H41G-1432
Understanding London’s Water Supply Tradeoffs When Scheduling Interventions Under Deep Uncertainty
Thursday, 17 December 2015
Poster Hall (Moscone South)
Ivana Huskova1, Evgenii Sergeevich Matrosov2, Julien J Harou2,3, Joseph R Kasprzyk4 and Patrick M Reed5, (1)University College London, London, United Kingdom, (2)University of Manchester, School of Mechanical, Aerospace and Civil Engineering, Manchester, United Kingdom, (3)University of Manchester, Manchester, United Kingdom, (4)University of Colorado at Boulder, Boulder, CO, United States, (5)Cornell University, Ithaca, NY, United States
Abstract:
Water supply planning in many major world cities faces several challenges associated with but not limited to climate change, population growth and insufficient land availability for infrastructure development. Long-term plans to maintain supply-demand balance and ecosystem services require careful consideration of uncertainties associated with future conditions. The current approach for London’s water supply planning utilizes least cost optimization of future intervention schedules with limited uncertainty consideration. Recently, the focus of the long-term plans has shifted from solely least cost performance to robustness and resilience of the system. Identifying robust scheduling of interventions requires optimizing over a statistically representative sample of stochastic inputs which may be computationally difficult to achieve. In this study we optimize schedules using an ensemble of plausible scenarios and assess how manipulating that ensemble influences the different Pareto-approximate intervention schedules. We investigate how a major stress event’s location in time as well as the optimization problem formulation influence the Pareto-approximate schedules. A bootstrapping method that respects the non-stationary trend of climate change scenarios and ensures the even distribution of the major stress event in the scenario ensemble is proposed. Different bootstrapped hydrological scenario ensembles are assessed using many-objective scenario optimization of London’s future water supply and demand intervention scheduling. However, such a “fixed” scheduling of interventions approach does not aim to embed flexibility or adapt effectively as the future unfolds. Alternatively, making decisions based on the observations of occurred conditions could help planners who prefer adaptive planning. We will show how rules to guide the implementation of interventions based on observations may result in more flexible strategies.
