Unification of Reliability and Confidence Level in Reliability-Based Design : The Case of 2-Parameter Weibull Distribution

抄録

In reliability-based design and testing, reliability R(=1-P) and confidence level C(=1-γ)are usually used in a combined manner. In the previous paper, an attempt was made to use the mean reliability as a one-dimensional measure. In the present paper, this idea was applied to the determination of the design-allowable stress S_a given that strength follows a two-parameter Weibull distribution W(α, β). The factor k in the expressions of S_a=β^^^/k^1/α for the case where the parameter α is known and only the scale parameter β is unknown, and S_a=β^^^/k^{1/α^^^(/SUP) for the case where both αandβ are unknown was determined based on this idea, where α^^^ andβ^^^are the most likely estimates of α and β, respectively. The variations of k with sample size n were obtained by theoretical analysis and Monte Carlo simulation.}