Kurtosis control is widely touted as a way to increase the damage content of a random vibration test without changing the Power Spectral Density (PSD). However, when a peak stress definition of overtesting is constructed based on the standard linear material model, it is found that the most rapid vehicle for delivering damage without overtesting is never a test with increased kurtosis.
This presentation explains why adding kurtosis to a vibration specification instead of increasing the PSD automatically adds time, and therefore cost, to the test procedure. Further, it explains how different vendors might comply with the same kurtosis specification while delivering different amounts of fatigue damage and peak stress levels to the device under test. In effect, vendors using different controllers will perform different tests. This is not because one controller is right and the other is wrong; the underlying kurtosis specification is inherently imprecise with regard to fatigue damage.