Probabilistic fretting fatigue life prediction of Ti–6Al–4V

A probabilistic analysis of the fatigue life of specimens subject to fretting fatigue was developed. A mechanics based fretting life analysis was applied that employed the local stress gradient at the edge of contact. The random variables in the analysis included the initial crack size, coefficient of friction, partial slip slope, crack growth rate law, and the contact pad profile. The variation in pad profiles was determined through measurement of 77 machined pads. Distributions for the other random variables were obtained using previously generated test data. A probabilistic fatigue analysis was applied using Monte Carlo sampling to determine the statistics (mean and standard deviation) of the fatigue life prediction and was compared to fretting fatigue test data. Several qualitative and quantitative sensitivity methods were applied to the results including the calculation of the probabilistic sensitivities (partial derivatives of the fatigue life statistics with respect to the input probability density function parameters) via linear regression and finite difference.