Investigation of Fatigue of Metals Under Stress

AT PRESENT, I am connected with an investigation of the so-called fatigue of metals under stress. So far we have studied the more fundamental and simple case of the repeated stress, without the additional complexity of impact, which might bring in other factors. We feel that this investigation, which has been in active progress for a little over a year, has shown more conclusively than has been shown before that steel under repeated applications of stress, reversed from positive to negative, will not fail below some fairly clearly defined limiting stress that,, so far we can see, does not bear any definite relation to the ordinary elastic limit, being as large as the elastic limit in some cases and about one-half the elastic limit in others. I might be asked, first, what is fatigue. The old view is summed up "in the word "crystallization." The idea was that under repeated stress the material changed its crystalline structure. I do not know of any evidence in favor of the theory that metals materially change their crystalline structure under repeated stress. The crystals are broken under repeated stress. The second theory, advanced by Bauschinger, the German scientist, is that under repeated stress some inherent property of the material changes its elastic limit; it is an inherent property of the material, possibly some property of the amorphous cement between the crystal. The third view is that all fatigue of metals is the result of the spread of damage from little localized overstresses. I do not feel justified in stating my positive belief in the third view as against the second, but I have yet to run across a case of failure of steel or other metal, either in the laboratory or in service, that could not be explained by the gradual spread of damage from some nucleus. We have found, for example, that it is possible to subject a homogeneous steel, like Armco iron or a 0.90 per cent. carbon steel, that hag been thoroughly annealed, to one hundred million reversals f stress as high as the elastic limit of the steel with no sign of failure. We stopped the test at one hundred million because the machines Were required for other work. Forty-five days, continuous running at 1500 revolutions a minute is required to produce one hundred million repetitions. On the other hand, a complex structured steel, like a chrome-nickel steel that