A Hybrid Approach to Test-Analysis-Model Development for Large Space Structures

Abstract

A new finite element model (FEM) reduction method is presented for use in the generation of Test-Analysis-Models (TAM) in test-analysis correlation. The method addresses the concern that some current TAM methodologies, specifically the Modal TAM, are overly sensitive to differences between test mode shapes and analysis mode shapes. This sensitivity can result in large off-diagonal terms within the orthogonality and cross-orthogonality matrices used for test-analysis mode shape correlation. It has been hypothesized that the sensitivity is due to the Modal TAM’s poor representation of residual mode shapes and frequencies which are modes that are not targeted for identification. In many cases it has been observed that the less accurate static TAM often gives better off-diagonal correlation results. A new Hybrid TAM methodology is developed to combine the exact representation of the FEM target modes from the Modal TAM with the more accurate static TAM representation of the residual modes. The superior residual dynamics representation of the Hybrid TAM is demonstrated for both a simple spacecraft and a much more detailed representation of a Large Space Structure. Simulated test-analysis correlation results are presented for both examples where test mode shapes are represented by FEM target modes with noise modeled as a random linear combination of all FEM modes. Analysis indicates that the Hybrid TAM’s improved residual representation results in reduced sensitivity of the test-analysis correlation to model error.