Genetic Algorithm-Based Optimization in the Development of Tropospheric Ozone Control Strategies: Least Cost, Multiobjective, Alternative Generation and Chance-Constrained Applications (Air Quality Management)


A typical air quality management problem involves a geographical region with sources that emit pollutants into the atmosphere and receptors that are deleteriously affected by the resulting ambient concentrations or depositions. Strategies to control air pollution often entail modifying source emissions by mandating specific emissions reductions or control technologies, or through implementing economic incentives programs, such as market-based emissions trading programs, emissions charges, and subsidies. Mathematical optimization is a tool that potentially can be used to identify cost-effective control strategies that meet the desired air quality standards. Optimization-derived strategies provide low-cost benchmarks and may be good starting places for further analysis. In addition, various optimization techniques can be used to identify tradeoffs among design objectives, consider uncertainty explicitly when developing strategies, generate very different alternative management strategies, and simulate the results of emissions trading programs. Traditional optimization procedures, such as linear programming, nonlinear programming, and integer programming, have been applied to air quality problems involving nonreactive or linearly-reactive pollutants. Very little emphasis, however, has been directed toward problems involving nonlinearly-reactive pollutants, such as tropospheric ozone. A lack of applications in this area can be attributed to the difficulties in representing nonlinear chemistry in the strict objective and constraint formats required by traditional optimization procedures. The goal of this dissertation is to demonstrate how an alternative optimization technique, genetic algorithms (GAs), can be used to optimize control strategies for problems with nonlinearly-reactive pollutants. GA formulations are presented for least cost optimization, multiobjective optimization, stochastic optimization, and alternative generation. The least cost formulation is shown to outperform a published heuristic in a case study ozone control problem. The multiobjective and alternative generation formulations represent new GA techniques. The stochastic formulation appears to improve upon an existing GA technique by greatly reducing computational requirements for obtaining a solution in a case study problem. While this dissertation focuses on the application of these GA formulations for an air quality management problem, GAs have great potential for addressing many complex environmental problems that are not readily modeled with traditional optimization approaches.