This paper was produced for the 2019 NAFEMS World Congress in Quebec Canada

Resource Abstract

Frequency domain methods for analysis and testing are experiencing a resurgence due to improved computer processing methods and newly developed algorithms. Recent SAE and NWC17 papers have demonstrated both the accuracy and benefits of these frequency domain methods. Time based methods will continue to be used for non-dynamic applications and for other specific types of dynamic calculations involving transients and non-linear systems. But, between the frequency domain and the time domain there is another kind of analysis, which uses sine sweeps.



Sometimes these sine sweeps are mixed with random loads (sine on random) and other times they are applied individually. The sweep is basically a sine input of specified magnitude and frequency where the frequency is varied in a prescribed manner. The sweep can typically be linear, logarithmic, or some kind of dwell around resonances. The dwell portions can also sometimes be perturbed in order to ensure that changing system properties (changing dynamic resonance frequencies) do not cause a “miss” in the maximum resonance. Typically, these test specifications will involve multiple “events” where each event represents a specific type of loading and a specific loading direction. Load specifications can be single channel (common) or multi-channel (less common in the frequency domain). All of the above procedures can be applied in a test laboratory, and there is now considerable interest in applying exactly the same procedures in a CAE analysis environment.



This paper will explore the relationship between the sine sweep test and an equivalent PSD loading and show that under certain conditions the two load types can be made equivalent in terms of fatigue damage potential. The details of this “equivalence” will be investigated in terms of not just fatigue damage but also the level of localized plasticity at hot spot locations.



Secondly, the consolidation of multiple events (and multiple channel data) into simplified (enveloped) test specifications will be explored. Specifically, the concept of reducing a multiple input loading scenario (with channel correlation) into a simplified form whilst retaining damage equivalence. Traditional methods for loads enveloping usually enforce a much cruder damage equivalence because the system properties are unknown. In this paper, the system properties will be used in the enveloping procedure. Again, as far as the authors know, this is the first time such a study has been done.