The resilient moduli of five Canadian soils under wetting and freeze-thaw conditions and their estimation by using an artificial neural network model

摘要

The resilient modulus (M-R) is a key parameter used in the mechanistic-empirical methods for the rational design of pavement structures. In permafrost and seasonally frozen regions, the M-R of subgrade soils is significantly influenced by the variations in moisture content and temperature. The M-R typically reduces due to the weathering action associated with wetting and freeze-thaw (F-T) cycles, which contributes to the reorientation of soil particles, loss in suction and cohesion, and formation of cracks in the subgrade soils. In the present study, the M-R values of five Canadian soils that are widely used as pavement subgrades were determined under wetting and F-T conditions. The key findings from the extensive experimental investigation suggest: (i) the M-R values of the soils at their respective optimum water contents significantly reduce up to the critical F-T cycle, which is typically the first or second F-T cycles; (ii) there is little change in the M-R values from the critical to the tenth F-T cycle; (iii) the percentage of reduction in the measured M-R at the optimum water content after the critical F-T cycle is strongly related to the soils plasticity index; (iv) the measured M-R values are typically low for the specimens subjected to wetting, and the effect of F-T cycles on these specimens is insignificant; and (v) the effect of stress levels on the M-R values is dependent on the initial water contents of the specimens and soil types. In addition, an artificial neural network (ANN) model was proposed and validated for estimating the M-R of the tested soils taking account of various influencing factors. Both the experimental data and the developed ANN model provide valuable information for the rational design of pavements in Canada.