Application of Conventional Mathematical and Soft Computing Models for Determining the Effects of Extended Aging on Rutting Properties of Asphalt Mixtures

Pavement performance prediction is a mounting task due to the many varied influencing factors particularly aging which varies with time, weather, production, type of pavement and etc. This paper presents a conventional mathematical model named Superpave model, Artificial Neural Network (ANN), and Supporting Vector Machine (SVM) techniques to predict the effects of extended aging on asphalt mixture performance measured in terms of rutting properties determined from the dynamic creep test. The accuracy of each method was compared to select the most reliable technique that can be used to forecast the rutting behavior of asphalt mixtures subjected to different aging conditions. The results indicated that the Superpave model was only reliable at lower temperatures, while ANN and SVM techniques showed the capability of precise prediction under all conditions. The overall results showed that the ANN was the most promising technique that can be adopted to satisfactorily forecast the effects of aging on rutting properties of all mixtures. The developed model can be embraced by the pavement management sector for more precise estimation of the pavement life cycle subjected to different aging conditions which can be used to design efficient pavement maintenance and rehabilitation plans.