摘要
Recently, a considerable emphasis has been laid to the reliability-based optimization model for water distribution systems. But a considerable computational effort is needed to determine the reliability-based optimal design of large networks, even of mid-sized networks. A new methodology which consists of two procedures is presented in this paper. The first procedure is that the optimal design is constrained only by the pressure heads at demand nodes, done in GRG2. Because the reliability constrains are removed from the optimal problem, a number of simulations do not need to be conducted and the computer time is greatly decreased. Then, the second procedure is a linear optimal search procedure. In this linear procedure, the optimal results obtained by GRG2 are adjusted by the reliability constrains. The results are a group of commercial diameters of pipes and the constraints of pressure heads and reliability at nodes are satisfied. Therefore, the computer burden is significantly decreased, and the reliability-based optimization is of more practical use.
Recently, a considerable emphasis has been laid to the reliability-based optimization model for water distribution systems. But a considerable computational effort is needed to determine the reliability-based optimal design of large networks, even of mid-sized networks. A new methodology which consists of two procedures is presented in this paper. The first procedure is that the optimal design is constrained only by the pressure heads at demand nodes, done in GRG2. Because the reliability constrains are removed from the optimal problem, a number of simulations do not need to be conducted and the computer time is greatly decreased. Then, the second procedure is a linear optimal search procedure. In this linear procedure, the optimal results obtained by GRG2 are adjusted by the reliability constrains. The results are a group of commercial diameters of pipes and the constraints of pressure heads and reliability at nodes are satisfied. Therefore, the computer burden is significantly decreased, and the reliability-based optimization is of more practical use.
