Fatigue damage accumulation under torsion and non-proportional push-pull interruption loading.

A new testing facility for fully reversed tension-torsion high cycle fatigue testing has been designed. The specimens used for the test programme were solid and made from a medium carbon steel. The test programme involved a tension-torsion multiaxial non-proportional loading sequence i.e. fully reversed torsion followed by a push-pull load interruption and then the continuation of the same torsion loading to failure. The push-pull load interruption represented a significantly low damage i.e. 4% damage according to Miner's linear damage theory, and was applied after different prior torsion cycle ratios. The tests were conducted with various interruption stress amplitudes all of which had fatigue lifetimes in the high cycle fatigue region.The torsion fatigue life was found to change significantly due to the application of push-pull load interruption which was considered to cause only a minor damage due to Miner's rule. Miner's linear damage theory cannot account for the predicted cumulative fatigue damage (Sigman/N[f]) for the push-pull interrupted torsion fatigue loading sequences used in the current test programme. The fatigue life was markedly enhanced when the interruption was applied at an early stage of torsion loading whilst the effect was less prominent when the interruption was applied at a later stage of torsion loading. At higher interruption stress amplitudes the torsion fatigue lifetime was reduced considerably and the damage summation was well below the unity predicted by the Miner's rule. The inability to predict damage accumulation by Miner's rule can be attributed to the complexity in the crack growth associated with the application of push-pull interruption.Crack growth equations to represent microstructural short crack (MSC) and the physically small crack (PSC) growth were determined for the material of the form;MSC - da/dN = C[m](d[i] - a).............................(1) and PSC - da/dN = C[p]a-D..................................(2). Material parameters for the models were derived using torsion and uniaxial constant amplitude fatigue S-N data, no crack coalescence, branching or re-initiation was considered.The crack growth model was able to predict the fatigue life in loading cases which were dominated by an uninterrupted crack growth. However, such a model was shown to significantly underestimate the torsion fatigue life in situations where the fatigue life was affected by secondary crack initiation due to the push-pull load interruption.