Abstract
Simple models representing a shaker and a test object are used to illustrate changes in test object response due to shaker dynamics and differences between the test and service environment. The degree of coupling is quantified in terms of ratios of the natural frequencies and the masses. Regions of overstress can depend on reproducing absolute rather than relative motion in a test. Shaker tests reprodusing output spectra observed in service, when shakerlji.xture impedance is higher than the impedance in service, is shown to cause overtest at frequencies below natural frequencies of the service environment.