Probabilistic Method for Predicting the Variability in Fatigue Behavior of 7075-T6 Aluminum

This paper assesses the applicability of a probabilistic model to estimating fatigue life variability for 7075-T6 aluminum. The test specimens for the experimental program were 7075-T6 aluminum single edge notch tension specimens tested under constant amplitude loading (cmax - 120 MPa, R = 0.01). The shortest observed fatigue life was 37,000 cycles and the longest 650,000 cycles. A plastic replication procedure was used to identify crack nucleation sites and monitor crack growth. The initiation sites were identified by examining the plastic replicas and the fracture surfaces of the failed specimens. In agreement with previous studies, the fatigue cracks formed in the center of the notch from material inclusions. For the prediction model, the distribution of inclusion sizes (0.2 (o.m2-45 jam2) within the material were used as the distribution of initial flaw sizes. It was assumed that the crack formation life was a small percentage of the total life, and thus life predictions were based entirely on crack propagation. The cumulative distributions for the fatigue lives from the experimental work and the numerical model were compared to assess the effectiveness of the model. While Experimental crack formation lives as long as 50-70% of the total life were observed in some of the- tests, the model predicted failures well in the shorter life regime. The predicted lives were conservative, and in the shortest life region, the predictions were within 15% of the observed lives.