Prediction of the Compressive Mechanical Behavior of Pozzolanic Cement Mortar Using a Coupled Thermodynamic–Numerical Modeling Approach

The use of pozzolanic materials, such as fly ash, microsilica, zeolite, and pumice, can reduce cement consumption and enhance the sustainability of cement-based materials. These materials influence cement hydration, setting time, and mechanical properties depending on their chemical composition and reactivity. Thermodynamic modeling using GEMS and the CEMDATA database enables the prediction of stable phases and hydration products. However, thermodynamic models cannot directly capture mesoscopic mechanical behavior, aggregate geometry, the interfacial transition zone (ITZ), or crack initiation and propagation. The Discrete Element Method (DEM) provides a complementary numerical approach for simulating the mesostructure of cement mortar and the mechanical interactions among its constituent phases. It also enables the investigation of nonlinear behavior, fracture, and crack development under loading. Therefore, this study combines thermodynamic modeling with DEM to establish a link between hydration characteristics, microstructural evolution, and the mesoscopic mechanical response of pozzolanic cement mortars, with particular emphasis on predicting compressive strength.

In this study, cement mortar mixtures consisting of a reference mixture and mixtures containing specified amounts of zeolite, pumice, and microsilica were prepared and investigated. First, thermodynamic hydration modeling was performed based on the chemical composition of cement and pozzolans, environmental conditions, and dissolved species. The hydration products, amounts of hydrated phases, and pH of the system were determined at different ages. Based on the modeling results, the mechanical properties of the cement paste and interfacial transition zone (ITZ) were subsequently determined. The effect of pozzolan content on the setting time was also investigated. In the second stage, the mechanical properties obtained from the thermodynamic model were used as input parameters for the discrete element model. Cubic mortar specimens were modeled in PFC2D by considering particle geometry, particle-size distribution, porosity, and the mechanical properties of the constituent phases, and were subjected to compressive loading. For model calibration and validation, the numerically predicted compressive strengths were compared with experimental results obtained at 7 and 28 days.

The thermodynamic modeling results showed that increasing the pozzolan content reduced the pore-solution pH and increased the C–S–H content up to 15% replacement. The initial and final setting times were also evaluated at 23 °C using calibrated C–S–H contents. Increasing zeolite and microsilica contents reduced both setting times, whereas increasing pumice content prolonged them. The numerical predictions agreed well with the experimental results, with a maximum relative error of approximately 3% for the final setting time of sample P10. The thermodynamic–numerical modeling approach predicted compressive strength with mean errors of 5.77% and 4.32% at 7 and 28 days, respectively, and overall MAE, RMSE, and R² values of 1.52 MPa, 1.73 MPa, and 0.95. The DEM model reproduced the experimental failure patterns and compressive stress–strain responses, with errors of approximately 8% in ultimate strain, 10% in peak strain, and 3% in peak compressive strength. The difference in energy absorption was approximately 3%. Furthermore, the compressive behavioral model derived from DEM simulations showed satisfactory agreement with the experimental results and the Kohees equation. Overall, the findings demonstrate the potential of the coupled thermodynamic–numerical framework to predict hydration characteristics, setting times, compressive strength, and mesoscopic mechanical behavior of pozzolanic cement mortars with satisfactory accuracy.