LAUSR

Abstract Chloride ingress is a severe problem for the durability of concrete structures worldwide, and the relevant researches are of great significance. According to the requirements on accurate and efficient prediction, the development of numerical simulation method is critical. Existing simulations on macroscale and mesoscale cannot balance solution efficiency and material variability well. In this paper, a refined method with cellular automata is proposed based on the mesoscopic features. The set of cell state variables is built up under the framework of cellular automata taking the effects of aggregate distribution, moisture transport, chloride diffusion and ion absorption into consideration, and the corresponding evolution rules are also developed. Due to the mutual relationship among the variables, the particular solution order is also specified in the method flow. The solution details such as the size of cell division and the length of time step are discussed and optimized, and the 10.0-mm cell size and 1.0-month time step is recommended for the stable and accurate calculation. To verify the accuracy of the method, the comparison between the refined method and the results of chemical tests is conducted, and the simulated relationship of average concentration to the depth is accurate. Compared with the macroscopic model and mesoscopic model, the concentration distribution caused by material variability is described more precisely in the model with cellular automata, and the efficiency is improved greatly as well.