In summary, the method developed by Kim and Dalzell (1979) 

 appears to be an accurate, rigorous method which could be 

 extended to include 3-D source representations. This was 

 indicated as being possible by Dr. J. Bendat . Moreover, 

 such a procedure when combined with an appropriate mechanical 

 model would yield the response statistics directly. 



Following another approach of a general deterministic nature, 

 there seemed to be general agreement that the Newman approxi- 

 mation (1974), in which only the diagonal elements are 

 employed for determination of the time varying drift force 

 from the mean drift force, is a suitable and appropriate 

 engineering tool. 



However, there did not seem to be clear and definitive agree- 

 ment on the relative importance of the radiation potential 

 as compared with the diffracted potential in determination of 

 the mean drift force for either inertia or restoration force 

 dominated systems. 



In order to avoid the difficulties associated with determina- 

 tion of the second order force (which depends upon the first 

 order motions), several possibilities were discussed. It was 

 pointed out that the second order force could be evaluated by 

 employment of a double convolution integral corresponding to 

 that contribution generated by the radiation potential. 



It is believed that this discussion could have benefited from 

 more available time. For example, consider the case of an 

 inertia dominated system such as a tanker. For such a system, 

 the ship is very unresponsive to the exciting waves. Therefore 

 the motions at frequencies corresponding to the exciting wave 

 periods are very small. As a result, although the ship has a 

 large surface area, the wave system radiated (by the ship 

 motions) is small as compared with the incident and/or diffract- 

 ed wave systems. Therefore, one might conclude that that part 

 of the second order drift force attributable to the radiation 

 potential could be approached by an iterative method (as 

 suggested, and, incidentally, as it has been done). 



On the other hand, for a restoration-force dominated system, 

 one might arrive at a completely different conclusion. 



All in all, there were a number of aspects of this topic that 

 should have been explored in more detail. 



3. In summary, I believe that it was realized that the problem 

 could be approached in two fundamentally different ways. 



a. A deterministic (SEADYNE/DSSM) approach in which 

 the output statistics are obtained from multiple 

 simulations (i.e., essentially "Monte Carlo"). 



157 



