Through this paper is presented a different close up to the Optimal Power Use Surface (OPUS) methodology involving the use of Integer Linear Programming (ILP). In this approach, some steps from the former procedure are modified in order to obtain Water Distribution Systems (WDS) designs more efficiently. In this case, OPUS is carried out until the optimal flow distribution is found and then the final design is calculated through ILP. This is done by decomposing the network into an open-tree structure in which nodes are removed and duplicated while maintaining the same hydraulic characteristics from the initial looped system. The objective of this variation is to avoid the redundancy in the scheme, which promotes hydraulic inefficiency to the process of optimization. In brief the understanding of the hydraulic principles and the ways in which energy is distributed and dissipated in the system are used to set-up an equivalent cost-efficient water network, one which can be used on a ILP optimization. In the present work, results are obtained by testing this methodology within four benchmark problems (Hanoi, Balerma, R28 and Taichung). Comparing the results with the designs achieved by diverse methodologies, we can establish that the methodology accomplishes results with similar constructive costs but requiring much less iterations. This methodology represents an alternative to exhaustive meta-heuristic proceedings, requiring a lower computational capacity and attaining low-cost WDS as well.
展开▼
