Enhanced remedial amendment delivery through fluid viscosity modifications: Experiments and numerical simulations

Journal of Contaminant Hydrology 101 (2008) 29–41

Contents lists available at ScienceDirect

Journal of Contaminant Hydrology j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / j c o n h yd

Enhanced remedial amendment delivery through fluid viscosity modifications: Experiments and numerical simulations L. Zhong a,⁎, M. Oostrom a, T.W. Wietsma b, M.A. Covert b a b

Energy and Environment Directorate, Pacific Northwest National Laboratory, P.O. Box 999, Richland, WA 99354, United States Environmental Molecular Sciences Laboratory, Pacific Northwest National Laboratory, P.O. Box 999, Richland, WA 99354, United States

a r t i c l e

i n f o

Article history: Received 20 February 2008 Received in revised form 3 July 2008 Accepted 11 July 2008 Available online 29 July 2008 Keywords: Remedial amendment Enhanced delivery Enhanced sweeping Heterogeneous aquifer Shear thinning Numerical model

a b s t r a c t Low-permeability zones are typically bypassed when remedial fluids are injected into subsurface heterogeneous aquifer systems. Therefore, contaminants in the bypassed areas may not be contacted by the amendments in the remedial fluid, which may significantly prolong remediation operations. Laboratory experiments and numerical studies have been conducted to investigate the use of a shear-thinning polymer (Xanthan gum) to improve access to low-permeability zones in heterogeneous systems. The chemicals sodium mono-phosphate and the surfactant MA-80 were used as the remedial amendments. The impact of polymer concentration, fluid injection rate, and permeability contrast in the heterogeneous systems has been studied in a series of eleven two-dimensional flow cell experiments. The Subsurface Transport over Multiple Phases (STOMP) simulator was modified to include polymer-induced shear-thinning effects. The experimental and simulation results clearly show that using the polymer leads to an enhanced delivery of remedial amendments to lower-permeability zones and an increased sweeping efficiency. An added benefit of using the polymer is the stabilization of the displacing front when density differences exist between displaced and displacing fluids. The modified STOMP simulator was able to predict the experimental observed fluid displacing behavior well and might be used to predict subsurface remediation performance when a shearthinning fluid is used to remediate a heterogeneous system at larger scales. Published by Elsevier B.V.

1. Introduction Heterogeneity is often encountered in subsurface contamination characterization and remediation. It can be classified into three groups: trending, layered, and discontinuous (Domenico and Schwartz, 1990). Heterogeneity in aquifers induces fluid bypassing by creating preferential flow channels in the high-permeability pathways during subsurface fluid flooding, leaving the low-permeability zones untouched (bypassed). Such a heterogeneity-induced bypassing phenomenon places certain contaminated areas inaccessible to the remedial fluids and remedial amendments, thus inhibiting the success of remedial operations, as reported in pump-andtreat systems (Mackay and Cherry, 1989), surfactant-enhanced ⁎ Corresponding author. E-mail address: [email protected] (L. Zhong). 0169-7722/$ – see front matter. Published by Elsevier B.V. doi:10.1016/j.jconhyd.2008.07.007

aquifer remediation (Saenton et al., 2002), injection of bioamendments (Yang et al., 1994), and in situ redox manipulation (ISRM) (Vermeul et al., 2002). The adverse effect can considerably delay the completion and closure of a remedial operation and significantly increase the cost or simply make the remediation impossible to accomplish. Methods of forcing fluids into low-permeability flow paths have been developed and widely implemented to solve the heterogeneity-induced bypassing problem encountered during oil recovery in the petroleum industry (e.g., Johansen, 1979). Since the intent of petroleum reservoir engineers is to control the mobility of the injected fluid in the high-permeable zones so that the fluid can be pushed through the low-permeable zones to contact and mobilize the remaining oil in these zones, these methods are thus referred as “mobility controlled flood” (MCF) techniques in the petroleum engineering literature. Two methods of mobility control have been developed. One method

30

L. Zhong et al. / Journal of Contaminant Hydrology 101 (2008) 29–41

is to use a water-soluble polymer to increase the viscosity of the injectate so that the in situ pore pressure is raised, and crossflow between layers with different permeability occurs (Lake, 1989). The majority of the polymer solutions used in oil-field applications is non-Newtonian fluids exhibiting shear-thinning (pseudoplastic) behavior (Lopez et al., 2003). The other method is to use surfactant-foam floods to generate foam in highpermeable zones in situ; therefore, the injected fluid is forced into the low-permeable areas (Hirasaki et al., 2000). The mobility ratio, M, is a key parameter that can be manipulated to achieve improved sweep efficiency of lowpermeability zones. This ratio M is defined as the mobility of the displaced, or resident, phase over that of the displacing phase, where the mobility is equal to the effective permeability of each phase divided by the viscosity of that phase. The mobility ratio can be simplified as the ratio of the resident fluid viscosity, μR, to the displacing fluid viscosity, μD (Martel et al., 1998; Robert et al., 2006; Flowers and Hunt, 2007). M¼

μR : μD

ð1Þ

When the displacing fluid is more viscous than the resident fluid, M b 1, the displacement is considered to be favorable and stable displacement front is observed. When the displacing fluid is less viscous than the displaced fluid, M N 1, the displacing is typically unstable and the displacement is considered to be unfavorable. Density differences between miscible fluids may play a significant role in groundwater solute transport and aquifer remediation. Density effects have been postulated to control the plume propagation under several landfill sites (Kimmel and Braids, 1975, 1980; MacFarlane et al., 1983). Schincariol and Schwartz (1990) reported that density difference as low as 0.8 kg/m3 could cause unstable flow displacement in homogenous and layered media. Unstable behavior of dense solute plumes has been observed in 2-D flow models (Oostrom et al., 1992a,b). Zhang and Schwartz (1995) reported downward migration of the high-density leachate in an aquifer when the density difference was only 3.5 kg/m3. Vertical convective flow

due to density difference was cited as the cause of unexpected tracer losses during subsurface characterization (Sudicky et al., 1983; Freyberg, 1986; LeBlanc et al., 1991). The undesired migration of micro-emulsions, formed during certain surfactant flushes could be prevented by reducing its density with addition of alcohol to the remedial fluids (Kostarelos et al., 1998; Shook et al., 1998; Taylor et al., 2004). In horizontal flow, displacement instability caused by density contrast between fluids is termed gravity override when a lessdense fluid displaces denser fluid, and gravity underride when a denser fluid displaces a less-dense fluid (Lake, 1989). Both gravity override and underride result in inefficient displacements, which has been the subject of several petroleum engineering studies (e. g. Craig et al., 1957; Crane et al., 1963). Density-induced underriding was observed in 2-D flow cell dense-nonaqueous phase liquid (DNAPL) remediation studies when horizontal flows were forced through flow cells (Jawitz et al., 1998; Taylor et al., 2001). Density overriding was reported when ethanol–surfactant solutions were used to remediate perchloroethene (PCE) contamination (Taylor et al., 2004), when less-dense co-solvent mixture was used to displace water in porous media (Jawitz et al., 1998), and when ethanol flooding was used to mobilize trichloroethene (TCE) in 2-D flow cell experiments (Grubb and Sitar, 1999). At lower flow rates, the displacement front tilted more from the vertical position (Grubb and Sitar, 1999; Taylor et al., 2001). A tilt angle, β, measured from the horizontal downstream direction, may be used to describe the stability of the fluid displacement front. For density-induced override, β will be between 0° and 90°, and for displacements exhibiting gravity underride, β will be between 90° and 180°. The tilt angle, β, can be calculated using the following formula (Lake, 1989; Jawitz et al., 1998): tanβ ¼

qΔμ kgΔρ

ð2Þ

where q (m s− 1) is the Darcy flux, k (m2) is the media intrinsic permeability, g (m s− 2) is the gravitational constant, Δρ (kg m− 3) is the density difference between the two fluids, and

Fig. 1. Schematic of packing configurations and sampling ports locations. The labels 20/30, 30/40, 40/50, and 70 refer to the grade level of the Accusand. Ports P1 through P10 are sampling ports located in the porous medium and E1 through E8 are effluent sampling ports.

L. Zhong et al. / Journal of Contaminant Hydrology 101 (2008) 29–41 Table 1 Accusand properties Accusand grade

Bulk density

Porosity

Particle diameter d50

Hydraulic conductivity

(m × 10− 3)

(m s− 1 × 10− 3)

0.37 0.34 0.30 0.42

0.713 a 0.532 a 0.539 a 0.123

2.50 a 1.49 a 0.72 a 0.15

(kg m− 3) 20/30 30/40 40/50 70 a

1.67 1.75 1.85 1.54

From Schroth et al. (1996).

Δµ (kg m− 1 s− 1) is the viscosity difference between the two fluids. Jawitz et al. (1998) used Eq. (2) to calculate the tilt angle in their flow cell experiments. They concluded that the deviation of the angle from the value calculated from Eq. (2) was attributed to the influence from the flow cell boundaries. The tilt angle changes observed by Grubb and Sitar (1999) and Taylor et al. (2001) could be explained using Eq. (2). Viscosity increasing for the displacing fluid was proposed in order to reduce the downward migration (gravity underriding) of micro-emulsions during surfactant flushing (Shook et al., 1998; Kostarelos et al., 1998). However, to our knowledge, no viscosity manipulating of displacing fluids aimed to reduce gravity instability was reported in the subsurface remediation literature. Mobility-controlled flooding using polymers may be applied to subsurface remediation to overcome heterogeneity-induced bypassing problems, so that amendments can be delivered to low-permeability zones. The experiments conducted by Martel et al. (1998) provided a first systematic investigation of mobility control for environmental purposes, using Xanthan gum polymer for solution mobility control. Although Martel et al. (1998) investigated sweeping efficiency for heterogeneous sand packs, no remedial amendments were added to the polymer solutions. Robert et al. (2006) used surfactant solutions with Xanthan gum polymer to flush 2-D aquifer models with heterogeneously packed porous media to remove TCE. During the polymer floods, enhanced sweeping over the heterogeneous systems was observed, and improved contact between TCE ganglia and surfactant–polymer solutions was reported. The focus of the study was the optimization of TCE dissolution in the heterogeneous system without particular emphasis on enhanced remedial amendment delivery to the lower permeable zones.

31

In this work, a total of 11 flow cell experiments were conducted to 1) demonstrate enhanced sweeping efficiency and remedial amendment delivery, 2) demonstrate the elimination of density-induced underriding when shear-thinning solutions are used, and 3) test and verify a modified version of the STOMP simulator (White and Oostrom, 2006) considering polymer shear-thinning behavior. The amendment used in the experiments were sodium mono-phosphate (NaH2PO4), a chemical that is currently evaluated to aid in the remediation of uranium [U(VI)] at the U.S. Department of Energy (DOE) Hanford Site (Wellman et al., 2007), and the surfactant MA-80, a common amendment for DNAPL remediation (Dwarakanath et al., 1999; Zhong et al., 2001). 2. Materials and methods 2.1. Flow cell experiments Three sets of experiments were conducted in a 0.5-m-long, 0.4-m-high, and 0.05-m-wide flow cell. A schematic of the flow cell is presented in Fig. 1. The front and back for the nominal two-dimensional (2D) flow cell consisted of glass, while the frame was made out of stainless steel. Fluids were injected through eight evenly-spaced (0.05 m distance) inflow ports using separate Encynova (Broomfield, CO) high-precision piston pumps. Outflow also occurred through eight equally-spaced effluent ports, denoted as E1 through E8 in Fig. 1, connected to a constant-head chamber. Assigning the left-bottom corner of the flow cell denoted as (x,z) = (0,0), the water level in the constanthead chamber was kept at z = 0.4 m. The porous media used in the flow cell experiments were four grades of silica sand (20/30, 30/40, 40/50, and 70 mesh), obtained from Unimin Corporation (Le Sueur, MN). The main properties of the Accusand are listed in Table 1. For additional information of the sands, the reader is referred to Schroth et al. (1996). Three packing configurations were used for three sets of experiments (Fig. 1). Accusand 20/30 was used as the porous medium matrix in all experiments. Two lower-permeability zones were formed with Accusand 30/40, 40/50, or 70. The flow cell was packed under saturated conditions to avoid the formation of entrapped air. The average porosity and dry bulk density values listed in Table 1 were obtained gravimetrically. The packed systems had an average porosity of approximate 0.35 and a total “pore volume” (PV) of 3.5 × 10− 3 m3.

Table 2 Description of experiments Name

Packing configurations (Fig. 1)

Flood type

Flow rate (ml/min)

Displacing fluid composition a

Sampling

Exp-Ia-W Exp-Ib-W Exp-Ic-P Exp-IIa-W Exp-IIb-P Exp-IIc-P Exp-IId-P Exp-IIe-P Exp-IIIa-S Exp-IIIb-S Exp-IIIc-SP

I I I II II II II II III III III

Water Water Polymer Phosphate–water Phosphate–polymer Phosphate–polymer Phosphate–polymer Phosphate–polymer Surfactant Surfactant Surfactant–polymer

5 50 5 5 5 50 5 50 5 5 5

De-ionized water De-ionized water 600 ppm Xanthan, 400 ppm NaCl De-ionized water 200 ppm Xanthan, 400 ppm NaCl, 200 ppm NaH2PO4 200 ppm Xanthan, 400 ppm NaCl, 200 ppm NaH2PO4 600 ppm Xanthan, 400 ppm NaCl, 200 ppm NaH2PO4 600 ppm Xanthan, 400 ppm NaCl, 200 ppm NaH2PO4 4% MA–80, 4% IPA, 6000 ppm NaCl 4% MA-80, 4% IPA, 6000 ppm NaCl 600 ppm Xanthan, 4% MA-80, 4% IPA, 6000 ppm NaCl

No No No No Yes Yes Yes Yes No Yes Yes

MA-80: sodium dihexyl sulfosuccinate; IPA: isopropanol alcohol. a All solutions contain 100 ppm Brilliant Blue FCF dye.

— sampling — sampling — sampling — sampling — effluent — effluent

ports ports ports ports

32

L. Zhong et al. / Journal of Contaminant Hydrology 101 (2008) 29–41

Table 3 Fluid properties of flooding solutions Composition a

Solution name

Water Polymer Phosphate Phosphate–200 ppm polymer Phosphate–600 ppm polymer Surfactant Surfactant–polymer

De-ionized water 600 ppm Xanthan, 400 ppm NaCl 200 ppm NaH2PO4, 400 ppm NaCl 200 ppm Xanthan, 200 ppm NaH2PO4, 400 ppm NaCl 600 ppm Xanthan, 200 ppm NaH2PO4, 400 ppm NaCl 4% MA-80, 4% IPA, 6000 ppm NaCl 600 ppm Xanthan, 4% MA-80, 4% IPA, 6000 ppm NaCl

Density at 22 °C

Viscosity at shear rate of 0.1 s− 1, 22 °C

Water–TCE interfacial tension

(kg m− 3)

(kg s− 1 m− 1 × 10− 3)

(N m− 1 × 10− 3)

0.998 1.000 1.000 1.000 1.000 1.005 1.008

1.002 44.1 1.002 12.4 44.1 1.08 48.9

34.5 NA NA NA NA 0.44 0.44

MA-80: sodium dihexyl sulfosuccinate; IPA: isopropanol alcohol. a All solutions contain 100 ppm Brilliant Blue FCF dye.

Details of the 11 experiments are presented in Table 2. In experiment set I, no remedial amendment was used. Sodium mono-phosphate (NaH2PO4) and the surfactant sodium dihexyl sulfosuccinate (MA-80) were the remedial amendments used in experiment sets II and III, respectively. In experiment set III, the two lower-permeability zones were packed with 12% (by volume of the pore space) of the DNAPL trichloroethylene (TCE). The alcohol isopropanol (IPA) and salt (NaCl) were used in the surfactant solutions for experiment set III to achieve an optimum phase behavior between the surfactant solution and the DNAPL (Zhong et al., 2003). Xanthan gum (Kelco Oil Field Group, Houston), a watersoluble biopolymer, was used to modify the viscosity of the displacing fluids, which then are associated with shearthinning properties. The DNAPL TCE was used in the remediation experiments. To allow for visual observations, the dye Brilliant Blue FCF was added to color all the displacing fluids, while Oil Red O dye was used to color the TCE in the experiments. All chemicals, except for the Xanthan gum, were obtained from Aldrich Chemical Company (Milwaukee, WI). The experimental initial TCE distributions in experiment set III were created to investigate conditions at a certain time after a considerable TCE infiltration and re-distribution event. At that time, it was conceptualized that most of the TCE in the higher

permeability zones would be dissolved but the TCE in the lower permeability materials would still be present as a NAPL. It is recognized that these situations require substantial spills allowing entry into lower-permeability materials. Fluid properties of the flooding mixtures are listed in Table 3. The specific densities of fluids were determined with pycnometers. The interfacial tension (IFT) of displacing surfactant solutions with TCE was measured using the pendant drop technique, where TCE was used as the drop phase and each of the aqueous solutions as the continuous phase. The viscosity of surfactant–polymer solutions were measured using a low shear (0.1 s− 1) falling ball viscometer (Gilmont 100), pre-calibrated with water. For the higher shear rates, a Brookfield LV viscometer was used. The Xanthan solutions are shear-thinning fluids and the measured apparent viscosity–shear rate relations are shown in Fig. 2. The solution with 4.0% MA-80 surfactant, 600 ppm Xanthan, and 6000 ppm NaCl was used for the TCE removal experiments because this solution has a favorable phase behavior in contact with TCE and the lowest IFT (Zhong et al., 2003). All polymer solutions were filtered through 0.6 μm membrane before injection to remove microgels that might exist in the solution (Chauveteau and Kohler, 1984). Liquid samples were taken during selected experiments in experiment sets II and III using gas-tight syringes. Phosphate

Fig. 2. Measured apparent viscosity as a function of shear rate at room temperature for several Xanthan gum polymer solutions.

L. Zhong et al. / Journal of Contaminant Hydrology 101 (2008) 29–41

33

Table 4 Values of a, b, μ0, and μ∞ used to compute viscosity of Xanthan solutions listed in Table 3, according to Eq. (2) Solution name

a

b −1 −1

(kg m s Polymer Phosphate–200 ppm polymer Phosphate–600 ppm polymer Surfactant–polymer

−3

(PO3− 4 ) concentrations were obtained using an ion chromatograph while a liquid chromatograph system was used for effluent TCE concentration determinations. 2.2. Numerical simulations The water operational mode of the STOMP (Subsurface Transport Over Multiple Phases) simulator (White and Oostrom, 2006) with sequential electrolyte transport was used in this study. The applicable governing equations are the component conservation equations for water flow and electrolyte transport in the aqueous phase. In this STOMP mode, electrolyte concentrations do affect density and viscosity but the massconservation equation is not fully coupled with the water conservation equation. Instead, an operator splitting approach is used where the electrolyte transport equation is solved after convergence of the water conservation equation. The partial differential equations for flow and transport are discretized following the integrated-volume finite difference method by integrating over a control volume. The 50× 40 cm computational domain was discretized into 8000 uniform (0.5 cm× 0.5 cm) grid blocks. Using Euler backward time differencing, yielding a fully implicit scheme, a series of nonlinear algebraic expressions is derived. The algebraic forms of the nonlinear governing equations are solved with a multivariable, residualbased Newton-Raphson iterative technique where the Jacobian coefficient matrix is composed of the partial derivatives or the

μ0 −1 −1

× 10 )

27.83 8.71 27.83 30.08

μ∞ (kg m s

−0.19 −0.22 −0.19 −0.16

−3

× 10 )

1.002 1.002 1.002 1.002

(kg m− 1s− 1 × 10− 3) 44.1 12.4 44.1 48.9

governing equations with respect to the primary variables. The algebraic expressions are evaluated using upwind interfacial averaging to fluid density and mass fractions. User specified weightings (i.e., arithmetic, harmonic, geometric, upwind) are applied to the remaining flux components. For the simulations described in these papers, harmonic averages were used for all other flux components. The maximum number of NewtonRaphson iterations was eight, with a convergence factor of 10− 6. A total-variation diminishing (TVD) technique (Datta Gupta et al., 1991) was used to minimize artificial diffusion, preserve sharp displacement fronts while avoiding oscillations that commonly affect classical higher-order schemes. Because of the use of the TVD schemes and the operator splitting approach, the maximum allowable time step in the simulations is 60 s. Polymer fluid viscosity was computed as a function of shear rate and polymer concentration (Eq. (3)), following a method outlined by Lopez et al. (2003) where the effective viscosity, μeff (Pa s), is a constant for the Newtonian ranges at low and high shear rates, and follows a power-law model for the range where shear-thinning occurs:  
  C γb ; μ 0 μ eff ¼ max μ ∞ ; min a Co

ð3Þ

where μ∞ and μ0 (kg m− 1 s− 1) are the fluid viscosity at high (N100 s− 1) and low (b0.1 s− 1) shear rates, respectively; C/Co is the normalized polymer concentration; a (kg m− 1 s− 1) and b are

Fig. 3. Displacement behavior of Exp-Ia-W and Exp-Ic-P at several pore volumes (PV) and comparisons with predicted C/Co = 0.5 isocontours. A PV is calculated based on the total pore volume of the flow cell.

34

L. Zhong et al. / Journal of Contaminant Hydrology 101 (2008) 29–41

power-law model constants. For the simulations, fitted a and b values to the associated curve in Fig. 2 are obtained using a fixed value for the measured values of the shear rate at 0.11 s− 1 (Table 3). The used a, b, μ0, and μ∞ values are listed in Table 4. The shear rate in Eq. (3) is computed as αq γ ¼ pffiffiffiffiffiffi kn

ð4Þ

where α is a shape parameter constant, q is the Darcy velocity (m s− 1), k is the permeability (m2), and n the porosity. For the simulations reported in this paper, an α-value of 2.37 is used, which is consistent with the value reported by Martel et al. (1998) for sands having similar diameter. The TCE liquid behavior was not explicitly modeled, except for an adjustment to the aqueous permeability of the zones containing the TCE based on the Xanthan concentration. If the concentration of the Xanthan b0.1 ppm, it was assumed that the residual TCE saturation was 0.12. However, for Xanthan concentration N0.1 ppm, the TCE saturation was assumed to 0 as a result of enhanced solubilization and/or mobilization. Following Lenhard and Parker (1987), the associated aqueous permeability for the case where the TCE saturation is 0.12 is 0.87. 3. Results and discussion 3.1. Experiment set I: Sweep efficiency The experiments in set I (Table 2) were conducted to investigate the effects of Xanthan injection on sweep efficiency, defined as the portion of a subsurface volume contacted by the injected fluid at any stage of the flood (Neil et al., 1983). In this work, sweep efficiency is determined visually based on the swept area of the lower-permeability zones. Fig. 3 shows a comparison at three times between a water flood (Exp-Ia-W) and a polymer flood (Exp-Ic-P) at flow rate of 5 ml/min. For the water flood, flow bypassing to the lower-permeability zones is considerable, especially in the 40/50 zone. During the polymer flood, the displacing front is straighter in the matrix sand. The simulated fluid displacing fronts, represented by the C/Co = 0.5 isocontours of the dye for the water flood and the Xanthan gum for the polymer flood, are also included in Fig. 3. The simulations show a good match

Fig. 5. Example of simulated viscosity and shear rate distributions for Exp-Ic-P: a) viscosity, b) shear rate, and, c) shear rate and flow velocity in the 40/50 sand zone.

Fig. 4. Sweeping efficiency for Exp-Ia-W and Exp-Ic-P.

with the experimental results. Information on swept area as a function of the injected pore volume for a water and polymer flood is shown in Fig. 4. The plots show that the sweeping

L. Zhong et al. / Journal of Contaminant Hydrology 101 (2008) 29–41

35

Fig. 6. Fluid displacements for experiment set II at several pore volumes (PV) and comparisons with predicted C/Co = 0.5 isocontours.

efficiency is considerably better when the mobility-controlling fluid was used as the displacing fluid as opposed to a regular water flood. The relative improvements are more pronounce for the sand with the lowest permeability. Simulated viscosity and shear rate distributions for Exp-Ic-P at 0.55 PV are provided in Fig. 5. The simulation time corresponds with the 0.55-PV picture in the lower panel of Fig. 3. In broad terms, the viscosity distribution in Fig. 5a shows the displaced water, with a viscosity close to 0.001 kg m− 1 s− 1, at the right-hand side, the displacing polymer solution, with a viscosity around 0.030 kg m− 1 s− 1, at the left-hand side of the flow cell, and a transition zone between the water and the polymer solution. As expected, the viscosity of the moving shear-thinning polymer

solution is less than the 0.042 kg m− 1 s− 1, listed in Table 3. The reduced viscosity in the areas below, above, and in between the two lower-permeability zones (Fig. 5a) can be explained by the higher flow velocities and shear rates in these areas (Fig. 5b). In and just upstream of the lower-permeability zones the fluid viscosity is larger than in the matrix porous medium containing the polymer solution due to the lower flow velocities and associated shear rates. A detailed view of the computed shear rates and flow velocities for the 40/50 Accusand zone is shown in Fig. 5c. This figure shows that the highest flow rates and shear rates are in the transition zone from water to polymer solution, and that the lower shear rates can be found just upstream, in the left part, and just downstream of the lower-permeability zone.

36

L. Zhong et al. / Journal of Contaminant Hydrology 101 (2008) 29–41

Fig. 7. Influence of a) polymer concentration and sand type, and, b) flow rate on swept area (% swept of lower-permeability zones) for experiment set II. See Table 2 for the descriptions of each experiment.

Over time, the area with lower shear rates in this zone expands to the right. The lower shear rates just downstream of the zone do not have an effect on the viscosity. At this point in time (0.55 PV), that area is still occupied by “clean” water and per Eq. (3), its viscosity is not affected by shear rate alterations. The velocity/ shear rate and viscosity distribution simulation illustrated that the highest velocity and shear thinning occurred in the zone with the lowest permeability, supporting the laboratory observation that the sweeping enhancement was most pronounced in this zone.

3.2. Experiment set II: Phosphate delivery Pictures and simulated result of the five phosphatedelivery experiments (set II in Table 2) are shown in Fig. 6 for three PVs. The pictures show that the displacement fronts of the four polymer-flooding experiments were straighter than the water-flooding experiment (Exp-IIa-W). The pictures do not clearly show a major improvement in sweeping efficiency for the 30/40 lower-permeability zones for the polymer floods. For the 70-mesh sand, improvements can be

Fig. 8. Comparison of normalized phosphate concentrations for a) ports 2 and 4 of Exp-IIa-W and Exp-IIb-P, b) ports 7 and 9 of Exp-IIa-W and Exp-IIb-P, c) ports 2 and 4 of Exp-IIa-W and Exp-IId-P, and, d) ports 7 and 9 of Exp-IIa-W and Exp-IId-P. See Table 2 for the descriptions of each experiment.

L. Zhong et al. / Journal of Contaminant Hydrology 101 (2008) 29–41

Fig. 9. Measured and simulated normalized phosphate concentrations for Exp-IIa-P.

seen at 0.52 PV for both the 200 and 600 ppm polymer flushes. The efficiency improvement is more pronounced for the 600 ppm polymer flood because of additional, viscosityinduced, penetration at the top and bottom of the zone. Visual differences in sweeping results between the two injection rates (5 ml/min for Exp-IIb-P and Exp-IId-P; 50 ml/min for Exp-IIc-P and Exp-IIe-P) are minimal. The simulation results for all five experiments are in good agreement with the experimental observations. Quantitative information on sweeping efficiency for experiment set II is shown in Fig. 7. Fig. 7a indicates that differences in the 30/40-mesh zone are small but that for the 70-mesh zone, the sweeping-efficiency enhancement increased as a function of polymer concentration. Compared

37

to the water flood, the swept area was more than doubled for the 200 ppm Xanthan flood (Exp-IIb-P) and more than tripled in the 600 ppm Xanthan flood (Exp-IId-P). The effect of flushing flow rate on the sweeping enhancement is shown in Fig. 7b. Again, it is obvious that the 600 ppm polymer solution improved the sweeping efficiency more than the 200 ppm polymer solution. However, increasing the flow rate from 5 ml/min to 50 ml/min had a minor effect on the swept areas. Phosphate transport in the lower-permeability zones by water (Exp-IIa-W), 200 ppm (Exp-IIb-P), and 600 ppm (Exp-IId-P) polymer solutions at a flow rate of 5 ml/min is shown in Fig. 8. For both polymer concentrations, no phosphate-delivery enhancement compared to the water flood was observed in the 30/40-mesh sand, which is consistent with results shown in Figs. 6 and 7. However, for the lowerpermeability 70-mesh sand, phosphate arrival at ports P2 and P4 (red lines in Fig. 8a and c) for both polymer floods was much sooner than for the water flood, showing an improved delivery of the amendments. Although the experiment duration was not long enough to show complete breakthrough of the phosphate at ports P7 and P9, Fig. 8d shows that at port P9 for the 600 ppm polymer flood, phosphate breakthrough started to occur after 0.8 PVs. The faster arrival at this location compared to the other experiments is again consistent with the pictures shown in Fig. 6 and the swept area information in Fig. 7. Simulated and experimental phosphate concentrations at the sampling ports P2, P4, P7, and P9 are shown in Fig. 9 for Exp-IId-P (600 ppm polymer flood). A reasonable match was obtained for the phosphate concentrations in both the 30/40-mesh sand (P2 and P7) and 70-mesh sand (P4 and P9), indicating that the major flow and transport processes are implemented correctly in the STOMP simulator. 3.3. Experiment set III: Surfactant delivery to residual TCE zones In Fig. 10, pictures and numerical simulations for the delivery of the surfactant by water (Exp-IIIb-S) are shown in

Fig. 10. Fluid displacements for Exp-IIIb-S and Exp-III-b-SP at several pore volumes (PV) and comparisons with predicted C/Co = 0.5 isocontours. See Table 2 for the descriptions of each experiment.

38

L. Zhong et al. / Journal of Contaminant Hydrology 101 (2008) 29–41

the upper frames, while pictures and simulations of the delivery of surfactant–polymer solution (Exp-IIIc-SP) are presented in the lower frames. In general, it was observed that the surfactant–polymer solution reduced heterogeneity-induced flow bypassing, enhanced surfactant delivery to the lowerpermeability zones and improved DNAPL removal from the zones. The surfactant flood showed considerable density underriding, while the displacement front of the surfactant– polymer solution was rather stable. The pictures at 0.55 PV show that the density effects clearly reduced the supply of surfactant to the lower-permeability zones. Gravity underriding resulted in a non-uniform sweeping of the lower-permeability zones. For example, after 0.55 PVs, only about 25% of the upper zone and about 65% of the lower zone were swept during the water flood. When the polymer solution was used, the swept area percentage reached close to 100% and 85% in the upper and lower zones, respectively. The STOMP simulations showed good agreement at all times for both types of flooding. The interfacial tension (IFT) of both the surfactant and surfactant–polymer with TCE was measured to be 0.44 × 10− 3 N/m. At this relatively low IFT, trapped TCE was mobilized immediately after it was contacted by the surfactant solutions. However, the extent of TCE displacement is different between the two remedial solutions. A higher percentage of TCE was mobilized by the viscous polymer solution flushing. By comparing pictures at the same PVs in the upper and lower panel in Fig. 10, it can be seen that when the lower-permeability zones were swept by the remedial solutions, more TCE was still trapped in these zones contacted

with surfactant-only solution. A TCE pool was formed from the mobilized NAPL only in the system flushed by the viscous surfactant–polymer solution. At a given time, more TCE was removed from the lower-permeability zones by surfactant– polymer solution than by surfactant solution. The observed enhanced DNAPL mobilization by viscous fluids was also reported by Robert et al. (2006). The total trapping number (NT), quantifying the balance of buoyancy (gravitational), viscous, and capillary forces acting on trapped DNAPL globules, was used to predict the extent of DNAPL mobilization (Pennell et al., 1996; Li et al., 2007). The critical value of NT required to initiate DNAPL mobilization was between 2 × 10− 5 and 5 × 10− 5, while complete mobilization of trapped DNAPL was observed as (NT) approached 5 × 10− 4 (Pennell et al., 1996). In this work, the computed (NT) was 1.51 × 10− 3 and 8.0 × 10− 4 for surfactant–polymer flood and surfactantonly flood, respectively and complete TCE displacement was therefore expected for the surfactant–polymer flood. The flushing of the heterogeneous porous medium with the surfactant solution was repeated (Exp-IIIb-S; Table 3) to demonstrate reproducibility and to obtain effluent data, allowing comparison with effluent data from the surfactant– polymer flood (Exp-IIIc-SP). The shapes of the displacing were similar over time (results not shown), indicating the reproducibility of the experimental results. For the two surfactantonly floods (Exp-IIIa-S and Exp-IIIb-S), the measured tilt angles at 0.55 PV, based on visual observations, were approximately 150 to 160°. This value range corresponds well to a computed tilt angle of 157°, according to Eq. (2), for

Fig. 11. TCE concentrations at effluent sampling ports for a) Exp-IIIb-S, and, b) Exp-IIIc-SP.

L. Zhong et al. / Journal of Contaminant Hydrology 101 (2008) 29–41

39

3.4. Injection pressures

Fig. 12. Measured pressure differences across flow cell for several water flood and 600 ppm Xanthan flood experiments.

injection of 5 ml/min surfactant solution in a homogeneous 20/30-mesh sand. Although the IFT of both flushing solutions were almost the same, for the surfactant–polymer experiment mobilization was the primary mode of removal. For the surfactant-only floods, solubilization appeared to be the dominant removal process. The different removal processes were not only observed in the photographs (Fig. 10), but also in the behavior of the effluent concentrations (Fig. 11). For the surfactant-only flood, the highest concentration in effluent ports E1 and E2 was less than 550 ppm, indicating that no enhanced solubilization was achieved in the paths flow through these two effluent ports (Fig. 11a). The effluent maximum TCE concentrations increased with depth, with over 25,000 ppm measured in port E8. This concentration is much larger than the aqueous solubility of ~1100 ppm, indicating enhanced TCE solubilization through micro-emulsion formation (Zhong et al., 2001). The observations indicate that when TCE was dissolved in the surfactant solution, the density of the solution increased, resulting in “sinking” of the fluid toward the bottom of the flow cell. Micro-emulsion “sinking” behavior was also reported by Kostarelos et al. (1998). The TCE concentration plots shown in Fig. 11b are different than in Fig. 11a. For the surfactant–polymer flood, the highest concentrations were not seen at the bottom ports, but at the centrally located ports E3, E4, and E5 (Fig. 11b). In addition, removal of TCE as an emulsion only occurred over a limited amount of time (~ 1 PV), compared to the surfactant-only flush. The effluent concentration data support the visual observations shown in Fig. 10. To put the results of the surfactant experiments in perspective, it should be noted that the purpose of the various surfactant floods conducted in this study was to demonstrate enhanced TCE removal from lower-permeability zones through improved amendment delivery. It was not our intent to optimize remediation. We fully recognize the negative potential of TCE mobilization observed in the surfactant– polymer flood. Additional work is clearly needed to combine the work shown in this manuscripts with other remediation techniques that minimize downward migration of DNAPL after mobilization and enhanced dissolution (Miller et al., 2000; Ramsburg et al., 2004), or by imposing upward flow to capture the mobilized DNAPL (Longino and Kueper, 1995; Lunn and Kueper, 1997).

A concern related to the injection of the viscous remedial solutions is that injection pressures might be high for a practical implementation. The pressure drop during water and the most-viscous remedial solution injections is presented in Fig. 12. The pressure across the 0.5-m-long flow cell was less than 700 Pa (~ 0.099 psi) for the 600 ppm polymer solution injection at 50 ml/min, corresponding to a Darcy velocity of ~ 10.3 m/day. The shear-thinning nature of the polymer solution limits the increase in pressure with increasing flow rate. Through simple extrapolation, it can be shown that for an aquifer with a similar hydraulic conductivity, a 10-m treatment zone, and an injection rate producing a 10 m/day Darcy velocity at the well, the required injection pressure will be about 13,800 Pa which is approximately 2 psi (pounds/square inch). This computation shows that field implementation might be feasible for these types of solutions, although site-specific calculations need to be conducted. 4. Summary and conclusions Laboratory experiments and numerical studies have been conducted to investigate the use of a shear-thinning polymer (Xanthan gum) to improve access to low-permeability zones in heterogeneous systems. Experimental variables were polymer concentration, fluid injection rate, and permeability contrast. The reported studies build upon the work by Martel et al. (1998) by considering transport of remedial amendments and by comparison experimental results with numerical simulations. The STOMP simulator was modified to include polymer-induced shear-thinning effects. The experimental and simulation results clearly show that using polymer solutions leads to enhanced delivery of remedial amendments to lower-permeability zones and increased sweeping efficiencies. In addition, the viscous polymer caused stabilization of the displacing front when density differences existed between displaced and displacing fluids. It should be noted that the investigated permeability range in this paper has been rather limited. Future research should be directed towards evaluating this method for larger permeability contrasts and towards developing polymer cocktails that will allow effective penetration of remedial fluids into lowerpermeability materials. A numerical simulator with the ability to simulate shear-thinning behavior should be very helpful in the further analysis of the method. The STOMP simulator was able to predict the experimental observed single phase fluid displacing behavior well and might potentially be used to predict subsurface remediation performance when a shear thinning fluid is used to remediate a heterogeneous system at larger scales. However, before the simulator can be used for applications with NAPLs, the shear-thinning behavior has to be incorporated into the multiphase modes and tested against more detailed laboratory experiments. This enhanced remedial amendment delivery approach may be applied to aquifer remediation sites where the emplacement of remedial amendments such as nutrients and reactive reagents to lower permeability zones is required. For example, when (poly)phosphate will be injected into the 300 Area aquifer of the Department of Energy Hanford Site in

40

L. Zhong et al. / Journal of Contaminant Hydrology 101 (2008) 29–41

Washington to form phosphate minerals (autunite and apatite) for uranium sequestration (Wellman et al., 2007), this technique may be applied to deliver the phosphate amendment to the lower permeability lenses. Because Xanthan gum is an inexpensive biopolymer for which only relative small concentrations are needed, the additional cost of using this delivery approach is small. Acknowledgements This study was performed under support provided by Pacific Northwest National Laboratory (PNNL) through the Laboratory Directed Research and Development (LDRD) program. PNNL is operated by the Battelle Memorial Institute for the Department of Energy (DOE) under Contract DE-AC0676RLO 1830. The intermediate-scale experiments were performed in the Environmental Molecular Sciences Laboratory (EMSL), a national scientific user facility sponsored by the DOE's Office of Biological and Environmental Research and located at PNNL. Scientists interested in conducting experimental work in the EMSL are encouraged to contact M. Oostrom ([email protected]). References Chauveteau, G., Kohler, N., 1984. Influence of microgels in polysaccharide solutions on their flow through porous media. SPE J. 24, 361–368. Craig, F.F., Sanderlin, J.L., Moore, D.W., Geffen, T.F., 1957. A laboratory study of gravity segregation in frontal drives. Trans. AIME 210, 277–280. Crane, F.E., Kendall, H.A., Gardner, G.H.F., 1963. Some experiments on the flow of miscible fluids of unequal density through porous media. Soc. Pet. Eng. J. 3, 277–280. Datta Gupta, A., Lake, L.W., Pope, G.A., Sepehrnoori, K., 1991. High-resolution monotonic schemes for reservoir fluid flow simulation. In Situ 15, 289–317. Domenico, P.A., Schwartz, F.W., 1990. Physical and Chemical Hydrogeology. John Wiley and Sons, Inc., NY. Dwarakanath, V., Kostarelos, K.S., Pope, G.A., Wade, W.H., 1999. Anionic surfactant remediation of soil columns contaminated by nonaqueous phase liquids. J. Contam. Hydrol. 38, 465–488. Flowers, T.C., Hunt, J.R., 2007. Viscous and gravitational contributions to mixing during vertical brine transport in water-saturated porous media. Water Resour. Res. 43, W01407. Freyberg, D.L., 1986. A natural gradient experiment on solute transport in a sand aquifer. 2. Spatial moments and the advection and dispersion of nonreactive tracers. Water Resour. Res. 22, 2031–2046. Grubb, D.G., Sitar, N., 1999. Mobilization of trichloroethene (TCE) during ethanol flooding in uniform and layered sand packs under confined conditions. Water Resour. Res. 35, 3275–3289. Hirasaki, G.J., Jackson, R.E., Jin, M., Lawson, J.B., Londergan, J., Meinardus, H., Miller, C.A., Pope, G.A., Szafranski, R., Tanzil, D., 2000. Field demonstration of the surfactant/foam process for remediation of a heterogeneous aquifer contaminated with NAPL. Part I. In: Fiorenza, S., Miller, C.A., Oubre, C.L., Ward, C.H. (Eds.), AATDF Monographs, NAPL Removal: Surfactants, Foams, and Microemulsions. Lewis Publishers, Boca Raton, FL. Jawitz, J.W., Annable, M.D., Rao, P.S.C., 1998. Miscible fluid displacement stability in unconfined porous media: two-dimensional flow experiments and simulations. J. Contam. Hydrol. 31, 211–230. Johansen, R.T., 1979. Overview of selected oil recovery processes. J. Rheol. 23, 167–179. Kimmel, A.E., Braids, O.C., 1975. Preliminary finding of a leachate study on two landfills in Suffolk County, New York. J. Res. U.S. Geol. Surv. 3, 273–280. Kimmel, A.E., Braids, O.C., 1980. Leachate plumes in groundwater from Babylon and Islip Land fill, Long Island, NewYork. U.S. Geol. Surv. Prof. Paper 1085. Kostarelos, K., Pope, G.A., Rouse, B.A., Shook, G.M., 1998. A new concept: the use of neutrally-buoyant microemulsion for DNAPL remediation. J. Contam. Hydrol. 34, 383–397. Lake, L.W., 1989. Enhanced Oil Recovery. Prentice-Hall Inc., Englewood Cliffs, NJ. LeBlanc, D.R., Garabedian, S.P., Hess, K.M., Gelhar, L.W., Quadri, R.D., Stollenwer, K.G., Wood, W.W., 1991. Large-scale natural gradient tracer

test in sand and gravel, Cape Cod, Massachusetts, 1. Experimental design and observed tracer movement. Water Resour. Res. 27, 895–910. Lenhard, R.J., Parker, J.C., 1987. A model for hysteretic constitutive relations governing multiphase flow. 2. Permeability–saturation relations. Water Resour. Res. 23, 2197–2206. Li, Y., Abriola, L.M., Phelan, T.J., Ramsburg, C.A., Pennell, K.D., 2007. Experimental and numerical validation of the total trapping number for prediction of DNAPL mobilization. Environ. Sci. Technol. 41, 8135–8141. Longino, B.L., Kueper, B.H., 1995. The use of upward gradients to arrest downward dense, nonaqueous phase liquid (DNAPL) migration in the presence of solubilizing surfactants. Can. Geotech. J. 32, 296–308. Lopez, X., Valvatne, P.H., Blunt, M.J., 2003. Predictive network modeling of single-phase non-Newtonian flow in porous media. J. Colloid Interface Sci. 263, 256–265. Lunn, S.R.D., Kueper, B.H., 1997. Removal of pooled dense, nonaqueous phase liquid from saturated porous media using upward gradient alcohol floods. Water Resour. Res. 33, 2207–2219. MacFarlane, D.S., Cherry, J.A., Gillham, R.W., Sudicky, E.A., 1983. Migration of contaminants in groundwater at a landfill: a case study, 1. Groundwater flow and plume delineation. J. Hydrol. 63, 1–29. Mackay, D.M., Cherry, J.A., 1989. Groundwater contamination: pump-andtreat remediation. Environ. Sci. Technol. 23, 630–636. Martel, K.E., Martel, R., Lefebvre, R., Gelinas, P.J., 1998. Laboratory study of polymer solutions used for mobility control during in situ NAPL recovery. Groundw. Monit. Rem. 103–113 Summer. Miller, C.T., Hill, E.H., Moutier, M., 2000. Remediation of DNAPL-contaminated subsurface systems using density-motivated mobilization. Environ. Sci. Technol. 34, 719–724. Neil, J.D., Chang, H.L., Geffen, T.M., 1983. Waterflooding and improved waterflooding. In: Poettmann, F.H., Bond, D.C., Hocott, C.R. (Eds.), Improved Oil Recovery. Interstate Oil Compact Commission, Oklahoma City, OK, p. 363. Oostrom, M., Hayworth, J.S., Dane, J.H., Güven, O., 1992a. Behavior of dense aqueous phase leachate plumes in homogeneous porous media. Water Resour. Res. 28, 2123–2134. Oostrom, M., Dane, J.H., Güven, O., Hayworth, J.S., 1992b. Experimental investigation of dense solute plumes in an unconfined aquifer model. Water Resour. Res. 28, 2315–2326. Pennell, K.D., Pope, G.A., Abriola, L.M., 1996. Influence of viscous and buoyancy forces on the mobilization of residual tetrachloroethylene during surfactant flushing. Environ. Sci. Technol. 30, 1328–1335. Ramsburg, C.A., Pennell, K.D., Kibbey, T.C.G., Hayes, K.F., 2004. Refinement of the density-modified displacement method for efficient treatment of tetrachloroethene source zones. J. Contam. Hydrol. 74, 105–131. Robert, T., Martel, R., Conrad, S.H., Lefevre, R., Gabriel, U., 2006. Visualization of TCE revovery mechanisms using surfactant–polymer solutions in a two-dimensional heterogeneous sand model. J. Contam. Hydrol. 86, 3–31. Saenton, S.T., Illangasekare, H., Soga, K., Saba, T.A., 2002. Effects of source zone heterogeneity on surfactant-enhanced NAPL dissolution and resulting remediation end-points. J. Contam. Hydrol. 59, 27–44. Schincariol, R.A., Schwartz, F.W., 1990. An experimental investigation of variable density flow and mixing in homogeneous and heterogeneous media. Water Resour. Res. 26, 2317–2329. Schroth, M.H., Ahearn, S.J., Selker, J.S., Istok, J.D., 1996. Characterization of miller-similar silica sands for laboratory hydrologic studies. Soil Sci. Am. J. 60, 1331–1339. Shook, G.M., Pope, G.A., Kostarelos, K., 1998. Prediction and minimization of vertical migration of DNAPLS using surfactant enhanced aquifer remediation at neutral buoyancy. J. Contam. Hydrol. 34, 363–382. Sudicky, E.A., Cherry, J.A., Frind, E.O., 1983. Migration of contaminants in groundwater at a landfill: a case study, 4. A natural-gradient dispersion test. J. Hydrol. 63, 81–108. Taylor, T.P., Pennell, K.D., Abriola, L.M., Dane, J.H., 2001. Surfactant enhanced recovery of tetrachloroethylene from a porous medium containing low permeability lenses: 1. Experimental studies. J. Contam. Hydrol. 48, 325–350. Taylor, T.P., Rathfelder, K.M., Pennell, K.D., Abriola, L.M., 2004. Effects of ethanol addition on micellar solubilization and plume migration during surfactant enhanced recovery of tetrachloroethene. J. Contam. Hydrol. 69, 73–99. Vermeul, V.R., Williams, M.D., Szecsody, J.E., Fruchter, J.S., Cole, C.R., Amonette, J.E., 2002. Creation of a subsurface permeable reactive barrier using in situ redox manipulation. In: Naftz, David L., et al. (Ed.), Chapter 6 in Handbook of Groundwater Remediation using Permeable Reactive Barriers: Applications to Radionuclides, Trace Metals, and Nutrients. Academic Press, San Diego, CA, pp. 163–192. Wellman, D.M., Pierce, E.M., Richards, E.L., Butler, B.C., Parker, K.E., Glovack, J.N., Burton, S.D., Baum, S.R., Clayton, E.T., Rodriguez, E.A., 2007. Interim Report: Uranium Stabilization Through Polyphosphate Injection - 300 Area Uranium Plume Treatability Demonstration Project. PNNL-16683. Pacific Northwest National Laboratory, Richland, WA.

L. Zhong et al. / Journal of Contaminant Hydrology 101 (2008) 29–41 White, M.D., Oostrom, M., 2006. STOMP Subsurface Transport Over Multiple Phases, Version 4.0, User's Guide, PNNL-15782. Pacific Northwest National Laboratory, Richland, Washington. Yang, X.Q., Erickson, L.E., Fan, L.T., 1994. Dispersive convective characteristics in the biorestoration of contaminated soil with a heterogeneous formation. J. Hazard. Mater. 38, 163–185. Zhang, H., Schwartz, F.W., 1995. Multispecies contaminant plumes in variable density flow systems. Water Resour. Res. 31, 837–847.

41

Zhong, L., Mayer, A.S., Glass, R.J., 2001. Visualization of surfactant-enhanced non-aqueous phase liquid mobilization and solubilization in a twodimensional micromodel. Water Resour. Res. 37, 523–538. Zhong, L., Mayer, A.S., Pope, G.A., 2003. The effects of surfactant formulation on nonequilibrium NAPL solubilization. J. Contam. Hydrol. 60, 55–75.