DOE‐Based Sensitivity Analysis for Rational Constitutive Parameter Selection in Coupled Hydro‐Mechanical Simulations of Test in Unsaturated Soils

Abstract

ABSTRACT The Barcelona Basic Model (BBM) is the most well‐known elasto‐plastic constitutive model for describing the behavior of unsaturated soils. However, the selection of BBM parameters from laboratory data remains challenging due to the strong coupling and interdependence among the constitutive parameters governing hydro‐mechanical behavior. In this context, the Design of Experiments (DOE) methodologies provide a systematic approach for quantifying parameter influence and reducing the dimensionality of calibration problems. This paper presents a sensitivity analysis of the parameters used in the coupled hydro‐mechanical numerical simulation of a suction‐controlled oedometer test for three distinct soils, enabled by a computational framework designed to automatically execute and manage many simulations. The analyses were performed using both exhaustive Full Factorial (FF) designs and reduced Plackett–Burman (PB) matrices. The investigated variables included BBM mechanical parameters and hydraulic parameters, while the response variable was the simulated vertical strain during oedometric loading. The FF analyses were used as reference and PB matrices with different numbers of trials were subsequently assessed regarding their ability to reproduce the patterns obtained from the FF analyses with fewer simulations. The results showed that the soil stiffness parameters λ(0) and r , together with the historical variable p 0 * , had the greatest influence on vertical strain across all three soils. In contrast, hydraulic parameters had little to no effect on the response variable in this case. The results demonstrate that DOE methodologies provide an effective and computationally efficient framework for sensitivity assessment and parameter prioritization in coupled hydro‐mechanical simulations involving unsaturated soils.