30^ 























20 



.Experimental Histogram 



c 









u 



/ 







"V 





I- 



■^ 



-=5 



Theoretical Ravleiqh 



11 10 



/ 



■^^ 









N 



h 



s 



Distribution 







^ 

















vH^^ 



1.4 2.8 ,^ 4.2 ^^ 5.6 



Double Wave "Annplitude" In. 



4 8 ,^ 12 ^^ IG 20 



Double Pitch Amplitude, Deg 























20 





c 





1 — 1 



u 









i- 







■-sn 







D- 10 



/ 







\ 



















N 



R 









/p 

















=¥S=U_, 



O.e 1.2 1.8 ^^ 2.4 3.0 3.e 

 Double Heave Amplitude. In. 



the mean log value and E (or o-^). Variations in 

 the parameters for different series of data, and 

 hence in the lines plotted in Fig. 18 for example, 

 are presumably due to such factors as the service 

 in which the ship is engaged (i.e., average severity 

 of the weather), the average loading of the ship, 

 (i.e., always fuU or sometimes light), longitudinal 

 weight distribution, and in the case of huU-girder 

 stresses, the section modulus and other structural 

 design features of the hull itself, and even the 

 particular point at which the stress is measured. 

 It does not seem surprising that if sea data exhibit 

 a log-normal distribution (Figs. 13-15 of the 

 paper) the ship responses also should follow this 

 distribution. 



It is somewhat disappointing that Fig. 18 does 

 not extend to larger P values and hence larger 



Experimental Histogram 



-^Theoretical Royleigh 

 Distribution 



18 36 ^ 54 72 



Double Bending Moment Amplitude" In-Lb 



stresses, since there is some indication of depar- 

 tures from the lines at the top of the graph. Table 

 9 indicates that a number of stresses have been 

 recorded above 10 kips/sq in. 



Coming next to the application of distribution 

 patterns of motions, there can be no doubt that 

 they will be of great value for the design of ship- 

 borne equipment, stabilization systems, fire- 

 control equipment, and aircraft launching and 

 landing devices. As for hull-girder stresses, the 

 author has performed a service in calling attention 

 to the possible importance of endurance strength 

 to the ship-fracture problem, and the distribution 

 patterns are a necessary step in studying the 

 problem. He is quite right in giving attention 

 also to the extreme values that might be expe- 

 rienced over the life of the ship. In the writer's 

 opinion, this is one of the biggest problems facing 

 us in structural design at the present time. The 

 statistical approach to this problem is interesting, 

 but as he points out, is unsatisfactory. He says, 

 "a better method.... is by the use of the experi- 

 mentally obtained distribution of extreme values." 

 Presumably this means assembling much more 

 statistical data, with emphasis on high values, 

 and I am heartily in favor of this. However, I 

 object strenuously to the statement, "Rejection 

 of predicted values lying outside the range of 

 experience would provide a proper safeguard 

 against too extreme an estimate." This simply 

 begs the question, and if such a policy were fol- 

 lowed inevitably would doom many ships to the 

 bottom. 



Even with a great deal of data on high stresses 

 recorded at sea, it appears doubtful that a satis- 

 factory answer to the problem of expected maxi- 

 mum values can be obtained by statistical methods 

 alone. There are definite physical Umits to the 

 bending moment exerted on the hull by the sea; 

 wave steepness has an absolute maximum and the 



42 



