Stochastic Modeling of Fatigue Crack Propagation

Summary

Deterministic fatigue crack propagation models, such as the Paris-Erdogan model, cannot account for random variability of time-histories observed in experiments. Therefore, more realistic models are proposed in which a random process is introduced as a multiplicative factor to deterministic laws. This factor is assumed to be a train of randomly arriving pulses with random amplitudes, permitting simple computation of the statistical properties of the fatigue crack size at any given time. Approximate probability distributions can then be calculated using a general scheme called cumulant-neglect closure. Results obtained from retaining only the first two cumulants are compared with available experimental data and with previously calculated distributions when the crack size was treated as a Markov process.