Abstract
The strategy in single-objective deterministic water network design is to size and locate components to minimize capital cost and meet future peak demands at or above a minimum pressure. Increasingly, practitioners are turning to multi-objective stochastic design to balance the minimum-cost objective with hydraulic performance objectives. The aim of the current paper is to review single-objective deterministic and multi-objective stochastic network design and discuss practical considerations concerning their advantages and disadvantages of relevance to water industry professionals and practitioners. Key differences in data and computational requirements, comprehensiveness of analysis, and decision flexibility between the two approaches are illustrated with a complex, hypothetical network example. A Monte-Carlo simulation program was used to solve the multi-objective stochastic problem and generate a set of Pareto or near-Pareto solutions with pipe cost ranging between $9.3 and $17.4 million and hydraulic robustness ranging between 65.8 and 96.4 per cent. Results indicated a non-linear relationship between pipe cost and robustness typical of many systems and that a large premium must be paid to achieve marginal improvements in robustness beyond a value of 90 per cent. The MCS program was run for 30.6 h to test the hypothetical network against a broad range of demands to ensure a high level of hydraulic robustness. The Pareto curve allows the decision maker the opportunity to quickly assess trade-offs between pipe cost and robustness.
Get full access to this article
View all access options for this article.