Robust optimization using a variety of performance measures: A case study of water systems planning under climate and demographic uncertainty in Amman, Jordan.

摘要

The first part of this dissertation suggests that, among the various techniques used for uncertainty analysis in optimization modeling, robust optimization (RO) might be the best qualified for evaluation and control of the various risks of poor system performance resulting from the uncertainties in the data input to water resources problems. Due to the great deal of variety in the use of, and some amount of confusion regarding, the term RO in the water resources literature, Chapter 2 offers an updated and somewhat broader definition of the term for use in this field. Chapter 3 presents a simple application of RO to a water resources problem, and recommends a revised formulation designed to overcome well-known inadequacies in the base formulation.;The second part of the dissertation concentrates on a case study of the water system in the Greater Amman Area (GAA), Jordan. Chapter 4 presents the current state of the water resources system in the GAA, and projections for the likely effects on the system of climate and demographic change. Chapter 5 presents an RO model as applied to the water system in the GAA, which minimizes the expected cost to the utility of operating and expanding the current water system in the GAA by selecting the water system infrastructure and policy options most cost effective for the satisfaction of anticipated water requirements in the next 75 years, given the anticipated water available, under a range of climate and demographic change scenarios. The model considers the entire anthropogenic water cycle and contains a number of innovations which enable the evaluation of the cost effectiveness of water reuse relative to all other water supply options. Chapter 6 presents the results of the RO model developed for the GAA. It concentrates on an evaluation of the recommendations of the model results from the Base Case storyline (though other storylines are presented), and explores how those recommendations would change when considering aversion to two specific types of risk: the risk of water shortage (risk of low feasibility robustness); and the risk of cost overruns (risk of low solution robustness).