A so-called 'local probabilistic Paris relation method' was presented for measuring the random thresholds of long fatigue crack propagation. A check was made to the conventional method, in which the thresholds were measured statistically and directly by the test data. It was revealed that this method was not reasonable because the test data have seldom a unified level of crack growth rates. Differently,in the presented method the Paris-Erdogan equation was applied to model the local test data around the thresholds. Local probabilistic relations with both the survival probability and the confidence were established on a lognormal distribution of the stress density factors. And then, the probabilistic thresholds were derived from the probabilistic factors with a given critical level of growth rate. An analysis on the test data of LZ50 axle steel for the Chinese railway vehicles verifies that the present method is feasible and available.
Experimental study is performed on the probabilistic models for the long fatigue crack growth rates (da/dN) of LZ50 axle steel. An equation for crack growth rate was derived to consider the trend of stress intensity factor range going down to the threshold and the average stress effect. The probabilistic models were presented on the equation. They consist of the probabilistic da/dN-ΔK relations, the confidence-based da/dN-ΔK relations, and the probabilistic- and confidence-based da/dN-ΔK relations. Efforts were made respectively to characterize the effects of probabilistic assessments due to the scattering regularity of test data, the number of sampling, and both of them. These relations can provide wide selections for practice. Analysis on the test data of LZ50 steel indicates that the present models are available and feasible.
When the historic probabilistic S-N curves are given under special survival probability and confidence levels and there is no possible to re-test, fatigue reliability analysis at other levels can not be done except for the special levels. Therefore, the wide applied curves are expected. Monte Carlo reconstruction methods of the test data and the curves are investigated under fatigue life following lognormal distribution. To overcome the non-conservative assessment of existent man-made enlarging the sample size up to thousands, a simulation policy is employed to address the true production where the sample size is controlled less than 20 for material specimens, 10 for structural component specimens and the errors matching the statistical parameters are less than 5 percent. Availability and feasibility of the present methods have been indicated by the reconstruction practice of the test data and curves for 60Si2Mn high strength spring steel of railway industry.
