ALBERT

Notes: An assessment of a conventional soil vacuum extraction (SVE) system to remove perchloroethylene (PCE), a dense nonaqueous phase liquid (DNAPL), was conducted in a dewatered 9 × 9 m block of the surficial sand aquifer located at the Canadian Forces Base (CFB), Borden, Ontario. Approximately two years prior to the installation and operation of the SVE system, 1250 kg of PCE were released under saturated conditions into this experimental cell, forming a heterogeneous DNAPL distribution. Using ground penetrating radar and neutron probe data in conjunction with detailed soil core data, it was determined that approximately 600 kg of PCE were available for removal by the SVE system after dewatering. Performance assessment of this SVE system was based on soil core data and monitoring of in situ gas concentration and pressure, soil temperature, and moisture content.After the first 25 days of operation, the SVE system had removed 230 kg of PCE, during which time the off-gas concentration decreased by 70% from an initial concentration of close to 3500 ppmY After 432 days of operation, 63% (376 kg) of the estimated available mass was removed from the SVE target zone. In situ pressure measurements and sulphur hexafluoride tracer tests indicated that advective air movement occurred throughout the SVE target zone. Changes to the extraction well configuration and the use of short-screened extraction drive points did not increase the mass removal performance.After 250 days of operation, eight soil cores were removed from the test cell, and bulk soil concentration and soil moisture content measurements were performed at 0.05 m increments along the length of each core. These data revealed zones where significant quantities of PCE remained, even though the cumulative mass removal profile for the SVE system was nonzero asymptotic. In general, these zones coincided with higher moisture content zones reflective of the heterogeneous nature of this porous medium.This investigation provides insight into some of the limitations of conventional SVE technology to clean up a sand aquifer contaminated with a DNAPL, given a reasonably good estimate of the before-extraction mass conditions. In particular, the findings clearly show that the heterogeneous nature of the porous medium gives rise to a nonuniform moisture content, a heterogeneous pure phase distribution, and the development of preferential gas phase flow pathways. As a consequence, this will in general limit this technology to remove a considerable fraction of the DNAPL mass present. The results also shed light on the use of soil gas and bulk soil concentration data to provide a representative picture of the in situ mass distribution. And finally, an analysis of the system off-gas data showed that continuous rather than pulsed operation of the extraction system increased mass removal.

Notes: This paper presents an optimization analysis of the remedial pumping design for a contaminated aquifer located in Elmira, Ontario, Canada. The remediation task presented in the paper is to remove two ground-water contaminant species, NDMA (N-nitrosodimethylamine) and chlorobenzene, to such extent that the specified ground-water quality standards are met. The contaminants, NDMA and chlorobenzene, have different initial plume configurations and retardation characteristics. The required quality standard for NDMA is five orders of magnitude smaller than the initial peak concentration. The objective is to minimize total pumping, and the constraints incorporate ground-water quality requirements on the maximum and the spatially averaged residual concentrations, with contaminant source control being considered. On the combination of simulation and optimization, the results of this study indicate that the performance of an optimization algorithm based on gradient search is controlled by the specified cleanup levels, and that contaminant concentrations can be nonconvex and nonsmooth for some pumping schemes.