On The Reliability Associated With Varying Degrees Of Structural Usage Monitoring

Dynamic component fatigue lives are traditionally calculated using a deterministic process whereby usage is biased to a postulated composite-worst-case (CWC), load severity is biased to top of scatter (TOS) loads, and fatigue strength is based on a mean minus three sigma S-N curve. With the advent of structural usage monitoring systems (SUMS) the usage spectrum can be revised from the postulated CWC to a more realistic spectrum based on actual usage. A necessary constraint associated with this alternative lifing process is that its resultant retirement times meet or exceed a reliability of six nines (i.e. 0.999999 or 0.96). A significant measure of this reliability can be gained from a revised spectrum by adjusting the mean SUMS to a mean plus one standard deviation usage. In the probabilistic sense this level of reliability is expected to come from three stochastic sources: (1) usage severity, (2) load severity, and (3) fatigue strength. The required 0.96 reliability can be achieved using the SUMS usage results through a deterministic increase in the resultant spectrum severity, while making the load severity and fatigue strength random variables. This paper will show that fatigue lives calculated in this manner incur an approximately twofold increase in retirement time when compared with the traditional (CWC spectrum/TOS loads/mean minus three sigma S-N curve) lifing method. Gains in life of this magnitude (at 0.96 reliability) were demonstrated for two typical rotorcraft dynamic components