Prediction models of mixtures’ dynamic modulus using gene expression programming

Abstract

The dynamic modulus (E*) among asphalt mixtures’ mechanical property parameters not only is important for asphalt mixtures’ pavement design but also in determining asphalt mixtures’ pavement performance associated with pavement response. Based on the principle of gene expression programming (GEP) algorithm, this paper explored two different GEP approach models, namely: GEP-I and GEP-II to predict the E* of hot mix asphalt (HMA) and mixtures containing recycled asphalt shingles, respectively. In this paper, The GEP-I was developed from a large database containing 2750 test data points from 205 unaged laboratory-blended HMA mixtures including 34 modified binders, and the GEP-II model was developed using the E* database containing 1701 sets of experimental data from 4 different demonstration projects. Both the GEP-I model and GEP-II model were compared with other E* prediction models. A sensitivity analysis of each model parameter was conducted by correlating these parameters with dynamic modulus. Both the GEP-I model and GEP-II model showed significantly higher prediction accuracy compared with the existing regression models and could easily be established. It is expected that these two GEP models could lead to more accurate characterisation of the asphalt mixtures’ E*, resulting in better performance prediction.

Keywords:

• Hot mix asphalt
• dynamic modulus
• GEP
• RAS
• mixtures
• prediction

View correction statement:

Erratum