Reliability of Fracture Mechanics Approach to Fatigue

Current fatigue design of fracture-critical components, such as tendons and risers, requires dual fatigue life criteria to be satisfied. The S-N approach includes a safety factor (SF) of 10 on the life of the component, while the fracture mechanics (FM) approach includes a safety factor of 5 on the life through-thickness of an acceptable initial flaw. FM provides critical initial flaw sizes such that suitability of the selected NDE methods and weld acceptance criteria can be established. This paper pertains to a comparative fatigue life reliability study between those two approaches. The objective is to develop a rationale for the selection of a safety factor on fatigue life to use in FM calculations. A reliability-based methodology is proposed and implemented. The SFs for FM are obtained by targeting the reliability obtained in fatigue designs based on historically proven S-N damage approach. Random variables entering both approaches were characterized and a number of weld design cases devised to obtain reliabilities. One important variable is the distribution of initial flaw sizes. For this study, flaw distributions were developed from actual inspection records, accounting for the effects of probability of detection and sizing accuracy of the inspection system, as well as the flaw acceptance criteria during fabrication. Comparisons of reliabilities obtained for designs by both approaches for various quality S-N curves, stress spectra, pipe sizes, and initial flaw sizes indicate that there is ample scope to modify downward the current FM safety factor. However, given the limited scope of this study, it is recommended to asses the FM SF using reliability analysis on a project-specific basis.