Cementitious concrete structures serving in sulfate environments suffer from serious durability challenges caused by chemical sulfate attacks (CSA), which lead to the volume expansion, cracking, and spalling of concrete and the early failure of structures. CSA on concrete involves the behaviors of ion transport, chemical reactions, the crystallization of reaction products, microstructural damage to the cement matrix, and the macroscopic deterioration of concrete, namely the transport-chemo-mechanical behaviors. This paper first introduces the reaction products, such as gypsum, ettringite, brucite, and thaumasite, between sulfate and concrete under different environmental conditions and their formation mechanism. Then, aiming at the ettringite type CSA, the theories of volume increase and crystallization pressure are elaborated to explain it-induced concrete degradation. Additionally, the crystallization pressure theory is used to describe the cracking behavior in the microstructure slurry caused by the ettringite crystal filling pore. Finally, a series of transport-chemo-mechanical models for ettringite type CSA are displaced module by module. It includes the sulfate diffusion-reaction model, the free expansion of concrete, and equivalent expansive force in concrete related to the reaction behavior: the model for chemo-mechanical behavior in concrete caused by CSA. These models can be used to analyze the distribution of sulfate ions and the reaction product content, expansive stress and strain in the concrete, and the cracking and spalling degree of the concrete, which is beneficial to evaluate the durability of concrete structures serving permanently in a sulfate environment.

Mineral admixtures (MA), like fly ash (FA), silica fume (SF), and slag (S), are usually added to cement-based materials to improve their compactness and further enhance their mechanical properties, permeability resistance, and durability. In this study, low-field nuclear magnetic resonance (LF-NMR) is adopted to explore the evolution of the early hydration characterization of cement-based materials with MA by testing the transverse relaxation time T2. Meanwhile, the effect of MA on mechanical properties is analyzed by measuring compressive and flexural strength. The results show that, in the early hydration (0–7 days), the T2 distribution shows a trend of gradually moving to a short relaxation time and changes from a double peak to a single main peak. The decrease in T2i (main peak vertex) means that the evaporated water is gradually distributed in smaller pores with more motion constraints. However, the type and content of MA have little effect on T2i. Porosity gradually decreases in the period of early hydration. The addition of MA causes the porosity to decrease, and the order influence is FA > S > SF, i.e., the porosities of cement paste with 0%MA, 10%FA + 10%SF, 10%FA + 10%S, and 20%FA at 7 days are 48%, 44.5%, 40.7%, and 40.2%, respectively. Additionally, the addition of MA to cement-based materials also decreases the early strength, and the influence order is FA > S > SF, i.e., the compression strength values of cement paste with 0%MA, 10%FA + 10%SF, 10%FA + 10%S, and 20%FA at 7 days are 47.8 MPa, 40.1 MPa, 38.6 MPa, and 37 MPa, respectively.

In the background of sustainable development in the construction industry, recycled cement mortar (RCM) has emerged as a research hotspot due to its eco-friendly features, where mechanical properties serve as critical indicators for evaluating its engineering applicability. This study proposes an artificial neural network (ANN) model optimized by intelligent algorithms, including the GWO (grey wolf optimizer), PSO (particle swarm optimization), and a GA (genetic algorithm), to predict the compressive strength of recycled mortar. By integrating experimental and prediction data, we establish a comprehensive database with eight input variables, including the water–cement ratio (W/C), cement–sand ratio (C/S), fly ash content (FA), aggregate replacement rate (ARR), and curing age. The predictive performance of neural network models with different database sizes (database 1: experimental data of RCM; database 2: experimental data of RCM and ordinary mortar; database 3: model prediction data of RCM, experimental data of RCM, and ordinary mortar) is analyzed. The results show that the intelligent optimization algorithms significantly enhance the predictive performance of the ANN model. Among them, the PSO-ANN model demonstrates optimal performance, with R² = 0.92, MSE = 0.007, and MAE = 0.0632, followed by the GA-ANN model and the GWO-ANN model. SHAP analysis reveals that the W/C, C/S, and curing age are the key variables influencing the compression strength. Furthermore, the size of the dataset does not significantly influence the computation time for the above models but is primarily governed by the complexity of the optimization algorithms. This study provides an efficient data-driven method for the mix design of RCM and a theoretical support for its engineering applications.