Vertical void distribution and uniformity analysis of asphalt mixtures using three-stage exponential model.

Heterogeneous void distribution is a common phenomenon in asphalt mixtures and is closely linked to their performance. Therefore, it is necessary to propose a physically meaningful and accurately fitting vertical void distribution model for asphalt mixtures, so as to realize the quantitative evaluation of void distribution. In this study, virtual and actual specimens of asphalt mixture were established using discrete element method (DEM), laboratory compaction, and field coring methods. The vertical void distribution of different specimens was obtained through CT scanning and image processing technology. The impact of various factors, including gradation type, nominal maximum aggregate size, and specimen height on void distribution, was investigated. Subsequently, a three-stage exponential model was proposed to characterize the vertical void distribution in asphalt mixture. The physical significance of different fitting parameters in the model was elucidated. The results indicate that the vertical void distribution of DEM models and laboratory compaction specimens exhibited a bathtub shape, while the vertical void distribution of pavement core specimens exhibited a semi-bathtub shape. The void distribution of the mixture was affected by gradation type, nominal maximum aggregate size and specimen height. The vertical void distribution of asphalt mixture could be effectively characterized by the three-stage exponential model. In addition, the root mean squared error of the second stage data from the three-stage exponential model was also proposed to evaluate the void distribution uniformity. • A three-stage exponential model was proposed to characterize the vertical void distribution. • The feasibility of the three-stage model was analyzed through simulation and test results. • An evaluation method for the uniformity of vertical void distribution was proposed. • The potential application of the three-stage exponential model was discussed. [ABSTRACT FROM AUTHOR]