A Monte Carlo Technique Using Component Failure Test Data to Approximate Reliability Confidence Limits of Systems with Components Characterized by the Weibull Distribution
Abstract
The paper develops a Monte Carlo technique which, with a digital computer, determines confidence limits for system reliability of complex systems containing components characterized by the Weibull distribution. The component distribution shape and scale parameters are estimated by the method of maximum likelihood from component failure times while the location parameter is assumed known. The asymptotic distribution of these maximum likelihood estimators and a Monte Carlo simulation are used to determine confidence limits on system reliability. As an example, confidence limits are calculated for two systems of up to eight components in combinations of series and parallel configurations using 99, 499, 999, and 2999 simulations. Accuracy of the confidence limits is found to be satisfactory after being checked by a method using a double Monte Carlo technique.
Document Details
- Document Type
- Technical Report
- Publication Date
- Mar 01, 1972
- Accession Number
- AD0743633
Entities
People
- Robert G. Lannon
Organizations
- Air Force Institute of Technology