2025 年 23 巻 12 号 p. 571-581
Evaluating the responses of cementitious materials with complex microstructures through experiments alone requires substantial time and effort. To address this, a comprehensive multiscale analysis framework combining experiments and simulations was adopted to identify correlations between microstructural features and mechanical properties. The framework comprises four modules: fitting/normalization, phase identification, virtual specimen generation, and finite element method (FEM) simulation. X-ray computed tomography (CT) and nanoindentation data were normalized, and their correlation was established using cumulative distribution functions (CDF). Based on this relationship, mechanical properties were inferred from CT data, and virtual specimens were generated for subsequent FEM simulation. The FEM module was calibrated by comparing simulated results with experimental splitting tensile tests. High-resolution micro-CT enabled detailed microstructural characterization for virtual specimen generation. The results demonstrated that porosity and pore connectivity are strongly correlated with the mechanical behavior of cement paste. Overall, the proposed multiscale analysis framework offers a robust and scalable approach for microstructural analysis and is expected to enhance the efficiency of material characterization and development process.