This paper presents a reliability-based damage tolerance (RBDT) methodology that employs a systematic approach for probabilistic fracture-mechanics damage tolerance analysis subject to uncertainties in initial flaw, probability of detection (POD) and other random variables. By integrating a basic probabilistic analysis engine, a finite element analysis code, and a fatigue crack analysis code, the RBDT methodology has been implemented in a prototype software and used to assess the applicability and benefits of applying RBDT methodology to supplement the safe-life and deterministic-based damage tolerance design methods.
A new efficient and accurate probabilistic method, built on a two-stage conditional importance sampling approach, is presented. The first stage computes risk, without inspection, using the most probable point (MPP)-based importance-sampling technique combined with a new error-checking method. The second stage computes risk, with inspection, by simulating inspection and maintenance effects using the samples generated from the Stage 1 failure domain. The error-checking procedure addresses the major shortcoming in the MPP-based approximation methods and leads to a robust and efficient sampling method.
For inspection optimization and maintenance planning, this paper proposes a strategy to significantly speedup the optimization process by re-using the crack growth histories for risk and risk reduction computations without additional stress and life analyses.
The results from several demonstration examples suggest that the improved two-stage importance-sampling method is well suited for RBDT analysis, particularly with inspection planning.