Spark-Based Monte Carlo Parallel Simulation Approach for FTA

In the field of reliability and safety, Fault-tree analysis proved to be a useful analytic tool. Aimed at static or dynamic FTs, Monte Carlo simulation based on time-to-failure (TTF) may be an effective way to calculate FTs. Through computing the TTF of bottom events of FTs, the TTF of the system can be calculated easily. However, Monte Carlo simulation based on TTF is time-consuming because modern engineering systems have become more complex and larger in scale. Traditional computation framework, in individual computers or small scale computer cluster, obviously cannot undertake this tough task. To address this issue, this paper puts forward a parallel algorithm under the Spark parallel computing framework, and through the experiment proves that the parallel algorithm can quickly get the simulation results and have the advantage of calculating large-scale dynamic FTs. This is very meaningful in the engineering.