FORCE PULSE TESTING 

 OF SHIP MODELS 



W. E. Smith and W. E. Cummins 



David Taylor Model Basin 



Washington, D.C. 



INTRODUCTION 



In a recent paper [l] one of the authors proposed that a useful and revealing 

 way of treating oscillatory motions of a ship was to relate them to the transient 

 response to an impulse. The response to an arbitrary excitation would be ex- 

 hibited as a convolution integral over the past history of the excitation. The 

 idea was hardly original, as this device is widely used in the discussion of linear 

 systems. However, there seemed to be some reluctance by those working in the 

 field to treat the ship response in this fashion. Most writers preferred to re- 

 strict their attention to the frequency response function. 



There have been some exceptions to this trend, notably Fuchs and MacCamy 

 in the discussion of the motions of a floating block [2], Dalzell in the treatment 

 of destroyer motions in severe sea states [3], and the paper by Davis and Zarnick 

 for the present symposium [4]. However, all of these are concerned with re- 

 sponses to wave pulses or hypothetical wave impulses, and not the response to a 

 force or moment impulse. The present paper is concerned with this latter prob- 

 lem. As a matter of fact, the solutions to the two problems, the response to the 

 wave pulse or impulse and the response to a force or moment excitation, com- 

 plement each other very effectively. The first solution characterizes the total 

 wave- ship system, while the second enables us to construct the equations of mo- 

 tion and thus separate the effects of damping, added mass, coupling, and hydro- 

 dynamic memory. When both solutions are known, the wave excitation can be 

 determined, and one is then in a position to say not only what the ship does but 

 why it does it. The designer then has clues as to how to make changes in the 

 design in order to improve seakeeping qualities. 



One can discuss and even use the impulse response function without directly 

 measuring it, as it is simply the Fourier transform of the frequency response 

 function. If the latter is known for all frequencies, the impulse response func- 

 tion can be computed. But to directly determine the frequency response function, 

 one must measure the response to a set of frequencies at suitably close inter- 

 vals over the whole frequency range in which there is significant response. The 

 alternative approach is most attractive. That is, apply a known impulse or 

 equivalent excitation to the model and observe the response. The frequency re- 

 sponse function can then be computed, and we have replaced a time consuming 

 and expensive test program requiring many runs with a single run. This paper 

 is concerned with such measurements. 



439 



