Concrete durability has become a central concern in civil engineering as structures are increasingly exposed to complicated and aggressive environments. In practical service conditions such as marine tidal zones, cold regions, sulfate-rich soils, or industrial atmospheres, deterioration rarely occurs in isolation. Instead, chloride ingress, carbonation, sulfate attack, calcium leaching, freeze–thaw damage, and load-induced cracking often coexist and interact on multiple physicochemical and geometric levels. These complex processes alter transport behavior, pore structures, chemical equilibria, and mechanical integrity, resulting in highly nonlinear deterioration that accelerates beyond the sum of individual mechanisms. As conventional models cannot fully capture such synergistic effects, advanced numerical modelling has emerged as a vital tool for durability prediction under multi-deterioration scenarios.

This paper reviews recent progress in modelling durability degradation under multiple deterioration mechanisms, with emphasis on both chemically driven and physically driven coupling effects. In chemically dominated degradation, chloride ingress is recognized as the most rapid and detrimental process, and it is strongly modified by concurrent chemical reactions. The interaction between carbonation and chloride transport is particularly complex: carbonation can decompose chloride-binding phases, releasing previously bound chlorides, while simultaneously refining the pore structure through calcium carbonate precipitation. Numerical models incorporating carbonation rate, degree, and pore structure evolution have enabled more accurate quantification of chloride distribution in fully carbonated and partially carbonated regions. Similarly, for combined sulfate–chloride attack, competitive adsorption and expansion-induced microcracking play decisive roles. Thermodynamic equilibrium models and reaction-kinetic models have been used to capture the competition among ions, the formation of expansive products, and the resulting effects on transport properties. These approaches highlight the necessity of considering multi-ion coupling and electrochemical interactions when assessing deterioration severity. Calcium leaching represents another critical chemical mechanism that strongly influences chloride behavior. The dissolution of calcium-bearing hydrates coarsens the microstructure, increases pore connectivity, and reduces binding capacity, thereby accelerating chloride ingress. Multi-ionic transport frameworks have been adopted to simulate these processes, showing that electrochemical coupling initially promotes leaching but later tempers its progression as chemical gradients evolve. Such findings emphasize the importance of capturing time-dependent feedback within reactive transport models.

On the physically driven side, processes such as freeze–thaw damage and load-induced cracking profoundly alter the geometric pathways of ionic transport. Freeze–thaw cycles induce pore dilation, microcracking, and structural weakening, forming preferential channels for chloride ingress. Models coupling thermal transfer, moisture and ionic transport have shown that although higher salt concentrations reduce freezing rates, the overall deterioration remains accelerated due to increased permeability. The synergistic effects of freeze–thaw damage and chloride transport thus demand integrated modelling strategies capable of representing dynamic pore evolution. Load-induced cracking also creates high-permeability pathways that significantly influence harmful ions transport. Mesoscale lattice models, multi-ionic models, and other models have demonstrated that crack width, shape, and orientation govern diffusion and migration behavior. However, establishing fully bidirectional coupling, where chloride ingress promotes crack propagation and crack propagation further accelerates chloride ingress, remains an unresolved challenge. Existing methods rely on staged or quasi-coupled approaches, often assisted by statistical learning, yet true geometric updating during crack evolution is still computationally difficult. This limitation highlights a critical frontier for future numerical modelling.

Research on durability degradation under multiple deterioration mechanisms has made remarkable progress, yet several breakthroughs are needed to achieve reliable long-term predictions. First, advancing the coupling of multi-mechanism models is essential. The mutual interactions between crack propagation and ionic transport, as well as pore-structure evolution under multiple deterioration modes, requires future numerical models that can dynamically update geometry and transport properties over time. More comprehensive physical representations of environmental effects such as temperature, humidity, salt concentration, freeze–thaw intensity will further enhance predictive accuracy. Second, there is an urgent need to improve the geometric dimensionality and computational efficiency of multi-scale, multi-physics frameworks. High-resolution three-dimensional modelling is still limited for multi-ion systems due to nonlinear chemical reactions and the complexity of heterogeneous microstructures. Future work should focus on algorithm optimization, parallel computing, and innovative discretization schemes to enable large-scale simulations with realistic material characteristics. Third, the scope of durability research must expand across both spatial and temporal scales. Linking micro-level deterioration mechanisms with macro-scale structural performance remains a key challenge, particularly for reinforced concrete exposed to multiple aggressive agents. Early-age behavior, interfacial transition zone development, and long-term interaction between mechanical and chemical processes should also be considered into lifecycle-oriented predictive frameworks. Finally, as sustainable and alternative binders such as alkali-activated materials and high-performance composite cements become more prevalent, existing models must be adapted or reconceptualized to accommodate their unique microstructure, chemical composition, and transport properties. Understanding multi-deterioration behavior in these emerging materials will be essential for their safe and widespread engineering application. Overall, progress in numerical modelling, strengthened by data-driven techniques, multi-scale experiments, and advanced characterization methods, is expected to transform our ability to predict durability degradation under complex service conditions. These advances will ultimately support the development of more resilient, sustainable, and long-lasting concrete structures.

The durability of cementitious materials in coastal environments is fundamentally challenged via salt crystallization, and a prevalent deterioration mechanism alters pore structure and degrades transport properties. While total porosity is commonly considered, pore size distribution (PSD) proves more critical. It is indicated that large pores (i.e., 0.1–5.0 μm) facilitate salt solution ingress, while small pores (i.e., <0.1 μm) generate high crystallization pressures. However, the existing models often fail to capture the complex crystallization at wet-dry interfaces in cement-based materials. This study was to develop a comprehensive mathematical framework using log-normal distributions to characterize the initial PSD and model its evolution under salt attack. We established predictive methods for permeability and sorptivity that could account for pore structure evolution, with validation through independent experiments. In addition, this study also analyzed the performance variations under different testing conditions and investigated water absorption behavior during salt crystallization.

The research methodology was founded on theoretical modeling and third-party experimental validation. First, a mathematical model for pore structure evolution was proposed. The initial pore size distribution (PSD) of cementitious materials was described by a composite log-normal distribution function, dividing pores into four families, i.e., large interlayer pores, gel pores, small capillary pores, and large capillary pores. Two distinct pore-filling paths during salt crystallization were hypothesized, i.e., A uniform (U) filling model, where salt deposits uniformly across all pores with a constant corrosion amount, and a non-linear (NL) filling model, where crystallization is concentrated near the wet-dry interface, with a corrosion amount proportional to the initial pore radius. The position of this critical interface was determined based on ambient relative humidity (RH) using the Kelvin-Laplace equation.

Subsequently, a novel intrinsic permeability prediction model (i.e., the Pore Size Distribution-Permeability (PSD-K) model) was derived from the Hagen-Poiseuille equation, explicitly incorporating total porosity, pore tortuosity, and the full PSD. For unsaturated conditions, this model was extended to predict gas permeability via integrating the wet-dry interface parameter r_c. Further refinements accounted for gas slippage effects and the influence of water saturation based on the Van Genuchten-Mualem model, yielding an apparent gas permeability. Finally, a sorptivity and capillary rise height model was formulated via coupling the PSD-K model with capillary pressure described by the Young-Laplace equation. All models were rigorously validated with the published experimental data from various mortar specimens subjected to salt exposure and drying-wetting cycles.

The validation against MIP data from a PC specimen after Na₂SO₄ immersion reveals a porosity reduction from 25.9% to 16.9%. The nonlinear (NL) filling model demonstrates a superior capability in capturing localized pore structure evolution, compared to the uniform (U) model. For a calculated critical radius of 64.9 nm at RH of 96.5%, the NL model accurately predicts a pore modification in the size range of 20–100 nm, dominating transport processes. However, discrepancies occur in smaller (i.e., <20 nm) and larger (i.e., >100 nm) pore regions, primarily due to the oversimplified assumption of uniform hydration product deposition. This indicates a need for future models to incorporate non-uniform hydration effects for improved prediction accuracy.

After the establishment of a reliable pore structure evolution model, we further evaluate its impact on the key transport property indicator–permeability coefficient. The PSD-K model can predict an intrinsic permeability for six mortar groups, having reasonable agreement with gas intrusion data (i.e., 0.66–7.51×10^-17m² for M40–M60). However, the PSD-K and the Katz–Thompson (K–T) models both overestimate water-based measurements by 1–3 orders of magnitude, revealing fundamental test-method dependencies. This discrepancy stems from C-S-H gel swelling during water intrusion, which compresses pore space versus enhanced flow due to gas slippage effects in dried specimens. For unsaturated conditions, the modified model confirms a critical water saturation threshold (i.e., about 0.8), beyond which gas permeability decreases dramatically as a liquid phase blocks continuous pore networks. The PSD-K model effectively tracks the permeability-porosity relationship during salt crystallization, demonstrating the model capability in simulating the performance degradation under chemical attack.

Beyond permeability, capillary water absorption represents another crucial process driving salt solution ingress, whose evolution patterns are similarly investigated. The sorptivity model shows a correlation with the experimental data for mortar specimens with varying slag contents. The results indicate that the sorptivity of GBFS0 mortar (i.e., 39.4% porosity) exhibits approximately four times higher than that of GBFS50 mortar (i.e., 8.4% porosity) due to its greater volume of large capillary pores (i.e., >100 nm) that dominate absorption processes. The results of environmental analysis reveal a negative correlation between RH and sorptivity as a higher initial water saturation reduces capillary driving forces, particularly in finer pores. Furthermore, salt-induced porosity increases significantly an enhanced sorptivity, with this effect being markedly pronounced under high-humidity conditions (RH > 80%) where the wet-dry front occupies larger pores, making absorption more sensitive to pore volume changes. These findings can provide crucial insights for predicting salt damage risks in various service environments.

This study established a theoretical framework for quantifying salt crystallization impacts on cementitious materials. The key findings demonstrated that the nonlinear pore-filling model could predict pore structure evolution, particularly in the critical size range of 20–100 nm, with the MIP data confirming a porosity reduction from 25.9% to 16.9% and crystallization localization at wet-dry interfaces. The PSD-K model could predict an intrinsic permeability, showing a reasonable agreement with the gas intrusion data (i.e., ~7.51×10^-17 m² for M60) and identifying a critical water saturation threshold (i.e., ~0.8) for gas transport. The sorptivity model captured a fourfold difference between high- and low-porosity mortars, revealing that salt-induced porosity increase coupled with a high humidity (i.e., RH>80%) drastically enhanced water absorption. This integrated framework could provide an effective tool for durability assessment of concrete in salt environments.