Decomposition and Parallel Processing Techniques for Large Scale Electric Power System Planning Under Uncertainty

M. Avriel, and G. B. Dantzig, and P. W. Glynn

Proceedings of  NSF Workshop on Resource Planning Under Uncertainty, 3-34 (1990)

Our research is concerned with developing software tools for finding optimal resource expansion plans for large scale power systems. An acceptable expansion plan must be capable of reliably supplying demand in the face of uncertainty about future demand and availability of generators and transimission lines due to equipment failures. The incorporation of uncerntainty into the resource planning model of an electric power systems is essential.

Unfortunately, the number of possible contingencies that must be hedged against can run in the millions. There are no off-the-shelf algorithms available to solve the problem in general. We believe, nevertheless, that these problems, though very very large in size, can be solved by applying improved methods such as decomposition techniques in combination with importance sampling and the exploration of advanced computer technologies (parallel processors). The methodology described is being tested on a prototype stochastic Western Resource Planning Model for the west cost of the United States.