Investigation of the effectiveness of using an experiment to validate experimental substructure models

摘要

Experimental-to-analytical substructuring has been investigated as a way to avoid costly tests and analyses on large systems by experimentally characterizing certain subcomponents and then coupling them to an analytical model for the rest of the structure. Unfortunately, substructuring has sometimes proven difficult to implement because the substructuring calculations can be sensitive to modal truncation and experimental noise and often require the measurement of rotational motions at the interface. These issues have led to decreased confidence in substructuring calculations, causing it to remain underutilized in industry. This work proposes performing a small, inexpensive validation test to give confidence or reveal flaws in an experimentally derived model. This is only meaningful if the error in the validation test is characteristic of the error in an application of interest; this work explores this assumption. The validation test is evaluated in two scenarios which represent substructuring realities. The first scenario investigates measurement errors by applying normally distributed noise to a simple beam structure and evaluating how well the validation test compares to the application of interest. The second scenario simulates substructuring of a complicated engine-generator system with a statically indeterminate connection, where the effect of modal truncation is not easy to anticipate.