Void Ratio of Porous Asphalt Mixtures as a Performance Determinant in Semi-Rigid Pavements: An Experimental and Artificial Neural Network (Ann) Approach

Semi-rigid pavements are a new type of composite pavement constructed by injecting a highly flowable cement grout into the voids of a compacted porous asphalt skeleton with a void ratio ranging between 25% and 35%. One of the most critical factors influencing the performance of semi-rigid pavements is the adequacy of the void content within the porous asphalt mixture. Insufficient void space hinders the proper penetration of the cement grout, thereby adversely affecting the structural and mechanical performance of the semi-rigid pavement. In this study, the void content of porous asphalt mixtures forming the skeleton of semi-rigid pavements was investigated. For this purpose, samples were prepared using two different aggregate types—basalt and limestone—each with six different gradations, and nine different bitumen contents were employed for each combination, resulting in a total of 108 samples. Empirical calculation methods were used to determine the void ratios, and a machine learning approach, specifically an artificial neural network (ANN) model, was developed to predict these values. In the ANN model, ten input variables were utilized: six aggregate size gradations, apparent specific gravity of the aggregates, optimum bitumen content, maximum theoretical specific gravity of the asphalt mixture, and the bitumen content expressed as a percentage of the total mixture weight. The sole output variable was the void ratio. The model's predictions demonstrated a high level of agreement with the experimental results, yielding an R² value of 0.98. This strong correlation indicates that the developed ANN model is a reliable and effective method for estimating the void content in porous asphalt mixtures.