Failure Probability of an Optimally Designed Gravity Retaining Wall

Abstract

The paper presents an optimal design for a gravity retaining wall based on a mixed-integer constrained optimization. To solve this optimization problem, a real coded genetic algorithm was applied. The gravity retaining wall optimization model (OPT-GRW) includes a cost function of the wall, which is limited by design and resistance constraints. The deterministic optimization model, based on partial safety factors, was extended by an optimization approach in which the cost objective function was constrained by a defined target probability of failure. The paper presents an example for determining the most cost-effective gravity retaining wall for given design parameters. The presented case study shows the optimization process based on Eurocode 7, which takes into account part of the uncertainty of the soil parameters. As the study progresses, the need for optimization based on the probability of failure becomes crucial, as several designs exist with the same probability of failure, but only one is optimal.