136 THE BEHAVIOR OF RIVETED JOINTS UNDER STRESSES. 



were overstrained at these higher temperatures, even beyond the limits of duplicate 

 cold tests, when subsequently tested to destruction at atmospheric temperatures, re- 

 tained substantially the strength which they had when hot. There was some loss in 

 the ductility of the steel, but without approaching a state of brittleness. 



So much for the ultimate strength of riveted joints. Under some circum- 

 stances it is probable that the riveted joints in an engineering structure might for 

 a time resist loads approximating those which were required to effect rupture in the 

 testing machine. More commonly they probably would not, nor would they be ex- 

 pected to display the ductilty witnessed in the laboratory behavior of the joints. 

 Stated in other words, structures designed to have a factor of safety of five based 

 on the ultimate strength of these joints would not as a matter of fact in service pos- 

 sess such a margin in strength and safety. Attention must be given the behavior of 

 the joints under stress, and whether the working loads are constant or variable, di- 

 rect or reversed stresses; and, in the case of repeated stresses, how many repeti- 

 tions there will be and the maximum stresses involved. 



The examination of some stress-strain curves prepared from earlier tests show 

 that the joints in general take a very wide departure from the curve representing 

 the solid plate, this being noticeable at 15,000 pounds per square inch, and in some 

 joints as early as 10,000 pounds. This was true with joints having efficiencies of 70 

 to 80 per cent of the solid plate. Among the joints thus compared were double- and 

 triple-riveted butt joints, and quintuple joints in which the inner butt strap was 

 wider than the outer one. 



Under 15,000 pounds per square inch, the joints in general, among the strong 

 ones, displayed an extension one and a half times to over twice the extension of the 

 solid plate. These joints were of the type which are used in steam boiler con- 

 struction. Observations on the behavior of double-riveted lap joints on some steam 

 boilers which had been in service showed greater extension across the longitudinal 

 seams at the middle of the width of the sheets than in the vicinity of the girth 

 seams, referring, in the tests just mentioned, to the effect of hydrostatic pres- 

 sures applied just above the pressures at which the boilers had been worked. 



It is a feature of interest whether riveted seams retain their primitive state 

 under prolonged service stresses, or whether they do not slip and eventually display 

 increased extensions under lower loads than suggested by the laboratory tests. 



Cover plates and splice plates in bridge work do not immediately take up the 

 full stresses which are acquired along the length of the plates after several rows 

 of rivets have been passed. The loads are progressively taken up by these splicing 

 members. 



From the limited number of observations which have been made it cannot be 

 said that the rigidity of joints on actual structures is greater than would be ex- 

 pected, judging from the laboratory tests. If there is a difference, they are prob- 

 ably less rigid in actual structures. 



The frictional resistance due to the shrinkage of the rivets is apparently a fac- 

 tor in the early behavior of a joint. The shrinkage force which is available is, or 



