Characterization of the dynamic behavior of structures using the Kriging surrogate and experimental data

The finite element method (FEM) is a useful numerical tool for solving engineering problems in general. FEM simulates the behavior of several structures regarding the boundary conditions applied. However, the FEM solution presents differences when compared to experimental data. A systematic approach called finite element method update (FEMU) can reduce those differences. Here, the structural or material parameters change following an optimization problem. Thus, this study aims to characterize the dynamic behaviors of structures using a methodology of FEMU. This methodology reduces the differences between the numerical and experimental dynamic behavior of structures. Also, it uses the Kriging surrogate and experimental data to identify geometric parameters and boundary conditions. We perform the simulations and the experiments in two structures: a fixed-free beam and a frame structure. The objective function quantifies the square root of the sum of the weighted quadratic differences between numerical and experimental frequency response functions. We develop numerical models and routines of the methodology in the Abaqus and MATLAB software packages. Finally, the methodology characterizes the numerical dynamic behavior of the studied structures quite accurately regarding the experimental data.