Accelerating Stochastic Lightning Attachment Simulations for the Estimation of Lightning Incidence to Overhead Lines

Abstract

Lightning attachment can be modeled through a stochastic approach adopting a detailed representation of the lightning phenomenon. A fractal-based modeling technique can be used for this purpose, considering lightning discharge branched and tortuous behavior, as well as physical properties associated with downward and upward leaders’ inception and propagation. However, fractal-based simulations require substantial computational resources, especially for the accurate calculation of the electric field at all points of the discretized simulation domain at each simulation step. Thus, the considerable computational cost inhibits the extensive application of stochastic simulations for estimating lightning incidence to common structures and power systems. This work investigates optimization techniques for fractal-based simulations regarding total simulation time; these are applicable to both high-performance computing and personal computers. The proposed techniques consist of a C-MATLAB integration methodology, as well as a multi-color ordering algorithm enabling parallel execution using CPU and GPU programming. Applications associated with lightning incidence to overhead transmission lines are presented. The total simulation time is substantially reduced with respect to the original code. A reduction of up to 98% is achieved, enhancing the applicability of stochastic modeling to lightning incidence estimation problems.