HR: 09:45h
AN: H11B-08    [PDF]
TI: Consideration of Time as a Decision Variable in Subsurface Remediation Optimization
AU: * Endres, K L
EM: klendres@mtu.edu
AF: Department of Civil and Environmental Engineering Michigan Technological University, 1400 Townsend Dr, Houghton, MI 49931
AU: Mayer, A S
EM: asmayer@mtu.edu
AF: Department of Geological Engineering Michigan Technological University, 1400 Townsend Dr, Houghton, MI 49931
AU: Horn, J
EM: jhorn@nmu.edu
AF: Department of Computer Sciences Northern Michigan University, 1001 Glenn T. Seaborg Science Center, Marquette, MI 49855
AB: Remediation time frames are normally fixed by a number of management and regulatory issues without consideration of the interaction between remediation cost and the time constraint. This work looks at the implications of the time constraint by considering time as a decision variable in the optimization process. We utilize a multi-objective optimization of a hypothetical contaminated aquifer that results in a trade off curve of total remediation time vs. remediation costs. This curve allows decision makers to view the full range of options for time and cost. The cost function includes treatment, pumping and management costs. The multi-objective problem is formulated to minimize the design cost while also minimizing the remediation time. The Niched Pareto Genetic Algorithm (NPGA) has been modified to allow enforcement of water quality constraints. The addition of this constraint enforcement is developed by two methods. The first method initiates a penalty in the fitness values as the enforcement mechanism. The second uses the niching domination to apply the constraint. Each of these methods is innovative in remediation optimization work. Comparisons of the two methods are presented. Three sets of numerical computational experiments are performed to produce tradeoff curves of cost and total time. The experiments increase in computational effort as the complexity of the time variables increases. In each experiment the cost objective will be a function of pumping rate. The first experiment will use a single management period, where total time is the decision variable. The second will use multiple management periods of fixed length, where the number of management periods is the decision variable. The third will have the number of management periods and the length of the periods as decision variables. This method of investigation in to the impact of time as an optimization variable incorporates the full range of management possibilities. Comparisons of the three experiments are presented as tradeoff curves of total remediation time vs. cost.
DE: 1831 Groundwater quality
DE: 1869 Stochastic processes
DE: 1894 Instruments and techniques
SC: Hydrology [H]
MN: 2003 Fall Meeting