An improved decomposition algorithm for optimization under uncertainty

This paper proposes a modification to the decomposition algorithm of Ierapetritou and Pistikopoulos (1994) for process optimization under uncertainty. The key feature of our approach is to avoid imposing constraints on the uncertain parameters, thus allowing a more realistic modeling of uncertainty. A theoretical analysis of the earlier algorithm leads to the development of an improved algorithm which successfully avoids getting trapped in local minima while accounting more accurately for the trade-offs between cost and flexibility. In addition, the improved algorithm is 3-6 times faster, on the problems tested, than the original one. This is achieved by avoiding the solution of feasibility subproblems, the number of which is exponential in the number of uncertain parameters. (C) 2000 Elsevier Science Ltd. All rights reserved.