| description abstract | Rutting, one of the main failures in flexible pavements, is the result of permanent deformation aggregation in pavement layers under traffic loading. Rutting decreases the life of the pavement, and, by influencing control properties of vehicles, creates serious dangers for road users. Therefore, it is very important to predict the rutting potential of different types of asphalt mixtures (before construction and operation) based on the characteristics of the mixture ingredients (bitumen and aggregate), environmental conditions, and traffic loads. This study used genetic programming to represent flow number results of different asphalt mixtures. The models presented predict the flow number (as an index of rutting potential) based on parameters such as the index of aggregate particle shape and texture (particle index), bitumen rutting parameter (G*/sinδ), and the stress level. Experimental data were collected from studies conducted on materials (gradation, dynamic shear rheometer, and particle index) and dynamic creep tests of asphalt samples at different levels of stress and temperature. The genetic programming models were compared with a multiple linear regression model. The results demonstrated that the genetic programming model predicted the flow number of asphalt mixtures with better accuracy rather than the regression model. The results of statistical studies revealed that three parameters, particle index, rutting potential, and stress level, influence the flow number, and the stress level is the most significant parameter. | |
