Quantitative assessment of electrical resistivity tomography for monitoring DNAPLs migration – Comparison with high-resolution light transmission visualization in laboratory sandbox

Quantitative assessment of electrical resistivity tomography for monitoring DNAPLs migration – Comparison with high-resolution light transmission visualization in laboratory sandbox

Accepted Manuscript Research papers Quantitative assessment of electrical resistivity tomography for monitoring DNAPLs migration – Comparison with hig...

2MB Sizes 0 Downloads 17 Views

Accepted Manuscript Research papers Quantitative assessment of electrical resistivity tomography for monitoring DNAPLs migration – Comparison with high-resolution light transmission visualization in laboratory sandbox Yaping Deng, Xiaoqing Shi, Hongxia Xu, Yuanyuan Sun, Jichun Wu, André Revil PII: DOI: Reference:

S0022-1694(16)30742-9 http://dx.doi.org/10.1016/j.jhydrol.2016.11.036 HYDROL 21650

To appear in:

Journal of Hydrology

Received Date: Revised Date: Accepted Date:

25 July 2016 4 November 2016 18 November 2016

Please cite this article as: Deng, Y., Shi, X., Xu, H., Sun, Y., Wu, J., Revil, A., Quantitative assessment of electrical resistivity tomography for monitoring DNAPLs migration – Comparison with high-resolution light transmission visualization in laboratory sandbox, Journal of Hydrology (2016), doi: http://dx.doi.org/10.1016/j.jhydrol. 2016.11.036

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Quantitative assessment of electrical resistivity tomography for monitoring DNAPLs migration – Comparison with high-resolution light transmission visualization in laboratory sandbox Yaping Denga, Xiaoqing Shia*, Hongxia Xua, Yuanyuan Suna, Jichun Wua*, André Revilb a

Key Laboratory of Surficial Geochemistry of Ministry of Education, School of Earth Sciences

and Engineering, Nanjing University, Nanjing 210023, China b

ISTerre, CNRS, UMR CNRS 5275, Université Savoie Mont-Blanc, 73376 cedex, Le Bourget du

Lac, France *

Corresponding author

Emails: [email protected]; [email protected]; [email protected]

Highlights: 

Quantitatively assess resistivity method by comparing with light transmission visualization



Resistivity method successfully monitors the migration of DNAPLs in slight heterogeneous media.



Resistivity tomography based on smoothness consistently overestimates the area of plume.



The resistivity increase caused by the low-saturation DNAPLs is slight.

1

Abstract Real-time monitoring of dense non-aqueous phase liquids (DNAPLs) migration and distribution is essential for the decision of an effective remediation strategy. Light transmission visualization (LTV) has shown its accuracy and efficiency for measuring DNAPLs saturation and water content in the laboratory, but it cannot be implemented in three dimensional sandbox or field-scale sites. Recently developed electrical resistivity tomography (ERT) has been applied in monitoring the migration and distribution of DNAPLs in bench- and field-scale studies. However, the evaluation of the ability of ERT for monitoring DNAPLs migration by a direct comparison of ERT with high-resolution techniques such as LTV within an experimental system is still lacking. Two sandbox experiments with different permeability conditions are conducted to quantitatively assess the capability of ERT for monitoring the DNAPLs migration. During the injections, LTV method is used to visualize the DNAPLs migration and provide high-resolution saturation data while ERT method is applied to capture the change of resistivity. The results from the comparison between LTV and ERT methods show that ERT is successful in detecting the accumulation and flow bypassing phenomenon around the low-permeability lenses, as well as the penetration through the high-permeability lenses. There is a fair correlation between the resistivity and saturation with overall correlation coefficients above 0.6, except at last stage. However, using classical regularization techniques (based on smoothness), the area of DNAPLs plume determined by ERT is commonly overestimated. Compared to the plume around the low-permeability lenses, the plume around the high-permeability 2

lenses estimated by ERT is more extensive due to larger resistivity contrasts. In addition, ERT measurements indicate that the resistivity increase caused by the low-saturation DNAPLs is not apparent enough, which is likely to be covered up under the changing hydrogeochemical environments in field investigations. Keywords: Chlorinated hydrocarbons, Groundwater monitoring, Electrical resistivity tomography, Light transmission visualization, Heterogeneity

3

1. Introduction The remediation of sites contaminated with organic compounds has been a common environmental problem. These organic contaminants may exist in the subsurface in the form of slightly soluble and volatile non-aqueous phase liquids (NAPLs), as a result of oil pipeline leakage, or as a solute dissolved in water as in the case of landfill leachate (National Research Council 2013; Newell et al. 2014). Because of their relatively low viscosity, low interfacial tension and greater density, dense NAPLs (DNAPLs) like trichloroethylene (TCE) and tetrachloroethylene (PCE) can migrate under the influence of gravitational, viscous and capillary forces until they get immobilized as residual blobs along the migration pathways or pooled on the low permeability media (Esposito and Thomson 1999; Lerner et al. 2003). In addition, changes of the wettability of soil system might occur in the mature NAPLs contaminated sites due to the formation of biopolymers as a product of the activity of microorganism (Ron and Rosenberg 2001; Cozzarelli et al. 2001), which probably make the NAPL trapped in smaller spaces and coat the surface of the minerals (Mercer and Cohen 1990), increasing the difficulty in the displacement of NAPLs (Dwarakanath et al. 2002). These entrapped DNAPLs will act as a long-term contamination source which poses an unacceptable threat to groundwater system and human health (Pankow and Cherry 1996; Yang et al. 2012). Characterizing and monitoring the transport and configuration of DNAPLs play an important role in the decision and implementation of remediation strategies (Saenton et al. 2002; Yoon et al. 2008; Triplett Kingston et al. 2010). Conventional 4

site investigations have mainly relied on a sparse network of intrusive drilling and sampling approaches, which can only provide discrete vertical profiles (Griffin and Watson 2002; Peter et al. 2008). Recently, geophysical methods have been increasingly applied in hydrogeology as it can provide non-intrusive and continuous profiles (see reviews in (Revil et al. 2012; Loke et al. 2013; Binley et al. 2015). Electrical resistivity tomography (ERT) has been recognized as an attractive approach to image the subsurface process because of its sensitivity to relative changes in the saturation of the water phase (Binley and Kemna 2005) and its wettability (Revil et al. 2011). ERT has been applied to the monitoring of soil water content (Lehikoinen et al. 2009), saline water intrusion (de Franco et al. 2009) and tracer tests (Perri et al. 2012). Several investigations of DNAPLs with ERT have been conducted in sandbox experiments (Weller et al. 1996) and field-scale sites (Cardarelli 2009; Naudet et al. 2014). Nevertheless, the application of ERT is limited by the static detection due to the geological heterogeneities, which can produce resistivity variations sometimes stronger than those associated with the presence of the oil itself. With the development of time-lapse inversion algorithms, ERT has exhibited great potential in monitoring subsurface changes (Chambers et al. 2004; Breen et al. 2012; Power et al. 2014; Orlando and Renzi 2015). Qualitative evaluations have revealed that inverted resistivity tomography is able to track the evolution of plume with digital photos taken on the side of the tank (Mao et al. 2015; Hort et al. 2015). However, the uncertainty in the interpretation of resistivity anomalies still exists in the application of ERT. For instance, Chambers et al. (2004) observed that differential time-lapse 5

resistivity images revealed relatively weak anomalies associated with the transport of DNAPLs, which could only be seen during initial stages of their experiment. Also low resistivity anomalies were detected by ERT in mature NAPLs contaminated zones (Delgado-Rodríguez et al. 2014). Thus, further quantitative evaluation of the ability of ERT for characterizing DNAPLs source zone in bench-scale experiments, which allows quantitative measurement of DNAPLs saturation, is still required. The light transmission visualization (LTV) method has proved its accuracy and efficiency in measuring DNAPLs saturation and water content in laboratory studies. Bob et al. (2008) explored a modified LTV method to measure the undyed and dyed DNAPLs saturation in two-dimensional (2-D), two fluid phase systems. Strong correlations were found between the results obtained from this method and the known amounts of DNAPLs with an R2 value of 0.993 and 0.999 for the undyed and dyed DNAPLs, respectively. Grant et al. (2007) demonstrated that LTV method successfully tracked the migration and distribution of DNAPL in a heterogeneous porous medium. However, LTV requires that the porous medium must be translucent, which limits its application in natural conditions (Niemet and Selker 2001). The overall objective of this study is to assess the efficiency of ERT monitoring of DNAPLs migration in slightly heterogeneous media through an accurate and quantitative method (LTV). Two experiments are conducted in a sandbox. The first experiment is designed to investigate the accumulation and flow bypassing around the layered low-permeability lenses, while the second is used to investigate the process of DNAPLs penetrating through the high-permeability lenses. During the TCE 6

infiltration process, the LTV technique is applied to directly measure the TCE saturation and ERT system is used to map the distribution of resistivity in sandbox. Direct comparison between ERT and LTV imaging results is obtained to provide a quantitative measure of the capability of ERT to characterize DNAPLs migration in heterogeneous media. 2. Sandbox experiment

2.1 Experimental setup and materials

The sandbox used in this study, is 60 cm in length, 45 cm in height and 2.5 cm in width (Figure 1). It consists of an acrylic frame mounted with two stainless-steel aluminum external frames to avoid the deformation of the structure and two glass panes. Three ports spaced 11 cm apart are cut out of each side of the interior frame as inlets and outlets, respectively. Fluids are allowed to flow through the sandbox using high-precision peristaltic pumps (Niemet and Selker 2001). The contaminants (DNAPLs) are injected into the sandbox through an injection needle (3.5 cm below the top side of sandbox) to simulate the leakage of DNAPLs into subsurface. The DNAPLs used in this experiment is TCE, which is a typical substance of organic contaminated sites (Rothmel et al. 1998). Its properties are specified in Table 1. The TCE is dyed with Oil-Red-O at a concentration of 0.05 g/L to obtain a better view of its migration. Like most NAPLs, DNAPLs are generally highly resistive after their injection in a porous material (Lucius et al. 1992), which provides the feasibility for the application of ERT in TCE detection. Note that, however, the presence of biofilm 7

in NAPLs/DNAPLs sites may change the observed resistivity of these plumes from being resistive to conductive (Sauck et al. 1998; Atekwana and Atekwana 2010; Atekwana and Abdel Aal 2015). A total of 208 electrodes are installed at the surface of glass pane (Figure 1c). The electrodes are made of brass with 3 mm in diameter and 26 mm in length and only inserted 5 mm into the porous media. They are arranged in 13 parallel survey lines, placed 3 cm apart both in horizontal and vertical direction (Figure 1d). A total of 325 resistivity measurements are taken for each survey. Although this electrode array may be not representative of real conditions, it provides a relatively accurate measurement of resistivity data to study the mechanism of DNAPLs migration in heterogeneous media in laboratory research. The sandbox was filled with Accusand silica sands (Unimin Corporation, USA), which has been used in similar laboratory experiments (Saenton and Illangasekare 2013; Petri et al. 2015), with various grain sizes to create a heterogeneous porous structure as explained below. All sands were sequentially soaked with dilute nitric acid, rinsed with deionized water to remove fine particulates that might have a negative influence on LTV measurements, and then oven-dried at 45℃ for 48 hours, however, the sands were not sterilized, following the procedures of Schroth et al. (1996); Niemet and Selker (2001); Bob et al. (2008). The 40/50 mesh size sands were used as background porous media and 20/30, 60/70, 100/140 mesh size sands were filled as layered lenses within the background media to mimic the existence of heterogeneous features that may exist in field sites. The sandbox was filled by 8

stepwise pouring approximately 1 cm layer of sands through a funnel, and then tapping the glass pane with a wood hammer to achieve a compact pack. For the layered lenses, three aluminum strips were set at designed locations to avoid the mixing with background media, and then 20/30, 60/70, 100/140 mesh size sands were introduced into the space between these strips through a funnel. In addition, two fine grain sand layers (using 100/140 mesh size sands) with a thickness of 2 cm were placed at the bottom and top of the background media to avoid possible escaping from the bottom ports and backflow of DNAPLs through injection needle. Two heterogeneous porous structure with different permeability of layered lenses were constructed in the sandbox, labeled as Test 1 and Test 2, respectively. The mean grain size of sands is used as an indicator for the sand permeability. The threshold value for defining the low-and high-permeability sand is 0.362 mm corresponding to the background media (40-50 mesh sand) used in this study. The sand with a mean grain size larger than 0.362 mm corresponds to the high-permeability sand, while the sand with a mean grain size less than 0.362 mm is the low-permeability sand. The properties of the sands are specified in Table 2. Both tests used 40/50 mesh size sand as background media but different mesh size sands for the lenses. For Test 1, 100/140 mesh size sand were used to construct layered lenses to simulate the low-permeability case (Figure 1a). For Test 2, coarse grain sand (20/30 mesh size sand) were used to create the high-permeability layered lenses (Figure 1b). The LTV system consists of a thermoelectrically air-cooled charge-coupled device (CCD) camera (AP2E, Apogee Instruments, Auburn, California, USA), a light 9

source consisting of 6 fluorescent tubes (Panasonic, YZ18RR6500K) and a sandbox, to provide a high-resolution TCE saturation data. The CCD camera is kept fixed in a wooden enclosure to ensure that it could capture the light filtering through the sandbox from the light source. The sandbox is placed between the CCD camera and light source. The resistivity measurements are collected with a single-channel geo-resistivity meter (Geopen, made in China, http://www.geopen.net/). The electrode configuration used in this experiment is dipole-dipole, which is a commonly used array in similar laboratory tank experiments (Wang et al. 2010; Power et al. 2014; Orlando and Renzi 2015).

2.2 Experimental procedure

After the packing of sandbox, a peristaltic pump was used to pump salt water into the sandbox at a constant flux to ensure the porous media saturated and keep a constant inflow and outflow through the sandbox. The salt water consisted of a 0.01 mol/L solution of NaCl with a resistivity of 8.8 Ω·m at 20°C. After that, the initial state was recorded by the geo-resistivity meter and CCD camera simultaneously. The room temperature was kept constant (20 ± 2°C) controlled by air-conditioner during the experiment to avoid the possible influence of temperature variation on the measurement. During the infiltration experiment, 40 ml of TCE was injected into the sandbox at a constant rate of 0.5 ml/min through the injection needle using a Harvard Apparatus syringe pump and glass syringe (Sigma-Aldrich Corp., St. Louis, Missouri, USA). Both LTV and ERT methods were used to monitor this infiltration process. For

10

LTV measurements, saturation data was acquired at an interval of 1 minute continuously. For ERT measurements, the apparent resistivity data was collected at 5 times, 20 minutes, 40 minutes, 60 minutes, 80 minutes and 690 minutes for Test 1 and 20 minutes, 40 minutes, 60 minutes, 80 minutes and 705 minutes for Test 2. The data was obtained line by line from top to bottom. During the resistivity measurement, the number of survey lines was adjusted based on the position of TCE plume monitored by LTV method. The AGI 3D EarthImager (Advanced Geosciences Incorporation, USA) was then used to combine these parallel survey lines to form an integral view of the distribution of resistivity in the whole sandbox (Chambers et al. 2002; Thabit and Khalid 2016; Kim et al. 2016). The resistivity acquisition system recorded 325 measurements for each survey at a time. For each measurement, a 30 (±5) mA current was injected into the porous media and a 1 s duration of the injection current was chosen to obtain a better quality of the data. In order to reduce the total acquisition time and track the movement of DNAPL, only 1 stack was used during the ERT measurement. The resulting collection time was approximately 10 minutes on average, including the data acquisition time and the time needed for transition from line to line.

2.3 Electrical resistivity tomography

To interpret the apparent resistivity data obtained by ERT measurements, a forward problem is solved by using finite-difference to calculate the potential values. Then inverse methods are then used to back calculate the resistivity using the previously obtained potential values. An optimization method is applied to minimize

11

the difference between the measured and calculated resistivity until a convergence criteria is reached (Loke et al. 2013; Binley 2015). It is known that 2D and 3D inversion methods are based on different assumptions. In 2D inversion program, the energy is assumed to be dissipated only in the X-Z plane (below the survey line). While in 3D inversion program, the resistivity is allowed to have a three-dimensional distribution. For the present electrodes layout, the distribution of currents is inherently three-dimensional even though the acquisition line is a 2D profile. Though a full 3D inversion does offer a better information of subsurface, it is time consuming and computational demanding. A pseudo 3D survey (by combining several parallel 2D lines and using a 3D inversion approach) is able to provide reasonable results with dense in-line spacing (Bentley and Gharibi 2004; Yang and Lagmanson 2006). For each survey line, the value of factor ‘n’ (an integer for measurements of multiple distance of minimum electrode spacing) was fixed at 2, to avoid measuring beyond the boundary of the tank (with a width of 2.5 cm). A sensitivity analysis based on the present electrodes arrangement and measurement schedule showed that the overall sensitivity decreased towards the bottom of the image plane. The effect of the insulating boundaries on the resistivity measurement was small because of the relatively low sensitivity in the region close to the edge of the tank. The difference inversion approach, which is popular for solving time lapse problems (Miller et al. 2008; Breen et al. 2012; Seferou et al. 2013), is used for the data inversion here. Sequential inversion algorithm uses the background data as the 12

reference model and inverts for the differences between the background and subsequent data sets (Labrecque and Yang 2001). It converges fast and can resolve small changes of conductivity over time. The “Difference inversion” module in AGI EarthImager 3D is utilized for inverting the apparent resistivity data obtained from combined parallel 2D lines. Firstly, the background data is inverted to construct a reference model for the sequential inversion. During the apparent resistivity inversion, a homogeneous model is used as the starting model without providing any prior information regarding the porous media heterogeneity. After that, a difference inversion between the monitored and reference data is carried out using the same command file.

2.4 Light transmission visualization Based on the Beer’s law (Ryer and Light 1997) and Fresnel’s law (Griffiths 1989), LTV method can quantify the absorption and interfacial refraction of light between different phases for the total thickness of the porous medium (Niemet and Selker 2001). In oil-water two-phase system which is the case in this study, oil saturations can be calculated by the following equations (Bob et al. 2008): Is

Iin

2K pw

(1)

e(- p dp kp )

where Is is the light intensity under water saturated condition, Iin is the incident light intensity, τpw is the transmission factor at the sand particles-water interface, K is the number of pores across the thickness of the sandbox, αp is the sand particles absorption coefficient, dp is the diameter of sand particles, kp is the number of sand

13

particles across the thickness of the sandbox. The expression for Ioil which is the light intensity under oil saturated condition: I oil

Is

2K wo

e(

(2)

d K)

o o

where τwo is the transmission factor at the water-oil interface, αo is the dyed oil absorption coefficient, do is the average pore diameter. Then the oil saturation can be written as: So

(ln I s ln I ) (ln I s ln I oil )

(3)

where So is the oil saturation and I is the emergent light intensity. More detailed illustrations on calculating procedures can be found in Bob et al. (2008). 3. Results In order to clearly mask the resistive anomaly aroused by the presence of TCE, the percent differences of resistivity between the time lapse and background models are mapped. High sensitive region of the image plane below the survey line is considered to evaluate the ability of ERT method. The sensitivity analysis shows that the apparent resistivity is relatively sensitive to the region close to the electrodes. Therefore, slices of resistivity distribution at 0.61 cm below the electrodes plane are extracted for the following analysis.

3.1 Temporal comparison between ERT and LTV images

Figure 2 shows a comparison between the LTV and ERT data with the presence of low- and high-permeability lenses, respectively. A threshold value of 10 percent differences of resistivity (determined by trial and error) is applied to the resistivity 14

tomography to separate the background media from TCE plumes. The results show that ERT is able to capture the real-time movement of the DNAPLs plumes and also reveal the different migration patterns under different permeability conditions. For the experiment with relatively low-permeability lenses (Test 1), the results show that the resistivity images are able to reflect the pooling and flow bypassing behavior around the low-permeability lenses. Both ERT and LTV method images consistently show that the TCE plume reaches the first layer at the first time step (t1) and then accumulates on the first layer and moves around it to reach the second layer (t2) (Figures 2a and b). At t3=60 and t4=80 minutes, the saturation images indicate that the TCE plume accumulates on the all three layered lenses which are shown as remarkable resistive anomalies in corresponding regions in resistivity images (Figures 2c and 2d). At last time step t5=690 minutes, the TCE plume pools on the bottom of the sandbox (Figure 2e). It is worth noting that the change of resistivity above lenses does not correspond well with saturations, which will be discussed further in the next section. For the experiment with high-permeability lenses (Test 2), the ERT and LTV images consistently reflect the process of TCE penetrating through the high-permeability lenses (Figure 2). At initial stage, the saturation and resistivity images show similar responses to Test 1 (Figure 2f). During later injections, the ERT image exhibits that TCE permeates into and fills in the whole high-permeability lens, which is in accordance with the saturation image (Figure 2g). Afterwards, both ERT and LTV images show TCE migrates through the high-permeability lens to reach the 15

bottom of the sandbox (Figures 2h and 2i). It should be noted that the area of resistive anomaly is significantly more extensive than the actual size of the plume in the region of high-permeability lenses filled with TCE. Similar with Test 1, at last time step t5=705 minutes, there is an overall decrease in resistivity in the whole cell though marked resistive anomalies are observed around the high-permeability lens and at the bottom of the sandbox. Figure 2 demonstrates that ERT successfully locates the position of the plume and depicts the evolution of its general shape. It is able to monitor the pooling and flow bypassing behavior of TCE plume when encountered the low-permeability media, as well as penetration through the high-permeability lens during the migration process. However, the area of the plume detected by ERT is more extensive than LTV method. In addition, both experiments show decreased resistivity in the whole area at last stage.

3.2 Statistical analysis

The plots of measured TCE saturation versus corresponding resistivity ratio for Test 1 and Test 2 are presented in Figure 3. The resistivity ratio is calculated by subtracting the log of the resistivity (non-thresholded resistivity data is used here). As shown in Figure 3, the correlation coefficient is not very good at the end of the experiment which can be expected from aforementioned qualitative analysis. Therefore, the data at time t5 is only inserted on the side of corresponding figure to obtain a better understanding of the relationship between the resistivity and saturation

16

at other times. For Test 1 (Figure 3a), there is an apparent liner trend connecting the resistivity and saturation. The correlation coefficients are larger than 0.6 which indicates a strong correlation between the saturation and resistivity (except for the last time step t5). Figure 3b presents the average saturation of TCE and resistivity ratio for the whole source zone. As expected, the resistivity increases with increasing saturation. Figure 3c highlights the effects of high-permeability lens which is highly saturated with TCE on the correlation between resistivity and corresponding saturation. The experiment with low-permeability lenses apparently gives better correlation than that with high-permeability lenses. As shown in Figure 3c, amounts of datum points deviate from the main trend that connects resistivity and saturation at the third and fourth time step (time steps t3 and t4). During this period, the high-permeability lens is filled with TCE (Figures 2h and 2i), which leads to a larger resistivity contrast on the edge of the TCE plume. Olayinka and Yaramanci (2000) showed that model misfit increased as the resistivity contrast increased. This probably results in the apparent decrease in corresponding correlation coefficients. Figure 3d indicates that resistivity increases with increasing average saturation. Figure 3 suggests that the resistivity correlates well with saturation for the correlation coefficients are generally larger than 0.6, except for the last stage t5. With TCE saturation less than 0.4, the resistivity increase aroused by the presence of TCE is smaller than expected for the pure TCE is generally highly resistive (~ 106 Ω·m). This is probably one of the reasons that ERT fails to detect the existence of DNAPLs 17

in contaminated sites since such small increase of resistivity may be covered up by the changing hydrogeochemical environment due to microbial activity (Cassidy et al. 2001; Atekwana et al. 2004). For more detailed discussions can be seen in next section. In addition, better correlation between the resistivity and saturation for the low-permeability case is observed when compared with the high-permeability case. This quantitative analysis of the correlation between resistivity and saturation supports the qualitative observations obtained from Figure 2. Figures 4 and 5 present the variations of saturation and resistivity change of TCE in depth in the assumed boreholes BH1 and BH2, which are located 21.75 cm and 15.75 cm (Figure 2) away from the left side, respectively. For the experiment with the low-permeability lenses, the curves of resistivity rise up to a peak at the depth around 4.5 cm from the top, which corresponds to the accumulation of TCE on the first low-permeability lens. During the TCE infiltration process, both resistivity and saturation increase over time though a few irregular datum points are observed (Figures 4a, 4b and 4c). Two peaks are observed in Figure 4d, which are located at the depth around 4.5 cm and 22.5 cm, respectively. Similarly, these two peaks correspond to the accumulation of TCE on the first and third low-permeability lenses, respectively. For the experiment with high-permeability lenses, both curves have two peaks which are situated at the depth around 7.5 cm and 16.5 cm, respectively (Figures 5a and 5b). The first peak indicates the accumulation of TCE on the first low-permeability lens and the second refers to the filling in the high-permeability lens 18

(Figures 2g and 2h). There is a good agreement between resistivity and saturation curves for the first and second peak, while the resistivity curve shows some deviations from the saturation curve in the deepest part of the sandbox (Figure 5c). The resistivity curves display similar trends with the saturation curves, again quantitatively indicating good correlation between the resistivity and corresponding saturation. The peaks in resistivity curves always correspond to a relatively high TCE saturation which can be caused either by an accumulation on the low-permeability lens or filling in the high-permeability lens.

3.3 Estimation of DNAPLs saturation

Electrical conductivity of a partially saturated porous material (water being the wetting phase) is described by the sum of a bulk conductivity term and a surface conductivity term, which is also saturation-dependent (Waxman and Smits 1968; Revil 2012). This second contribution is associated with electrical conduction in the electrical double layer coating the surface of the grains. Usually for sands and clean sands and clean sandstones, surface conductivity is very small and can be neglected for the salinity used in the present study. With this assumption, we therefore estimate the saturation, SnERT , using the second Archie equation (Archie 1942): SnERT 1 (

o 1/ n

(4)

)

t

where

o

is the initial resistivity of the water-saturated medium (background

resistivity),

t

is the real-time ERT measured resistivity of partially saturated medium,

n is saturation exponent, commonly close to 2 for clean sand (clay-free) (Archie 1942; 19

Hearst et al. 2000). Longeron et al. (1989) found that the value of n depended on the saturation range and direction of water saturation variation. The experimental results showed that n equaled to 2.08 for the water saturation ranging from 0.28 to 1 during drainage (water saturation decrease) for sandstone in an oil/water system. Ulrich and Slater (2004) also observed that the saturation exponent ranged between 1.1 and 2.7 for individual sample. Therefore, we used n=2.0 in this experiment. Archie’s law has gained its popularity in the application of oil contamination problem since its formulation in 1942. Chambers et al. (2004) and Power et al. (2014) used Archie’s law (with saturation exponent n=2.0) to estimate the residual DNAPL saturation and remediated DNAPL volume reasonably well. Though the surface conduction is not considered in this study, it probably plays an important role in the ERT measurement for field investigations, especially with the presence of clays (Monego et al. 2010). Figures 6 and 7 show the comparison between calculated saturations from ERT data (thresholded data is used here, corresponding to the distribution of resistivity in Figure 2) and measured saturations from LTV data obtained from Test 1 and Test 2, respectively. It can be observed that both experiments overall underestimate the TCE saturation especially at early stages, but successfully locate the primary plume and resolve its general shape. However, both of the two experiments fail to resolve the plumes at final stage. In addition, the area of the calculated plume is slightly more extensive than that of the actual size of the plume. For Test 1, the configuration of TCE plume is relatively complicated with the presence of low-permeability lenses. Regularization used for the resistivity tomography imposes smoothness over the final 20

resistivity tomogram. As a consequence, some of the details in the oil fingering may be lost. In other words, some of the details in the oil patches can be merged together in the final resistivity tomogram. For Test 2, the calculated saturation images exhibit the evidence of TCE infiltrating into the high-permeability lens and flowing through it afterwards. The most significant observation is the remarkable overestimation of saturation and area of TCE plume around the high-permeability lens, which can be attributed to the effects of relatively high resistivity contrast on the inversion for TCE plume in corresponding region (Olayinka and Yaramanci 2000). Comparison between the calculated and actual volume of injected TCE provides a way for quantifying the estimation of TCE saturations in Figures 6 and 7. Figure 8 shows the full-field volumes calculated by LTV and ERT methods plotted against the actual injected volumes for Test 1 and Test 2. It can be observed that the calculated volumes of full-field plume using ERT data roughly match the volumes of injected TCE measured by LTV method for both Test 1 and Test 2, suggesting the calculated TCE saturation should be credible. For Test 1, the volumes calculated by ERT are very close to the volume results by LTV, though an obvious overestimation is observed at 30 ml injected (t3). This may be caused by the apparent overestimation of TCE plume area (see in Figure 6c). For Test 2, the volumes calculated by ERT are underestimated when compared with the volumes by LTV at early stages due to the smoothing of the resistivity model in inversion scheme. However, the calculated volumes by ERT are approximate to the actual volumes at later injections. This could be a consequence of both underestimation of TCE saturation and overestimation of 21

plume area (as previously shown). 4. Discussions Based on the difference of electrical properties of the subsurface, ERT is able to identify different materials. As it provides a noninvasive, continuous and rapid, as well as relatively low-cost way to investigate the spatial distribution of the subsurface, it is now applied for environmental problems, such as the survey of organic contamination. However, the results from the inversion of ERT data can be uncertain and non-unique, especially in heterogeneous media. Therefore, further evaluation of the ability of ERT for characterizing DNAPLs source zone is extremely necessary. This study provides an experimental evaluation of ERT method through comparison with LTV method. The result shows that ERT is successful in capturing the real-time movement of DNAPLs and also well reveals the different migration patterns under different permeability conditions in a sandbox. Pooling and flow bypassing phenomenon of TCE plume are observed through ERT in the experiment with low-permeability lenses. Also, ERT reproduces the process of TCE penetrating through the high-permeability lens. Statistical analysis suggests a good correlation between the resistivity and saturation with overall correlation coefficients above 0.6, except for the last stage. The correlation coefficient can be improved if any prior information is available for the inversion. In this study, in order to keep the inversion unbiased by such information that may also not be available in field condition, the experimental data are inverted without any prior information. Although the ERT method successfully characterizes the migration of TCE in slightly heterogeneous 22

media, however, there are still some problems that are relevant to practical applications deserving further discussions. ERT consistently overestimated the area of primary plume especially in the presence of high-permeability lenses (Figures 2 and 6~7). The reasons for this are likely to be threefold. For one thing, as previously mentioned, LTV method is a relatively accurate technique with high spatial resolution. In this case, the spatial resolution of LTV method is 0.04-0.05 cm, while that of ERT is 1.5 cm. As a result, the images obtained by ERT are rougher than LTV. The spatial resolution of ERT mainly depends on the electrode spacing and measurement scheme, which suggests that improved electrode array design is necessary to enhance the resolution of the field-scale ERT measurements. For another, the inversion algorithm used in this study is based on a smoothness constraint, which has been widely applied in 2-D and 3-D resistivity inversion models for its good convergence and stability (Binley 2015). However, such inversion approach may not work well when large contrasts in resistivity exist in the subsurface (de Groot-Hedlin and Constable 2004). For example, it is unable to correctly identify the edges of TCE plume with large resistivity contrasts, and provides a more extensive area of plume. To enforce sharp contrasts in the inversion, one way is to use a L1 norm optimization method (Loke et al. 2003; Johnson et al. 2010). Thus, the low- and high-permeability experiments are also inverted using L1 norm optimization method as shown in Figure 9. However, the inversion results using a L1 norm do not show a clearly superior resolution than the present results using a L2 norm (Figure 2). In particular, the present inversion scheme 23

shows better consistence at initial stages. This is probably because both smooth and sharp boundaries are present during the evolution of plumes which is usually the case in most field investigation. In order to enhance the interpretation of inversion results, the image guided inversion based on GPR reflections (e.g. (Zhou et al. 2014) will be used for the problem in the future. This technique requires more prior information than the present technique. In addition, ERT method is sensitive to the entire fluid, including water, both free and dissolved phases of DNAPLs (Orlando and Renzi 2015), while LTV is more sensitive to the free DNAPLs. This may also lead to the overestimation of the area of plume measured by ERT when contrasted with LTV. Though there is a good fit between the resistivity and saturation images at earlier stages, the resistivity images are of relatively poor qualities in the end of the experiment (Figure 2). As shown in Figures 2e and 2j, the resistivity images only characterize the primary anomalies with high TCE saturation and the average value of resistivity is lower than that of the previous stages. Orlando and Renzi (2015) also observed a similar phenomenon which indicated a decrease in the resistivity within the whole cell after contaminant release. Since the air-conditioner has been opened to control the room temperature, this decrease is hardly caused by the variation of temperature. Though the wettability alteration of soil system may lead to the change of its petrophysical properties, the wettability of the porous media is not likely to change significantly during the experiment. Firstly, the clean quartz sands used in the experiments are commonly considered to be strongly water wet due to the hydrophilic nature of silica sand particles. Secondly, while the change of soil wettability may 24

occur following NAPLs spills to the subsurface due to the polar compounds in oil that sorb strongly to charged mineral surfaces (Dwarakanath et al. 2002; Al-Raoush 2009) or biofilm formation or bio-surfactant adsorption as a product of microbial activities(Karimi et al. 2012; Sarafzadeh et al. 2013), the work from Powers et al. (1996) showed that the extent of change in the wettability of the quartz surfaces was strongly correlated to the composition of the organic phase and neat solvents, such as the pure TCE we used here did not have a significant impact on the wettability of quartz sands. Besides, wettability study of alteration of sandstone showed that bacteria solution could only change the originally water wet cores to a state of wettability that was slightly less water wet than the initial wettability (Afrapoli et al. 2009; Zargari et al. 2010). Therefore, the significant decrease of resistivity values cannot be aroused by the wettability alteration of porous media. We thus attribute this to the possible biodegradation or photolysis of TCE, which stimulates chemical reactions and increases the conductivity of fluid in porous medium (Acworth 2001; Li et al. 2004; Aal et al. 2006). This phenomenon apparently increases the difficulty in mapping the spatial configuration of organic plume over long-term investigations. Therefore, it is necessary to introduce one more geophysical techniques, such as induced polarization (IP) method which is sensitive to the properties at pore-scale, in DNAPLs contaminated sites. While the DNAPLs appears resistive in the experiments, the increase of resistivity caused by low-saturation DNAPLs is not very significant in this study. For example, the TCE plume with a saturation of approximate 0.3 only results in a 0.25 25

resistivity ratio ( equals to around 75 % increase of resistivity) for Test 1 though pure DNAPLs is rather resistive (~106 Ω·m). Schneider and Greenhouse (1992) estimated the bulk resistivity of a sandy aquifer with a porosity of 40 % as a function of DNAPLs saturation. The results indicated that the bulk resistivity increased very slowly with DNAPLs saturation below a DNAPLs saturation of ~ 45 %. However, common practice for ERT survey of DNAPLs contaminated sites is to seek highly resistive zones. Therefore, such small increase of resistivity is probably omitted by investigators or covered up under changing hydrogeological conditions, which is partly responsible for the failure of DNAPLs detection in field-scale survey. For example, Sauck et al. (1998) and Atekwana et al. (2000) found that the regions contaminated with free / residual LNAPL showed low resistivity instead of high resistivity in a field site. The modification of local hydrogeochemical conditions aroused by the biodegradation of LNAPL in the subsurface was likely the primary cause. Thus, in order to accurately characterize the spatial distribution of DNAPLs plumes, it is necessary to construct a model which considers the variations of hydrogeochemical conditions due to the changes in pore fluids chemistry, pore configuration, mineral surface area related to microbial activities in mature contaminated sites (see reviews in (Atekwana and Slater 2009; Atekwana and Atekwana 2010).

.

5. Conclusions The study directly visualizes the DNAPLs migration through LTV method and quantitatively evaluates the capability of ERT for characterizing the transport of 26

DNAPLs in heterogeneous media. Two sandbox experiments are conducted to observe the DNAPLs migration in porous media with layered low-permeability lenses and high-permeability lenses, respectively. The results show that ERT successfully display the pooling and flow bypassing phenomenon in the experiment with low-permeability lenses and track the process of TCE penetrating through the high-permeability lenses. Besides, the correlation between the resistivity and saturation data is strong (with overall correlation coefficients above 0.6) during injections. However, the area of DNAPLs plume estimated by ERT is more extensive than actual one, particularly with the presence of high-permeability lenses. In addition, there is a significant decrease in average resistivity in the whole sandbox at last stage, which is probably caused by the changes of chemical environment of pore water due to the biodegradation or photolysis of DNAPLs. Though the DNAPLs is highly resistive, the increase of resistivity aroused by the presence of low-saturation DNAPLs is slight, which is likely to be covered up under complex hydrogeological conditions. This laboratory experiment exhibits the ability of ERT to monitor the DNAPLs migration in heterogeneous media with difference inversion method. However, in order to improve the accuracy of ERT monitoring of DNAPLs transport and establish a reliable petrophysical relationship between resistivity and saturation, the variation of bulk conductivity should be taken into consideration. Acknowledgements This work was financially supported by the National Nature Science Foundation 27

of China grants (No. U1503282, 41030746, and 41172206). References Aal, G., Slater, L., Atekwana, E., 2006. Induced-polarization measurements on unconsolidated sediments from a site of active hydrocarbon biodegradation. Geophysics 71(2): H13-H24. http://dx.doi.org/10.1190/1.2187760. Acworth, R., 2001. Physical and chemical properties of a DNAPL contaminated zone in a sand aquifer. Q. J. Eng. Geol. Hydroge. 34(1): 85-98. Afrapoli, M.S., Crescente, C., Alipour, S., Torsaeter, O., 2009. The effect of bacterial solution on the wettability index and residual oil saturation in sandstone. J. Petrol. Sci. Eng. 69: 255-260. http://dx.doi.org/10.1016/j.petrol.2009.09.002 Al-Raoush, R.I., 2009. Impact of wettability on pore-scale characteristics of residual nonaqueous

phase

liquids.

Environ

Sci

Technol.

43:

4796-4801.

http://dx.doi.org/10.1021/es802566s Archie, G.E., 1942. The electrical resistivity log as an aid in determining some reservoir characteristics. Trans. Am Ind Min Metall Pet Eng. 146(01): 54-62. http://dx.doi.org/10.2118/942054-g. Atekwana, E.A., Abdel Aal, G.Z., 2015. Iron biomineralization controls on geophysical signatures of hydrocarbon contaminated sediments. J. Earth. Sci. 26(6): 835-843. http://dx.doi.org/10.1007/s12583-015-0611-2. Atekwana, E.A., Atekwana, E.A., 2010. Geophysical signatures of microbial activity at hydrocarbon contaminated sites: A Review. Surv. in Geophys. 31(2): 247-283. http://dx.doi.org/10.1007/s10712-009-9089-8. 28

Atekwana, E.A., D. Dale Werkema, J., Duris, J.W., Rossbach, S., Atekwana, E.A., Sauck, W.A., Cassidy, D.P., Means, J., Legall, F.D., 2004. In‐situ apparent conductivity measurements and microbial population distribution at a hydrocarbon



contaminated

site.

Geophysics

69:

56-63.

http://dx.doi.org/10.1190/1.1649375 Atekwana, E.A., Sauck, W.A., Werkema Jr, D.D., 2000. Investigations of geoelectrical signatures at a hydrocarbon contaminated site. J. Appl. Geophys. 44(2–3): 167-180. http://dx.doi.org/10.1016/S0926-9851(98)00033-0. Atekwana, E.A., Slater, L.D., 2009. Biogeophysics: A new frontier in Earth science research. Rev. Geophys. 47: 1-30. http://dx.doi.org/10.1029/2009RG000285. Bentley, L.R., Gharibi, M., 2004. Two-and three-dimensional electrical resistivity imaging at a heterogeneous remediation site. Geophysics 69(3): 674-680. http://dx.doi.org/10.1190/1.1759453. Binley, A., 2015. 11.08–Tools and Techniques: Electrical Methods. In: Treatise on Geophysics, 2nd edn. Elsevier, Oxford, pp. 233-259. Binley, A. et al., 2015. The emergence of hydrogeophysics for improved understanding of subsurface processes over multiple scales. Water Resour. Res. 51(6): 3837-3866. http://dx.doi.org/10.1002/2015WR017016. Binley, A., Kemna, A., 2005. DC Resistivity and Induced Polarization Methods. In: Rubin, Y., Hubbard, S.S. (Eds.), Hydrogeophysics. Springer Netherlands, Dordrecht, pp. 129-156. http://dx.doi.org/10.1007/1-4020-3102-5_5.

29

Bob, M.M., Brooks, M.C., Mravik, S.C., Wood, A.L., 2008. A modified light transmission visualization method for DNAPL saturation measurements in 2-D models.

Adv.

Water

Resour.

31(5):

727-742.

http://dx.doi.org/10.1016/j.advwatres.2008.01.016. Breen, S.J., Carrigan, C.R., LaBrecque, D.J., Detwiler, R.L., 2012. Bench-scale experiments to evaluate electrical resistivity tomography as a monitoring tool for geologic CO2 sequestration. Int. J. Greenhouse Gas Contl. 9: 484-494. http://dx.doi.org/10.1016/j.ijggc.2012.04.009. Cardarelli, E., 2009. Electrical resistivity and induced polarization tomography in identifying the plume of chlorinated hydrocarbons in sedimentary formation: a case study in Rho (Milan-Italy). Waste Manage. Res. 27(6):595-602. http://dx.doi.org/10.1177/0734242X09102524. Cassidy, D.P., Jr. DDW, Sauck, W., Atekwana, E., Rossbach, S., Duris, J., 2001. The effects of LNAPL biodegradation products on electrical conductivity measurements.

J.

Environ.

Eng.

Geoph.

6:

47-52.

http://dx.doi.org/10.4133/JEEG6.1.47 Chambers, J., Ogilvy, R., Kuras, O., Cripps, J., Meldrum, P., 2002. 3D electrical imaging of known targets at a controlled environmental test site. Environ. Geol. 41(6): 690-704. http://dx.doi.org/10.1007/s00254-001-0452-4. Chambers, J.E., Loke, M.H., Ogilvy, R.D., Meldrum, P.I., 2004. Noninvasive monitoring of DNAPL migration through a saturated porous medium using electrical

impedance tomography. 30

J.

Contam.

Hydrol.

68(1–2): 1-22.

http://dx.doi.org/10.1016/S0169-7722(03)00142-6. Cozzarelli, I.M. et al., 2001. Progression of natural attenuation processes at a crude-oil spill site: I. Geochemical evolution of the plume. J. Contam. Hydrol. 53(3–4): 369-385. http://dx.doi.org/10.1016/S0169-7722(01)00174-7. de Franco, R. et al., 2009. Monitoring the saltwater intrusion by time lapse electrical resistivity tomography: the Chioggia test site (Venice Lagoon, Italy). J. Appl. Geophys. 69(3–4): 117-130. http://dx.doi.org/10.1016/j.jappgeo.2009.08.004. de Groot-Hedlin, C., Constable, S., 2004. Inversion of magnetotelluric data for 2D structure

with

sharp

resistivity

contrasts.

Geophysics

69(1):

78-86.

http://dx.doi.org/10.1190/1.1649377. Dean, J.A., 1985. Lange's handbook of chemistry. McGrawHill, New York Delgado-Rodríguez, O. et al., 2014. Joint interpretation of geoelectrical and volatile organic compounds data: a case study in a hydrocarbons contaminated urban site. Geofísica

internacional

53(2):

183-198.

http://dx.doi.org/10.1016/s0016-7169(14)71499-0. Dwarakanath, V., Jackson, R.E., Pope, G.A., 2002. Influence of wettability on the recovery of NAPLs from alluvium. Environ. Sci. Technol. 36(2): 227-231. http://dx.doi.org/10.1021/es011023w. Esposito, S.J., Thomson, N.R., 1999. Two-phase flow and transport in a single fracture-porous medium system. J. Contam. Hydrol. 37(3–4): 319-341. http://dx.doi.org/10.1016/S0169-7722(98)00169-7. Grant, G.P., Gerhard, J.I., Kueper, B.H., 2007. Multidimensional validation of a 31

numerical model for simulating a DNAPL release in heterogeneous porous media.

J.

Contam.

Hydrol.

92(1–2):

109-128.

http://dx.doi.org/10.1016/j.jconhyd.2007.01.003. Griffin, T.W., Watson, K.W., 2002. A comparison of field techniques for confirming dense nonaqueous phase liquids. Ground Water Monit. R. 22(2): 48-59. http://dx.doi.org/10.1111/j.1745-6592.2002.tb00312.x. Griffiths, D., 1989. Introduction to Electrodynamics; 2nd edn., Prenticehall. Inc. Englewood Cliffs, NJ. Guard, U.C., 1978. CHRIS hazardous chemical data. US Gov. Print Off., Washington, D.C. Hearst, J., Nelson, P.H., Paillet, F.L., 2000. Well logging for physical properties: a handbook for geophysicists, geologists and engineers: West Sussex, England. John Wiley and Sons Ltd. Hort, R.D., Revil, A., Munakata-Marr, J., Mao, D., 2015. Evaluating the potential for quantitative monitoring of in situ chemical oxidation of aqueous-phase TCE using in-phase and quadrature electrical conductivity. Water Resour. Res. 51(7): 5239-5259. http://dx.doi.org/10.1002/2014WR016868. Johnson, T., Versteeg, R., Ward, A., Day-Lewis, F., Revil, A., 2010. Improved hydrogeophysical characterization and monitoring through parallel modeling and inversion of time-domain resistivity andinduced-polarization data. Geophysics 75: WA27-WA41. http://dx.doi.org/10.1190/1.3475513 Karimi, M., Mahmoodi, M., Niazi, A., Al-Wahaibi, Y., Ayatollahi, S., 2012. 32

Investigating wettability alteration during MEOR process, a micro/macro scale analysis.

Colloid.

Surface.

B.

95:

129-136.

http://dx.doi.org/10.1016/j.colsurfb.2012.02.035 Kim, J.-H., Tsourlos, P., Karmis, P., Vargemezis, G., Yi, M.-J., 2016. 3D inversion of irregular gridded 2D electrical resistivity tomography lines: Application to sinkhole mapping at the Island of Corfu (West Greece). Near Surf. Geophys. 14(3): 275-285. http://dx.doi.org/10.3997/1873-0604.2016009. Labrecque, D.J., Yang, X., 2001. Difference inversion of ERT data: a fast inversion method for 3-D in situ monitoring. J. Environ. Eng. Geoph. 6(2): 316–321. http://dx.doi.org/10.4133/jeeg6.2.83. Lehikoinen, A., Finsterle, S., Voutilainen, A., Kowalsky, M.B., Kaipio, J.P., 2009. Dynamical inversion of geophysical ERT data: State estimation in the vadose zone.

Inverse

Probl.

Sci.

En.

17(6):

715-736.

http://dx.doi.org/10.1080/17415970802475951. Lerner, D., Kueper, B., Wealthall, G., Smith, J., Leharne, S., 2003. An illustrated handbook of DNAPL transport and fate in the subsurface. Environment Agency, Almondsbury, Bristol. Li, K., Stefan, M.I., Crittenden, J.C., 2004. UV photolysis of trichloroethylene: Product study and kinetic modeling. Environ. Sci. Technol. 38(24): 6685-6693. http://dx.doi.org/10.1021/es040304b. Loke, M., Acworth, I., Dahlin, T., 2003. A comparison of smooth and blocky inversion methods in 2D electrical imaging surveys. Exploration Geophysics 34: 33

182-187. http://dx.doi.org/10.1071/EG03182 Loke, M., Chambers, J., Rucker, D., Kuras, O., Wilkinson, P., 2013. Recent developments in the direct-current geoelectrical imaging method. J. Appl. Geophys. 95: 135-156. http://dx.doi.org/10.1016/j.jappgeo.2013.02.017 Longeron, D.G., Argaud, M.J., Feraud, J.P., 1989. Effect of overburden pressure and the nature and microscopic distribution of fluids on electrical properties of rock samples.

SPE

Formation

Eval.

4(2):

194-202.

http://dx.doi.org/10.2118/15383-PA. Lucius, J.E., Olhoeft, G.R., Hill, P.L., Duke, S.K., 1992. Properties and hazards of 108 selected substances. 1992 ed. edn. Open-File Report, U.S. Geological Survey. Mao, D. et al., 2015. Resistivity and self-potential tomography applied to groundwater remediation and contaminant plumes: Sandbox and field experiments. J. Hydrol. 530: 1-14. http://dx.doi.org/10.1016/j.jhydrol.2015.09.031. Mercer, J.W., Cohen, R.M., 1990. A review of immiscible fluids in the subsurface: Properties, models, characterization and remediation. J. Contam. Hydrol. 6(2): 107-163. http://dx.doi.org/10.1016/0169-7722(90)90043-G. Miller, C.R., Routh, P.S., Brosten, T.R., McNamara, J.P., 2008. Application of time-lapse ERT imaging to watershed characterization. Geophysics 73(3): G7-G17. http://dx.doi.org/10.1190/1.2907156. Monego, M. et al., 2010. A tracer test in a shallow heterogeneous aquifer monitored via time-lapse surface electrical resistivity tomography. Geophysics 75(4): WA61-WA73. http://dx.doi.org/10.1190/1.3474601. 34

National Research Council, 2013. Alternatives for Managing the Nation’s Complex Contaminated Groundwater Sites. In: National Research Council (Ed.), Washington, D. C. Naudet, V., Gourry, J.-C., Girard, F., Mathieu, F., Saada, A., 2014. 3D electrical resistivity tomography to locate DNAPL contamination around a housing estate. Near

Surf.

Geophys.

12(3):

351-360.

http://dx.doi.org/10.3997/1873-0604.2012059. Newell, C.J., Kueper, B.H., Wilson, J.T., Johnson, P.C., 2014. Natural Attenuation Of Chlorinated Solvent Source Zones. In: Kueper, H.B., Stroo, F.H., Vogel, M.C., Ward, H.C. (Eds.), Chlorinated Solvent Source Zone Remediation. Springer New York,

New

York,

NY,

pp.

459-508.

http://dx.doi.org/10.1007/978-1-4614-6922-3_13. Niemet, M.R., Selker, J.S., 2001. A new method for quantification of liquid saturation in 2D translucent porous media systems using light transmission. Adv. Water Resour. 24(6): 651-666. http://dx.doi.org/10.1016/S0309-1708(00)00045-2. Olayinka, A.I., Yaramanci, U., 2000. Assessment of the reliability of 2D inversion of apparent

resistivity

data.

Geophys.

Prospect

48(2):

293-316.

http://dx.doi.org/10.1046/j.1365-2478.2000.00173.x. Orlando, L., Renzi, B., 2015. Electrical permittivity and resistivity time lapses of multiphase DNAPLs in a lab test. Water Resour. Res. 51(1): 377-389. http://dx.doi.org/10.1002/2014WR015291. Pankow, J.F., Cherry, J.A., 1996. Dense chlorinated solvents and other DNAPLs in 35

groundwater: History, behavior, and remediation. Waterloo Press: Waterloo, Canada. Perri, M., Cassiani, G., Gervasio, I., Deiana, R., Binley, A., 2012. A saline tracer test monitored via both surface and cross-borehole electrical resistivity tomography: comparison

of

time-lapse

results.

J.

Appl.

Geophys.

79:

6-16.

http://dx.doi.org/10.1016/j.jappgeo.2011.12.011. Peter, A., Miles, B., Teutsch, G., 2008. Estimation of emission from an LNAPL contaminated zone considering groundwater recharge. Environ. Geol. 55(2): 321-337. http://dx.doi.org/10.1007/s00254-007-0978-1. Petri, B.G. et al., 2015. Effect of NAPL Source Morphology on Mass Transfer in the Vadose

Zone.

Ground

Water

53(5):

685-698.

http://dx.doi.org/10.1111/gwat.12284. Power, C., Gerhard, J.I., Karaoulis, M., Tsourlos, P., Giannopoulos, A., 2014. Evaluating four-dimensional time-lapse electrical resistivity tomography for monitoring DNAPL source zone remediation. J. Contam. Hydrol. 162: 27-46. http://dx.doi.org/10.1016/j.jconhyd.2014.04.004. Powers, S., Anckner, W., Seacord, T., 1996. Wettability of NAPL-contaminated sands. J.

Environ.

Eng-ASCE.

122:

889-896.

http://dx.doi.org/

10.1061/(ASCE)0733-9372(1996)122:10(889) Revil, A., 2012. Spectral induced polarization of shaly sands: Influence of the electrical double layer. Water Resour. Res. 48(2): 1-23. http://dx.doi.org/ 10.1029/2011wr011260. 36

Revil, A., Karaoulis, M., Johnson, T., Kemna, A., 2012. Review: Some low-frequency electrical

methods

hydrogeology.

for

subsurface

characterization and

Hydrogeol.

J.

monitoring

20(4):

in

617-658.

http://dx.doi.org/10.1007/s10040-011-0819-x. Revil, A., Schmutz, M., Batzle, M.L., 2011. Influence of oil wettability upon spectral induced polarization of oil-bearing sands. Geophysics 76(5): A31-A36. http://dx.doi.org/10.1190/geo2011-0006.1. Ron, E.Z., Rosenberg, E., 2001. Natural roles of biosurfactants. Environ. Micbiol. 3(4): 229-36. http://dx.doi.org/10.1046/j.1462-2920.2001.00190.x. Rothmel, R.K., Peters, R.W., St. Martin, E., DeFlaun, M.F., 1998. Surfactant foam/bioaugmentation technology for in situ treatment of TCE-DNAPLs. Environ. Sci. Technol. 32(11): 1667-1675. http://dx.doi.org/10.1021/es970980w. Ryer, A., Light, V., 1997. Light measurement handbook. International Light, Newburyport, MA. Saenton, S., Illangasekare, T.H., 2013. Effects of incomplete remediation of NAPL-contaminated investigations.

aquifers: Appl.

Experimental Water

and

Sci.

numerical 3(2):

modeling 401-414.

http://dx.doi.org/10.1007/s13201-013-0090-5. Saenton, S., Illangasekare, T.H., Soga, K., Saba, T.A., 2002. Effects of source zone heterogeneity on remediation

surfactant-enhanced

end-points.

J.

NAPL

Contam.

Hydrol.

http://dx.doi.org/10.1016/S0169-7722(02)00074-8. 37

dissolution

and

59(1–2):

resulting 27-44.

Sarafzadeh, P., Hezave, A.Z., Ravanbakhsh, M., Niazi, A., Ayatollahi, S., 2013. Enterobacter cloacae as biosurfactant producing bacterium: Differentiating its effects on interfacial tension and wettability alteration Mechanisms for oil recovery during MEOR process. Colloid. Surface. B. 105: 223-229. http://dx.doi.org/10.1016/j.colsurfb.2012.12.042 Sauck, W., Atekwana, E.A., Nash, M.S., 1998. High conductivities associated with an LNAPL plume imaged by integrated geophysical techniques. J. Environ. Eng. Geoph. 2: 203-212. Schneider, G.W., Greenhouse, J.P., 1992. Geophysical detection of perchloroethylene in a sandy aquifer using resistivity and nuclear logging techniques. Proceedings of the Symposium on the Application of Geophysics to Engineering and Environmental Problems, EEGS, pp. 619-628. Schroth M, Istok J, Ahearn S, Selker J (1996) Characterization of Miller-similar silica sands for laboratory hydrologic studies. Soil Sci. Soc. Am. J. 60(5): 1331-1339. http://dx.doi.org/ 10.2136/sssaj1996.03615995006000050007x Seferou, P. et al., 2013. Olive-oil mill wastewater transport under unsaturated and saturated laboratory conditions using the geoelectrical resistivity tomography method

and the

FEFLOW model.

Hydrogeol.

J.

21(6):

1219-1234.

http://dx.doi.org/10.1007/s10040-013-0996-x. Thabit, J.M., Khalid, F.H., 2016. Resistivity imaging survey to delineate subsurface seepage of hydrocarbon contaminated water at Karbala Governorate, Iraq. Environ. Earth Sci. 75(1): 1-7. http://dx.doi.org/10.1007/s12665-015-4880-y. 38

Triplett Kingston, J.L., Dahlen, P.R., Johnson, P.C., 2010. State-of-the-practice review of in situ thermal technologies. Ground Water Monit. R. 30(4): 64-72. http://dx.doi.org/10.1111/j.1745-6592.2010.01305.x. Ulrich, C., Slater, L., 2004. Induced polarization measurements on unsaturated, unconsolidated

sands.

Geophysics

69(3):

762-771.

http://dx.doi.org/10.1190/1.1759462. Wang, S., Lee, M., Park, M.K., Kim, J.-M., 2010. Box experiments on monitoring the CO2 migration in a homogeneous medium using electrical resistivity survey. Geosci. J. 14(1): 77-85. http://dx.doi.org/10.1007/s12303-010-0009-1. Waxman, M., Smits, L., 1968. Electrical conductivities in oil-bearing shaly sands. Soc Petrol. Eng. J. 8(02): 107-122. Weller, A., Gruhne, M., Seichter, M., Börner, F.D., 1996. Monitoring hydraulic experiments by complex conductivity tomography. Digital Library Home, 1: 209-228. http://dx.doi.org/10.4133/1.2922452. Yang, X., Lagmanson, M., 2006. Comparison of 2D and 3D electrical resistivity imaging methods. 19th EEGS Symposium on the Application of Geophysics to Engineering and Environmental Problems, pp. 585-594. Yang, Z., Niemi, A., Fagerlund, F., Illangasekare, T., 2012. Effects of single-fracture aperture statistics on entrapment, dissolution and source depletion behavior of dense

non-aqueous

phase

liquids.

J.

Contam.

Hydrol.

133:

1-16.

http://dx.doi.org/10.1016/j.jconhyd.2012.03.002. Yoon, H., Werth, C.J., Valocchi, A.J., Oostrom, M., 2008. Impact of nonaqueous 39

phase liquid (NAPL) source zone architecture on mass removal mechanisms in strongly layered heterogeneous porous media during soil vapor extraction. J. Contam.

Hydrol.

100(1–2):

58-71.

http://dx.doi.org/10.1016/j.jconhyd.2008.05.006. Zargari, S., Ostvar, S., Niazi, A., Ayatollahi, S., 2010. Atomic Force Microscopy and Wettability Study of the Alteration of Mica and Sandstone by a Biosurfactant-Producing Bacterium Bacillus thermodenitrificans. Journal of Advanced

Microscopy

Research

5:

143-148.

http://dx.doi.org/10.1166/jamr.2010.1036 Zhou, J., Revil, A., Karaoulis, M., Hale, D., Doetsch, J., Cuttler, S., 2014. Image-guided inversion of electrical resistivity data. Geophys. J. Int. 197: 292-309. http://dx.doi.org/10.1093/gji/ggu001

40

Tables Table 1 Fluids properties for experiments Properties Density (g·cm-3) Absolute viscosity (cP) Interfacial liquid tension (N·m-1) Water solubility (mg·l-1) Resistivity (Ω·m) a

Dean (1985)

b

Guard (1978)

c

Lucius et al. (1992)

d

Measured data

TCE 1.4679a 0.566a 0.0345b 1,100c 106c

Water 1.000 1.000

8.8d

Table 2 Properties of the porous media used in the experiment sands mean grain size (mm) density (g/cm3)

20-30 mesh 0.730 2.56

40-50 mesh 0.362 2.56

41

60-70 mesh 0.250 2.56

100-140 mesh 0.130 2.56

Figures

Fig. 1. (a) Sketch of the experimental setup for Test 1. (b) Sketch of the experimental setup for Test 2. (c) Position of the electrodes on the back of the sandbox. (d) Picture of the experimental apparatus.

42

Fig. 2. Comparison between LTV (left) and ERT (right) data at the same time for Test 1 (a-e) and Test 2 (f-j). Data acquired after TCE injection: 20 min; 40 min; 60 min; 80 min; 690 min for Test 1 and 20 min; 40 min; 60 min; 80 min; 705 min for Test 2. LTV images show the distribution of TCE saturation. Sn stands for the TCE saturation using the LTV data. ERT images show the percentage difference between sequential data and base data. Black lines indicate the position of the assumed borehole BH1 foe Test 1 and BH2 for Test 2.

43

Fig. 3. (a) and (c) Scatter plot of measured TCE saturation versus resistivity ratio (log ρ/ρ0), ρ is the real-time resistivity of partially saturated medium, ρ0 is the initial resistivity of water-saturated medium, and (b) and (d) average of saturation and resistivity ratio acquired from Test 1 (left) and Test 2 (right) at each monitoring time step for the whole source zone, r1~r5 indicate the correlation between the saturation and resistivity for 5 time steps.

44

Fig. 4. Measured saturation and resistivity curves in the assumed borehole BH1. Sn stands for the TCE saturation. Data acquired after TCE injection: (a) 20 min; (b) 40 min; (c) 60 min; (d) 80 min.

Fig. 5. Measured saturation and resistivity curves in the assumed borehole BH2. Sn stands for the TCE saturation. Data acquired after TCE injection: (a) 40 min; (b) 60 min; (c) 80 min.

45

Fig. 6. Comparison between the calculated saturations from ERT data with saturation exponent n=2 in Archie’s law (upper panels) and measured saturations from LTV data (lower panels) for Test 1 acquired at (a) 20 min; (b) 40 min; (c) 60 min; (d) 80 min; (e) 690 min.

Fig. 7. Comparison between the calculated saturations from ERT data with saturation exponent n=2 in Archie’s law (upper panels) and measured saturations from LTV data (lower panels) for Test 2 acquired at (a) 20 min; (b) 40 min; (c) 60 min; (d) 80 min; (e) 705 min.

46

Fig. 8. Plots of total volumes calculated by LTV and ERT method for Test 1 (a) and Test 2 (b) versus actual injected volumes.

47

Fig. 9 The inversion results of Test 1 (left) and Test 2 (right) using L 1 norm optimization method. Data acquired after TCE injection: 20 min; 40 min; 60 min; 80 min; 690 min for Test 1 and 20 min; 40 min; 60 min; 80 min; 705 min for Test 2.

48