Accurate identification of pavement materials that are susceptible to moisture damage with the use of advanced conditioning and test methods and the use of machine learning techniques

Abstract

In order to prevent premature failure in hot mix asphalt (HMA) pavements due to moisture ingress, it is necessary to identify mixes that are susceptible to moisture damage during the mix design stage. The objective of this paper is to present a method that utilizes a conditioning process along with suitable destructive and nondestructive tests and machine learning techniques to accurately assess the moisture damage potential of HMA mixtures. The scope of work included conditioning with the Moisture Induced Stress Tester (MiST) and testing of a set of HMA mixes with known field performance, and analysis of the data using machine learning techniques. Input variables obtained from a destructive test, the indirect tensile strength test and a non-destructive test, the ultrasonic pulse velocity test, were used to identify the good and poor performing mixes. Various machine learning models were used and it was found that the Support Vector Machine and the Naïve Bayes methods were able to classify the mixes with high accuracy. Use of indirect tensile strength, ultrasonic pulse velocity test and classification based machine learning technique are recommended for the development of accurate models and research in pavement engineering.