From Equation (97) it is seen that the slope of the pj versus p^ curve 

 at the point pj = pp is equal to unity, as it should, only if Bj^ = Bq. Then, 

 and only then, does the curve defined by Equation (93) become tangent to 

 the pp curve in the elastic buckling range. Thus, in the one extreme 

 which corresponds to the elastic buckling range. Equation (93) for the in- 

 elastic buckling pressure p, reduces to the elastic buckling pressure p^, 

 as it should. In the other extreme which corresponds to the axisymmetric 



collapse range, which conversely implies that p ->oo, we see that Equation 



E 



(93) reduces to ('^„h/R)B , which is a hoop- stress relation at midbay 

 between adjacent ring frames. As was indicated before, what to use for 

 B is a question which has still not been completely resolved in the mind 

 of the author. The question that needs to be answered is: "What is an 

 appropriate hoop-stress criterion, if any exists, which adequately pre- 

 dicts cLxisyminetric collapse precipitated by yielding for the broad range 

 of geometry which is of interest?" Some work to clarify this point is 



presently underway at the Model Basin; one possible approach toward re- 



38 

 solving this question has been suggested by Pulos and Hom in which 



empirical curves have been fitted through experimental data from many 

 structural model tests and the trends toward asympotic values predicted 

 by the various collapse criteria noted. 



Equation (93) may appear to be somewhat nonrigorous to the more 

 theoretically inclined stress analysts. A truly rigorous approach to solv- 

 ing the inelastic panel -buckling problem of ring- stiffened cylindrical 



65 



