Full probabilistic design of earth retaining structures using generalized subset simulation

Reliability-based design (RBD) of earth retaining structures has attracted considerable attentions due to several distinctive elements, such as correlated load and resistance, multiple correlated failure modes, etc. These elements can be rationally considered using full probabilistic design approaches based on Monte Carlo simulation (MCS). However, MCS-based full probabilistic design approaches are often criticized because of a lack of computational efficiency and resolution at small probability levels. This paper develops an efficient MCS-based full probabilistic design approach, which directly makes use of Generalized Subset Simulation (GSS) to perform RBD of earth retaining structures. The proposed method allows simultaneously approaching failure domains of different possible designs in a prescribed design space by a single GSS run and generating their corresponding failure samples in an efficient manner. As a result, a large number of failure samples can be generated for numerical representation of failure domains of possible designs. Then, the failure probabilities of possible designs are calculated for identifying feasible designs and the design point of each design can be approximately determined as the most probable failure sample of the design according to its probabilistic interpretation. Finally, the proposed approach is illustrated through two design examples of gravity retaining wall.