next up previous
Next: References Up: Power Systems Transient Stability Previous: Estimates of Potential

Conclusions

A recent IEEE committee report [41] by a task force of the Computer and Analytical Methods Subcommittee of the Power Systems Engineering Committee states that

Except for those analytical procedures that require repeat solutions, like contingency analysis, there are no obvious parallelism inherent in the mathematical structure of power systems problems,
We believe that this view of parallelism in power systems problems illustrates the need for closer coordination with the computational science research community. There are problems in other disciplines that illustrate the existence of parallelism in the solutions of differential-algebraic equations, the central component of the transient stability problem. The critical point is the granularity where parallelism exists and the level of sophisticated techniques that are required to extract that parallelism for particular parallel processing hardware.

Various computational science research topics for the transient stability problem have been proposed in this paper. The goal of this paper has been to describe various research areas where the computational science academic community can interact with the power systems engineering community to improve the quality and performance of power systems transient stability analysis simulations. Speedups of over 1000 appear possible for large multiprocessors simulating the transient stability of large interconnected power systems, while speedups of over 100 appear reasonable for individual medium-sized power utility companies. Real-time or faster-than-real-time transient stability simulations will require both highly parallel computers and better overall algorithms to get the computational speedup required for this grand computing challenge, but the benefits of improved system reliability should yield substantial payoffs for electrical utility company profits, environmental impact, and customer satisfaction,

Acknowledgment: We thank Ernst Hairer, Alan Hindmarsh, Alvin Leung, Nancy McCracken, Linda Petzold, and Tony Skjellum for their assistance in preparing this paper.



David P. Koester
Sun Oct 22 16:35:54 EDT 1995