Predictive modelling of the M_R of subgrade cohesive soils incorporating CPT-related parameters through a soft-computing approach

Abstract

Performance of pavements is significantly influenced by the characteristics and behaviour, e.g. compaction and deformation, of their underlying layer as subjected to different loadings. This can be experimentally or theoretically estimated through the subgrade soil resilient modulus (M_R) parameter. The M_R found by laboratory tests does not completely describe the actual in situ conditions of subgrade soils. Additionally, almost all of the predictive models proposed in the literature relate the M_R parameter to lab-related soil indices and conditions of stresses applied in laboratory tests. This study explores the feasibility of producing a new M_R equation using key soil lab and in situ related parameters for the pavements subgrade cohesive soils through gene expression programing (GEP) approach. A database comprising several experimental data acquired by conducting cone penetration test and corresponding resilient moduli tests on various cohesive subgrade soils is employed for modelling of the pertinent M_R factor and a new empirical mode is suggested. In order to examine the proposed model from both precision and engineering perspectives, various validation and verification analyses are done. The results present that the strength of GEP approach and the proposed formula for indirect estimation of the M_R of pavements subgrade soils.

Keywords:

• pavements
• subgrade layer
• cohesive soils
• indirect estimation
• soft computing
• gene expression programming