Complex Electrical Resistance Tomography of a Subsurface PCE Plume

A controlled experiment was conducted to evaluate the performance of complex electrical resistivity tomography (CERT) for detecting and delineating free product dense non aqueous phase liquid (DNAPL) in the subsurface. One hundred ninety liters of PCE were released at a rate of 2 liters per hour from a point 0.5 m below ground surface. The spill was conducted within a double walled tank where saturated layers of sand, bentonite and a sand/bentonite mixture were installed. Complex electrical resistance measurements were performed from 4 vertical electrode arrays, each with 10 electrodes spaced between 3 m and 0.5 m depth. Data were taken before the release, several times during, and then after the PCE was released. Magnitude and phase were measured at 1 and 64 Hz. Data from before the release were compared with those during the release for the purpose of imaging the changes in conductivity resulting from the plume. Conductivity difference tomographs showed a decrease in electrical conductivity as the DNAPL penetrated the soil. A pancake-shaped anomaly developed on the top of a bentonite layer at 2 m depth. The anomaly grew in magnitude and extent during the release and borehole television surveys data confirmed the anomaly to be free-product PCE whose downward migration was stopped by the low permeability clay. The tomographs clearly delineated the plume as a resistive anomaly. Images showing phase changes caused by the spill are also presented. The phase changes at 64 Hz suggest that the DNAPL spill increased the induced polarization (IP) effect of the clay layers. INTRODUCTION Large volumes of hydrocarbons have contaminated numerous aquifers. The contaminated soil contain both dissolved phase hydrocarbon as well as free product. Current contaminant detection methods require the extensive use of core drilling and chemical analysis of soil samples to map the extent of the subsurface plumes. Relatively recent work by Olhoeft (1992), Borner et al (1993) and others suggest that complex resistance measurements may provide a method to detect organic contamination in soils. The proposed CERT may minimize the need for drilling and for chemical analyses of large numbers of soil samples if it can be shown to be capable of defining regions contaminated with free product DNAPL. We describe here the results of a controlled field experiment where complex electrical resistivity tomography (CERT) was used to detect and locate complex resistivity changes during a PCE spill conducted in a layered soil medium. This work was performed at the Oregon Graduate Institute of Science and Technology (OGI) in Beaverton, Oregon because the facilities there include a unique double-wall tank filled with soil into which a contaminant can be legally released at a scale sufficiently large to see physical phenomena at a realistic scale. The 10 m square and 5 m deep tank is instrumented for geophysical and hydrological studies. Layers of sand, bentonite and sand/bentonite mixtures were constructed within the tank to simulate a layered soil system (refer to figure 1 for details). Soil pore water consisted of a dilute NaCl solution with a resistivity of 40 ohm-m. The DNAPL was dyed so that its location could be determined using borehole television surveys. A more complete description of the OGI facility is given by Johnson et al., 1992. Three boreholes in the tank were each instrumented with 10 electrodes. The reconstruction model assumes a resistivity distribution constant in the direction perpendicular to the plane defined by the electrodes although electrical potential is modeled three dimensionally to permit point electrodes. The soil-air interface is a boundary of zero normal current flow. After reconstruction, the tomographs are spatially smoothed to give the appearance of a continuous resistivity distribution. The mesh elements were approximately square and there were two elements between each electrode along the boreholes. The algorithm used for these inversions is described by LaBrecque et al. (1995). Data were collected using a frequency domain, multichannel resistivity meter. The surveys were performed at five frequencies between 1 and 1024 Hz. Each data set consisted of 660 data points, and half of the data set consisted of the reciprocals for the other half. Data collection time for each data set at a single frequency was about 15 minutes.