Application of the BFGS quasi-Newton method to slope stability analysis

Abstract

Mana computer programs have been developed for solving slope stability problems. Since slope stability problems can be characterized as optimization problems, many optimization techniques can be used for searching for the lowest safety factor for a given problem and the corresponding critical slip surface. Most of the slope stability programs use the direct search method which requires only the function value (i.e., safety factor value). In this thesis, a new optimization technique, the Broyden (1970), Fletcher (1970), Goldfarb (1970), and Shanno (1970) (BFGS) quasi-Newton optimization method, is used in conjunction with the STABR program of Lefebvre (1971) to solve slope stability problems. This method of optimization requires the function value and the first derivative value, which can be found by the finite difference method. A new program CSLIP3, incorporating the BFGS technique, is used to solve a variety of realistic slope stability problems. It is determined that CSLIP3 is reliable and efficient.