HR: 09:15h
AN: H11B-06    [PDF]
TI: Calibration of Biokinetic and Biological Parameters for a Groundwater Bioremediation Model using Heuristics and Function Approximation Optimization
AU: * Mugunthan, P
EM: pm78@cornell.edu
AF: School of Civil and Environmental Engineering, Cornell University, 220 Hollister Hall, Ithaca, NY 14853 United States
AU: Shoemaker, C A
EM: cas12@cornell.edu
AF: School of Civil and Environmental Engineering, Cornell University, 220 Hollister Hall, Ithaca, NY 14853 United States
AU: Regis, R G
EM: rgr6@cornell.edu
AF: School of Operations Research and Industrial Engineering, Cornell University, 206 Rhodes Hall, Ithaca, NY 14853 United States
AB: Heuristics and function approximation optimization methods were applied in calibrating biological and biokinetic parameters for a computationally expensive groundwater bioremediation model for engineered reductive dechlorination of chlorinated ethenes. Multi-species groundwater bioremediation models that use monod type kinetics are often not amenable to traditional derivative based optimization due to stiff biokinetic equations. The performance of three heuristic methods, Stochastic Greedy Search (GS), Real Genetic Algorithm (RGA), Derandomized Evolution Strategy (DES), and, Function Approximation Optimization based on Radial Basis Function (FA-RBF) were compared on three-dimensional hypothetical and field problems. GS was implemented so as to perform a more global search. Optimization results on hypothetical problem indicated that FA-RBF performed statistically significantly better than heuristic based evolutionary algorithms at a 10% significance level. Further, this particular implementation of GS performed well and proved superior to RGA. These heuristic methods and FA-RBF, with the exception of RGA, were applied to calibrate biological and biokinetic parameters using treatability test data for enhanced bioremediation at a Naval Air Station in Alameda Point, CA. All three methods performed well and identified similar solutions. The approximate simulation times for the hypothetical and real problems were 7 min and 2.5 hours respectively. Calibration of such computationally expensive models by heuristic and function approximation methods appears promising.
DE: 1829 Groundwater hydrology
DE: 1831 Groundwater quality
DE: 1832 Groundwater transport
DE: 3230 Numerical solutions
DE: 3260 Inverse theory
SC: Hydrology [H]
MN: 2003 Fall Meeting