Improving the Efficiency of Grounding System Analysis Using GPU Parallelization

Abstract

Safety analyses of the effectiveness of large grounding systems are often hampered by the lengthy computation times. Using even the simplest image models, the evaluation of touch and step voltages can require from several minutes to several hours of computations on modern CPUs. Our analysis shows that substantial reduction of computation times can be achieved by utilizing GPU parallelization. In this paper we provide basic steps in the implementation of GPU parallelization on the simplest equipotential model for grounding analysis in homogeneous earth, and we test the effectiveness of this approach in different scenarios.