Parallel Computing and Localization Techniques for Faster Power System Dynamic Simulations

摘要

Dynamic simulation studies are used to analyze the behavior of power systems after adisturbance has occurred. This type of simulation is essential when the system is operatingclose to its stability limits or its behavior is dictated by complex control and protectionschemes modifying its trajectory. These simulations can be computationally very demanding,especially if performed over a time interval of several minutes. In this paper, new shared-memory parallel computing techniques to increase the performance of large-scale powersystem dynamic simulations are described. The algorithms presented achieve this by utilizingthe parallel processing resources available in modern, inexpensive, multi-core machines. Inaddition, the localized response of power systems after a disturbance is exploited to furtheraccelerate simulations without decreasing accuracy. The medium-scale model of a real powersystem and a realistic large-scale test system have been used for the performance evaluationof the proposed methods.