Comparative analysis of GPU Approaches for Power Flow Calculation in CUDA

摘要

Power Flow (PF) calculation is widely used in the analysis of Electrical Power System (EPS), providing voltage and current values on the bus system as well as the loss of active and reactive power. Its computation, however, can become costly, as the number of existing buses in network increases. This study aims to assess the implementation of PF's main steps in one of its most used calculation methods for numerical solution, the Newton-Raphson NR) algorithm, under a General Purpose GPU (GPGPU) architecture. The main steps treated are the calculation of the admittance and Jacobian matrixes, and updating of voltages in the buses; as they present greater possibility of parallelization in order to reduce the execution time of the whole process. The application is tested comparing the computation time of its serial implementation and two GPU-based approaches.