PARAGON: an approach for parallelization of power system contingency analysis using Go programming language

With the recent emphasis on N-k contingency analysis (CA) and expansion of power systems, the computational demand of CA has greatly increased to ensure power system security. The serial execution platforms are proving to be insufficient to cater to this demand. This paper presents PARAGON, an efficient approach for parallelization of power system contingency analysis using Go programming language. PARAGON uses Google's Go language to parallelize CA over multiple cores using goroutines. The main contribution of this paper is implementation of work-stealing scheduler on widely used multicore processors using concurrent programming language, which can be easily adopted by power system control centers to accelerate legacy codes. Simulations have been performed using a large power system with hundreds of contingencies, and contingency analysis has been parallelized using 2, 4, 8, 12, and 16 cores. The results show that PARAGON provides large computational advantage over serial execution and also outperforms a conventional scheduling technique, namely master-slave scheduling.