computed on the basis of the Hencky-Von Mises criterion for a monolithic 

 thick cylinder of equivalent weight. 



The one-diameter-long model (OV-3) was tested to determine wheth- 

 er an increase in strength for a titanium hxill similar to that observed for 

 the steel model, OV-1, could be realized as a consequence of the effects 

 of strain hardening. In addition, it was felt that this test of Model OV-3 

 would provide information on the feasibility of constructing models by slip- 

 ping one shell over the other, and also would indicate whether this novel 

 technique involved a loss of structural strength, if any existed, compared 

 to the method utilized for Model OV-1, where the outer shell of the sand- 

 wich section was physically joined to the webs by slot welding. 



The four-diameter-long model {OV-4) was intended to determine the 

 collapse pressure and mode of failure for the overall cylindrical compart- 

 ment of the original design. This model was considered necessary since 

 the general-instability strength could not be determined by testing short- 

 length cylinders such as Models OV-1 and OV-3. 



MODEL OV-2 



By the use of the elastic analysis of Reference 3, it was possible to 

 predict high bending stresses in both shells between adjacent webs of the 

 original sandwich design; in addition, the webs were found to be under- 

 stressed. These results indicated that an improved design could be ob- 

 tained, without increasing the weight of the structure, by a redistribution 

 of the material. Several improved designs were analyzed. One of these 

 designs is shown in Figure 2b and is represented by Model OV-2. Calcu- 

 lations indicated that the OV-2 configuration represented a more balanced 

 design than that originally proposed. That is, the spread between the max- 

 imum and the minimum shell stresses is reduced in an attempt to more 

 uniformly stress the entire sandwich cross section. It should be noted, 

 however, that a "balanced stress design" need not represent the optimum 

 structure for a given weight when the instability modes of failure are also 

 of significance, as in the present design of a long hull compartment. Figure 

 7 presents a comparison of the computed stresses in the critical regions 

 of a typical sandwich section for each of the three designs investigated. 



10 



