Probabilistic Stress-life (p-S-N) Curve Experiment Scheme Design And Verification

In consideration of the dispersion of the fatigue test datas, a group of S-N curves under different reliabilities could be obtained, named as probabilistic stress-life curve. This curve has been widely used to represent the fatigue properties of materials and parts. For the materials subjected to alternating loads, p-S-N curve is very important for design, which is a kind of more comprehensive stress-life relationship.The standard test method of p-S-N curve is the group test method. On the full consideration of the randness and dispersion of the fatigue test datas, so many samples were used in this method. The test results of the group test method could fulfill certain statistical requirements, reflect the characteristics of the population truthfully, but the high accuracy and confidence of this method were at the cost of long times, burdensome works and high costs. Owing to the disadvantages of the group test method, some classical fatigue theories and mathematical statistics tools were adopted and some efficient or simplified test methods were introduced in order to reduce the workload. Among these methods, some were built on the S-N curves that calculated by the single-point method, others have used inadequate samples in the fatigue experiment. Consequently, the accuracy and confidence began to decline and cannot fulfill the voice of engineer.In this paper the performance-life probability mapping principle was applied, the sample which material properties in a certain probability percentile of its population distribution, the life at any stress level will be at the level corresponding to the life of the population. Thus, the mapping relationship exist in the life distribution that under different stress levels and this one-to-one correspondence relation also exist in the life samples. According to this correspondence, the fatigue test datas that under different levels could be converted to a specified stress level, forming samples gathered and information fusion effect, then the life distribution parameters could be estimated.In this paper the p-S-N curve experiment scheme that base on the performance-life probability mapping principle was introduced, named as new-style small sample method. This method based on the principles that the fatigue life following lognormal distribution and there was a linear relationship between the stress level and the mean value or standard deviation of the normal population. Besides, the variance statistics test principle was also necessary. Fin-ally, the p-S-N curves that with high confidences could be successively conducted by only a small quantity of samples.