A comparison of some methods for predicting creep strain and rupture under cyclic loading

Abstract

There are many good reasons for attempting to predict creep behaviour under conditions of varying stress and temperature from data derived from tests perfomed at constant stress and temperature. This Report starts by describing the most straightforward hypotheses at present used for this purpose. Computed results for cyclic variations have shown that (i) the "strain hardening" and "life fraction" hypotheses predict very similar rupture times, (ii) the times to a given creep strain do not depend on the frequency of the cycles or the sequence of loading within the cycles providing there are several (10 or more) cycles involved, (iii) when a substantial proportion (more than about two-thirds) of the creep life shows a "tertiary" behaviour the "time hardening" hypothesis predicts the shortest rupture times for the same cyclic loading. A method is demonstrated for evaluating effective mean stresses or temperatures for any cyclic conditions according to either strain or time hardening hypotheses.