Efficient improvement for the estimation of the surface of free energy asphalt binder using Machine Learning tools

Abstract

The Surface Free Energy (SFE) of a material is defined as the energy needed to create a new surface unit under vacuum conditions. This property is directly related to the resistance to fracture and recovery of material and the ability to create strong adhesion with other materials. This value can be used as a complementary parameter for the selection and optimal combination of materials for asphalt mixtures, as well as in the micromechanical modelingof fracture and recovery processes of said mixtures. This document describes the results of the implementation of the use of machine learning and Random Forest prediction techniques for the estimation of surface free energy based on data from previous studies. The experimental samples were twenty-three asphalt binders used in a Strategic Highway Research Program (SHRP). A decrease of 54% and 82% in the mean absolute error (MAE) and the mean square error (MSE), respectively was found for the new model built. While the model fits better with a 12% improvement, according to the adjusted determination coefficient, the accuracy and the score of the model also increases notably in 2% and 55%, respectively.