Parallel Newton type algorithms for transient stability computation were tested on machines with two different parallel architectures. The discretized nonlinear differential equations are solved together with the nonlinear algebraic equations for each time step. A parallel version of the very dishonest Newton (VDHN) method, which is the fastest sequential algorithm for transient stability simulation, and a successive over relaxed (SOR) Newton, which is inherently parallel, are tried on the local memory iPSC/2 and shared memory Alliant machines. Higher speedups than previously reported for transient stability analysis are obtained but the main thrust of the paper is to explore the match between the algorithms, their implementation and machine architectures. For example, the less parallel but sequentially faster VDHN runs faster on the hypercube (iPSC/2) whereas the more parallel SOR-Newton requires data sharing more often because of the extra iterations and does better on the Alliant. The implementation on the hypercube requires significant manual programming to schedule the processors and their communication whereas the compiler in the Alliant recognizes parallel steps but only if the software is properly coded. The paper presents these various considerations together with the test results.
ASJC Scopus subject areas
- Energy Engineering and Power Technology
- Electrical and Electronic Engineering