Novel Spanning-Tree Matrix Approach to Model and Optimize Large-Scale, Tree-Shaped Water Distribution Networks

Abstract

There exist many criteria for the optimal design of water distribution networks. One of the most common criteria is to design the optimal cost water distribution network while satisfying the hydraulic design constraints. This study was carried out to propose a novel computational method named Spanning-Tree Matrix Approach that can model large-scale tree-shaped water distribution networks. A case study was tested to demonstrate the use of the Spanning-Tree Matrix Approach model coupled with the Honey-Bee Mating Optimization algorithm to find the combination of pipe diameters that minimizes the cost of the network. The results show that the Spanning-Tree Matrix Approach is successful in modeling a tree-shaped water distribution network of any size. Moreover, proposed Spanning-Tree Matrix Approach has the flexibility to be adapted to any desirable governing equation or design criteria being imposed, and the element of simplicity to output desired constraint evaluations into a modern stochastic optimization algorithm (i.e., Genetic Algorithm, Simulated Annealing, Ant-Colony Optimization, Honey-Bee Mating Optimization, etc.) for the network optimization purpose.