In practical engineering applications, the use of low-carbon concrete materials is in line with the principles of sustainable development and helps to reduce the impact on the environment. Creep effects are particularly critical in the research on such materials. However, traditional characterization methods are time-consuming and often fail to account for the interactions of multiple factors. This study constructs a time-series database capturing the behavioral characteristics of low-carbon concrete materials over time. Three temporal prediction models—Artificial Neural Network (ANN), Random Forest (RF), and Long Short-Term Memory (LSTM) networks—were retrained for creep prediction. To address limitations in model architecture and algorithmic frameworks, an enhanced Adaptive Crowned Porcupine Optimization algorithm (ACCPO) was implemented. The improved performance of the ACCPO was validated using four diverse benchmark test functions. Post-optimization results showed remarkable improvements. For ANN, RF, and LSTM, single-metric accuracies increased by 20%, 19%, and 6%, reaching final values of 95.9%, 93.9%, and 97.8%, respectively. Comprehensive evaluation metrics revealed error reductions of 22.6%, 7.9%, and 8% across the respective models. These results confirm the rationality of the proposed temporal modeling framework and the effectiveness of the ACCPO algorithm. Among them, the ACCPO-LSTM time series model is the best choice.

Non-uniform corrosion cracking in reinforced concrete buildings constitutes a fundamental difficulty resulting in durability failure. This work develops a microscopic-scale multi-species electrochemical phase field model to tackle this issue. The model comprehensively examines the spatiotemporal coupling mechanisms of the full “corrosion-rust swelling-cracking” process by integrating electrochemical reaction kinetics, multi-ion transport processes, and a unified phase field fracture theory. The model uses local corrosion current density as the primary variable to accurately measure the dynamic interactions among electrochemical processes, ion transport, and rust product precipitation. It incorporates phase field method simulations of fracture initiation and propagation in concrete, establishing a bidirectional link between rust swelling stress and crack development. Experimental validation confirms that the model’s predictions about cracking duration, crack shape, and ion concentration distribution align well with empirical data, substantiating the efficacy of local corrosion current density as an indicator of electrochemical reaction rate. Parametric studies were performed to examine the effects of interface transition zone strength, oxygen diffusion coefficient, protective layer thickness, reinforcing bar diameter, and reinforcing bar configuration on cracking patterns. This model’s multi-physics field coupling framework, influenced by dynamic corrosion current density, facilitates cross-field interactions, offering sophisticated theoretical tools and technical support for the quantitative analysis, durability evaluation, and protective design of corrosion-induced cracking in reinforced concrete structures.

Fully graded concrete exhibits significantly enhanced mechanical properties due to optimized aggregate gradation. However, the effects of coarse aggregate morphological characteristics at the microscale—such as axial ratio (Ar) and angularity coefficient (Ac)—on fracture behavior remain insufficiently understood and require further investigation. This study investigates the fracture behavior of fully graded concrete using a peridynamic approach. First, a multilinear constitutive model of concrete considering nonlocal effects and damage evolution is constructed. Second, aggregate modeling via the Tensile-Concave Synthesis Method in peridynamics, and different property bonds and random bond parameters are introduced to characterize the heterogeneity and initial defects of concrete. Finally, the effects of loading rate, aggregate randomness, and the parameters Ar and Ac on the uniaxial tensile properties of concrete are systematically studied. The results demonstrate that peridynamics accurately captures the entire process of crack initiation and propagation in fully graded concrete. Uniaxial tensile behavior exhibits strong rate dependence. Different random aggregate models result in variations in the peak tensile strength of concrete. With the increase in Ar, the peak tensile stress gradually decreased by 4.15%, whereas the elastic modulus increased by 14.31%. As Ac increased, the peak stress exhibited an overall trend of first increasing by 5.14%, followed by a decrease of 3.8%. Therefore, in numerical simulations, the influences of loading rate and aggregate randomness should not be overlooked. Moreover, to enhance the strength of fully graded concrete, the proportion of aggregates with large Ar and Ac values should be minimized.

This study proposes a physical-prior and residual-learning method for creep prediction of low-carbon concrete and integrates it into a multi-objective optimization framework for mix proportion design. After cleaning and normalizing 1066 creep test records, the B4 model served as the physical prior backbone, and a sparse variational Gaussian process (SVGP) was used to fit the residuals between measured and theoretical values. Global linear calibration, a Matérn-5/2 kernel with automatic relevance determination, and length-scale regularization were used to suppress kernel collapse and overfitting. The B4+SVGP model was compared with random forest, extreme gradient boosting, and a backpropagation neural network under a unified dataset, feature set, and training protocol, and Shapley additive explanations (SHAP) were used to analyze mixture effects on creep compliance. The B4+SVGP model that performed best under record-level validation was then embedded in NSGA-II to optimize net carbon emissions and 1000-day creep compliance simultaneously; TOPSIS was used to derive mix proportions under carbon-emission-oriented, balanced, and creep-oriented weight preferences. Under a unified record-level 10-fold cross-validation protocol, B4+SVGP achieved favorable prediction accuracy (R² = 0.9660) and retained a relative advantage in the target mix proportion region. Loading duration, age at loading, and 28-day compressive strength were the most influential factors, consistent with the physical mechanisms of the B4 model. Under strength and mix-feasibility constraints, the Pareto front exhibited an L shape with an inflection at approximately 205 kg/m³ net carbon emissions. The balanced (5:5) scenario yielded a candidate with a water-to-binder ratio of 0.4227, a binder content of 300.03 kg/m³, and fly ash/slag/silica fume replacements of 6.76%/6.70%/14.82%; it reduced net carbon emissions by approximately 18.7% while increasing creep compliance by only approximately 2.1%. The resulting mixtures are computational candidates whose fresh-state properties, compressive strength, and long-term creep behavior require experimental verification before engineering recommendation.