waves into which the cylinder buckles (also a function of the relative dimensions) is a vital 

 factor in the determination of general-instability strength. Thus it is clear that the problem 

 would be greatly simplified if a few fundamental parameters could be obtained by combining 

 some of the dimensions in a rational manner. 



It is shown in Bryant's paper^ that the general-instability pressure can be treated 

 approximately as the sum of two terms, one involving the strength of the shell and the other 

 the strength of frame per unit length of shell. This latter term was defined as the moment of 

 inertia about the centroid of a section comprising one frame plus a length of shell equal to 

 one frame spacing. Numerical results indicate that Kendrick's second solution in Reference 2 

 can be broken down fairly successfully in this way. However, a slightly different parameter 

 used by Bijlaard^ in treating the same problem lent itself better to a graphical presentation and 

 gave less scatter in the results. Bijlaard expresses frame strength as the moment of inertia /^ 

 about the centroid of the combined section of a frame plus an effective length L^ of shell. L^ 

 is taken to be 1.57 yfWh so long as the frame spacing L, exceeds 2 \fRh. For smaller values 

 of Lf, L can be determined from Table 46 of Reference 6. The quantity /^ can be written 



/ = ^^— + /. + -^— [1] 



1 + — i 



Where R is the radius to the median surface of the shell, 



h is the shell thickness. 



At is the frame area. 



If is the moment of inertia of the frame about its own centroid, and 



e is the distance from the median surface of the shell to the centroid of the frame. 



With the parameters as defined above, the general-instability pressure p^^ of a stiffened 

 cylinder can be expressed as the sum of two terms: 



''" = ''-(t'4'")*''/(7^'") ''' 



where n is the number of circumferential waves, 



L^ is the bulkhead spacing, and 



p is a linear function of h/R. 

 The quantity v can be determined readily from Figure 1 in which p^- ' is plotted against 



Lfj/R for values of n of 2, 3, 4, and 5. Similarly, Figure 2 shows the variation of ?y with 

 / /LfR^ for the different values of n. These curves were drawn to fit a large number of 

 calculated points.* The same information is presented more concisely in Figure 3 where the 

 results from Figures 1 and 2 are combined in one graph having the coordinates 



♦Since only values for p were obtained from the calculations, p^^ was plotted against 1^/LcR , and p^ and 

 pj- were determined by extrapolation to I /LrR = 0. 



3 



