Risk Factor Based Design of Cantilever Retaining Walls

Abstract

Sensitivity analysis of geotechnical random variables on potential failure modes (overturning, sliding, bearing capacity and eccentricity) of a cantilever retaining wall reveals that high sensitivity of a particular variable on a particular mode of failure does not necessarily imply a remarkable contribution to the overall failure probability. The present paper aims to combine probability of failure (P f ) of each failure mode and sensitivity of the random variables to these failure modes and introduces a new factor, called Probabilistic Risk Factor (R f ) for each random variable. P f is calculated by Monte Carlo Simulation and sensitivity analysis of each random variable is calculated based on normalized F-Statistics value. R f is a reduction factor which takes into account the variations of random variables and hence can be directly implemented in design by the designers. The random variables (friction angle and unit weight of backfill soil; and friction angle, unit weight and cohesion of foundation soil), when divided by R f and applied in design, yield a structure which is safe against variations of the random variables. It is observed that R f of friction angle (φ 1 ) of backfill increases and cohesion (c 2 ) of foundation soil decreases with an increase in variation of φ 1 , while R f for unit weights (γ 1 and γ 2 ) of both the soil and friction angle of foundation soil (φ 2 ) remains almost constant. Finally, design guidelines for different variations of φ 1 are provided based on the proposed methodology, which proves to be cost effective.