The ice melting performance of three types of deicers, including sodium chloride, calcium chloride and sodium acetate, were tested in laboratory under different temperature conditions, and their effects on asphalt mixture were evaluated from the point of the stripping resistance of asphalt mixture. Unsaturated Marshall samples were exposed to freeze-thaw cycling while immersed in the deicer solutions of different concentrations. After the freeze-thaw cycles, Cantabro tests were performed, and Cantabro loss was adopted to characterize the stripping resistance of asphalt mixture. The test results show that calcium chloride has the best comprehensive ice melting performance, and all deicers have detrimental effect on the stripping resistance of asphalt mixture at different degrees. The damage degree depends on deicer types and their concentration in the solution. Deicer solutions with about 2% concentration cause the greatest loss of stripping resistance due to serious freeze-thaw damage. Sodium acetate causes greater loss of stripping resistance than sodium chloride and calcium chloride at the same concentration.
To investigate the validity of two dynamic modulus predictive models（ Witczak 1-37 A viscosity-based model and Witczak 1-40 D shear modulus-based model） in the context of Jiangsu, and evaluate the effect of different mixture design variables（ aggregate gradations, binder type, and volumetric properties） on dynamic modulus E＊, asphalt mixtures commonly used in the local surface layer, including Sup-13 and AC-13, are prepared in the laboratory and their dynamic modulus E＊values are predicted based on the above mentioned models. The corresponding asphalt tests, including viscosity and dynamic shear modulus tests, are also carried out to obtain the prediction model parameters. The test results showthat binder type and asphalt content have a significant impact on dynamic modulus.There is a good correlation between the E＊values based on above two predictive models and the measured E＊, while a relatively lower bias can be expected from Witczak 1-37 A model. The test results can be used for the calibration of dynamic modulus with higher accuracy.
