeatedly analyzed for different values of coefficient of variation.