PLASTIC FAILURE OF CYLINDRICAL SHELLS. 



635 



19 



Fig. 28. 



Test A on Cylinder No. 1. 



to the two rings adjacent to the central one through the medium of the 

 shell, and plastic deformation of the central ring was accompanied by 

 elastic deformation of those adjacent to it. At D, all three rings were in 

 the range of plastic deformation and 

 ultimately complete collapse occurred 

 at the load represented by E. 



This explanation suggests that 

 in the fully stiffened cylinders the 

 presence of further rings would give 

 more sections of the curve similar to ; 

 CD and DE. This was borne out ; 

 by the later test on cylinder No. 2, - 

 although the efficacy of the remaining 

 rings was small compared with that 

 of those adjacent to the load point. 



The appearance of cylinder No. 

 after test OB is shown in Fig. 27 

 (facing p. 16). 



Test No. lA was the first made on 

 a complete cylinder and the results are shown in Fig. 28. The maximum 

 load carried was only 5,500 lb. compared with 9,000 lb. by the incomplete 

 cylinder No. 0, which had only three stiffening rings and very inadequate 

 end plates compared with those in the standard cylinder. That result was 

 surprising until the load was removed and the surface of the cylinder 

 exposed. It was found that a circumferential split had developed in the 

 shell against the central ring under the loading ram. The split was a 

 characteristic tensile fracture and left no room for doubt that the end 

 plates had restrained the free distortion of the shell sufficiently to produce 

 large tensile stresses in the shell plate. That was consistent with tests 

 already described which were made on small cylinders. 



When cylinder No. 1 was turned over in the sand bed and the un- 

 damaged surface loaded (Test No. IB), no split developed ; possibly 

 because the split from the previous test gave sufficient flexibility to relieve 

 the longitudinal tensile stresses. The load-displacement curve for Test 

 No. IB is shown in Fig. 29 and the two sets of contours of the damaged 

 surfaces in Figs 30 and 31. Figs 32 (facing p. 17) show cylinder No. 1 

 after tests, with the contours painted on it. 



During the progress of Test No. IB, the alteration in overall length of 

 the top generator of the cylinder was measured at intervals ; the maximum 

 shortening was about 0-20 inch. 



As a result of those tests it was decided that cylinder No. 2 should be 

 modified so that the ends were entirely unstiffened and the end disks 

 were therefore removed. The load-displacement curve is shown in Fig. 33 

 and exhibits no new features. Very large distortions were obtained 

 without any sign of cracking in the shell, although one or two of the spot- 



