Article ID: | iaor20081414 |
Country: | United Kingdom |
Volume: | 37 |
Issue: | 3 |
Start Page Number: | 291 |
End Page Number: | 305 |
Publication Date: | Apr 2005 |
Journal: | Engineering Optimization |
Authors: | Keedwell Edward, Khu Soon-Thiam |
Keywords: | heuristics: genetic algorithms, programming: multiple criteria |
The New York city tunnel network problem is a classical case study for water distribution network rehabilitation or upgrading. However, standard problem formulations only offer single solutions based on fixed pressure head and demand requirements. This problem is re-examined through the use of multiple criteria optimization and a framework for interpreting these results is proposed. Results from previous studies were closely examined to highlight areas that can be improved. A 6-objective optimization problem was then formulated and results were compared with optimization using a single-objective genetic algorithm (GA) and 2-objectives optimization using a non-dominated sorted genetic algorithm-II. Although the initial runtime for the 6-objective optimization was long compared with single-objective GA, it provided solutions which span the whole spectrum of possibilities (in terms of pressure head requirements and cost). It is concluded that the proposed 6-objectives formulation can indeed offer more design choices, hence flexibility, for designers and decision makers.