Effect of the employed soil constitutive model on the response of large-span soil steel bridges to soil and truck loading

Abstract

Current design codes and standards recommend refined analysis methods to accurately predict the behaviour of large-span soil-metal bridges. Consequently, the FE modelling of these structures must employ an appropriate constitutive model for the soil capable of mimicking its performance under various loading conditions. This study investigates the influence of different soil constitutive models on the performance of soil-steel arch bridges with spans up to 32.4 m. A comprehensive analysis was performed using a validated 3D finite element model to evaluate the suitability of three widely adopted soil models: the Mohr-Coulomb (MC) model, the Hardening Soil (HS) model, and the Hardening Soil model with small-strain stiffness (HSs). The results indicate that the choice of soil model significantly affects the predicted structural response. The MC model consistently overestimates the settlement and lateral displacement of the foundations, particularly for larger spans, while underestimating upward deformations of the culvert crown during backfilling. Under traffic loading, the MC model underestimated the outward bending moment at the culvert crown by up to 83 % for smaller spans but showed a reduced discrepancy for larger spans. Overall, the HS and HSs models provided more accurate predictions of the vertical and lateral displacements, bending moments, and thrust forces. The findings underscore the importance of selecting an appropriate soil model in finite element analyses to ensure reliable design and safe performance of soil-steel composite structures.