Model Reduction and Sensor Placement Methods for Finite Element Model Correlation

The issue of model reduction is one that must often be overcome in order to perform the necessary checks as part of the spacecraft Finite Element Model (FEM) validation process. This work compares different reduction methods; specifically the popular and long-standing Guyan method, and the potentially more accurate System Equivalent Reduction Expansion Process (SEREP). The influence of sensor set location on the quality of the reduced model has also been considered, and the commonly applied methods to maximize kinetic energy and effective independence have been applied. These investigations have taken the form of studies involving two large, unique, scientific spacecraft. The computational results are compared with experimental results that are also detailed in the paper. The findings highlight the potential issues with the accuracy of a Guyan reduced model in replicating the full system dynamics, even with a reasonably large sensor set. It is shown that this can be improved slightly in some circumstances through implementation of sensor set placement optimization techniques. The SEREP method is shown to have the benefit of being more accurate at replicating the full system behavior than the more traditional Guyan method, while also producing higher diagonal values in cross-orthogonality comparisons between FEM and test.