Fatigue life of metals with particular reference to nonlinear damage accumulation under variable amplitude loading

Fatigue in engineering components and structures subjected to cyclic loading is characterised by the process of accumulation of damage over a period of time. This phenomenon of damage accumulation could potentially lead to catastrophic failures. Unfortunately the occurrence of such failures is difficult to predict due to a limited understanding of the fatigue process particularly under variable amplitude loading (VAL). Fatigue life of metals is usually determined from linear damage rule (LDR) that offers sufficient level of conservatism and accuracy, particularly for high cycle fatigue (HCF) loading blocks typified by magnitudes of stresses well within elastic limit and period exceeding millions of cycles. However, for low cycle fatigue (LCF) associated with a relatively small number of high amplitude load cycles causing bulk plastic strain, the damage process is affected by the amplitude and sequence of plastic strain. Past attempts to apply the LDR to predict fatigue life for LCF have been largely unsuccessful. Irrespective of the various modifications proposed, the LDR has failed to capture the nonlinear accumulation of damage observed for certain load sequences within VAL spectra, where at least a portion of the loads cause plastic strain. Amongst the variety of LCF life prediction models proposed in the literature, those based on the energy approach of continuum damage mechanics (CDM) provide significant scope to further the development as they conform to the principles of thermodynamics and hysteretic energy dissipation. This thesis presents an energy based nonlinear damage accumulation (NLDA) model that utilises the CDM concept and a cumulative damage parameter to account for nonlinear damage accumulation under VAL. The fatigue life predictions of the NLDA model for metals subjected to constant, high amplitude, strain controlled loading, compare well with the experimental data reported in the literature. The nonlinear capability of the NLDA model has been demonstrated using a well established two-step block -loading test.