Use of a genetic algorithm to perform reliability analysis of unsaturated soil slopes

Rainfall induced landslides are a major cause of disturbance to transport networks in many parts of the world. In slopes where the water table is some depth below the ground surface, negative pore water pressure (suctions) develop in the near surface soils which contribute significantly to their overall stability. However, these suctions are transient and reduce as water percolates into the slope (and a wetting front develops) during periods of heavy or prolonged rainfall.

In this paper the development of a model for determining the reliability of a slope in which the soil properties are considered as random variables is presented. By transforming the variables into polar coordinates the complexities associated with defining the limit state function, which have affected many previous attempts at probabilistic analysis of slopes, can be overcome. The minimisation problem is solved in a powerful and efficient Genetic Algorithm environment.