Probabilistic fatigue life prediction considering the uncertainty of S-N curve

The traditional fatigue life prediction methods based on the S-N curve recognize the parameters in life prediction models can be determined by experiment, simulation and fitting. That is to say, in a given loading environment, various parameters are treated as different constants, and then substitute these parameters into the model, the fatigue life can be obtained. This kind of methods treats the parameters in the model as constants, which belong to deterministic life prediction method. Deterministic life prediction method is suitable for fatigue life prediction and evaluation where the test data is sufficient. However, in practice, because of the restrictions on development cycles and there may be a shortage of funds, combined with technical difficulties, it has great difficulty to carry out a large number of trials. At the same time, due to different test specimens, along with test operations and data reading depend on the accuracy of test equipment and subjective judgment of test personnel, which increases the uncertainty of the fatigue life. S-N curve is the basic parameter of the material or structure life prediction and it is gained mainly through a large number of fatigue tests and fitting analyses. Due to the influence of uncertainty factors, which results in dispersion of fatigue life under constant load in different degrees, so as to the uncertainty of S-N curve. Therefore, in order to keep consistent with the actual situation of the life prediction, taking the uncertainty of S-N curve into account is needed. This article summarizes the uncertainty factors that affect fatigue life of welded joint, the polynomial chaos theory is introduced into the fatigue life prediction, a welded joint cumulative damage model considering the uncertainty S-N curve and probabilistic fatigue life prediction method is built combined with nonlinear cumulative damage model.