Dense nonaqueous phase liquid (DNAPL) contamination poses a major threat to the groundwater supply. Delineating and removing the DNAPL source is an essential step that renders remediation successful and lowers the estimated remediation time and cost significantly.

DNAPL sources, however, are unlikely to be identified and their extent adequately defined using field investigative techniques and strategies that focus exclusively on locating separate phase DNAPL. The target DNAPL location is generally too small and filamentous to identify efficiently via borings or geophysical methods, even using state-of-the-art field techniques. On the other hand, the plume emanating from a DNAPL source is typically quite large and identification of its extent and the concentration topology may, depending on the nature of the groundwater flow field, require the collection of considerable field data typically including water quality, soil-gas concentration, and to some degree lithological and permeability information.

This project addressed the issue of identifying and delineating DNAPL by creating an optimal search strategy to obtain information regarding the magnitude and location of a DNAPL at its source. Specifically the project developed, tested, and evaluated a computer-assisted analysis algorithm to identify the location and geometry of a DNAPL source. Prior to a detailed site investigation, the tool can be used to identify where to initially sample the subsurface to determine the DNAPL source characteristics and then to update the sampling strategy for the field as the investigation proceeds.

The technical approach exploited the concept that DNAPL is indicated by the presence of a DNAPL species concentration in excess of a specified value attributable to dissolution as described by formulae based on Raoult's law. The computer-based search strategy developed in this project uses groundwater flow and transport modeling under uncertainty, a linear Kalman filter to combine modeling information and field data, and an optimization algorithm. The algorithm indicates where, and if necessary when, to sample groundwater quality in order to define the location of the DNAPL containing area identified with the pre- specified concentration of the target compound. The search strategy is interactive and provides real-time information on the location and approximate shape of the source and the plume simultaneously.

The three-dimensional stochastic model was developed and successfully tested using various synthetic example problems of increasing complexity. The effectiveness of the search strategy in identifying a DNAPL source was demonstrated at two field sites— Anniston Army Depot in Alabama and Hunters Point Naval Shipyard in California—where the contaminant of interest was trichloroethene.

In combining water quality information with expert knowledge, this tool will assist groundwater professionals in defining the DNAPL source. The search strategy can provide the best estimate of the true source location given existing water quality information. It also has the capability to identify where the best sampling location is if groundwater professionals decide to take new water quality samples. The information of each newly selected sample can be used in real time to update the model and to select the next optimal sampling location until the algorithm converges to a solution. Because sampling is conducted in a cost-effective manner, the overall investigatory costs of finding a DNAPL source are minimized. (Project Completed - 2009)