Applying Results of Seakeeping Research 



DISCUSSION 



W. A. Swaan 



Netherlands Ship Model Basin 



Wageningen, Netherlands 



The paper gives a clear review of the possibilities of applying the results 

 of presently available and future seakeeping research. I do agree with the au- 

 thor in his conclusion about the need for more data on the sea, model series in 

 waves and criteria for seagoing performance. Especially the lack of sea data is 

 a great obstacle in providing useful behaviour predictions for new ship designs. 



It appears doubtful to use the expression "response amplitude operator" for 

 the power coefficients because it is something essentially different from the 

 "response amplitude operator" for ship motions. In the case of the power co- 

 efficient the result of the described procedure is a mean value; the zero fre- 

 quency component of the power in irregular seas. In the case of ship motions 

 and bending moments the results have an oscillatory character with a zero 

 mean. Therefore it might be better to indicate the power coefficient as "re- 

 sponse operator" and leave the expression "response amplitude operator" for 

 oscillating phenomena. 



The comparison of the wave bending moment in Fig. 5 for short- crested and 

 long-crested seas is very illuminating. It can serve as a warning against using 

 long-crested irregular seas in an overconfident way. 



Figure 6 shows the highest expected vertical bending moment in 10,000 

 cycles. Because the ships in this diagram have lengths between 300 ft and 1300 

 ft and speeds from 6 knots to 18 knots one would expect the time interval cov- 

 ered by these 10,000 cycles to be a function of ship length. This involves a dif- 

 ferent risk when small and large ships are compared. Because the author does 

 not convert the spectrum to the frequency of encounter it is not quite obvious in 

 which way one can reach some definite conclusion about the time covered by 

 10,000 cycles. 



There is another difficulty involved in using the highest expected value in 

 10,000 cycles. The relation given in the paper between the spectrum area and 

 the value of the average highest amplitude is only valid for a narrow band spec- 

 trum in which no negative maxima and positive minima occur when the zero 

 level is taken at the mean position. It seems therefore much easier to discard 

 the use of bending moment amplitudes altogether and return to the Gaussian 

 distribution of the bending moment values when they are determined at constant 

 time intervals. Because the variance of this Gaussian distribution is equal to 

 the area of the spectrum it is not difficult to determine the percentage of time in 

 which a certain bending moment is exceeded. From this it follows that for a 

 narrow spectrum the average highest amplitude in 10,000 oscillations is equiva- 

 lent to the deviation (absolute value) which will be exceeded during 0.0009% of 

 the time or 3/4 seconds per day. The number of times it occurs is left undefined 



213 



