An application of different surrogate models to model updating of a heritage structure

Abstract

Abstract The preservation of historic structures is of primary importance due to their significant cultural value. However, destructive tests as well as invasive interventions should be as much limited as possible. In this contest, Finite Element model updating can be applied as a non-destructive method to indirectly estimate selected unknown parameters through numerical model calibration. To this aim, an inverse problem has to be solved by minimizing the gap between corresponding experimental and numerical estimates of modal parameters. However, the application of Finite Element model updating is usually associated with large computational efforts. These can be reduced through the choice of a suitable surrogate model. Douglas-Reid method and Response Surface method are frequently applied to this aim due to their relatively simple quadratic formulations. In spite of the similarity of the mathematical formulation of the surrogate models in the two cases, the methods show some peculiarities affecting their application and, in some cases, their performance. The present paper illustrates an application of Douglas-Reid method and Response Surface method to a historical building, reporting a comparative assessment of the two approaches in terms of computational time and model updating results.