Synthesis of separation sequences by ordered branch search

An efficient and algorithmic procedure is developed for the synthesis of multicomponent separation sequences. The procedure involves list processing of the prossible separation subproblems followed by an ordered branch search to find the optimal sequence with respect to system structure. The technique is conveniently represented by an and/or directed graph. The distribution of sequence costs for a separation problem is considered. When a wide distribution exists, only a minimum of separators need be analyzed to find the optimal sequence. Even when a narrow distribution of sequence costs exists, not all sequences must be developed. The importance of near optimal sequences is examined, and the search procedure is extended to find those sequences whose costs are within a specified factor of the cost of the optimal sequence.