Low-Cost Analysis of Load Flow Computing Using Embedded Computer Empowered by GPU

Online power flow analysis is considered an essential tool for the smart grid operation due to the dynamic operating conditions of loads, market prices, distributed generation, and demand response. However, regarding microgrid operation, the cost of hardware and software to execute online load flows can significantly impact the project budget, mainly if commercial simulation tools are employed. Hence, in this paper, an algorithm based on the Newton-Raphson (NR) method to execute load flows using low-cost Embedded Computers (ECs) is presented. To optimize the convergence time of the NR method, a vectorization in the computation of the matrix, Jacobian matrix, and power injected flows is developed to execute these tasks in parallel using Graphics Processing Units (GPU). The proposed algorithm is evaluated using Python and an embedded computing board NVIDIA Jetson Nano with different test systems. As a result, the convergence time shows that the proposed algorithm is feasible for online power flow analysis in microgrids using a low-cost architecture.