J. Jatskevich, O. Wasynczuk, N. Mohd Noor, Purdue University; E. A. Walters, PC Krause and Associates, Inc; C. E. Lucas, Purdue University; P. T. Lamm, U.S. Air Force Research Laboratory
Recent advancements in computer networking have enabled the interconnection of inexpensive desktop computers to form powerful computational clusters that can be effectively utilized when simulating electric power systems. In order to maximize the computational gain, it is necessary to identify the computational tasks that can be performed concurrently and to optimally distribute those tasks among the available processors. In this paper, the electrical power system is viewed as a collection of interconnected dynamical subsystems each described by a set of differential/algebraic equations. The tasks that can be performed concurrently are identified and a new approach of optimally distributing the corresponding calculations is set forth. The effectiveness of the proposed approach is demonstrated by distributing a detailed simulation of the Western System Coordinating Council (WSCC) three machine nine-bus system on an SCI-based network composed of three personal computers. The simulation includes the effects of network and stator transients. Using the Runge-Kutta-Fehlberg integration algorithm, a 206% improvement in simulation speed was achieved.
Proceedings 14th Power Systems Computation Conference, June 24-28, 2002, Sevilla, Spain.