OPTIMAL WATER QUALITY MANAGEMENT STRATEGIES FOR URBAN WATERSHEDS USING MACRO-LEVEL SIMULATION MODELS LINKED WITH EVOLUTIONARY ALGORITHMS

• Main Author: Tufail, Mohammad
• Format: Others
• Published: UKnowledge 2006
• Subjects: Evolutionary Algorithms|Optimization|Macro-level Simulation Models|Water Quality|Watershed management
• Online Access: http://uknowledge.uky.edu/gradschool_diss/313; http://uknowledge.uky.edu/cgi/viewcontent.cgi?article=1316&context=gradschool_diss

Urban watershed management poses a very challenging problem due to the varioussources of pollution and there is a need to develop optimal management models that canfacilitate the process of identifying optimal water quality management strategies. Ascreening level, comprehensive, and integrated computational methodology is developedfor the management of point and non-point sources of pollution in urban watersheds. Themethodology is based on linking macro-level water quality simulation models withefficient nonlinear constrained optimization methods for urban watershed management.The use of macro-level simulation models in lieu of the traditional and complexdeductive simulation models is investigated in the optimal management framework forurban watersheds. Two different types of macro-level simulation models are investigatedfor application to watershed pollution problems namely explicit inductive models andsimplified deductive models. Three different types of inductive modeling techniques areused to develop macro-level simulation models ranging from simple regression methodsto more complex and nonlinear methods such as artificial neural networks and geneticfunctions. A new genetic algorithm (GA) based technique of inductive modelconstruction called Fixed Functional Set Genetic Algorithm (FFSGA) is developed andused in the development of macro-level simulation models. A novel simplified deductivemodel approach is developed for modeling the response of dissolved oxygen in urbanstreams impaired by point and non-point sources of pollution. The utility of this inverseloading model in an optimal management framework for urban watersheds isinvestigated.In the context of the optimization methods, the research investigated the use of parallelmethods of optimization for use in the optimal management formulation. These includedan evolutionary computing method called genetic optimization and a modified version ofthe direct search method of optimization called the Shuffled Box Complex method ofconstrained optimization. The resulting optimal management model obtained by linkingmacro-level simulation models with efficient optimization models is capable ofidentifying optimal management strategies for an urban watershed to satisfy waterquality and economic related objectives. Finally, the optimal management model isapplied to a real world urban watershed to evaluate management strategies for waterquality management leading to the selection of near-optimal strategies.