Abstract
Anti-Freezing Asphalt Pavement (AFAP) has good snow-melting performance and is used widely in many countries around the world. The objective of this study was to analyze AFAP’s long-term performance and predict its snow-melting ability. Two types of anti-freezing stone matrix asphalt (SMA) mixtures (SMA-13 with Iceguard and SMA-13 with Mafilon) were prepared with the Marshall method. Water stability, high-temperature stability, low-temperature crack resistance, and freeze-thaw split tests were conducted to evaluate mixtures’ performance. Meanwhile, the effect of anti-freezing filler, asphalt content, and soaking temperature on the salt dissolution of anti-freezing asphalt mixtures was analyzed, and the snow-melting ability of AFAP was predicted based on the Back Propagation (BP) neural network. The results illustrated that water stability of anti-icing asphalt mixture reduced, and the dynamic stability after short-term aging was improved. The tensile strain and tensile strength ratio of the anti-icing asphalt mixture reduced after long-term aging and soaking in water. In addition, the salt dissolution rate increased with the increase of anti-freezing filler content and the decrease of asphalt content. The research conducted suggests that the BP Neural Network can be utilized to predict the snow-melting ability of the anti-freezing asphalt mixture, and the regression coefficient of the predicted and measured salt dissolution was higher.