Exploring the Interactions of Chloride Deicer Solutions with Nanomodified and Micromodified Asphalt Mixtures Using Artificial Neural Networks

Abstract

The objectives of this research are to modify an asphalt mixture with two materials—nanoclay and carbon microfiber—and to investigate the interactions of chloride deicer solutions with nano- and/or micromodified and unmodified asphalt mixtures in terms of indirect tensile strength (ITS) and fracture energy. Artificial neural networks (ANNs) were used in this study to establish predictive models and quantify the complex cause-and-effect relationships between the nano- or micromodification and conditioning of asphalt mixtures and the resulting mechanical properties. Four influential variables (nanoclay content, microfiber content, deicer type, and deicer dilution ratio) were collectively examined to predict the ITS and fracture energy of asphalt mixtures, and a back-propagation neural network of three layers with seven or nine hidden nodes was employed respectively. The established ANN models were then successfully used for numerical investigations on the parameters affecting the asphalt properties. The addition of polysiloxane-modified montmorillonite and/or carbon microfiber (both at less than 2% by weight of asphalt binder) can enhance the tensile strength fracture energy of asphalt concrete mixtures and reduce their moisture susceptibility and cracking risk, and such benefits are especially significant when the asphalt concrete is conditioned in water or chloride-based deicer solutions. This evaluation makes it possible to design asphalt mixtures for a desired level of ITS or fracture energy in the absence or presence of common chloride-based deicer solutions.