Factors Affecting Bank Formation during Surfactant-Enhanced ...

Environ. Sci. Technol. 1999, 33, 2440-2446

Factors Affecting Bank Formation during Surfactant-Enhanced Mobilization of Residual NAPL CLINTON S. WILLSON Department of Civil and Environmental Engineering, 3507 CEBA, Louisiana State University, Baton Rouge, Louisiana 70803 JOY L. HALL

†

AND CASS T. MILLER*

Department of Environmental Sciences and Engineering, 104 Rosenau Hall, CB #7400, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina 27599-7400 PAUL T. IMHOFF Department of Civil and Environmental Engineering, 137 DuPont Hall, University of Delaware, Newark, Delaware 19716-3120

Two-dimensional flow cell experiments were used to investigate the flow dynamics and factors affecting tetrachloroethylene (PCE) mobilization and bank formation in an otherwise water-saturated porous medium. Aqueous phase injection rates and flow cell angles were varied to control both buoyancy and viscous forces, and both macroscopic- and pore-scale images were captured and analyzed to determine the effects of these forces on PCE transport characteristics. Results were interpreted in terms of a nondimensional bank number, NBa, which relates the forces on the trapped nonaqueous phase liquid (NAPL) ganglia parallel to the flow direction to those forces perpendicular to the flow. NBa, was found to predict bank formation well except for NBa ≈ 1, where other characteristics may have been important, such as droplet coalescence. Pore-scale observations showed that the mobilized PCE moved through the porous medium as noncoalesced droplets and that some of the trapped NAPL was mobilized through a dissolution/mobilization process.

Introduction In recent years, laboratory and field studies have demonstrated the effectiveness of using surfactants to remove residual nonaqueous phase liquid (NAPL) from unconsolidated porous media (1-5). Surfactants can enhance the remediation of NAPL-contaminated sites by increasing the aqueous phase solubility of the NAPL via micellar emulsification and/or microemulsification or mobilizing separate phase NAPL (4, 6, 7). The effect of viscous forces on oil mobilization (8, 9) and the formation and stability of oil banks during water flooding (10, 11) was examined in early research on enhanced oil recovery. Recently, the combined effect of buoyancy and viscous forces on NAPL mobilization was studied to deter-

mine the conditions necessary for NAPL mobilization during surfactant flushing (12). To our knowledge, however, no work has focused on the formation and stability of NAPL banks under conditions where buoyancy forces are important. The dynamics of the process are important because of the desire to remove free product from the subsurface efficiently and safely. The objectives of this work were to investigate the effect of buoyancy and viscous forces on the formation and stability of NAPL banks during surfactant flushing.

Background Surfactant-enhanced remediation of contaminanted subsurface systems has received considerable attention over the last few years because of the significant potential for such flushing solutions to dramatically increase the rate at which residual NAPL contaminants are removed from a system (7, 5, 12). Surfactants are used to decrease the required flushing volume primarily by solubilizing and/or mobilizing trapped NAPLs. Solubilization of contaminant molecules can occur in surfactant micelles or single-phase microemulsion systems (13). NAPL is mobilized as discrete droplets when surfactants lower the nonaqueous-aqueous phase interfacial tension (8). A third transport mechanism that is important for some surfactant/NAPL systems is the formation and transport of macroemulsionsssmall NAPL droplets (diameter > 1 µm) suspended in the aqueous phase that do not coalesce quickly and may be transported through the porous medium (4). The dominant NAPL transport mechanism in any surfactant remediation system depends upon the phase behavior of the system and the balance of capillary, viscous, and buoyancy forces. Several system parameters affect the phase behavior of water/NAPL/surfactant systems: fractional composition of these three components, ionic strength, temperature, and concentrations of any additional alcohols or solvents added (15). Complex phase behavior typically results from these dependencies, which can lead to more than one operative transport mechanism. Surfactant mobilization of residual NAPL has sometimes relied upon the formation of a Winsor type III middle phase microemulsion (1, 14, 4, 12). Within a type III system, the interfacial tensions between the middle phase and the aqueous and nonaqueous phases reach a minimum. NAPL is solubilized in the middle phase and mobilized as a nonaqueous phase due to the low interfacial tensions. While NAPL mobilization can be efficient in such systems, conditions necessary to form this middle phase can be difficult to determine, set, and maintainsparticularly in natural settings. However, recent laboratory and field work have shown that significant NAPL mobilization can occur through a reduction in interfacial tension without the formation of a Winsor type III system (15, 16). The balance of forces on a trapped NAPL ganglion determines the conditions under which mobilization occurs. Early enhanced oil recovery research into the mobilization of residual NAPL focused primarily on the role of viscous forces (8, 9). Recently, the combined effect of buoyancy and viscous forces was considered, and a dimensionless trapping number, NT, was developed (12)

NT) xN2Ca + 2NCaNB sin R + N2B

(1)

with * Corresponding author phone: (919)966-2643; fax: (919)966-7911; e-mail: [email protected]. † Current address: Black and Veatch, P.O. Box 33396, Raleigh, NC 27636. 2440

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 33, NO. 14, 1999

NCa )

q aµ a σna cos θ

10.1021/es980427u CCC: $18.00

(2)

 1999 American Chemical Society Published on Web 06/11/1999

and

NB )

∆Fkkra σna cos θ

(3)

where R is the angle the flow makes measured in the clockwise (CW) direction from the horizontal axis that is directed 90° counterclockwise from the direction of gravity; NCa is the capillary number, which relates viscous to capillary forces; qa is the Darcy velocity of the aqueous phase; µa is the viscosity of the aqueous phase; σna is the interfacial tension between the nonaqueous and the aqueous phase; θ is the contact angle between the NAPL and the porous media; NB is the bond number, which relates buoyancy to capillary forces; ∆F is the density difference between the nonaqueous and the aqueous phase; k is the intrinsic permeability of the porous medium; and kra is the relative permeability to the aqueous phase. Note that the definition for the angle R in the expression for NT is valid for a NAPL denser than the aqueous phase. In one-dimensional column experiments, the critical value of NT required to initiate NAPL mobilization was between 2 × 10-5 and 5 × 10-5, while tetrachloroethylene (PCE) was completely displaced as NT approached 1 × 10-3 (12). In addition to predicting the onset of mobilization, NT has also been used in models predicting the residual NAPL saturation after surfactant flushing. (17). While establishing the conditions under which mobilization occurs is important, efficient surfactant-enhanced remediation is often associated with the formation and movement of a NAPL bank, a moving body of NAPL that is continuous across the entire interface between the flushing solution and the resident aqueous phase. Theoretical and experimental studies have shown that neutrally buoyant NAPL ganglia that have a lower viscosity than the displacing fluid when mobilized move with an average velocity larger than that of the displacing fluid and form a NAPL bank, if sufficent NAPL is present (18). As the NAPL bank moves down gradient, it collides with trapped ganglia that merge with the bank resulting in efficient removal of residual NAPL from the porous medium. However, the results from these studies can only be applied to the mobilization of NAPLs in systems where the buoyancy forces are small or where the flow is in the same direction as gravity. The influence of buoyancy forces on the movement of fluid fronts has been studied by researchers in petroleum engineering. For example, relations describing gravity tonguing or underrunning during water-displacing oil processes were developed for conditions where buoyancy forces are important (10, 11, 19). These relations were derived assuming two-phase flow, vertical equilibrium, and the absence of a transition zone, i.e., only one phase or component flowing at any point in the reservoir. While the results from this research are useful for predicting the inclination angle of a surfactant front relative to the dip angle or bedding plane as the front displaces water in a porous medium, these results cannot be used to address the question of NAPL bank formation, which is the focus of this investigation. In addition to capillary, viscous, and buoyancy forces, the formation of a NAPL bank during surfactant flushing may depend upon other physical and chemical properties that may vary with time and spatial location: the viscosity ratio between NAPL and the aqueous phase (18); the wettability of the porous medium (20); interfacial charge (20); pore geometry (18); the size of mobilized droplets (18); and coalescence of emulsified NAPL droplets (18, 21, 22). Coalescence behavior is affected by film thickness and viscosity of the interfacial film between the nonaqueous and aqueous phases, surface diffusion of the surfactant within the film, surfactant adsorption onto the film, surface shear

viscosity and elasticity, and surface dilatatial viscosity and elasticity (23). Additionally, other factors may be important, including the residual NAPL saturation, the NAPL ganglia size distribution, the angle of inclination of the displacing surfactant front with the bedding plane, and the aspect ratio of the residual NAPL zone with respect to the mean aqueous phase flow direction, i.e., the length of the contaminated zone in the flow direction divided by the width perpendicular to the flow direction. In this study we focused on the influence of viscous and buoyancy forces on the formation of NAPL banks.

Experimental Design Materials. A uniform porous medium was created by sieving glass beads (Cataphote, Jackson, MS) to obtain 35-42 mesh sieve fractions (0.355-0.425 mm). HPLC-grade PCE (Fisher Scientific, FairLawn, NJ) was selected as the representative dense NAPL for all experiments and dyed with Oil Red O (Sigma Chemical Co., Chicago, IL) at 0.27 g/L to permit visual observation. The surfactant solution was a 1:1 mixture by weight of sodium diamyl sulfosuccinate (aerosol AY, C14H25O7SNa) and sodium dioctyl sulfosuccinate (aerosol OT, C20H37O7NaS). These anionic, food-grade surfactants were used as received from the manufacturer (Cytek, West Paterson, NJ); they significantly reduced interfacial tension (12), and their physical and chemical properties remained constant over the time scales of our experiments. Solutions of 1% by weight aerosol AY/OT were prepared for all experiments with deionized (DI) water (model 21RC1, Dracor Water Systems, Inc., Durham, NC). To ensure consistent pH between experiments, the DI water was equilibrated with atmospheric CO2 prior to mixing the surfactant solutions. One drop of green food dye (C.F. Sauer Company, Richmond, VA) was added to every 100 mL of surfactant solution for visualization. Chemical and Physical Property Measurements. Surfactant solution densities were measured at 26 °C using a density meter (model DMA 48, Anton Paar, Graz, Austria). Time-dependent viscosities were measured at 26 °C in triplicate with a falling ball viscometer (Haake Instruments, Paramus, NJ) equipped with a jacketed glassware water bath (series 9500, PolyScience Co., Niles, IL). For comparative purposes, the falling ball viscometer was also used to measure the viscosity of the PCE-in-water macroemulsions that formed during the surfactant flush. The macroemulsions were created by adding 50 mL of 1% aerosol AY/OT to 50 mL of PCE in a glass vial, shaking vigorously, and siphoning off the milky macroemulsions. The pH of the influent surfactant solution was measured prior to each of the experiments with an expandable ion analyzer equipped with a combination pH electrode and an automatic temperature control probe (model EA 940 Orion Research, Boston, MA). The effluent pH was measured with pH paper with 0.5 pH unit increments (Micro-Essential Laboratory, Brooklyn, NY), since soluble PCE interfered with the pH electrode. The interfacial tension between PCE and 1% solution of aerosol AY/OT was measured using a spinning drop tensiometer (SITE 04, Kruss, Charlotte, NC) after equilibrating the fluids overnight. Because the water/PCE/ surfactant system contained a mixture of separate phase PCE, PCE trapped in micelles, and PCE contained in macroemulsions, the apparent solubility of PCE in 1% aerosol AY/OT was determined by adding 5 mL of PCE to 5 mL of a 1% aerosol AY/OT solution in 11 mL vials. The vials were capped with Teflon-lined septa and shaken (Junior Orbital Shaker, Lab-Line Instruments, Melrose Park, IL) for a minimum of 24 h at 26 + 0.5 °C. The equilibrated surfactant solutions were then removed and centrifuged (IEC HN-SII centrifuge, Damon, Needham Heights, MA) at approximately 1300g for 30 min to separate the emulsified PCE, and duplicate samples VOL. 33, NO. 14, 1999 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2441

FIGURE 1. Experimental setup for PCE mobilization experiments. In the case shown here, the horizontal axis is aligned with the x-axis of the cell. of the aqueous phase were added to methanol and analyzed for PCE concentrations by high-pressure liquid chromatography (HPLC; Waters, Milford, MA). The 1-3 mL of surfactant solution removed from the 11 mL vials for sample analysis was replaced with new solution, and the above steps were repeated until the PCE concentrations in each of the surfactant solutions remained stable for 3 consecutive days, thus ensuring equilibrium values. The coalescence behavior of the water/PCE/surfactant system was studied in glass vials using a simple qualititative procedure (22). Flow Cell Methods. Flow cell experiments were conducted at 25 ( 1 °C in a 20 × 1.5 × 0.1 cm3 borosilicate glass cell with tapered ends (see Figure 1). Pressure drop across the cell was measured every 0.5 s by connecting the ends of the cell to a switching device (Solenoid Controller CTLR2/S2-S6, Scanivalve, San Diego, CA) that was in turn connected to a 1 psi pressure transducer (Omega Engineering, Stamford, CT). The inlet of the cell was connected to a syringe infusion pump (model 22, Harvard Apparatus, South Natick, MA) that injected the surfactant solution into the cell at a specified flow rate. During flushing experiments, the outlet was connected to a sealed vial with a small air opening to collect the effluent during sampling periods. Prior to the initial packing, all beads were soaked in a 1 M solution of nitric acid for a minimum of 12 h, rinsed with tap water 3 times, and then rinsed with DI water 3 times. The beads were then dried in an Isotemp oven (model 630G, Fisher Scientific, Pittsburgh, PA) at 85 °C and packed tightly in the cell. Stainless steel wool was packed in each end of the cell to prevent the beads from moving during an experiment. The same glass bead packing was used for all of the experiments. With the cell oriented vertically, the medium was saturated with water by pumping a minimum of 10 pore volumes (PV) of deaired, DI water upward into the cell. Once saturated, tracer tests showed that the flow was uniform across the cell after a distance of about 1 cm from the inlet. PCE was then injected vertically upward into the cell at 0.225 mL/min. PCE flow was stopped to leave at least a 1 cm PCE-free region at the end of the cell. Mass balances indicated that the PCE saturation, Sn, i.e., fraction of PCE filling the pore space, in the PCE-invaded regions was approximately 0.80-0.85. Then, approximately 5 PV of water was pumped downward through the cell at 0.18 mL/min, leaving residual ganglia of trapped PCE. The PCE-contaminated steel wool was removed and replaced with clean steel wool, and the distribution of residual PCE was recorded using a N90s 35 mm camera (Nikon Corp., Japan). The 35 mm slides were later analyzed to ensure that the morphology of the residual NAPL ganglia (size, distribution, and total number of trapped ganglia) remained similar between experiments, confirming similar wetting conditions 2442

9


for all experiments. The average residual PCE saturation was determined in the PCE-contaminated region from the PCE mass eluted during surfactant flushing. After creating the residual NAPL, water was pumped through the cell at a flow rate equal to that of the planned experiment, while measuring the temperature and the pressure drop across the cell. From these measurements, the effective aqueous phase permeability, ke, of the contaminated media was calculated. To facilitate the development of a uniform front, surfactant was first pumped upward into the vertically oriented cell through the inlet end and into the steel wool. The cell was then turned to the desired angle and the surfactant solution injected at a specified flow rate. Experiments were run at various Darcy velocites, from 12 to 38 cm/min, and angles, ranging from 0 to 90° measured CW from the horizontal axis (see Figure 1). After completion of the flushing experiment, the total effluent was weighed, and after vigorous mixing three independent samples were analyzed for PCE concentration using HPLC. The density of the effluent was also measured using the density meter and used with the mass measurement to calculate the total effluent volume. These measurements were used to determine the total mass of PCE that left the system. Digital images were collected using a CCD camera (Panasonic Corp.) attached to either a 35 mm lens (Nikon Corp., Japan) or a SMZ-U stereomicroscope (Nikon Corp., Japan). Image data were transmitted continuously to a PC where real-time image grabbing was performed using the Optimas 5.0 software package (Optimus Corporation, Bothell, WA). Slides were also captured using the N90s 35 mm camera. The flow images were monitored using a video monitor (HR Trinitron, Sony Corp., Japan) and recorded and stored on videotape for further analysis (AG-MD830 VCR, Panasonic Corp.). Digital images from each of the experiments were analyzed to determine the first and second moments of dyed PCE observed along the wall of the cell about a rectangular coordinate system, the origin of which was at the bottom inlet corner of the cell (see Figure 1). The first moment of the PCE shows the centroid of the PCE mass, and the second moment indicates the spread of PCE about the centroid, i.e., the variance. If a NAPL bank forms during flushing, the centroid of the PCE should increase and the variance decrease in the x direction as a function of time, compared to the case in which a bank does not form. Moment analysis was performed after 0.7 PV of surfactant had been injected into the cell. PCE did not exit the cell by 0.7 PV in any of the experiments, and at this point a bank had already formed or would not form at all.

TABLE 1. Experimental Conditions and Parameters q [cm/min] R [deg] Sn k × 106 [cm2] kra NB × 103 NCa × 103 NT × 103 NBa

MC1

MC2

MC3

MC4

MC5

MC6

MC7

MC8

MC9

MC10

MC11

12 0 21 0.93 0.59 3.7 1.5 4.0 0.41

12 20 21 1.3 0.42 3.6 1.5 4.4 0.81

29 0 18 1.1 0.54 3.8 3.7 5.3 0.96

17 20 18 1.3 0.41 3.6 2.1 4.8 0.99

31 0 22 1.0 0.54 3.7 3.9 5.4 1.05

29 0 22 0.93 0.55 3.4 3.7 5.0 1.08

38 0 22 1.3 0.42 3.6 4.9 6.1 1.34

17 60 19 1.1 0.47 3.5 2.19 5.5 2.95

17 71 18 1.1 0.43 3.3 2.1 5.3 4.91

17 80 18 1.1 0.52 3.9 2.1 6.0 8.84

17 90 20 1.1 0.48 3.6 2.1 5.7 ∞

Without the packing being removed, the cell was cleaned between experiments by flushing it with methanol and soaking it in a 1 M nitric acid solution. At this point, the intrinsic permeability, k, was measured following the same procedures used earlier to determine the effective permeability, ke. The aqueous phase relative permeability, kra, at the initial residual PCE conditions was determined from kra ) ke/k.

Results Chemical and Physical Properties. In order to investigate only the impact of viscous and buoyancy forces on NAPL bank formation, all of the pertinent chemical and physical properties of the PCE and surfactant were carefully quantified. While several surfactant mixtures were tested, the 1% solution of aerosol AY/OT was chosen because the chemical and physical properties remained constant over the time scales of the experiments, thus providing a well-defined, consistent system. For example, the dynamic viscosity of the 1% solution of aerosol AY/OT increased from 0.970 ( 0.009 cP (95% CI) when initially mixed to 0.989 ( 0.007 cP after 24 h, while the dynamic viscosity of a 4% solution of the same surfactants increased from 1.54 ( 0.005 cP to 3.18 ( 0.007 cP over the same time period. The density of the 1% surfactant solution was 0.9987 g/cm3 with an estimated systematic error less than (0.0002 g/cm3 at 26 °C. Multiple density measurements indicated that the random error in this measurement was negligible. The initial dynamic viscosity of 0.970 ( 0.009 cP (95% CI) at 26 °C was used in all calculations. The dynamic viscosity of a PCE-inwater macroemulsion solution was compared with the corresponding dynamic viscosity of the PCE-free surfactant solution, and the hypothesis that the mean viscosities determined for each solution were the same was not rejected at a significance level of 0.01. The solubility of PCE in the surfactant solution was 7736 ( 331 mg/L at 26 °C; the solubility of PCE in water is 221 ( 3 mg/L at 20 °C (25). The interfacial tension between one sample of the surfactant solution and PCE was 0.09 dyne/cm, which is identical to the value of 0.09 dyne/cm for PCE and a 4% by weight solution of the same surfactants reported in an earlier study (12). At 25 °C the interfacial tension between water and non-dyed PCE is 34.5 dyne/cm, the density of PCE is 1.62 g/cm3, and the dynamic viscosity of PCE is 0.89 cP (26). The influent pH of the surfactant solution ranged from 5.4 to 6.1. The pKa value for the surfactant solution was determined by acid-base titration and was less than 2. Based on the low pKa value, pH changes over the observed range were not expected to affect coalescence behavior in these systems. The pH of the influent surfactant solution and the effluent were also compared for three separate experiments, confirming that pH did not change by more than 0.5 pH units during the experiments. Macroscopic Results. Parameters for the 11 mobilization experiments are summarized in Table 1. NT for all of the experiments was above the minimum value needed for

mobilization (12). Two macroscopic regimes were observed after the residual NAPL was contacted with the surfactant flushing solution: (1) NAPL bank formation and transport of PCE in the mean flow direction, i.e., in the x direction; or (2) downward mobilization and collection of PCE in a pool or mound at the bottom of the cell. When a NAPL bank formed, NAPL ganglia mobilized and subsequently formed a bank across the entire interface between the surfactant solution and the resident aqueous phase at the flushing front. When a bank did not form, mobilized NAPL accumulated at the bottom of the cell in a pool or mound that slowly moved out of the system. In all cases, pore-scale images showed that some NAPL was still trapped after the surfactant front had passed. These NAPL droplets were eventually mobilized after dissolution reduced their size. Bank Number. The magnitude and direction of the viscous and buoyancy forces acting on a trapped NAPL ganglion will influence the initial direction of movement. From a twodimensional force balance on a dense NAPL ganglion in a homogeneous, isotropic porous medium, a bank number, NBa, can be defined as

NBa )

NCa + NB sin R NB cos R

(4)

where R is the same angle used in the definition of NT. The bank number is the ratio of the forces acting parallel to the mean flow direction to the forces acting perpendicular to the flow, at the onset of NAPL movement. For a NAPL less dense than the aqueous phase, the sign changes in the numerator of eq 4. Table 1 shows the NBa’s for each experiment. Low NBa. In the two experiments where NBa e 0.8 (MC1, MC2), the NAPL was mobilized perpendicular to the mean flow direction. As each experiment progressed, nearly all of the mobilized NAPL moved to the cell bottom where it pooled with little or no NAPL swept out of the cell by the surfactant front. As flushing continued, the NAPL pool migrated along the cell bottom and began leaving the system after approximately 3.5 PV of surfactant was injected. Under these conditions, nearly complete downward mobilization resulted from the dominating influence of gravity. High NBa. In the four experiments with NBa g 3 (MC8MC11), most of the mobilized NAPL moved along the cell with little mobilization in the y direction, resulting in a steep, sometimes diffuse bank. For this set of experiments, the combination of buoyancy and viscous forces resulted in a force on the ganglia directed at an angle 10 min) at specific locations in the cell showed NAPL ganglia smaller than 2-3 pore diameters migrating through the picture while larger ganglia remained trapped. In several cases, the larger trapped ganglia shrank in size until the capillary trapping force was overcome and they too were mobilized. Mobilized NAPL ganglia ranged in size from less than one to several pore diameters, sizes that correspond to drop traffic flow and small ganglion dynamics, respectively (27). Collisions between droplets occurred on a regular basis, but at no time was coalescence between colliding droplets observed. Even when the mobilized NAPL droplets accumulated in a NAPL pool they did not coalesce. Close-up images of the pools showed larger (≈2-3 pore diameters) NAPL ganglia in

the bottom of the pool with smaller NAPL droplets (≈1 pore diameter) rolling down gradient along the top of the pool. These observations are consistent with the study of this water/ PCE/surfactant system in glass vials, where after vigorous mixing NAPL droplets did not coalesce after 24 h.

Discussion This work examined the impact of viscous and buoyancy forces on NAPL bank formation in a well-characterized water/ NAPL/surfactant system in a homogeneous porous medium. Because the definition of NBa relies upon general physical and chemical properties, it should be applicable to any set of conditions. However, natural groundwater/NAPL/surfactant systems are much more complex and the physical and chemical properties harder to quantify. NBa was calculated for three of the surfactant-enhanced mobilization experiments from Pennell et al. (12): the experiment shown in Figure 2a-d on page 1332 corresponds to NBa ) ∞ and resulted in a steep but diffuse bank; for the experiment shown in Figure 5a,b on page 1334, NBa ) 0.5 and NAPL was mobilized downward with no bank formation; and the experiment in Figure 5c,d on the same page resulted in the formation of a bank with NBa ) 0.97. These results are consistent with the data presented in this work. The absence of coalescence in the water/NAPL/surfactant system may, in part, explain the instability of bank formation for cases where NBa ≈ 1. Whether or not the NAPL is coalesced, i.e., continuous, affects how easily the NAPL flows through the system. Noncoalesced drops must move through the higher viscosity aqueous phase, while continuous NAPL flows more easily. Several other factors that may affect bank formation were listed earlier. One is the system aspect ratio, which in these experiments is the ratio of the cell length to the cell height. Systems with large aspect ratios promote bank formation, since mobilized NAPL can move only a small distance orthogonal to the mean flow direction. Aspect ratios in natural systems depend upon the layering of geological materials of significantly different permeability. The ratio of the horizontal to vertical correlation scales measured for ln(k) fields is one measure of aspect ratios in the field. In a review of several field investigations (28), the ratio of correlation scales varied from 2.7 to 40, a range including the aspect ratio of 13.3 used in this study. Thus, the results from this investigation should apply to some field sites, although further study is required to establish the influence of system aspect ratio on bank formation. Finally, it is interesting to compare the behavior of dissolving NAPL ganglia in this investigation with the behavior in two water flushing experiments (29, 30). When water was flushed through sandy media to dissolve residual NAPL, the NAPL ganglia remained trapped as they shrank in size through dissolution with no measurable droplet mobilization (29, 30). However, in a study in sandy media where surfactants were present, emulsified NAPL droplets were mobile (31), just as observed here. The transport of emulsified droplets is affected by electrostatic forces. Due to the anionic nature of the surfactants in this work, the PCE droplets were negatively charged. The glass beads were also negatively charged due to the ionic strength of the surfactant solution (32). Thus there was an electrostatic repulsive force between the emulsified PCE droplets and the glass beads in this study that probably did not occur in the water flushing systems.

Acknowledgments The work was supported by Grant 5 P42 ES05948-02 from The National Institute of Environmental Health Sciences and Grant DAAL03-92-G-0111 from The Army Research Office, Research Triangle Park, NC. The authors thank Edward H. VOL. 33, NO. 14, 1999 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2445

Hill III, Oliver Pau, and Xin Fu for assistance with the measurements.

Supporting Information Available Additional images and movies from the experiments are available from the Journal’s Web page. This material is available free of charge via the Internet at http://pubs.acs.org.

Literature Cited (1) Baran, J. R.; Pope, G. A.; Wade, W. H.; Weerasooriya, V.; Yapa, A. Environ. Sci. Technol. 1994, 28, 1361-1366. (2) Brown, C. L.; Pope, G. A.; Abriola, L. M.; Sepehrnoori, K. Water Resour. Res. 1994, 30 (11), 2959-2977. (3) Grimberg, S. J.; Miller, C. T.; Aitken, M. D. Environ. Sci. Technol. 1996, 30, 2967-2974. (4) Okuda, I.; Mcbride, J. F.; Gleyzer, S. N.; Miller, C. T. Environ. Sci. Technol. 1996, 30, 1852-1860. (5) Sabatini, D. A.; Knox, R. C.; Harwell, J. H. EPA/600/S-96/002; Surfactant-Enhanced DNAPL Remediation: Surfactant Selection, Hydraulic Efficiency, and Economic Factors United States Environmental Protection Agency: Ada, OK, 1996. (6) West, C. C.; Harwell, J. H. Environ. Sci. Technol. 1992, 26 (12), 2324-2330. (7) Pope, G. A.; Wade, W. H. Surfactant-Enhanced Subsurface Remediation: Emerging Technologies; Sabatini, D. A., Knox, R. C., Harwell, J., Eds.; American Chemical Society: Washington, DC, 1995. (8) Ng, K. M.; Davis, H. T.; Scriven, L. E. Chem. Eng. Sci. 1978, 33, 1009-1017. (9) Morrow, N. R.; Chatzis, I.; Taber, J. J. SPERE 1988, 2, 927-934. (10) Dietz, D. N. Akademie van Wetenschappen Proc. V 53-B; Amsterdam, 1953; p 83-92. (11) Hawthorne, R. G. Pet. Trans. AIME 1960, 219, 81-87. (12) Pennell, K. D.; Pope, G. A.; Abriola, L. M. Environ. Sci. Technol. 1996, 30, 1328-1335. (13) Jawitz, J. W.; Annable, M. D.; Rao, P. S. C.; Rhue, R. D. Environ. Sci. Technol. 1998, 32, 523-530. (14) Shiau, B. J.; Sabatini, D. A.; Harwell, J. H.; Vu, D. Q. Environ. Sci. Technol. 1996, 30, 97-103. (15) Martel, R.; Gelinas, P. J.; Desnoyers, J. E., J. Contam. Hydrol. 1998, 29 (4), 319-346.

2446

9


(16) Martel, R.; Lefebvre, R.; Gelinas, P. J. J. Contain. Hydrol. 1998, 30 (5), 1-31. (17) Delshad, M.; Pope, G. A.; Sepehrnoori, K. J. Contam. Hydrol. 1996, 23 (4), 303-327. (18) Constantinides, G. N.; Payatakes, A. C. J. Colloid Interface Sci. 1991, 141 (2), 486-504. (19) Crane, F. E.; Kendall, H. A.; Gardner, G. H. F. SPEJ 1963, 3, 277-280. (20) Donaldson, E. C.; Chilingarian, G. V.; Yen, T. F. Enhanced Oil Recovery II: Processes and Operations; Elsevier: Amsterdam, 1989. (21) Wasan, D. T.; Shah, S. M.; Chan, M.; Sampath, K.; Shah, R. Chemistry of Oil Recovery 115; American Chemical Society: Washington, DC, 1979. (22) Rudin, J.; Wasan, D. T. Recent Advances in Separation TechniquessII; American Institute of Chemical Engineers: New York, 1980. (23) Wasan, D. T. Chem. Eng. Educ. 1992, 26 (2), 104-112. (24) Shah, D. O.; Schechter, R. S. Improved Oil Recovery by Surfactant and Polymer Flooding; Academic Press: New York, 1977. (25) Imhoff, P. T.; Frizzell, A.; Miller, C. T. Environ. Sci. Technol. 1997, 31, 1615-1622. (26) Cohen, R. M.; Mercer, J. W. EPA/600/SR-93/022; DNAPL Site Evaluation; U.S. EPA Robert S. Kerr Environmental Research Laboratory: Ada, OK, 1993. (27) Avraam, D. G.; Payatakes, A. C. J. Fluid Mech. 1995, 293, 207236. (28) Gelhar, L. W. Stochastic Subsurface Hydrology; Prentice Hall: Englewood Cliffs, NJ, 1993. (29) Imhoff, P. T.; Jaffe, P. R.; Pinder, G. F. Water Resour. Res. 1994, 30 (2), 307-320. (30) Powers, S. E.; Abriola, L. M.; Weber, W. J., Jr. Water Resour. Res. 1994, 30 (2), 321-332. (31) Soma, J.; Papadopoulos, K. D. Environ. Sci. Technol. 1997, 31, 1040-1045. (32) Huang, C. P. Adsorption of Inorganics at Solid-Liquid Interfaces; Anderson, M. A.; Rubin, A. J., Eds.; Ann Arbor Science: Ann Arbor, MI, 1981.

Received for review April 27, 1998. Revised manuscript received March 31, 1999. Accepted April 5, 1999. ES980427U