This project evaluated the benefits and utility of applying transport optimization codes to groundwater pump-and-treat systems in optimizing extraction/injection rates and locations. Optimization potentially can result in reduced cleanup time and reduced life-cycle costs. Three Department of Defense pump-and-treat systems were evaluated (two existing systems and one in the design phase). Three mathematical formulations were developed for each site, consisting of an objective function to be minimized and a set of constraints to be satisfied. Two modeling groups applied transport optimization algorithms, and one group applied trial-and-error to serve as a control. The objective of the project was to demonstrate the cost benefit of transport optimization codes in comparison to traditional trial-and-error optimization methods.

Technology Description

Transport optimization codes couple transport models with nonlinear mathematical optimization to allow a more rigorous evaluation of potential pumping strategies (i.e., using mathematical algorithms instead of manual iteration). The transport optimization codes demonstrated use a variety of heuristic global optimization methods that include simulated annealing, genetic algorithms, outer approximation, and tabu search. These global methods often require intensive computational effort but have become more practical for application on personal computers as speeds have increased. The project used two simulation optimization packages: SOMO3, developed at Utah State University, and MGO, developed at the University of Alabama. Specific codes/algorithms applied were at the discretion of the modeling groups.

Demonstration Results

In every case, the groups applying the optimization algorithms found improved solutions relative to the trial-and-error group. The solutions found were 5% to 50% better than those obtained using trail-and-error, with a typical improvement of about 20%. Because multiple sites and multiple formulations for each site were evaluated, there is a high degree of confidence in the conclusion that the application of optimization algorithms provides improved solutions for problems posed in the manner demonstrated (i.e., mathematical formulations with an objective function to be minimized/maximized and a series of constraints). The optimization algorithms were able to evaluate far more alternatives and also suggested alternatives that were not otherwise obvious. At all three sites, the potential cost savings outweighed the expected costs of applying the technology. The estimated cost of applying transport optimization algorithms is approximately $25,000 to $60,000 for up to 2 constituents, simulations up to 2 hours long, up to 3 formulations. The cost for the trial-and-error group was approximately $30,000 per site, although that group reported that it would have performed fewer simulations if not done within the context of this demonstration project. Thus, it is assumed that for comparable projects trial-and-error may cost approximately $20,000 to $25,000. Therefore, the premium for applying the transport optimization may be as little as zero or as much as $40,000.

Implementation Issues

The potential life-cycle cost savings associated with the application of transport optimization algorithms will almost certainly exceed the premium of up to $40,000 at most modest to complex pump-and-treat sites. Obviously, for sites with high costs and/or high durations, such as a yet-to-be constructed pump-and-treat system where fewer cost and design parameters are fixed, the potential life-cycle cost savings becomes more significant. For example, at the Former Blaine Naval Ammunition Depot in Hastings, Nebraska, potential cost savings of approximately $10 million were identified relative to the trial-and-error solutions. The potential savings over the lives of the projects far exceed the additional costs for applying the simulation optimization methods. Computational complexity still poses challenges for transport simulation optimization, and expertise is required in the posing and solving of the problems. (Project Completed - 2004)