| Despite the significant advancements that have taken place in computer technology, the lengthy execution times associated with a single-computer implementation of a detailed simulation of large-scale systems, including typical electrical power systems, remain unacceptable using even the most advanced state-of-the-art computer workstations. In this thesis, a discussion of the characteristics or attributes of typical power systems that ultimately limit the size or the complexity of the system that can be simulated is presented. In addition, several existing techniques that have been applied to improve the computational speed of such large-scale system simulations are set forth. While each of these approaches has yielded some improvement, they also involve approximations or have other associated limitations. A new simulation approach, referred to as the Parallel-Rates Integration (PRI) technique, that yields significant improvement in the computational speed of power-electronic-based systems has been developed. In this new approach, the overall system is viewed as a collection of interconnected subsystems. The structure allows for an unlimited number of subsystems to be included into a simulation and readily distributed across any number of networked computers. Each subsystem interacts with the other subsystems through the necessary exchange variables. The subsystems can be electrical, mechanical, hydraulic, or any combination thereof. The principle features associated with this technique are improved computational performance as well as a compatibility with established approaches. These features along with others described herein will enable this PRI technique to be used to implement system-level simulations wherein the constituent subsystem models can be developed by inter-disciplinary researchers. |
