Artificial Neural Network Modeling for Dynamic Modulus of Hot Mix Asphalt Using Aggregate Shape Properties

Abstract

Over the past few years, many regression-based and artificial neural network (ANN)-based models have been developed to estimate the dynamic modulus of hot mix asphalt (HMA). These models use the gradation of aggregates and the volumetric properties of compacted samples as input variables to the model. However, none of these models use aggregate shape parameters (i.e., angularity, texture, form, and sphericity) in the development of the model. Recently, researchers have expressed concerns that the shape parameters of aggregates need to be considered in the estimation of dynamic modulus. The primary objective of this study was to develop an ANN-based model for the estimation of dynamic modulus of HMA using aggregate shape parameters. The dynamic modulus of 20 different HMA mixes composed of various sources, sizes, types of aggregates, and different volumetric properties were measured in the laboratory. The shape parameters of different sizes of coarse and fine aggregates were measured with an automated aggregate image measurement system (AIMS). An ANN-based model was developed to consider the following input variables: aggregate shape parameters (i.e., angularity, texture, form, and sphericity), frequency, asphalt viscosity, and air voids of compacted samples. A sensitivity analysis of each model parameter was conducted by correlating these parameters with dynamic modulus. It is expected that this study will be helpful in predicting the dynamic modulus of HMA using aggregate shape parameters.