Evaluation of Factors Influencing the Compaction Characteristic of Recycled Aggregate Asphalt Mixture

Abstract

Aggregate and air void distribution determined by compaction commonly affects damage appearance and development inside asphalt mixture and is related to asphalt pavement durability and quality. The main objective of this study is to investigate the recycled aggregate (RA) effect on asphalt mixture compaction behavior under different engineering conditions. First, the aggregate fragmentation caused by compaction effort was simulated using the superpave gyratory compactor. In this regard, the influences of aggregate type and RA content were investigated. Second, the indoor experiment scheme was determined using the Taguchi method to obtain compaction data of recycled aggregate asphalt mixture (RAAM). Finally, a genetic algorithm-based backpropagation (GA-BP) artificial neural network (ANN) model using the 216 data sets of the indoor experiment was developed to predict and explore the relative contribution of engineering-conditions-related parameters to RAAM compaction difficulty. The results showed that the aggregate particles suffer fragmentation mainly in the early compaction of recycled aggregate asphalt mixture. The effect of RA on aggregate fragmentation during the compaction process is not statistically significant. The 10-14-1 GA-based BP ANN model developed in this study is an effective method in predicting the compaction energy consumption of RAAM with a correlation coefficient (R2) of 98.59% and a mean-squared error value of 0.6266. The gradation shape, NMAS, FI3d, AI3d, and T3d and incorporated content of recycled aggregate have a considerable positive correlation with the compaction difficulty. The limitation of this study is that the compaction difficulty prediction model is developed according to indoor test data. Therefore, the model’s applicability to field pavement projects required further practical verification.