Normally these stresses were less than the membrane stresses in the 

 spherical hiill. Any mismatch which might occur on the three prototype 

 hulls would cause very local bending stresses in an area reinforced by the 

 increased thickness of the penetration insert. Since these stresses are 

 normally less than the membrane stresses or are very local, they should not 

 adversely affect the strength of the hulls. 



Throughout the analysis, reference has been made to available 

 stress-strain curves. It has been assumed that the stress-strain curves 

 presented in Figure 8 are representative of the material used in the hull. 

 As additional data becomes available, they should be compared with the data 

 used in this analysis. If significant differences are reported from other 

 sources the predicted collapse depths can easily be adjusted using the 

 geometry presented in Table 2 and Equations [8] and [9]. It should be 

 mentioned that preliminary data on the weld material indicate that the 

 strength of the welds may be 5 percent lower than that of the hull material. 

 In view of the stability of these hulls, it is not felt that this would 

 affect the collapse of these spheres since this is a very local effect. 



Although the results of this analysis differ little from the 

 results obtained using the nominal geometry of the spherical pressure 

 hulls, it should be eitphasized that this is only because the hulls have 

 been fabricated so accurately. For example, if the hull had been built 

 to an out-of-roundness tolerance of ±l/8 in., the collapse depths for the 

 three hulls would range from 12,600 to 13,400 ftT Further, if normal 

 boiler code tolerances of ±1 percent of the outside diameter were obtained 

 in fabrication this range would be reduced even further, that is, the 

 collapse depths would range from 6000 to 6400 ft. Thus the effect of 

 neglecting initial imperfections and the inqjortance of determining the 

 exact shape of spherical pressure hulls is evident. 



"For these calculations it is assumed that the maximum allowable out-of- 

 roundness A (0.25 in. and 1.64 in.) occurs over the critical arc length. 



35 



