NUMERICAL EVALUATION OF GUARANTEED AVAILABILITY.

Steady-state availability and expected availability over a finite time interval are two commonly used system evaluation measures. These measures are not adequate when there is a penalty associated with not meeting a prespecified (or guaranteed) level of availability over a finite time interval. In such a case, the probability of not meeting this availability requirement must be evaluated. Therefore, a complete distribution of availability over a finite interval is needed. In general, there are no closed solutions for this distribution. A numerical approach for continuous-time Markov-chain models which allows component level modeling, Coxian failure and repair distributions, time-dependent failure and repair rates, and deferred repair and nondeferred repair strategies to be handled is introduced. This approach is illustrated by several examples.