GPU-Based DC Power Flow Analysis Using KLU Solver

The Power Flow (PF) analysis is the backbone of power systems operation, scheduling, and planning studies. The main objective in PF analysis of power systems is to obtain the voltage magnitude and phase angle at each bus. Computational burden is a major concern in PF analysis since numerical methods, such as Newton-Raphson (NR) and Gauss-Seidel (GS) methods, are used to iteratively solve such a nonlinear problem. By increasing the size and complexity of power systems, the computational burden correspondingly increases. This paper aims at presenting a Graphics Processing Unit (GPU)-based Direct Current (DC) PF analysis using the KLU solver to minimize the computational burden. The presented method is a GPU-based sparse solver based on the LU decomposition and can be used for fast matrix calculation. The obtained results clearly show a speed-up of approximately
$\times 10$
compared to MATPOWER, which is broadly used for PF analysis of power systems.