Managing population and drought risks using many‐objective water portfolio planning under uncertainty

Abstract

[1] This study contributes a many-objective analysis of the tradeoffs associated with using the portfolio planning approach for managing the urban water supply risks posed by growing population demands and droughts. The analysis focuses on four supply portfolio strategies: (1) portfolios with permanent rights to reservoir inflows, (2) adaptive options contracts added to the permanent rights, (3) rights, options, and leases, and (4) rights, options, and leases subject to a critical reliability constraint used to represent a maximally risk averse case. The portfolio planning strategies were evaluated using a 10 year Monte Carlo simulation of a city in the Lower Rio Grande Valley (LRGV) within Texas. Our solution sets provide the tradeoff surfaces between portfolios' expected values for cost, cost variability, reliability, surplus water, frequency of using leases, and dropped (or unused) transfers of water. Using an additional severe drought scenario, this work shows that leases and options can reduce the potential for critical supply failures when urban supply systems must contend with unexpected and severe extremes in both demand and water scarcity. In summary, this paper contributes a framework that couples interactive visualization and many-objective optimization to innovate urban water portfolio planning under uncertainty. The many-objective analysis of the LRGV case study shows that effective water portfolio planning can simultaneously improve the costs, efficiency, and reliability of urban water supply while ensuring adaptability and resiliency to future changes.Citation: Kasprzyk, J. R., P. M. Reed, B. R. Kirsch, and G. W. Characklis (2009), Managing population and drought risks using many-objective water portfolio planning under uncertainty, Water Resour. Res., 45, W12401,

Evolutionary algorithms and other metaheuristics in water resources: Current status, research challenges and future directions

“…In contrast to the objective space, the decision space does not always have a clear topology, as many practical engineering problems contain mixed decision spaces (combinations of real and discrete decision dimensions) and complex decision tree structures (such as those arising from planning applications, e.g. Kasprzyk et al (2009), which require more general approaches to defining topology that are meaningful for behavioural analysis. A second issue concerning the spatial dimension is the characterisation of "spread".…”

“…The WRM optimisation approaches where uncertainties are handled in a more explicit way have appeared relatively recently (e.g. Cui and Kuczera, 2005;Deb and Gupta, 2006;Gopalakrishnan et al, 2003;Kasprzyk et al, 2009;Kourakos and Mantoglou, 2008;Labadie, 2004;Mortazavi et al, 2012;Smalley et al, 2000;Tolson et al, 2004;Vasquez et al, 2000).…”

“…More recently, different EAs were used (e.g. Cui and Kuczera, 2005;Deb and Gupta, 2006;Gopalakrishnan et al, 2003;Kasprzyk et al, 2009;Kourakos and Mantoglou, 2008;Smalley et al, 2000;Tolson et al, 2004;Vasquez et al, 2000). The original deterministic optimisation problem objectives are modified, typically by: (a) introducing some measure of robustness (or risk, resilience, reliability, etc.)…”

Optimal Design of Water Distribution Systems Using Many-Objective Visual Analytics

“…Recent studies have demonstrated the use of many-objective visual analytics in water supply risk management (Kasprzyk et al 2009;Kasprzyk et al 2012) and groundwater monitoring network design (Kollat et al 2011), and have yielded new design insights and demonstrated the potentially highly negative consequences that could result from lower dimensional formulations. The increase in the number of objectives brings new challenges to multiobjective optimization: deterioration of search ability, exponential increase in non-dominated Pareto approximate solutions and difficulty in solution visualization.…”

“…Popular multiobjective genetic algorithms, such as NSGAII (Deb et al 2002), lose their effectiveness as the number of objectives increases, although they have been shown as effective for multi-objective WDS design (Farmani et al 2003). The epsilon Nondominated Sorted Genetic Algorithm II (ε-NSGAII) (Kollat and Reed 2006) has been demonstrated as effective and efficient for solving many objective problems (Kasprzyk et al 2009;Kollat et al 2011, Hadka and Reed in press, Reed et al in press).…”