Water Journal : Water Journal August 2012
governance refereed paper technical features 54 AUGUST 2012 water algorithm identifies low-flow showerheads as the most cost-effective solution leading to improved cost efficiency. To encourage greater uptake and larger water savings, rebate levels increase, leading to reduced cost efficiency. Towards the right of the Pareto curve, the least cost-efficient measures -- indoor end-use connected rainwater tanks -- become part of the Pareto optimal combination of rebates. Subsidising only the most cost-efficient of measures -- low-flow showerheads -- achieves limited volumes of yield. The two curved lines in Figure 2 represent all non-dominated combinations of policy for South Australia (lower) and Western Australia (upper). The Western Australian frontier lies further within the optimal space -- more efficient solutions are able to be achieved due to higher observed uptake of measures at lower rebate levels compared to the observed South Australian program. The two horizontal lines represent estimates of the long-run marginal cost of supply -- that is, the levelised cost associated with the next possible expansion in supply -- in Western Australia ($1.08/kL) and South Australia ($2.40/ kL). Points on the Pareto frontier that lie below these lines represent policy configurations that would induce sufficient uptake in order to deliver water savings lower than the cost of supply expansion. The intersection of the curved Pareto frontier and the horizontal lines of long- run marginal cost represent economically efficient and Pareto-optimal solutions. For South Australia, this corresponds to rebate policy, which applies rebates of half of the retail cost to indoor appliances and pool covers, delivering a program cost of $101.3m and an estimated water saving of 42.55GL for a 10-year period. Applying an optimisation algorithm that evaluates the entire Pareto front in one run facilitates posterior articulation of preferences from policymakers. They are able to identify the various configurations of policy that are able to meet objectives at either constrained budgets or for target water savings. On this chart we also model reference points of past Western Australian and South Australian programs. These correlate to the water savings and cost that could have occurred given the actual numbers of uptake, evaluated with our water savings dataset and cost model. These reference points exclude the water savings achieved by, and costs incurred from, measures our demonstration did not include due to lack of data, but which were part of the programs -- e.g. garden measures, groundwater bores, etc. This allows for a 'like-for-like' comparison. Immediately evident on this chart is the fact that these two reference points lie within the dominated area of the Pareto curve. The horizontal distance at which these points lie above the curve, multiplied by the water savings achieved, represents the potential cost savings that could have been realised if rebate choices and prices were optimally formulated. For South Australia, this represents removing rebates for 1000L tanks, reducing the rebate incentive for washing machines and most tank rebate options, and increasing the rebate incentive for showerheads and pool covers to $60 and $300 respectively. Conclusion This research develops and demonstrates a new framework for Australian and international policymakers to formulate optimal rebate policies for water-saving measures to reduce household water consumption. Multi-objective genetic algorithm optimisation techniques were successfully applied for the first time in this context to allow policymakers to develop a Pareto-efficient frontier of optimal policy mixes. A review of historical rebate programs produces new knowledge about the drivers and expected response of Australian consumers to rebate programs. Applying behavioural end-use stochastic simulation, this research demonstrates a new method of estimating water savings from retrofit and rainwater tank installation measures. Through the application of a formal optimisation framework, our results demonstrate that improved economic and environmental outcomes can be achieved. Current methods of evaluating program success -- through ex-post economic analysis -- are plagued, as they are retrospective and often rely on poor or exaggerated water savings estimates. Least-cost planning methods that identify the levelised cost of measures in isolation are useful only to formulate policy when budgets are unconstrained. Our framework improves on both through ex-ante estimation of yield and cost with an uptake model, improved estimates of water savings and consideration of measures in combination. This paper won the Undergraduate Water Prize at Ozwater'12 in May. Acknowledgements The Authors wish to acknowledge Professor Graeme Dandy and Dr Mark Thyer from the School of Civil, Environmental and Mining Engineering at the University of Adelaide for their research supervision and support. The Authors Augusta Lane, Ronnie Ling, Catriona Murphy (email: catriona.murphy@ smec.com) and Ian Usher graduated with degrees in Civil & Environmental Engineering at the University of Adelaide in 2011. Catriona and Ian are Graduate Water Engineers at SMEC and GHD Adelaide respectively. Augusta is completing a Bachelor of Science in Geology and will next year start as an engineer with SKM Adelaide. Ronnie is completing a Bachelor of Petroleum Engineering and will join ExxonMobil in Melbourne in 2013. References Barton A (2003): Investigation into the Characteristics of Residential Water Use In Adelaide, In: Association, AW (ed.) AWA SA Branch Regional Conference. Adelaide, SA. Breen P, Coleman J, Deletic A, Duncan H, Fletcher T, Gerhart T, Grant A, Holt P, Inman M, Maheepala S, Mitchell G, Sharma AK, Stewart J, Rahilly M & Vieritz A (2006): Integrated Urban Water Modelling Review Stage 2 Report In: CRC, E. (ed.). Australia. Coombes PJ & Kuczera G (2003): Analysis of the Performance of Rainwater Tanks in Australian Capital Cities. 28th International Hydrology and Water Resources Symposium, Wollongong NSW: The Institution of Engineers, Australia. Deb K, Pratap A, Agarwal S & Meyarivan T (2002): A Fast and Elitist Multi-Objective Genetic Algorithm: NSGA-II. IEEE Transactions on Evolutionary Computation, Vol 6. Gillingham K (2009): Economic efficiency of solar hot water policy in New Zealand. Energy Policy, 37, pp 3336--3347. Loh M & Coghlan P (2003): Domestic Water Use Study. In: CORPORATION, W (ed). Millock K & Nauges C (2010): Household Adoption of Water-Efficient Equipment: The Role of Socio- Economic Factors, Environmental Attitudes and Policy. Environmental & Resource Economics. Olmstead SM & Stavins R (2008): Comparing price and non-price approaches to urban water conservation. Harvard University. Roberts P (2005): Residential End Use Measurement Study -- Final Report. Yarra Valley Water. Savic DA, Bicik J & Morley MS (2011): A DSS generator for multiobjective optimisation of spreadsheet-based models. Environmental Modelling Software, 26, pp 551--561. Thyer M, Micevski T, Kuczera G & Coombes P (2011): A Behavioural Approach to Stochastic End Use Modelling. Presented at Ozwater'11, Adelaide, SA. White SB & Fane SA (2002): Designing Cost Effective Water Demand Management Programs in Australia. Water Science and Technology, Vol 46 (6--7), pp 225--232.
Water Journal September 2012-1
Water Journal July 2012