A computationally efficient procedure for calibrating model parameters of multiple specimens

Model calibration can be a very intensive and time-consuming task, especially when dealing with non-linear and large finite element models. The computational effort further increases when multiple specimens have to be calibrated. This is typical of laboratory experiments where several specimens made with the same and/or different constituent materials are tested. This paper proposes a calibration procedure aimed at reducing the computational effort of multiple specimen model calibration. The calibration procedure combines the robustness of a surrogate-assisted evolutionary algorithm with the exploitation of a database collecting the results of the previously calibrated specimens. In this research, the proposed procedure is applied to the calibration of the parameters of a cohesive crack model for fiber-reinforced concrete specimens. The benefit of the proposed procedure is shown by comparing the results with those obtained from the same calibration method but without accounting for the previous results in the calibration of a new specimen.