Fatigue reliability analysis using nondestructive inspection

In this paper, a comprehensive approach is developed to integrate computational reliability methods and nondestructive inspection (NDI) for fatigue reliability evaluation. Uncertainties existing in the performance and outcome of NDI are considered through numerical statistical measures, and the mathematical relationships among these measures are derived. A new method is proposed to estimate the probability of detection (POD) and confidence interval using these relationships. Also, a new method of using the NDI results to update the reliability estimation is presented, which accounts for POD, measurement accuracy, false call probability, and the probability that a very large flaw may not be detected. The application of the proposed methods is illustrated through an example fatigue problem and inspection techniques with different POD characteristics are compared.