Identifying the parameters of a hydro-mechanical model for internal erosion occurring in granular soils by using an enhanced backtracking search algorithm

Abstract : Due to the complexity of the hydro-mechanical behaviour of soils subjected to internal erosion, a high number of parameters are usually required for the erosion models and the constitutive models. This aspect makes it difficult to determine by trial-error their relevant values from laboratory tests. To address this issue, an efficient optimisation-based procedure for identifying the parameters of a recently developed hydro-mechanical model for internal erosion using an enhanced backtracking search algorithm (so-called MBSA-LS) has been proposed. The MBSA-LS incorporates two points: (1) modifying the mutation of the original Backtracking Search Algorithm (BSA) and (2) incorporating an efficient differential evolution (DE) as a local search to improve the optimisation performance. A mono-objective framework with six different criteria has been proposed to identify the parameters related to the interlocking effect and the erosion process. The proposed procedure was successfully applied to identify the parameters from the erosion tests of Hong Kong-Completely Decomposed Granite mixture (HK-CDG). All results demonstrated that coupling the MBSA-LS and the hydro-mechanical erosion model could efficiently solve the issue of parameter identification accounting for both the mechanical behaviour and internal erosion.