A Decomposition Strategy for Global Optimum Search in the Pooling Problem

Modeling of refinery and petrochemical processes that involve pooling of streams often results in nonconvex nonlinear relationships. In a number of cases these nonlinearities can make it difficult to determine the global optimum. A new search technique is presented in this paper that involves the decomposition of the original nonconvex problem into two specially structured subproblems. A series of these subproblems is then solved in an attempt to determine the global optimum. Suitable transformations can be combined with a judicious decomposition scheme leading to subproblems which can attain their respective global optimum solutions at each iteration. The effectiveness of the proposed approach is demonstrated using two small examples of the pooling problem proposed by C. A. Haverly in 1978 and one larger separation/blending/pooling problem. INFORMS Journal on Computing, ISSN 1091-9856, was published as ORSA Journal on Computing from 1989 to 1995 under ISSN 0899-1499.