Environ. Sci. Technol. 2005, 39, 8963-8970
Field Evaluation of In Situ Source Reduction of Trichloroethylene in Groundwater Using Bioenhanced In-Well Vapor Stripping M A R K N . G O L T Z , * † R A H U L K . G A N D H I , ‡,§ STEVEN M. GORELICK,‡ GARY D. HOPKINS,‡ L A U R E N C E H . S M I T H , ‡,| B R I A N H . T I M M I N S , ‡,# A N D PERRY L. MCCARTY‡ Systems and Engineering Management, Air Force Institute of Technology, 2950 Hobson Way, Building 641, Wright-Patterson Air Force Base, Ohio 45433-7765, and Stanford University, Stanford, California 94305
Two technologies in combination, cometabolic bioremediation and in-well vapor stripping, were applied to reduce trichloroethylene (TCE) concentrations in groundwater at a contaminant source area without the need to pump contaminated groundwater to the surface for treatment. The vapor-stripping well reduced source TCE concentrations (as high as 6-9 mg/L) by over 95%. Effluent from the well then flowed to two bioremediation wells, where additional reductions of approximately 60% were achieved. TCE removal was extensively monitored (for research and not regulatory purposes) using an automated system that collected samples about every 45 min at 55 locations over an area of approximately 50 × 60 m2. During 4.5 months of system operation, total TCE mass removal was 8.1 kg, 7.1 kg of which resulted from in-well vapor stripping and 1.0 kg from biotreatment. The system reduced the average TCE concentration of about 3000 µg/L in the source-zone groundwater to about 250 µg/L in water leaving the treatment zone, effecting greater than 92% TCE removal. A 6 month rebound study after system operation ceased found TCE concentrations then increased significantly in the treatment zone due to diffusion from the fractured rock below and perhaps other processes, with mass increases of about 1.5 kg in the lower aquifer and 0.3 kg in the upper aquifer.
Introduction Remediation of dense nonaqueous phase liquids (DNAPLs) that serve as sources of dissolved groundwater contaminants is one of the most difficult challenges faced in managing the problem of subsurface contamination (1). DNAPLs are separate phase organic compounds that, because they are denser than water, may be found as a separate phase below * Corresponding author phone: (937)255-3636 x4638; fax: (937)656-4699; e-mail:
[email protected]. † Air Force Institute of Technology. ‡ Stanford University. § Present address: Mars & Co., San Francisco, CA 94111 | Present address: Sanitation Districts of Los Angeles County, Whittier, CA 90607 # Present address: Enzyme Technologies, Inc., Portland, OR 97230 10.1021/es050628f CCC: $30.25 Published on Web 10/14/2005
2005 American Chemical Society
the water table. Trichloroethylene (TCE) and tetrachloroethylene (PCE) are the most commonly encountered DNAPLs in the subsurface (1). These compounds slowly dissolve into groundwater flowing past the DNAPL source, resulting in dissolved contaminant plumes that can be extensive in both time and space, with near-source dissolved concentrations that exceed drinking water quality standards by several orders of magnitude (1). In this study, an in situ source reduction technology was evaluated. The technology, termed bioenhanced in-well vapor stripping (BEHIVS), is based upon a combination of two previously demonstrated in situ technologies, in-well vapor stripping (IWVS) and in situ aerobic cometabolic bioremediation. IWVS is a method both for removing volatile organic compounds (VOCs) such as PCE or TCE from groundwater and for recirculating groundwater. Air is injected into a well, causing water to rise, and at the same time, VOCs are volatilized. The VOC-rich vapor is removed and treated using granular activated carbon. The rising water, then depleted in VOCs, is returned to the water table through an infiltration gallery. Pumped water is never brought to the surface. Both laboratory (2, 3) and field studies (4) have demonstrated that dissolved concentrations can be reduced by 90-93% within the cleanup zone. In situ bioremediation by aerobic cometabolism of TCE has also been demonstrated (5). In this application, toluene was injected and aerobically biodegraded in situ, resulting in the production of monoxygenase, which fortuitously degraded TCE. The mixing of the contaminant and nutrients and their delivery to indigenous microorganisms are perhaps the main challenges in successfully implementing in situ bioremediation (6). The well-to-well “conveyor-belt” circulation system used in the field study overcame these delivery and mixing challenges (5). The recirculation system, which consisted of a pair of dual-screened biotreatment wells, mixed oxygen and toluene with TCE and reduced TCE concentrations from over 1 mg/L in groundwater upgradient of the treatment system to approximately 20 µg/L downgradient of the system (5, 7, 8). It should be noted that there was an aquitard separating an upper and lower aquifer at the field demonstration site (5). The screens of the biotreatment wells were installed in the upper and lower aquifers, so that the aquitard inhibited the potential for short-circuiting of flow between the screens of a single biotreatment well. One concern with the recirculation system is that it may only be appropriate for use at sites where such specific geological conditions exist, although modeling studies have shown that recirculation between dual-screened treatment wells can be achieved so long as the ratio between horizontal and vertical hydraulic conductivity is on the order of 20:1 or greater, which is not unusual (9). In addition to questions about whether recirculation will be achieved, an important limitation of aerobic cometabolism itself is that it is restricted to relatively low (on the order of low mg/L) contaminant concentrations (10). BEHIVS is a combination of the two technologies described above. Here, the IWVS well pumps water in an upward direction, discharging it to the upper aquifer region, where it is drawn to the biotreatment wells, which in turn pump the groundwater downward to the lower levels of the aquifer (Figure 1). This causes recirculation between the upflow IWVS well and downflow biotreatment wells. IWVS is expected to reduce high TCE concentrations, such as those that would be encountered in a source area, to relatively low levels (in the high µg/L or low mg/L range) that can subsequently be VOL. 39, NO. 22, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
8963
FIGURE 1. Schematic cross-sectional view of the BEHIVS treatment system. treated by aerobic cometabolism to levels in the low µg/L. The objective of the field experiment reported here was to evaluate the extent to which concentrations within a TCE source area could be lowered by BEHIVS operation. It was not our intent in this research demonstration to install a system that resulted in groundwater quality that met regulatory potable water standards but rather to investigate the magnitude of in situ source reduction that could be achieved with just three source-area recirculation wells. By quantification of the source reduction achieved from three treatment wells, a full-scale remediation system could be scaled, designed, and implemented. Other objectives were to investigate the extent to which contaminant concentrations in the source area “rebounded” upon termination of system operation as well as to ascertain whether recirculation between the upflow IWVS well and downflow biotreatment wells could be achieved, even in the absence of a specific feature like the aquitard that had been present in the earlier implementation of a similar recirculating well system involving bioremediation alone (5).
Experimental Section Site Description. The field evaluation of the BEHIVS technology was performed at the source area of a plume of TCEcontaminated groundwater at site 19 of Edwards Air Force Base. The stratigraphy at site 19 consists of unconsolidated sediments overlying granitic bedrock. The sediments consist primarily of alluvial deposits with minor amounts of lacustrine silts and clays. The thickness of the alluvium ranges from approximately 12 m near the western portion of the site where the plume source area is, increasing toward the east to over 45 m. A zone of weathered bedrock approximately 4.5-6 m thick, grading into fractured competent rock, was encountered beneath the alluvium (11). Note that unlike the earlier study of in situ aerobic cometabolic bioremediation of TCE that was also conducted at site 19 in the downgradient plume (5), the plume source area in the current study had no aquitard present within the alluvium. Groundwater at site 19 occurs in the alluvium and fractured bedrock. Groundwater elevation contours show an east-southeast flow direction following an average hydraulic gradient of 0.0047, which gradually decreases downgradient. Two individual pumping tests were conducted at the site 19 source area to evaluate the characteristics of the shallow alluvial portion of the aquifer and the deeper fractured rock portion of the aquifer. Hydraulic conductivities estimated for the alluvial and fractured rock portions of the aquifer were 3.4 × 10-3 and 1.3 × 10-3 cm/s, respectively (11). Through the use of the hydraulic conductivity estimated for the alluvium, an average gradient of 0.0047, and an estimated effective porosity of 25%, the linear groundwater velocity at the site is approximately 6 cm/day. 8964
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 22, 2005
Installation of BEHIVS and Monitoring System. The BEHIVS system consisted of an upgradient IWVS well (VS1) and two biotreatment wells (BIO1 and BIO2) that were separated by a distance of 15 m and located about 8 m downgradient of the IWVS well. System operation was monitored using an automated system (automated sampling and analysis platform (ASAP)) that collected and analyzed samples about every 45 min around the clock at 55 locations covering an area of approximately 50 × 60 m2 (Figure 2). Each of the four components of the BEHIVS and monitoring system (IWVS well, biotreatment wells, ASAP, and monitoring well network) are described in detail below. In-Well Vapor Stripping Well Design. The 25 cm diameter IWVS well (VS1 in Figure 2) was installed in a 41 cm borehole produced by the air rotary casing hammer (ARCH) method. The IWVS well upper screen was 3.3-7.9 m below ground level (BGL) while the lower screen was 13-28.2 m BGL. A bentonite/concrete seal was placed between 9.1 and 11.9 m BGL to prevent short-circuiting of flow through the sand pack from the upper to lower screens. In addition to the IWVS well, three 5 cm monitoring wells were also installed within the borehole. These monitoring wells were screened at 7.2-7.8 m BGL, 13.0-13.6 m BGL, and 27.6-28.2 m BGL. An infiltration gallery consisting of three arms was installed as part of the IWVS system. The purpose of the gallery was to allow infiltration of water into the aquifer from the upper IWVS well screen. The gallery was constructed by trenching three 0.5 m wide by 4.3 m long by 4.6 m deep trenches starting 0.6 m from the center of the IWVS well at 120° angles. The trenches were filled with #3 Lonestar sand and then covered with plastic sheeting to prevent infiltration of water from the surface. Air was supplied through a 2.54 cm stainless steel pipe extending 8.5 m below the static water table, with the bottom 0.3 m of the pipe slotted to allow air introduction. Air was supplied to the IWVS well from a 10 horsepower (HP), threephase, blower (Koch). Carbon dioxide from a gas cylinder was added to the air entering the vapor stripper to maintain pH in the water exiting the vapor stripper at about 7.0 to avoid CaCO3 precipitation. Effluent air was treated via two granular activated carbon tanks in series with the effluent recycled back to the IWVS well. The monitoring system for the IWVS well consisted of two sampling pumps (one for the lower screen (VS1-L), one for the upper screen (VS1-U)), a Signet paddle wheel flow sensor for measuring water flow through the well, and two pressure transducers (in situ) for monitoring hydraulic heads in the lower and the upper screened sections. Biotreatment Well Design. Two biotreatment wells were installed each at a distance of 11 m from the vapor-stripping well. The 15 cm diameter biotreatment wells (BIO1 and BIO2 in Figure 2) were installed in 36 cm boreholes produced by
FIGURE 2. Plan view showing locations of the BEHIVS IWVS well (VS1), the two biotreatment wells (BIO1 and BIO2), and monitoring wells, with the active treatment zone delineated. an ARCH rig. The lower well screens were from 15.2-18.3 m BGL and the upper screens from 9.1-12.2 m BGL. A bentonite seal was installed between 12.8 and 14.3 m BGL to prevent short-circuiting of flow through the sand pack from the lower to the upper screens. Both biotreatment wells were operated in the “down flow” mode, pulling groundwater in from the upper screens and adding toluene, oxygen gas, hydrogen peroxide, and bromide (for tracer tests), before injecting the water into the aquifer through the lower screens. To create the flow, a 10 cm Grundfos environmental pump with a 0.5 HP Franklin threephase motor, located inside the well at the upper screen, was used. The pump motor speed was controlled by a variable frequency drive (ABB) connected to a Signet flow controller. Attached to the outlet of the pump was a block for chemical augmentation, followed by a static mixer. After the mixer, piping directed water through an inflatable packer to the lower screen, where it was injected into the aquifer. The monitoring system for each biotreatment well consisted of two sampling pumps (one for the lower screen, one for the upper screen), a Signet paddle wheel flow sensor for measuring water flow through the well, and two pressure transducers (in situ) for monitoring hydraulic heads in the lower and upper screened sections. Monitoring Well Network. We delineate an approximately 32 × 42 m2 active treatment zone consisting of nested wells closest to the IWVS and biotreatment wells (Figure 2). Overall, the active treatment zone consists of 31 monitoring locations in the upper and lower aquifer zones (nested wells TZ1 through TZ11, BIO1, BIO2, and VS1). Additional nested wells were installed upgradient (UG1-UG4) and downgradient (DG1-DG8) of the active treatment zone. The nested wells were composed of two 5 cm diameter wells installed in 25 cm diameter boreholes produced by an ARCH rig. In each of the boreholes, one well monitored the upper zone (8.810.4 m BGL) and the other monitored the lower zone. The location of the screen interval for the lower zone varied
somewhat for each monitoring well, but each covered a 1.5 m interval located between 15.2 and 18.3 m BGL. Each monitoring well was instrumented with a Grundfos RediFlo-2 pump set at 13.7 m from the top of the well casing for the lower zone well and approximately 9.8 m from the top of the well casing for the upper zone wells. The boreholes were backfilled with #3 Lonestar sand around the screen sections, with transition sands to interface with bentonite/concrete seals placed between the upper and the lower wells. In addition to the newly installed wells, there were three existing dual nested wells (TZ4, TZ8, and DG1), which were outfitted with down-well Grundfos RediFlo-2 pumps. When the four sample locations at the two biotreatment wells and the five sample locations at the IWVS well are included, a total of 55 sample locations were used to monitor the field site. On-Line Analytical System. The BEHIVS system performance was monitored using an automated sampling and analysis platform (ASAP), which collected and analyzed samples every 45 min around the clock. The ASAP was composed of (1) a sampling manifold to connect to each of the 55 locations via 1 cm stainless steel tubing, (2) a Grundfos interface manifold to select and control each of the down-well Grundfos pumps, (3) a liquid interface manifold connected to a trapping manifold for the preparation of volatile compounds for gas chromatography (GC) analysis, and (4) a high-performance liquid chromatography (HPLC) manifold for the preparation for direct reading of anion chromatograph samples. Additional components of the analytical system include a GC with tandem flame ionization and photoionization detectors for chlorinated hydrocarbon analysis, an HPLC pump and electroconductivity detector for ion chromatography, integrators for conversion of detector outputs to compound concentrations, and dissolved oxygen and pH probes. A central PC completed the analytical system, VOL. 39, NO. 22, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
8965
TABLE 1. Operational Schedule for the BEHIVS System at Site 19, Edwards Air Force Base vapor stripper (IWVS) day
description
air flow (L/min)
-320 -29 0 8 34 43 44 51 51 55 66 74 86 96 116 125 140 140 145 147 168 311
monitoring begins testing of IWVS BT pumping begins begin IWVS operation IWVS off, O2 inject begins toluene pulse addition IWVS on toluene pulse addition IWVS off BEHIVS stopped toluene continuous addition begins IWVS on BT flow doubled toluene increased IWVS off IWVS on BIO1 toluene off BIO2 remains on IWVS off BIO1 toluene on evaluation ends rebound study ends
0 0 0 2200 0 0 2200 2200 0 0 0 2200 2000 2000 0 1800 1800 1800 0 0 0 0
connecting to each of the other instruments to allow storage of analytical results in a database, real-time graphing of results, and remote access for control and database transfer. The ASAP system also provided automated calibrations and quality assurance/quality control (QA/QC) analysis of known standards (5). Schedule of Operation. BEHIVS operation began August 13 (day zero) and continued for about 5 months (145 days) (Table 1). However, the groundwater monitoring system was placed into operation 320 days before day zero, so an extensive database of groundwater quality was obtained prior to starting the BEHIVS system. On the basis of prior evaluations of IWVS and in situ cometabolic bioremediation at the site (5, 12), initial operation of the vapor-stripping well began at groundwater flow rates of about 30 L/min with the biotreatment wells pumping at 7.6 L/min, but without toluene injection. The objective was to have the IWVS system reduce TCE concentrations such that at the biotreatment well concentrations were in a range below the inhibitory level of 1 mg/L. This was achieved by day 34. From days 43 through 168, except for short periods, toluene, oxygen, and hydrogen peroxide were added at the biotreatment wells into the lower aquifer to create bioactive zones of TCE degradation. In addition to its role as a source of oxygen, the hydrogen peroxide was added to serve as a biocide to reduce microbial growth on the injection screens (5). On day 86, the biotreatment wells each pumped at an increased rate of 15 L/min, toluene was increased gradually to about 12 mg/L, and hydrogen peroxide addition began. Overall, approximately 50 kg of toluene and 150 kg of hydrogen peroxide were added to both biotreatment wells over the course of the experiment. The operation of the IWVS well was stopped on day 145. To conduct a bromide tracer study at BIO2, as discussed below, the biotreatment wells were operated without the IWVS well from days 146 to 168. Monitoring well sampling continued through day 311 to observe the extent to which contaminant concentrations in the source area “rebounded” upon termination of system operation. Bromide Tracer Studies. During the evaluation of system operation, bromide tracer tests were conducted to help define the recirculatory flow pathways as well as to allow comparison of reactive solute transport with transport of a conservative 8966
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 22, 2005
water flow (L/min) 0 0 0 30 0 0 30 30 0 0 0 30 27 27 0 23 22 20 0 0 0 0
biotreatment (BT) biowell nos.
water flow (L/min)
toluene added (mg/L)
1 and 2 1 and 2 1 and 2 1 and 2 1 and 2 1 and 2 1 and 2 1 and 2 1 and 2 1 and 2 1 and 2 1 and 2 1 and 2 1 and 2 1 2 2 1 and 2
0 0 7.6 7.6 7.6 7.6 7.6 7.6 7.6 0 7.6 7.6 15 15 15 15 15 15 15 15 0 0
0 0 0 0 0 10 0 3.8 3.8 0 5.7 5.7 6.4 12 12 12 0 12 12 10 0 0
tracer. Bromide tracer, at steady concentrations of approximately 230 mg/L, was injected at the IWVS well from day 87 through day 92, the BIO1 well from days 126 though 132, and the BIO2 well from days 151 to 160. Bromide concentrations throughout the site were subsequently monitored.
Results and Discussion Operating Conditions. The ASAP analytical system began operating 10 months prior to day 0 to establish preoperational conditions. The initial TCE concentrations in the shallow and deep zones were significantly different (Table 2). The average (initial) concentration in the upper aquifer zone based upon 985 samples was about 1300 µg/L, while in the lower aquifer zone the average concentration was about 5000 µg/L, based upon 970 samples. The range of average concentrations in the upper aquifer was from 460-2930 µg/ L, while in the lower aquifer the range was between 2480 and 8270 µg/L. During the 10 month baseline period of sample collection and analysis, concentrations at specific locations did not change significantly. This is reflected by the relatively low coefficient of variation for given sampling locations, averaging 0.22 in the upper aquifer (standard deviation ≈ 300 µg/L) and 0.17 in the lower aquifer (standard deviation ≈ 800 µg/L). The coefficient of variation reflects not only true variations in water quality with time at a given location but also measurement errors. The preoperational dissolved oxygen concentrations throughout the aquifer generally were quite low, especially in the lower aquifer (
