Multiscale model calibration by inverse analysis for nonlinear simulation of masonry structures under earthquake loading

The prediction of the structural response of masonry structures under extreme loading conditions, including earthquakes, requires the use of advanced material descriptions to represent the nonlinear behaviour of masonry. In general, micro- and mesoscale approaches are very computationally demanding, thus at present they are used mainly for detailed analysis of small masonry components. Conversely macroscale models, where masonry is assumed as a homogeneous material, are more efficient and suitable for nonlinear analysis of realistic masonry structures. However, the calibration of the material parameters for such models, which is generally based on physical testing of entire masonry components, remains an open issue. In this paper, a multiscale approach is proposed, in which an accurate mesoscale model accounting for the specific masonry bond is utilised in virtual tests for the calibration of a more efficient macroscale representation assuming energy equivalence between the two scales. Since the calibration is performed offline at the beginning of the analysis, the method is computationally attractive compared to alternative homogenisation techniques. The proposed methodology is applied to a case study considering the results obtained in previous experimental tests on masonry components subjected to cyclic loading, and on a masonry building under pseudo-dynamic conditions representing earthquake loading. The results confirm the potential of the proposed approach and highlight some critical issues, such as the importance of selecting appropriate virtual tests for model calibration, which can significantly influence accuracy and robustness.