A Global Optimization Approach for Parameter Identification in Contaminant Transport ModelingEPA Grant Number: R823136
Title: A Global Optimization Approach for Parameter Identification in Contaminant Transport Modeling
Investigators: Zheng, Chunmiao , Wang, Pu P.
Institution: University of Alabama at Birmingham
EPA Project Officer: Hiscock, Michael
Project Period: October 1, 1995 through September 30, 1997
Project Amount: $133,661
RFA: Exploratory Research - Chemistry and Physics of Water (1995) RFA Text | Recipients Lists
Research Category: Water , Land and Waste Management , Engineering and Environmental Chemistry
This project is intended to explore a new approach for parameter identification in groundwater flow and contaminant transport modeling based on a global optimization technique-simulated annealing (SA) or its variant. Specifically, the project will develop two-phased SA algorithms capable of: (1) incorporating all prior information on the parameters to be estimated using geostatistical concepts, and (2) finding the "best" values of the parameters to match the measured head and concentration data based on inverse modeling analysis. The comprehensive databases from field tracer experiments recently conducted at the Columbus Air Force Base in Mississippi will be used in this research.
The research project represents a first effort to explore the potential of SA as a powerful approach for identification of optimal groundwater flow and transport parameters. It also represents a first attempt to investigate the conceptual and computational issues relevant to parameter identification using massive data sets from three-dimensional field experiments in a highly heterogeneous aquifer. Some of the issues to be investigated include incorporation of prior information, search of global minimum, coupling of flow and transport in inverse analysis, computational efficiency of the global optimization approach, and use of parallel computation.
It is expected that the project will lead to the establishment of a new conceptual framework and SA-based mathematical algorithms for parameter identification in complex, three-dimensional flow and transport systems. Mapping a groundwater parameter identification problem onto the SA framework is not straightforward. It requires a great deal of exploratory study, including the determination of the hierarchical levels, the partition of the variables, an efficient mechanism to generate and update the solutions, and the selection of effective and efficient annealing parameters. The results from this study are likely to have potentially wide-ranging impact on parameter identification in groundwater flow and contaminant transport modeling.