A spark-based parallel simulation approach for repairable system

Fault-tree analysis is a useful analytic tool for the reliability and safety of complex system. However, fault tree is not suitable for repairable system. In this paper, we will propose a new method called TTF (time to failure) and TTM (time to maintenance) to analyze repairable system. Nevertheless, Monte Carlo simulation may be time consuming. In order to reduce simulation time, a parallel algorithm based on Spark will be used in this paper. Spark-MapReduce is the latest parallel computation framework. In situations where the amount of data is prohibitively large, we will propose a parallel algorithm for repairable system analysis to quickly get the simulation result. In this article, we propose a parallel algorithm to speed up through the experiment, we prove that the parallel algorithm has a superior performance on large scale models, and under Spark-MapReduce framework, researchers can concentrate on algorithm itself. It has significant benefits on reliability or availability assessment issues because it can free researchers, who are non-computer professional researchers, from parallelization and computational frame.