Review of Numerical Simulation for Chloride Transport in Cement-Based Materials: Progress, Bottlenecks and Future Directions

Abstract Chloride-induced corrosion is the dominant cause of durability degradation in concrete structures, and numerical modeling is an effective approach to reveal chloride transport mechanisms and predict structural long-term service performance. Based on standardized literature retrieval from Web of Science, Scopus and CNKI (2000 to early 2026), this review systematically summarizes the numerical modeling progress of chloride transport in cement-based materials from five key aspects: macroscopic theoretical models, meso- and micro-scale simulation techniques, multi-scale and multi-physics coupling theories, environmental erosion mechanisms, and performance optimization of modified concretes, including ultra-high performance concrete, graphene-modified concrete and recycled aggregate concrete. A unified macroscopic/mesoscopic/microscopic classification criterion is explicitly formulated within this review, which defines model scale according to the explicit discretization of heterogeneous mesophases rather than the complexity of chemical reactions, thereby clarifying the macroscopic nature of multi-ion coupling models and resolving the long-standing scale ambiguity in existing studies. This paper thoroughly analyzes the regulatory effects of typical complex environments, including marine exposure, freeze–thaw cycles and wet–dry alternations, on chloride ingress, and summarizes the anti-chloride mechanisms of various modified cementitious materials. Three core bottlenecks limiting the engineering application of current numerical models are identified, namely insufficient adaptability of fully coupled CHTM models under extreme environments, imperfect long-term performance verification systems for emerging modified materials, and unclear quantitative cross-scale correlations restricting model engineering transformation. Accordingly, prospective research directions are proposed, focusing on extreme-environment multi-field coupling models, long-term dynamic evolution simulation frameworks, and engineering-oriented reduced-order cross-scale methods. Particular attention is paid to the application potential of data-physics hybrid modeling based on physics-informed neural networks. Different from existing reviews that merely summarize single-scale models or experimental phenomena, this work establishes a complete logical chain spanning fundamental transport theories, multi-scale computation and material engineering applications. The inherent limitations are also clarified: machine learning methods are only prospected without systematic elaboration, and long-term multi-field experimental data for novel modified concrete remain insufficient. This review provides reliable theoretical support for the durability design and full-life-cycle evaluation of concrete structures.