Considerations to reduce modal analysis test variability

In the design and development of automobiles, the test results on prototype hardware are typically used to verify and tune finite element models and support design decisions. Often the underlying assumption is that the test results represent the true answer although it is well known that there is some inherent variability in any test procedure. Therefore, it is prudent to conduct tests in such a way as to minimize variability and also to quantify the test variability. That variability can depend on many factors including the test setup, data acquisition, and the effects of nonlinearities. As part of an experiment designed to quantify test-to-test variability in the modal analysis of automobiles, typical modal test procedures were scrutinized for ways to reduce variability. Specifically, several topics were addressed including input force level and waveform, selection of number and location of inputs, the shaker attachment method, roving vs. non-roving accelerometers, calibration issues, and signal processing considerations. This paper discusses these issues and shows the results from an analysis of variance on some of the modal parameters.