Parallel Computing of Load Flow Analysis

This paper presents a simple and an efficient approach for load flow analysis of a radial distribution network by using parallel computation. By introducing CUDA (Computationally Unified Device Architecture), a platform is provided for performing the parallel implementation of the analysis.The analysis starts by determining the order of the nodes present in the distribution network. To generate the order of the nodes, Breadth First Search is used in parallel. Further, the current and the voltages present at each node of the network are evaluated. This procedure is repeated till the convergence criterion is achieved. CUDA provides an efficient utilization of the computational power of GPU, by which the performance and throughput of sequential load flow algorithm, has been enhanced dramatically. With the immense support of many GPUs, CUDA helps load flow analysis to efficiently manage and manipulate a large blocks of data of the distribution networks. Moreover there has been a lot of reduction in the execution steps as compared to the serial computing.