A comparison of two confidence bound methods to estimate reliability

The central focus of this paper is a description of a coverage method for determining the validity of two lower confidence bound (LCB) procedures that were specifically designed for estimating the reliability of complex, repairable systems that give rise to grouped, continuous data. This coverage validation method was applied to (1) a stochastic confidence interval (CI) procedure that employs parametric bootstrap methodology and (2) a deterministic CI procedure that adheres to a set of closed-form equations. A systematic application of this coverage validation method to these two CI procedures provided LCB coverage solutions that clearly demonstrate that the bootstrap procedure is a more robust, stable method over a wide range of scenarios and is, therefore, the preferred method to use.