MODELS, ANALYSIS, AND APPROXIMATIONS FOR SYSTEM RELIABILITY AND AVAILABILITY,
Abstract
Probabilistic models are formulated for the computation of the system reliability in terms of the component reliabilities. Emphasis is placed on obtaining bounds and approximations that can be used to simplify the necessary computation. In evaluating the usefulness of these results the concept of 'goodness' is introduced. The system structure is modeled in terms of a reliability graph. From the minimal cut sets of the reliability graph, good bounds are obtained for the high reliability region, and from the minimal tie sets, good bounds are obtained for the low reliability region. For the special case of the k out of n structure, useful approximations are developed for the system reliability based on the normal and Poisson approximations, and for the special case of 'chain type' structures, the Weibull reliability function is shown to provide a good approximation to the system reliability. Two models are considered for the analysis of non-repaired systems with component dependence. The first is the state dependent hazard model where the component hazards are assumed to be conditional only on the set of failed components and not on the times at which these failures occur. The second case treated is where the component hazards are functions of the present time and times of all previous failures. Mathematically, the first case is treated as a Markov process and the second case is treated by means of multiple integrals. Finally, analog and Monte Carlo simulation are considered. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Jun 01, 1967
- Accession Number
- AD0671474
Entities
People
- Martin Messinger
Organizations
- New York University Tandon School of Engineering