Determination of Reliability Using Event-Based Monte Carlo Simulation Part II

Analytic determination of reliability of general systems becomes increasingly difficult as the system complexity increases. Inclusion of s-dependent components and/or allowing repair of failed components makes such analysis virtually impossible. This paper extends the Monte Carlo simulation procedure of Kamat and Riley to estimate the reliability of such systems. A computer program finds all minimal tie-sets from a coded reliability flow graph and performs the simulation. In each replication, the s-dependency of failures and/or repair of failed components is incorporated in the generation of pseudorandom time-to-failure through modification of the failure rate for each affected component. A search through the minimal tie-sets identifies the success or failure of the system at each value of required time of operation. The computer program estimates the system reliability as a function of time with approximate individual (for each time point) 95 percent symmetric s-confidence limits.