Enhanced Contact of Cosolvent and DNAPL in Porous Media by

Oct 26, 2002 - Graham, W. D.; Wood, A. L.; Enfield, C. G. Water Resour. Res. 1997, 33 ... (16) Wood, A. L.; Lee, T. R.; Enfield, C. In Proceeding of t...
0 downloads 0 Views 815KB Size
Download PDF
Environ. Sci. Technol. 2002, 36, 5238-5244

Enhanced Contact of Cosolvent and DNAPL in Porous Media by Concurrent Injection of Cosolvent and Air SEUNG-WOO JEONG, A. LYNN WOOD,* AND TONY R. LEE U.S. Environmental Protection Agency, Office of Research and Development, National Risk Management Research Laboratory, Subsurface Protection and Remediation Division, 919 Kerr Research Drive, Ada, Oklahoma 74820

Dense nonaqueous phase liquid (DNAPL) contamination is a major environmental problem. Cosolvent flooding is proposed as a remedial alternative to water flooding. The efficacy of cosolvent flooding is a function of the degree of contact between the injected remedial fluid and the resident DNAPL. Poor contact may result from remedial fluids traveling in preferential flow paths which bypass trapped DNAPL. Thus, the motivation for this study was to use the preferential flow of air in porous media to enhance contact between the injected cosolvent and resident DNAPL. The study evaluated concurrent injection of cosolvent and air to improve the spatial extent of DNAPL removal in porous media. A 70% ethanol/30% water (v/v) cosolvent was injected simultaneously with air into a micromodel containing residual tetrachloroethylene (PCE). Double drainage displacement was observed as a dominant DNAPL removal mechanism in the initial period of the cosolventair flooding (i.e., gas displaced PCE that displaced water). The residual PCE residing in the preferential paths traversed by air was readily displaced. In addition to this initial PCE mobilization, air flowing through the preferential flow paths displaced cosolvent from these paths into other flow paths and facilitated dissolution of PCE.

Introduction The contamination of groundwater by dense nonaqueous phase liquids (DNAPLs) such as trichloroethylene (TCE) and tetrachloroethylene (PCE) has become a national concern. These chemicals have limited aqueous solubilities, which may allow them to persist in the subsurface for many decades. Therefore, the presence of residual DNAPL poses a significant long-term threat to groundwater quality. Because conventional treatment methods such as water flooding have shown little success at removing DNAPLs, cosolvent flooding has been suggested as an alternative remediation strategy. DNAPL contaminants are removed by two processes: enhanced dissolution and mobilization (1). The efficacy of cosolvent flooding is a function of the degree of contact between the injected remedial fluid and the resident DNAPL (2). Poor contact occurs because of nonuniform flow of remedial fluids and nonuniform distribution of trapped DNAPL. Nonuniform fluid displacement can be a result of spatial variability of * Corresponding author phone: (580)436-8552; fax: (580)436-8529; e-mail: [email protected]. 5238

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 23, 2002

hydrodynamic properties of the porous medium or differences between fluid properties of resident and displacing fluids (3). Remedial fluids may predominantly travel through more permeable regions of a contaminated formation, preferentially removing DNAPL in these regions and bypassing much of the DNAPL in other regions (2, 4). Air having low viscosity and density clearly exhibits preferential flow in air sparging systems for volatile organic compound removal (5-7) and tertiary gas flooding for residual oil recovery (8). However, these preferential flow processes may be important for remediation techniques and systems may potentially be designed to take advantage of these processes to improve contact between trapped DNAPL and remedial fluids. For example, concurrent injection of cosolvent and air may facilitate DNAPL and cosolvent displacements during flooding because air flowing through the preferential flow paths may displace cosolvent from these paths into less permeable paths. The goal of this study was to evaluate the impact of co-injected air on the dynamics of cosolvent displacement in order to enhance DNAPL displacement and dissolution. A glass micromodel was used to assess the efficiency and spatial extent of DNAPL removal by immiscible displacement and dissolution. Many studies have used micromodels to quantify changes in the volumes of selected NAPL blobs. However, current image analysis techniques permit the quantification of NAPL saturation within a larger pore network (9-11). The use of a micromodel allows direct visualization and quantification of changing DNAPL volume as the fluids propagate. Both local and average DNAPL saturation can be monitored and quantified using image analysis.

Material and Methods Micromodel. A micromodel is an artificial two-dimensional network model of interconnecting pores and throats. The micromodel in this study was made of etched glass and was fabricated at U.S. EPA’s Robert S. Kerr Environmental Research Center, Ada, Oklahoma. The micromodel consists of a pore network and influent and effluent reservoirs including inlet and outlet ports. The pore network patterns were created using CorelDraw software. The network used in this study consists of mutually perpendicular channels with throats and circle-shaped pores. The void space (pores and throats) through which fluids flow is represented by the black regions in the micromodel shown in Figure 1a, while grains of soil are simulated by the white (glass) portions. An average pore body radius of 0.28 mm was determined from the area of a circle equal to the cross sectional area of a pore body. An average pore throat radius was 0.12 mm. The pore volume (PV) and porosity of the porous medium excluding the influent and effluent reservoirs were 0.95 mL and 0.64, respectively. The intrinsic permeability of the micromodel was 2.43 ( 0.09 × 10-7 cm2 (i.e., similar to well sorted sand, according to “Ranges of intrinsic permeabilities” of ref 12). The effective permeability of the micromodel at an average PCE saturation of 0.32 was 3.58 ( 0.05 × 10-8 cm2. Experimental Setup. The experimental setup is shown in Figure 1b. The apparatus consisted of the micromodel, syringe pumps (ISCO 100DM), a stereoscope (Zeiss SV11), and a video camera (Zeiss ZVS-47E). Three syringe pumps were used to inject fluids through the micromodel. Live images of the micromodel were recorded with a VCR (Panasonic AG-1980), viewed on a SONY Trinitron monitor, digitized using a frame grabber board (TCI SE, Coreco Inc.), and analyzed with image analysis software (Optimas v.6.5, 10.1021/es0157705 CCC: $22.00

 2002 American Chemical Society Published on Web 10/26/2002

FIGURE 1. A micromodel configuration (a) and experimental setup (b).

TABLE 1. Properties of Fluids

PCE water 70%(v/v) ethanol air

densitya (g/cm3)

viscositya,b (cP)

interfacial tension with red dyed-PCEa,c (dyn/cm)

1.627c 1.001 0.895 1.184 × 10-3 f (18)

0.84 (17) 0.913 ( 0.002 2.331 ( 0.003 1.85 × 10-2 f (18)

39.10 ( 0.40d 2.71 ( 0.06d 29.04 ( 0.12e

a Room temperature. b Ubbelohde calibrated viscometer, ASTM D445. c Saturated with 0.5 g Oil-Red-O/Liter PCE. Nouy ring method. f 25 °C.

Media Cybernetics Inc.). The micromodel was mounted on a Semprex motorized motion stage. Teflon tubing was used to make the connections from the syringe pumps to the micromodel. Pressure gradients during flooding were monitored with a pressure transducer (Cole Parmer Model J-7354) and stored in a computer containing data logging software (LabView v.6.0, National Instruments, Inc.). Effluent from the micromodel was collected in a flask filled with charcoal. Experimental Procedure. The micromodel was vertically positioned, flushed by carbon dioxide, and then saturated with water at an injection rate of 0.02 mL/min until breakthrough was reached. If gas bubbles were still trapped in the micromodel, methanol (Sigma Chemical Co.) was injected by syringe into the micromodel to remove them. Ten pore volumes of water were injected at a rate of 0.2 mL/min to displace the methanol. PCE was chosen as the DNAPL for this study. PCE (99.9+%, Sigma Chemical Co.) was dyed with 0.5 g/L of Oil-Red-O (Fisher Scientific). Dyed PCE was injected into the water-saturated micromodel at a rate of 0.02 mL/min until breakthrough at the outlet was observed. The micromodel was horizontally positioned and then flushed with water at a rate of 0.2 mL/min until a stable residual PCE volume was established. The average residual

d

Drop volume method. e du

PCE saturation was 0.34 ( 0.04. The quantification method of PCE saturation is explained in the following paragraph. A 70% ethanol/30% water (v/v) cosolvent was injected at a rate of 0.02 mL/min (q ) 0.96 m/day) in either Cosolvent-Air (CA) flooding or cosolvent (CS) flooding. Air for CA flooding was continuously injected from a syringe pump at a rate of 0.1 mL/min (a superficial velocity of 5.76 m/day). Therefore, alternating flows of cosolvent and air were formed inside the inlet tube. One-sixth of the total flow rate was cosolvent. Selected properties of fluids used in this study are given in Table 1. Flooding experiments were conducted in the horizontal flow pattern. PCE saturation (SPCE) is the ratio of entrapped DNAPL volume to total pore volume. Assuming a uniform pore thickness within the micromodel (measured as 0.3 mm by photomicrograph), SPCE was obtained from the ratio of the total area of red colored PCE to the total void area. A transparent film with regular grids was used to divide the pore network into 100 cells of equal area (see Figure 2). The film was placed on the micromodel and images were taken via the stereoscope, captured with a frame grabber, and analyzed using an image analysis software package. Each cell area was 115.2 mm2. The total PCE blob area in the VOL. 36, NO. 23, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

5239

FIGURE 2. Quantification of PCE saturation using an image analysis technique (A transparent film with regular grids was used to divide the pore network into 100 cells of equal area) and a schematic of PCE blobs in the micromodel; A is observed area of blobs; P is perimeter of blobs. micromodel was obtained by the summation of all PCE blob areas from each cell. Thus, each data point shown in Figure 3 represents the analysis of several thousand PCE blobs in images from all 100 cells that comprised the micromodel pore network. Measurements of Specific Interfacial Area and Count of PCE Blobs. Areas, perimeters, and count of PCE blobs were quantified with the aid of image analysis software (Figure 2). Assuming that PCE blobs have a rectangular cross-section in the micromodel, volume of PCE blobs was obtained by multiplying blob area to height of the blob (equivalent to pore thickness). The contact area between the cosolvent and PCE was determined by 2A + Ph. A is observed area of blobs. P is perimeter of blobs and h is the height of the blob. Specific contact area was calculated as the ratio of total contact area (Σ(2A+ Ph)) to total blob volume (ΣAh).

FIGURE 3. Normalized PCE saturations measured during cosolventair (CA) flooding and cosolvent (CS) flooding (horizontally laid setup): 70%(v/v) ethanol was injected with a cosolvent: air ratio of 1:5; SPCE,i is the initial PCE saturation. (a) Results are shown as a function of total pore volume of fluids injected. (b) Results are shown on the basis of equivalent volumes of injected cosolvent.

Results and Discussion Air Phase Movement and PCE Mobilization during CA Flooding. The observed rates of PCE removal as a result of CA or CS flooding are shown in Figure 3. All replicate data of CA floods and CS floods were plotted in Figure 3. The results are displayed on the basis of total volume of injected phases (cosolvent and air, Figure 3(a)) and injected cosolvent (Figure 3b). Approximately 40 PV of a mixture of cosolvent and air were required for 80% removal of the initial PCE; however, one-sixth (6.7 PV) of this would be cosolvent. Flushing with an equivalent amount of cosolvent (6.7 PV) without air removed about 25% of the PCE. The approximate 27% removal in the initial period of the CA flooding can be attributed to PCE mobilization by immiscible displacement. A sufficient entrance pressure is required for a mixture of injected fluids (cosolvent and air) to penetrate the pore throats of porous media. Thus, the injected fluids accumulated in the injection fluid reservoir until the entrance pressure was reached (i.e., approximately 4.1 kPa/m). Figure 4 shows pressure responses during CA and CS floodings and shows high pressure responses in an initial period of the CA flood. The injected air then flowed through preferential flow paths and displaced PCE and water 5240

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 23, 2002

FIGURE 4. Pressure responses during CA and CS flooding and change in air saturation during CA flooding. to reach breakthrough. Pore-scale observation clearly shows air invasion into a selected cell (Figure 5b). After injection of 0.09 PV of the remedial fluids, only air had reached the selected cell. Cosolvent had not yet reached the selected cell likely because only 0.02 PV of cosolvent had been injected.

FIGURE 5. Images taken from a selected cell of the micromodel during CA flooding: (a) initial image; (b) after 0.09 PV (air 0.07 PV + cosolvent 0.02 PV) flood; (c) after 17 PV (air 14 PV +cosolvent 3 PV) flood;( d) after 82 PV (air 69 PV + cosolvent 13 PV) flood; (e) after 100 PV (air 83 PV + cosolvent 17 PV) flood.

FIGURE 6. Images taken from a selected cell of the micromodel during cosolvent flooding: (a) initial image; (b) after cosolvent 14 PV flood; (c) after cosolvent 17 PV flood. The residual PCE in the preferential flow paths traversed by air was readily displaced. The preferential flow paths are formed due to capillary fingering and irregular distribution of PCE, as already mentioned in the Introduction. Displaced free-phase PCE was observed in the effluent tube attached to the outlet port. As shown in the microscopic image, double drainage (airfPCEfwater) or direct drainage of PCE by air was a dominant displacement mechanism in the initial period of the CA flooding. The rate of free-phase PCE elution from the micromodel decreased with time as a result of depletion of PCE in the preferential flow paths. The PCE mobilization can be attributed to interfacial tension reduction and immiscible displacement by air. To assess PCE removal mechanisms under air flow conditions, two more experiments were conducted; concurrent injection of water and air and air only. Only averages of the flood results are shown in Figure 3. Even though water and air were injected concurrently, the pore network was exposed to an alternating flow of these fluids because of mixing in the delivery ports. Therefore, concurrent injection of water and air was tested to examine the effects of alternating exposure to these immiscible fluids on PCE removal in the absence of significant dissolution. Flooding conditions were the same for both water-air and cosolventair systems. The initial rates of PCE displacement were similar for water-air and cosolvent-air flooding. However, after the

immiscible displacement in the initial period of water-air flooding, there was no significant reduction in PCE saturation. The results suggest that PCE dissolution is an important removal mechanism during CA flooding. To investigate the influence of volatilization on PCE removal the micromodel was sparged with air after flushing with cosolvent. Residual PCE in the micromodel was prepared with the same procedures mentioned in the previous section, then flushed by 2 pore volumes of the cosolvent, and air flooded at the same injection rate used in CA flooding. About 13% of the initial PCE saturation was removed by air flooding. Most of this removal occurred in the initial period of the flooding by the immiscible displacement mechanism. After this initial immiscible displacement, there was no significant reduction in PCE saturation. Thus, these results suggest that removal of PCE by volatilization was negligible in the CA flooding when compared with dissolution and free-phase displacement. Therefore, based on the results of Figures 3-5, the first PCE removal or redistribution mechanism was immiscible displacement by alternating flows of air and cosolvent. Figures 5 and 6 illustrate the dynamics of PCE removal for CA and CS flooding, respectively, from a selected cell which is located in the middle of the porous medium as shown in Figure 2. In CS flooding, local PCE mobilization was observed in the first period of flooding probably due to VOL. 36, NO. 23, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

5241

FIGURE 7. PCE saturation distribution during CA and CS flooding: (a) and (a*): initial PCE saturation distributions; (b) and (b*): after 6.7 PV CS flood and 40 PV CA flood (6.7 PV cosolvent), respectively; (c) and (c*): after 17 PV CS flood and 100 PV CA flood (17 PV cosolvent). the reduced interfacial tension conditions. But, this PCE mobilization resulted in little change in PCE saturation and contributed little to the PCE removal efficiency from the micromodel. After the initial PCE displacement, gas bubble trains which propagated through the preferential flow paths were quickly 5242

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 23, 2002

segmented into smaller air bubbles due to capillary forces under water-wetting conditions (Figure 5c). Although gas bubbles intermittently flowed after gas breakthrough, little PCE displacement was observed and contributed little PCE removal from the micromodel. The trapped air bubbles were moved by two displacement mechanisms: aggregation with

other flowing air bubbles and imbibition of cosolvent. The first mechanism may occur when the pressure of the displacing fluid becomes greater than the capillary pressure, and the small bubbles trapped in the preferential flow paths are aggregated with the flowing air phase. The result was intermittent flow of gas bubbles through preferential flow paths in the pore network. The mobilized gas formed relatively long trains (over several pores) of linked air bubbles. The trapped air bubbles were also mobilized by imbibition of cosolvent. This air phase movement by imbibition occurred only under the concurrent injection of cosolvent and air because air flow is continuously followed by cosolvent flow. However, air bubbles were displaced only short distances (few pore-lengths) by this mechanism. The importance of concurrent injection is shown in the difference in the PCE removal by water+air flood and air flood alone (Figure 3a). The water+air flood displaced more PCE in the initial period of flooding than air injection alone. Air flood produced air flow channels in the porous medium to reach breakthrough after which there was no change in the distribution of these channels. In concurrent flow of water and air, the air phase was segmented into air bubbles by co-injected water and then mobilized through more pores by either imbibition or drainage mechanisms. Air displacement pathways changed during the course of CA flooding. These changes were likely a result of the capillary fingering phenomena and PCE redistribution. Dissolution of PCE by cosolvent will affect gas phase flow patterns. As the PCE blob sizes decreased in either pore bodies or pore throats, the gas phase flowed through the space previously occupied by PCE (Figure 5d,e). The flowing air phase was trapped and then broken or segmented into smaller bubbles. However, the gas phase saturation in the porous medium was relatively constant during the CA flooding, as shown in Figure 4. The results indicate that air flows continuously through the micromodel without accumulation. Enhanced Dissolution of Residual PCE Blobs. The series of images shown in Figure 7 illustrates the two-dimensional PCE removal patterns between CA and CS flooding systems. Figures 7 shows changes in PCE saturation in each area of the micromodel. In CS flooding, there was relatively little change in the residual PCE entrapped in approximately half of the micromodel even after 17 pore volumes of CS flooding. The rate of dissolution of PCE was limited due to poor contact with cosolvent. Although more spatially uniform removal was observed with CA flooding, some preferential removal along preferential paths are shown in Figure 7b*. PCE was displaced by airflow channeling as mentioned earlier, and the PCE near the preferential flow paths were then gradually dissolved by the cosolvent. Replicate experiments for the two flooding techniques showed similar removal patterns. The decrease in PCE saturation during CA flooding may be attributed to both PCE displacement by air and the dissolution of PCE by cosolvent. As mentioned in the previous section, mobilization of free-phase PCE was observed in the initial period of CA and water-air flooding but negligible dissolution occurred with water-air. The observed PCE removal rate after initial free-phase PCE mobilization was greater with CA flooding than with CS flooding (Figure 3). This enhanced removal can be attributed to improved contact between the cosolvent and trapped PCE throughout the porous medium. Air phase continuously flowed through the porous medium either in preferential flow paths or newly developed paths. Since cosolvent does not travel in pores occupied by air, the injected cosolvent is distributed into other flow paths. Therefore, more PCE blobs are contacted by cosolvent during CA flooding, resulting in enhanced dissolution. This study measured specific contact areas of PCE blobs during CA and

FIGURE 8. PCE blob morphology during flooding: (a) specific interfacial areas and surface areas and (b) blob counts. CS flooding and compared them to show the degree of contact. Figure 8a shows specific contact areas and contact areas of PCE blobs during CA and CS floodings. Figure 8b shows PCE blob counts. The specific contact area during CA flooding increases, while that of CS flooding remains relatively constant. The results indicate that PCE was segmented into small blobs in the beginning period of the CA flooding and then the small blobs gradually shrink with additional flooding, causing the specific contact area to increase. Results from contact area and PCE blob count also validate gradual shrinkage of PCE blobs during CA flooding. Contact areas of PCE blobs in CS flooding were always larger than those in CA flooding. The results indicate that there were large PCE ganglia during cosolvent flooding, as shown in pore scale images (Figure 6). In CA flooding, large PCE ganglia are segmented into smaller ganglia by snap-off and breaking mechanisms near the pore necks. The number of PCE blobs increased after one pore volume of cosolvent injected as shown in Figure 8b. However, the contact area and the number of PCE blobs decrease after flushing with a few pore volumes of remedial fluid. Facilitated dissolution of PCE throughout the porous medium occurred during CA flooding. Capillary Number Analysis. Capillary number (NCa) is a dimensionless number representing the ratio of viscous force to capillary force (13). This study defined Capillary number as follows

NCa )

kkrw∇Φ σ cos θ

(1)

where k is the intrinsic permeability of the medium, krw is VOL. 36, NO. 23, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

5243

pattern may be considered for application. For co-air injection to be effective in the field, air must be distributed throughout the volumetric extent of the DNAPL-contaminated zone, and vertical flow patterns may be the most effective means of achieving this objective. Vertical flow patterns may be achieved by using multiple vertical circulation wells, which were recently used to induce veridical flow in a chemical flooding demonstration (16). This study used a 2-dimension-physical model to assess the spatial extent of DNAPL removal by concurrent injection of cosolvent and air. Based on the results of this study, larger scale, 3-dimensional studies are needed to evaluate the utility of this approach for field application.

Acknowledgments FIGURE 9. PCE desaturations and Capillary number varations versus cosolvent pore volumes (PV) injected. All data are obtained from the horizontal flood setup of the micromodel. Three CA flood results and one CS flood result are shown in this figure. the relatively permeability of water phase (cosolvent), Φ is the potential (i.e. Φ ) P + Fgz, where P is the pressure, F is the density, g is the gravitational acceleration constant, and z is the vertical distance), σ is the interfacial tension between two phases, and θ is the contact angle between two phases (160° at PCE phase to 70% ethanol). The relative permeability of water phase was calculated by Corey’s equation. Cosolvent saturation was approximated using the sum of the PCE and gas phase saturation during CA flooding. This study used a horizontal setup and treated the pressure gradient as the potential gradient. Figure 9 shows variations in Capillary number and PCE saturation during flooding. Data points of the replicated experiments shown in Figure 3 are separately shown in Figure 9 (PCE sat. ratio 1, PCE sat. ratio 2, and PCE sat. ratio 3). CA floods exhibit higher NCa than a CS flood, ranging from 1 × 10-5 to 4 × 10-5 for CA floods and 4 × 10-6 to 1 × 10-5 for CS floods. The range of Capillary numbers indicated that the capillary force is still a dominant force during CA flooding. Organic phases or nonwetting fluids are mobilized in porous media as total trapping number approaches 10-5 to 10-4 (14). A complete removal of residual NAPL is achieved at a total trapping number greater than 1 × 10-3. The total trapping number is the sum of the Capillary number and the Bond number (ratio between gravitational and capillary forces). Because this study used a horizontal setup, the Capillary number becomes the total trapping number. The gas saturation during CA flooding was approximately 0.20. The gas bubbles trapped in flow paths reduce the relative permeability of the water or cosolvent and exhibited increased pressure. Because of dominant capillary forces, gas flowed intermittently through specific paths as viscous forces exceeded capillary forces. Results also imply that the dominant PCE removal process during CA flooding is dissolution by cosolvent. Results of analysis in Capillary number are consistent with observations and removal processes discussed in the previous sections. Extension of Flood System. Cosolvent mixtures may exhibit gravity override in a horizontal flushing system because the cosolvent mixtures are less dense than the water being displaced (15). To take advantage of the buoyancy properties of air and cosolvent, a vertically upward flow

5244

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 23, 2002

The U.S. Environmental Protection Agency through its Office of Research and Development funded and managed the research described here. This work was performed while S.-W.J. was a National Research Council Research Associate with the National Risk Management Research Laboratory, Subsurface Remediation and Protection Division. This document has not been subjected to Agency review and therefore does not necessarily reflect the views of the Agency, and no official endorsement should be inferred.

Literature Cited (1) Rao, P. S. C.; Annable, M. D.; Sillan, R. K.; Dai, D.; Hatfield, K.; Graham, W. D.; Wood, A. L.; Enfield, C. G. Water Resour. Res. 1997, 33, 2673-2686. (2) Sillan, R. K.; Annable, M. D.; Rao, P. S. C.; Dai, D.; Hatfield, K.; Graham, W. D.; Wood, A. L.; Enfield, C. G. Water Resour. Res. 1998, 34, 2191-2202. (3) Lenormand, R.; Touboul, E.; Zarcone, C. J. Fluid Mech. 1988, 189, 165-187. (4) Imhoff, P. T.; Gleyzer, S. N.; McBride, J. F.; Vancho, L. A.; Okuda, I.; Miller, C. T. Environ. Sci. Technol. 1995, 29, 1966-1976. (5) Ji, W.; Dahmani, A.; Ahlfeld, D. P.; Lin, J. D.; Hill, E. I. Ground Water Monit. Remed. 1993, 13, 115-126. (6) Clayton, W. S. Ground Water Monit. Remed. 1998, 18, 134-145. (7) Marulanda, C.; Culligan, P. J.; Germaine, J. T. J. Hazard. Mater. 2000, 72, 179-215. (8) Oren, P. E.; Billiotte, J.; Pinczewski, W. V. SPE Form. Eval. 1992, 7, 70-78. (9) Jeong, S.-W.; Corapcioglu, M. Y.; Roosevelt, S. E. Environ. Sci. Technol. 2000, 34, 3456-3461. (10) Zhong, L.; Mayer, A.; Glass, R. J. Water Resour. Res. 2001, 37, 523-537. (11) Detwiler, R. L.; Rajaram, H.; Glass, R. J. Water Resour. Res. 2001, 37, 3115-3131. (12) Fetter, C. W. Applied Hydrogeology, 3rd ed.; Macmillan College Publishing Company: New York, 1994. (13) Larson, R. G.; Davis, H. T.; Scriven, L. E. Chem. Eng. Sci. 1981, 36, 75-85. (14) Pennell, K. D.; Pope, G. A.; Abriola, L. M. Environ. Sci. Technol. 1996, 30, 1328-1335. (15) Jawitz, J. W.; Annable, M. D.; Rao, P. S. C. J. Contam. Hydrol. 1998, 31, 211-230. (16) Wood, A. L.; Lee, T. R.; Enfield, C. In Proceeding of the 3rd International Conference on Remediation of Chlorinated and Recalcitrant Compounds; Battelle Press: OH, in press. (17) Lide, D. R. Properties of organic solvents; CRC Press Database Inc.: Boca Raton, FL, 1996. (18) Munson, B. R.; Young, D. F.; Okiishi, T. H. Fundamentals of fluid mechanics, 2nd ed.; John Wiley & Sons: New York, 1994.

Received for review October 30, 2001. Revised manuscript received August 30, 2002. Accepted September 9, 2002. ES0157705