Novel optimization modeling applications for environmental and ...

This dissertation presents three studies that demonstrate the use of optimization modeling techniques for gaining insight into complex environmental and energy systems and providing decision support. The first uses multi-objective linear programming and hydrological modeling to improve decisions made by land use planners when faced with conflicting economic and environmental objectives. The second uses linear programming theory and statistical modeling to approximate the response of large scale linear programming based policy models to new sets of inputs, thus cutting down on the number of large scale model runs required to achieve a desired policy solution. The final study uses stochastic dynamic programming, Monte Carlo simulation, and linear programming, to assess the efficacy of traditional types of commitment and dispatch modeling in power systems with increasing levels of renewable energy integration, and proposes an improved modeling approach. The three studies share the following themes. Each encourages the efficient use of resources to maximize economic benefit while adhering to environmental constraints. Although the first two can be applied to many sectors, each of the three studies has applications in the electricity industry. Together, each study could represent a different aspect of efficient electricity system planning under environmental constraints； from local land use planning of new power installations in the first study, to long term national power system policy planning in the second study, to efficient operation of power systems that include renewable sources of energy in the third. The second chapter presents the land use planning model, and an application to the Chagrin watershed in the Lake Erie basin. The methods in this chapter are used to explore the tradeoffs in land development between economic objectives, represented by proxies such as proximity to existing development, and ecological objectives, represented by the hydrological impact of land use change on the watershed. The third chapter presents the methodology developed for metamodeling of large scale linear programming-based policy models. This study has been applied to the US Environmental Protection Agencys Integrated Planning Model IPM) to predict the response of system costs and emissions to changes in policy instruments and system assumptions. Examples of policy instruments that IPM can be used to investigate include potential future carbon dioxide caps or taxes, and measures to limit other pollutants such as mercury and nitrogen oxides. Examples of system assumptions include the costs and capacity constraints of technology like carbon capture and storage or nuclear energy, and expected fuel costs. The methodology is not limited to IPM and is applicable to approximating the response of any convex program. The fourth chapter develops a stochastic dynamic programming formulation for electricity system unit commitment and dispatch in systems with uncertain supply from renewable energy sources. An application is presented for the Netherlands electricity system in which the marginal value of additional wind power is quantified, accounting for ramp limits and uncertainty in wind forecasts.