Interpreting a Pareto set of operating options for water grids: a framework and case study

Multi-objective optimisation is being increasingly applied in water supply management to identify optimal operating options. However, a key challenge in the implementation of multi-objective optimisation is interpreting the large and multidimensional Pareto-optimal set. This paper shows how cluster, visual and post-optimisation analysis can aid the decision maker in addressing this challenge. This is demonstrated for a case study based on South East Queensland Water Grid, Australia, as part of a broader operational planning framework. Firstly, cluster analysis identifies a smaller set of representative options to aid in visual analysis. Secondly, visual analysis techniques are used to identify the trade-offs between objectives, the relationships between decision variables and objective performance, and to shortlist promising operating options. Finally, post-optimisation analysis techniques identify efficient operating options from the Pareto set, based on decision-maker preferences. Together these techniques can be used to identify a shortlist of operating options, for further consideration using multicriteria analysis.