644 



28 PEPPARD AND CHTTTY ON EXPERIMENTS ON THE 



There was a difiference in ultimate strength between the material for 

 the two batches of cylinders (Nos. to 6 and Nos. 7 to 11 respectively). 

 The first batch gave approximately 23 tons per square inch and the second 

 batch approximately 26 tons per square inch. 



All specimens gave a value of Young's Modulus of 14,000 tons per 

 square inch, an ultimate extension of 14 to 15 per cent., and a reduction 

 in area at fracture of 40 per cent. 



Conclusions. 



Ab indicated earlier the investigation was made primarily to supple- 

 ment research into the behaviour of submarine shells subjected to under- 

 water explosions, but the following general conclusions upon the plastic 

 failure of cylindrical shells may be drawn from the results of the experi- 

 ments. 



1. The effect of removing the load from a cylinder under test and then 

 re-loading it was very slight. There was little or no looping of the load- 

 displacement diagram either in the small cylinders or the large stiffened 

 shells. 



2. The stiffness of the end plates had a marked effect on the type of 

 distortion. When those plates had negligible resistance to bending out 

 of their plane the tests strongly supported the hypothesis advanced by 

 Taylor. The resemblance of his models to the actual deformed shells was 

 striking and that visual evidence was confirmed by the quantitative 

 agreement of the experimental results with calculations based on his 

 hypothesis. The agreement held for cylinders deformed either by static 

 or dynamic loading. 



3. The stiffness of the end plates also governed the type of distortion 

 resulting from hydrostatic testa on cylinders stabilized against failure over 

 one-half of their depth. When the ends were flexible failure occurred by 

 the development of a single dent, but when they could resist bending out 

 of their original planes, failure was multi-lobed. The lobes developed 

 successively — not simultaneously. There is evidence (supported later by 

 tests under explosive forces), for assuming that if the ends are not com- 

 pletely flexible, but are capable of resisting some degree of bending out 

 of their planes, failure begins by multi-lobed deformation, but subsequent 

 bending of the ends leads to the incipient lobes merging into a single dent 

 showing the boundary between initial lobes as ridges. 



4. In cylinders with flexible ends no actual fractures of the shell 

 occurred, but with stiff ends longitudinal stresses were induced sufficient 

 to cause such fractures. They were circumferential and were character- 

 istically tensile. 



5. The load-displacement curves for the internally stiffened cylinders 

 showed evidence of the progressive extension of the region of plastic 

 straining from the loaded ring to the adjoining sheet and then to rings 



