The site used in this study is a manufacturing facility located on 65 ha in Marshall County, in Middle Tennessee (fig. 4). The manufacturing facility has been in operation since 1937 and is currently used to assemble air conditioning and heating equipment. A degreaser was installed in 1973 near the south end of the main manufacturing building. The chlorinated solvent TCE was piped underground from an above-ground storage tank to the degreaser. In 1980, the underground pipe connecting the storage tank to the degreaser ruptured and released an estimated 13,000 liters of TCE into a piping trench, allowing TCE to migrate into sanitary sewer pipes and the soil. The highest concentration of TCE (950,000 mg/L) was detected in 1987, when a 61-cm column of dense nonaqueous phase liquid (DNAPL) was measured in a shallow well screened near the top of the bedrock (Wolfe and others, 1997). In 1997, TCE was detected in a shallow well at a concentration (171 mg/L) indicating that DNAPL is still present beneath the manufacturing building.
Approximately 50 ground-water monitoring wells have been completed at the site (fig. 4). Shallow wells are screened near the top of bedrock (as much as 6 m into the bedrock) in a shallow water-bearing zone. Deeper wells range in depth from 18 to 67 m below land surface and are screened in a karst aquifer that consists of several water-bearing zones in the bedrock. Additional well completion data are given in table 2. Remedial activities at the site include construction of two ground-water collection trenches on the top of the bedrock in 1989, excavation and thermal treatment of 1,150 m3 of contaminated soil near the above-ground storage tanks in 1995, installation of several ground-water pump-and-treat wells, and construction of an aeration weir in Snell Branch, a small stream that flows through the study site (fig. 4). Air strippers are used to remove chlorinated solvents from water collected from the recovery trenches and the pump-and-treat wells.
The site is underlain by the Ordovician-age Lebanon, Ridley, and Pierce Limestones (Crawford, 1992; Crawford and Ulmer, 1994; Farmer and Hollyday, 1999) (table 3). The Ridley Limestone is susceptible to dissolution, especially where exposed at land surface. Sinkholes are common, and streams flowing off the Lebanon Limestone lose water and sink into the Ridley Limestone. Caves and cave streams primarily develop at contacts between the aquifer and confining units (Crawford, 1992). The Ridley Limestone is exposed just north of the study site (Wilson and Luther, 1963), and Snell Branch sinks into the formation and flows to a spring approximately 2.7 km northeast of the site (Crawford, 1992). Water yields are low unless a well intersects a well-developed conduit or cave stream in the Ridley Limestone (Crawford and Ulmer, 1994).
The general flow direction of the shallow ground water is north-northwest toward Snell Branch as indicated by water levels in shallow wells (fig. 5). Water levels in the shallow water-bearing zone do not appear to be affected by the pump-and-treat wells. Along the western side of the manufacturing building, shallow ground water flows toward a trough in the bedrock surface (fig. 6). This trough is downgradient of the original TCE spill and appears to be the main route for horizontal transport of chlorinated ethenes in the shallow water-bearing zone (fig. 7).
Well-driller logs and well-pressure test data from unpublished TDEC-DSF files indicate that the karst aquifer at the site consists of at least three distinctive water-bearing zones located along bedding planes in the upper part of the Ridley Limestone (fig. 8). The first water-bearing zone is located at an altitude of approximately 208 m above sea level and was detected during the construction of wells 3D and 12D and the 10D wells. The second water-bearing zone is located at an altitude of approximately 200 m above sea level. The third zone is located just above the thin-bedded member of the Ridley Limestone at an altitude of 190 m above sea level. The second and third zones were intersected during the construction of several of the deep wells. Water-bearing zones were not detected in the lower part of the Ridley or Pierce Limestones.
Well-driller logs for wells 12D, 10D-A, 10D-B, and 10D-C, which are located near Snell Branch, indicate the presence of several highly fractured zones between the altitudes of 190 and 217 m above sea level. Driller logs for the three 10D wells (fig. 4) describe a highly fractured zone at an altitude of 216 m above sea level that is hydraulically connected to Snell Branch as well as several zones of fractures between the altitudes of 190 and 208 m above sea level that are hydraulically connected. During the construction of well 12D, an uncompleted well boring intersected a mud-filled cavity 1.2 m in height at an altitude of 208 m above sea level that was hydraulically connected to Snell Branch. The extensive vertical fracturing beneath Snell Branch, which runs approximately north 40° east through the study site, may be a result of jointing. Geologic investigations described in unpublished Tennessee Division of Superfund site files indicate that the local joint orientation is north 40° east and north 50° west.
Well-completion data indicate that the water-producing zones are not continuous beneath the site, and in some areas fractures appear to be isolated from the major zones of ground-water flow. For example, well 11D intersects the three water-producing zones in the karst aquifer, yet the well produced less than 3 L/min. Other wells intersected water-bearing zones and produced abundant water.
Because of the slight differences in water levels (fig. 9) when the pump-and-treat wells are operating and because the wells are screened in different water-bearing zones, determining the general direction of ground-water flow in the upper part of the Ridley Limestone is difficult. Since well 11D is screened in fractures isolated from the major zones of ground-water flow, changes in the water level of this well lag behind changes in other wells, and the water levels in 11D may be significantly higher or lower than water levels in other deep wells (fig. 9).
Pump-and-treat wells completed in the upper part of the Ridley Limestone (wells 9D, 13D, and 14D) draw down water levels in many of the deep wells and affect local ground-water flow (fig. 9). The pumping of well 9D (approximately 100 L/min) affects water levels in all of the deep wells located south of Snell Branch. The pumping of wells 13D and 14D (approximately 20 L/min at alternating intervals) affects a much smaller area north of Snell Branch (fig. 9). The pump-and-treat operations in the karst aquifer help to remove chlorinated solvent and contain the contaminant on site.
Ground-water tracing studies documented in unpublished TDEC-DSF site files verify that conduits between wells 3D and 9D and between wells 12D and 9D are hydraulically connected. Sodium chloride was transported from well 3D to well 9D in 18 hours (average velocity 8 m/h) and rhodamine WT was transported from well 12D to well 9D in 28 hours (5 m/h). During these ground-water tracing studies, the pumping rate of well 9D was approximately 64 L/min. The reports were unclear on how much water was used to flush the tracers into the conduits. These tracer results indicate that ground water could flow relatively fast in the karst aquifer at this site but do not address the issue of ground-water storage.
Results from additional ground-water tracing studies demonstrate that some water-bearing zones are isolated from major zones of ground-water flow. These water-bearing zones may store water and contaminants for relatively long periods. Fluorescein (2.3 kg) was injected into well 8D on November 15, 1991, and the well was flushed with water (750 L/min) for 26.5 hours. During the fluorescein study, the pump-and-treat wells were turned off and nearby springs were monitored for fluorescein. Fluorescein was detected in some nearby springs; however, the results were inconclusive because of the lack of background samples from the springs. The deep wells at the study site were not sampled for fluorescein until 1996, 5 years after the dye injection. During those 5 years, pump-and-treat wells and other remediation efforts were active at the site. Unpublished TDEC-DSF files report that fluorescein was still present in water samples collected from deep wells during two separate sampling events in 1996. The dye was detected at significant concentrations in water samples from wells 3D (11 to 18 micrograms per liter), 4D (0.54 to 1.2 micrograms per liter), 5D (0.07 to 0.22 micrograms per liter), and 8D (23 to 37 micrograms per liter).
| Next | Title Page | Table of Contents | List of Illustrations | List of Tables | Conversion Factors |