Application of heuristic optimization to groundwater management
Matott, Loren Shawn
MetadataShow full item record
Subsurface flow and contaminant transport models are often used in solving groundwater management problems. Automated optimization involving such models is becoming commonplace, and researchers are increasingly encountering problems for which standard gradient-based search algorithms are inadequate. Such cases have motivated an interest in the use of more robust, but computationally expensive, heuristic algorithms. The research reported in this dissertation advances the state-of-practice of heuristic optimization in groundwater management by applying a variety of heuristic methods to three groundwater management problems: (1) optimization of pump-and-treat containment systems, (2) optimization of multi-layered sorptive barrier systems, and (3) calibration of reactive transport models involving nitrate contamination. These studies were facilitated by the development of a new open source software package for model-independent multi-algorithm optimization, which includes special tools for calibration and model ranking and selection. Overall, the optimization studies make several important research contributions by (1) suggesting methods and guidelines for the effective selection and use of heuristic algorithms, (2) investigating techniques for reducing the computational demand associated with heuristic algorithms, (3) providing general insight into the behavior of the selected problems, (4) utilizing modeling techniques, remediation constraints, and/or parameter representations not previously applied to the selected problems, and (5) introducing a novel optimization software package to the research community.