On the prediction of impact noise, IV: Estimation of noise energy radiated by impact excitation of a structure
This series of papers [1-3] sets out to provide a practical method of predicting the noise radiated by a machinery-type structure as a result of an impact upon it. As such noise energy may vary from 10 -8 to 10 -3 of the mechanical energy used in the machine, it is necessary to validate the accuracy of the "energy accountancy equation" which has been developed  to permit this prediction in terms of those items which can be measured, viz., the force derivative during impact, the ppint response term, the calculated "modified" radiation efficiency, the damping factor, and the average thickness of the machinery components. To do so, and to represent typically a gearbox casing, this paper sets out to calculate, measure and compare the noise radiated from a damped plate, held rigidly in a frame, the impact being produced by a sphere (whose natural frequencies are well above those being examined) striking the plate either directly, through a lumped mass which is either attached directly to the centre of the plate, or through a resilient and damped interlayer. As Skudrzyk  has indicated that the mean point response of a curved casing can be treated as a flat plate under conditions which apply to gearbox casings, the representation referred to in this paper can be taken as typical of such examples as a gearbox casing excited by out-of-balance shaft loads or backlash, with concentrated masses or some degree of isolation added to the shaft housing area. It can also be seen to apply to close but semi-isolated covers or panels surrounding diesel engine frames, drop hammer tups and the like. It is concluded that the "energy accountancy" equations do predict the very large changes in noise radiation throughout the important frequency ranges with accuracies of ±2 dB in one-third octave bands, even on configurations giving 50 dB reductions (see Figure 19).