Artificial Neural Network Based Backcalculation of Conventional Flexible Pavements on Lime Stabilized Soils

Conventional flexible pavements built on lime stabilized soils (CFP-LSS) were studied for the backcalculation of pavement layer moduli from nondestructive Falling Weight Deflectometer (FWD) testing. The validated ILLI-PAVE finite element program was used in pavement structural analyses by taking into account the effects of nonlinear layer modulus behavior, i.e., stress hardening for granular materials and stress softening for fine grained soils, and lime stabilization on pavement responses. Various pavement geometries were analyzed with different layer material properties. The computed surface deflections due to typical FWD loading scenarios were collected in a database to develop Artificial Neural Network (ANN) models for predicting the pavement layer moduli and critical pavement responses. Comparisons of the estimated layer moduli with ILLI-PAVE results produced very low mean absolute percentage errors to validate the ANN based backcalculation of pavement layer moduli for CFP-LSS.