Predicting the Dynamic Modulus of Asphalt Mixtures Containing Recycled Asphalt Shingles using Artificial Neural Networks

Abstract

The hot mix asphalt (HMA) containing recycled asphalt shingles (RAS) are substantially different in composition and properties compared with virgin HMA, leading it hard to predict the performance of asphalt mixture containing RAS. This paper explored an ANN model to predict the E* of asphalt mixture containing RAS. In this paper, the ANN model was developed using the E* database containing 1701 sets of experimental data from four different demonstration projects. A sensitivity analysis of each model parameter was conducted by correlating these parameters with dynamic modulus. The developed ANN model was compared with the Iowa model and the developed ANN model showed significantly higher prediction accuracy than the Iowa model. The results show that ANN has great potential to be used as a tool to predict the dynamic modulus (E*) of asphalt mixture containing recycled asphalt shingles.