Tearing Algorithms for Large-Scale Network Programs

Abstract

In a recent survey taken among more than 50 electric utilities and power pools, more than 30 percent expressed a need for load flow and stability programs that would be capable of representing more than 2000 buses and 3000 branches. In fact, more than 20 percent indicated a need for 3000 bus programs within the next two years, and some wanted 4000 bus programs. When using methods for network solution such as the Newton-Raphson method, the amount of computer memory required by a 2000 bus program, even when sparsity is effectively exploited, is near the limit of very large computers. There is, then, an obvious need for tearing techniques that can be applied to such solution methods as the Newton-Raphson method, the Y-bus method, the factorized Z-bus method, and other equivalent methods.