Composite reliability evaluation by sequential Monte Carlo simulation on parallel and distributed processing environments

This paper describes two parallel methodologies for composite reliability e valuation using sequential Monte Carlo simulation, The methodologies are ba sed on coarse grain asynchronous implementations. In the first methodology, a complete simulation year is analyzed on a single processor and the many simulated years necessary for convergence are analyzed in parallel. In the second methodology, the adequacy analysis of the system operating states wi thin the simulated years is performed in parallel and the convergence is ch ecked on one processor at the end of each simulated year. The methodologies are implemented on a 10 nodes IBM RS/6000 SP scalable distributed memory p arallel computer and on a network of 8 IBM RS/6000 43P workstations. The re sults obtained in tests with actual power system models showed high speedup and efficiency on both parallel platforms.