High Performance Computing (HPC) for Advanced Power System Studies

This MSc dissertation presents a probabilistic study for advanced modern power systems based on a high performance computing (HPC) method supported by the Computational Shared Facility at the University of Manchester. The main outcome of this research is the contingency-based probabilistic load flow analysis of the IEEE 9 & 39 bus system as well as the use of parallel processing on HPC for running advanced application programs efficiently.In recent years, with the advent of smart grid and widely application of renewable energy the modern inter-connected power systems are becoming more complex and more uncertainties are introduced. In order to obtain further information about the effect of the uncertainty for power systems, probabilistic studies are developed to perform the stability and security research. From the precious research it is known that some probabilistic system variables such as wind speed, load demand and solar irradiation follow Weibull, Normal and Beta distribution respectively. In this project, load demand and power generation are both assumed following the Normal distribution as the probabilistic input of the system.There will have a huge burden on standard PC because of the hundreds of thousands computation involved with the probabilistic study. The High Performance Computing (HPC) is introduced as a powerful computation technique which enables large-scale computation tasks to be split to a number of small jobs which could be run simultaneously on parallel cores and get the results quickly. Big data obtained from HPC needs to be visualised to a suitable layout form to display what the results are obtained and then to perform the power system load flow analysis, N-1 as well as N-2 contingency analysis.