Finite Element Model Tuning with Spatially-Dense 3D Modes

Abstract

An investigation into varying the spatial density of three-dimensional (3D) modes in order to accurately tune a finite element (FE) model to larger numbers of modes is conducted. This FE model tuning approach is evaluated on a 60,000+ FE degree of freedom (DOF) model of the United States Air Force Academy’s fifth small satellite, FalconSAT-5 (FS-5), structural engineering model (SEM). The evaluation of this FE model tuning approach starts with the collection of natural frequencies and 3D modes extracted from scanning laser Doppler velocimeter (LDV) frequency response function measurements at 2,165 closely-spaced points on the surface of the test article. The measured modes and their associated natural frequencies serve as target values in a gradient-based tuning approach. The FE model is tuned to have differences with the measured natural frequencies less than 2% and in many cases cross-orthogonality values greater than 0.90. Using both a QR-decomposition and the Triaxial Effective Independence (EfI3 + ) sensor selection strategies, the effect of adding more tuning points on tuning accuracy is studied. This study demonstrates that increasing the number of experimental data points using the EfI3 + sensor selection strategy for FE model optimization results in increased tuning accuracy.