Laboratory and field evaluation of asphalt concrete modulus and critical strain inputs for M-E flexible pavement design in the state of Illinois

This study aims to update the current algorithm for modulus prediction of hot-mix asphalt (HMA) in the state of Illinois-USA. Numerically, laboratory and field-obtained modulus and strain levels were connected to advances in asphalt materials and current design criteria. Four mix designs were selected and combined with seven different binders for modulus measurements. The reference experimental results were compared to the current Illinois Department of Transportation (IDOT) modulus algorithm, traditional regression models, a Bayesian Neural Network (BNN) model, and field measurements obtained using the traffic speed deflectometer. Results showed that the current IDOT algorithm under-predicts modulus and BNN (840 data points, R² = 0.986) outperformed the Witczak and Hirsch models. Field modulus and critical strains were sensitive to temperature and showed considerable variability, but confirmed that HMA in perpetual pavement sections performs under low strain levels. These findings guided considerations for an updated design method: realistic modulus prediction can be achieved using data-driven methods and less conservative strain criterion might be considered.