Reliability analysis of interconnected power systems using Monte Carlo simulation on ncube parallel computer Conference Paper uri icon

abstract

  • Reliability analysis of power systems using Monte-Carlo simulation is a computation intensive problem. The parallel processing techniques provide a natural solution to this problem. In this paper, parallel implementation of Monte-Carlo simulation on ncube parallel computer is discussed. If the Monte-Carlo simulation has to be performed for a fixed number of iteration (times), then the task would be allocating the total number of iterations among various processors. At the end of the simulation, the results can then be aggregated. In this case, the synchronization and communication overhead involved in parallelizing the problem is minimum and the efficiency that can be obtained is the best. But in actual practice, Monte-Carlo simulation is not performed for a fixed number of iterations, but is stopped based on some convergence criterion. After a certain number of iterations the results are checked for convergence. The convergence checking if employed in parallel algorithms will increase the overhead in synchronization and communication. An algorithm which minimizes these overheads on ncube, which is a distributed memory MIMD architecture is discussed in this paper. Speedup, efficiency and various issues related to this parallel implementation is discussed for a sample system.

published proceedings

  • Proceedings of the American Power Conference

author list (cited authors)

  • Gubbala, N., & Singh, C.

complete list of authors

  • Gubbala, N||Singh, C

publication date

  • December 1993