A procedure for formulation of multi-objective optimisation problems in complex water resources systems

Abstract

The recently completed Wimmera Mallee Pipeline Project (WMPP) provides reticulated water to 36 towns and about 6000 farms across an area of approximately 2 million hectares in Western Victoria (Australia). This new pipeline has replaced an open channel distribution system and has vastly improved efficiencies in the supply of water, with water savings being returned to the environment, existing consumptive use and new development. One of the challenges for managers of these water recovery projects is to determine the most effective or optimal operational strategy to meet the needs of all water users. The study area supplies a subset of the Wimmera Mallee Pipeline and includes two separate river systems, namely the Glenelg River and the MacKenzie River which flow South-ward and North-ward from the Great Dividing Range respectively. Both river systems have their own unique environmental and socio-economic attributes which are indicative of those relating to the broader Wimmera-Mallee Water Supply System. In Victoria, these often conflicting interests to water have traditionally been addressed through a consultative process supported by surface water simulation modelling. Simulation models attempt to represent all the major characteristics of a system and are therefore tailored to examine “what if?” scenarios. Whilst such models are highly effective in demonstrating the effect of changes in system operation, the modelling process is limited to finding one solution at a time for a given set of conditions. Moreover, such traditional approaches have been challenged by the need to develop sustainable water management plans which attempt to meet the need of all users by searching for the optimal operating rules. Optimisation models have also proven to be effective tools but unlike simulation models are characterised by a numeric search technique and are better suited to address “what should be?” questions. However, the lack of popularity in optimisation models has been due to the complexity in their development, computational effort, and subsequently the simplification in problem specification. In recent times there has been growing interest in linking optimisation techniques with simulation models in order to build on the strengths of both modelling approaches in the search for optimal solutions. The general structure of this combined modelling technique provides for an iterative process; simulation outputs are used to quantify the effect of candidate solutions which are in turn passed to the search engine to find optimal solutions. The aim of this study is to develop a generalised procedure for the formulation of multi-objective optimisation problems relating to multi-reservoir systems with complex operating rules. Importantly the procedure has been developed for problems that are intended to be solved using a combined optimisationsimulation modelling technique. For the present study, the procedure will be used to formulate a sample multi-objective problem for the optimisation of operation of the study area. The procedure is applied in case study form, detailing the various components of the problem, both in mathematical terms and also the necessary qualitative information derived from stakeholder participation. The outcomes of this paper demonstrate: • the importance of on-going stakeholder participation in providing higher level qualitative information as part of (a) the problem formulation process in order to explicitly account for all interests to water, and (b) the optimisation process in order to enable decision makers to make the necessary trade-offs between choosing one optimal solution over another; and • the need to systematically identify the relevant system operating rules that control the movement of water within the simulation model.