Document Type


Journal/Book Title/Conference

American Geophysical Union Fall Meeting

Publication Date



Contamination sites associated with light non-aqueous phase liquids {LNAPL) are numerous and represent difficult cleanup problems. Remediation methods for cleanup of LNAPL fluids in subsurface systems are continuously evolving with the development of various technologies for pump.-and~treat, soil venting, and in-situ bioremediation. Evaluating the effectiveness of remediation techniques as well as attempting to improve their efficiency has been a focus of many researchers, These efforts have included the development of computer simulation models to predict and analyze the fluid movement, entrapment, and mobilization of three~phase systems in porous media. The capability of computer models that not only simulate but optimize remediation processes are in great need. Simulation/optimization (S/0) models allow engineers to optimize processes and to make cost-effective design and management decisions when evaluating remediation strategies. An innovative approach is presented to optimize pumping to immobilize and recover the free product of a floating contaminant consisting of LNAPL's. This determines the best pumping strategy to capture and remove the free oil product not left behind as residual oil. The approach combines detailed simulation, statistical analysis, and operations research optimization. This modeling technique derives regression equations describing system response to unit pumping stimuli. The statistical approach develops regression equations that represent free oil area within the system versus pumping. These equations are derived for multiple time steps and used in the S/0 model to determine the optimized pumping rates required to minimize free oil area and entrapped oil while maximizing free oil recovery and immobilizing the plume via a pump and oil recovery system. When compared with standard simulation techniques, this approach can save substantial computational time and improve remediation strategy design. Utilized is a modified version of US/REMAX, a linear or nonlinear simulation/optimization (S/0) model. US/REMAX can be used to analyze and optimize a variety of groundwater management problems. The modified version incorporates system responses generated externally using ARMOS plus regression analysis. ARMOS is a Z~D areal multiphase flow model. It has the capability of simulating spill, leakage, redistribution, or recovery of LNAPL materials. Application to a representative field problem illustrates the S/O model utility for problem analysis and remediation design. Potential applications of such an S/0 model are numerous providing optimized strategies for recovery of LNAPL spills.