The evaluation of material properties with complex microstructures is challenging due to the uncertainties caused by the microstructural features. In this study, a multiscale framework was developed to effectively evaluate the tensile strength of cement paste by combining real and virtual experiments. The statistical uncertainty of the tensile strength was efficiently reduced using a Bayesian updating approach. The real cement paste microstructures were obtained from high-resolution micro-CT and used to train the extended generative adversarial network (GAN) to reconstruct virtual multiphase microstructures [1]. Latin hypercube sampling (LHS) was used to generate characteristics for virtual microstructures, which were used in the GAN model to reconstruct virtual microstructures for simulation. The tensile strength of the virtual microstructures was evaluated using a phase-field fracture model [2], where the input material parameters were determined from the multiscale correlations between micro-CT image features and nanoindentation experiments. To implement Bayesian updating, real experiment data were used to estimate the prior tensile strength distribution parameters, and additional virtual experiment data points were used as observed data. The uncertainties of the updated posterior strength distributions were significantly reduced. Conducting real experiments to evaluate material properties requires considerable time and effort. Using the multiscale framework of the virtual experiment to complement real experimental data through Bayesian updating can efficiently reduce the effort and increase the accuracy. It is confirmed that with further development, the proposed framework can be used to accelerate the development of new cementitious materials and the evaluation of existing materials. REFERENCES [1] S.-W. Hong, S.-Y. Kim, K. Park, K. Terada, H. Lee, T.-S. Han, Mechanical property evaluation of 3D multi-phase cement paste microstructures reconstructed using generative adversarial networks, Cement and Concrete Composites, 152:105646, 2024. [2] C. Miehe, L.-M. Schänzel, H. Ulmer, Phase field modeling of fracture in multi-physics problems. Part I: Balance of crack surface and failure criteria for brittle crack propagation in thermo-elastic solids, Computer Methods in Applied Mechanics and Engineering, 294:449–485, 2015.