Abstract: The purpose of this paper is to develop the optimization procedures that reduce the dimensionality of the nonserial converging branch networks in dynamic programming. Space and computational complexities of the single converging branch system are analyzed and compared with two other methods. An efficient algorithm for the multi-converging branch network is developed by identifying the levels of branches. The space complexity of the procedure is also examined and compared. The determination of the best main serial chain is considered based on the complexity analysis.