Multi-level curvature-based parametrization and model updating using a 3D full-field response

Model updating improves the correlation between the response of the real structure and the response of the finite-element (FE) model; however, the selection of the updating parameters (parametrization) is crucial for its success. Using full-field modal shapes, a large number of parameters can be updated, e.g., the Young's moduli of all the finite elements; however, the structural response is not necessarily sensitive to an arbitrary parameter, making the optimization problem ill-conditioned. Additionally, the computation of the full sensitivity matrix is not feasible for relatively large FE models. Not all locations are equally important for model updating; at locations of the highest mechanical loads, more focus is required. In this research, the updating parameters are based on the curvature of the 3D full-field experimental shape, where locations with high curvature are associated with high sensitivity. The assumption is initially researched with the Euler-Bernoulli beam elements and second-order tetrahedrons. The proposed method is investigated on numerical and real experiments, where successful updating was confirmed. With the proposed parametrization and updating approach, a geometrically complex structure is parametrized and the parameters updated without significant user input, generalizing the model-updating procedure.

Automated summary

Model updating improves the correlation between the real structure response and the finite-element model response. The selection of updating parameters is crucial for success. Full-field modal shapes can update a large number of parameters, but the structural response may not be sensitive to arbitrary parameters, and computation of the full sensitivity matrix is not feasible for large FE models. This research investigates updating parameters based on the curvature of 3D experimental shape, and successfully applies the proposed method to numerical and real experiments.