A case study for demonstrating the application of U.S. EPA s monitored natural attenuation screening protocol at a hazardous waste site

Transcription

1 Journal of Contaminant Hydrology 59 (2002) A case study for demonstrating the application of U.S. EPA s monitored natural attenuation screening protocol at a hazardous waste site T. Prabhakar Clement a, *, Michael J. Truex b, Peter Lee c a Department of Environmental Engineering, University of Western Australia, Nedlands, WA 6009, Australia b Battelle Pacific Northwest Division, Richland, WA 99352, USA c NPC Services Inc., 2401 Brooklawn Drive, Baton Rouge, LA 70807, USA Received 11 June 2001; received in revised form 12 February 2002; accepted 13 February 2002 Abstract Natural attenuation assessment data, collected at a Superfund site located in Louisiana, USA, are presented. The study site is contaminated with large quantities of DNAPL waste products. Source characterization data indicated that chlorinated ethene and ethane compounds are the major contaminants of concern. This case study illustrates the steps involved in implementing the U.S. EPA s [U.S. EPA, Technical protocol for evaluating natural attenuation of chlorinated solvents in ground water, by Wiedmeier, T.H., Swnason, M.A., Moutoux, D.E., Gordon, E.K., Wilson, J.T., Wilson, B.H., Kampbell, D.H., Hass, P.E., Miller, R.N., Hansen, J. E., Chapelle, F.H., Office of Research and Development, EPA/600/R-98/128] monitored natural attenuation (MNA) screening protocol at this chlorinated solvent site. In the first stage of the MNA assessment process, the field data collected from four monitoring wells located in different parts of the plume were used to complete a biodegradation scoring analysis recommended by the protocol. The analysis indicates that the site has the potential for natural attenuation. In the second stage, a detailed conceptual model was developed to identify various contaminant transport pathways and exposure points. The U.S. EPA model and BIOCHLOR was used to assess whether the contaminants are attenuating at a reasonable rate along these transport paths so that MNA can be considered as a feasible remedial option for the site. The site data along with the modeling results indicate that the chlorinated ethene and chlorinated ethane plumes are degrading and will attenuate within 1000 ft down gradient from * Corresponding author. Department of Civil Engineering, Auburn University, AL 36830, USA. address: clement@ce.auburn.edu (T.P. Clement) /02/$ - see front matter D 2002 Elsevier Science B.V. All rights reserved. PII: S (02)

2 134 T.P. Clement et al. / Journal of Contaminant Hydrology 59 (2002) the source, well before reaching the identified exposure point. Therefore, MNA can be considered as one of the feasible remediation options for the site. D 2002 Elsevier Science B.V. All rights reserved. Keywords: Modeling; Groundwater contamination; Bioremediation; Biodegradation; Natural attenuation; Reactive transport 1. Introduction Waste disposal operations began at the Petro-Processors Inc. (PPI) Brooklawn site (known as Brooklawn site) between 1968 and 1970 and continued until During this period, various types of hazardous waste material were disposed at the site. The material included dense nonaqueous phase liquids (DNAPLs) that originated from chlorinated solvent manufacturing plants and from other refineries. Based on a preliminary site investigation, which was completed by NPC Services, a draft work plan for implementing remedial activities was developed in A hydraulic containment system, and an active source recovery system coupled with the treatment of the extracted water were selected as remedial strategies. Microcosm tests recently performed using the sediment samples collected at the Brooklawn site indicated that the soil microbes have the potential to degrade various chlorinated compounds (Acar et al., 1995; Constant et al., 1995; Clover et al., 1998; Pardue, 1999; Truex et al., 2001). Therefore, monitored natural attenuation (MNA) appears to be one of the additional remedial alternatives available for managing the dissolved plumes at the site. Previously published natural attenuation studies indicate that the biological activity required for degrading many chlorinated organic compounds are ubiquitously present in most anaerobic aquifers (Semprini et al., 1995; Bradley and Chappelle, 1997; Lorah and Olsen, 1999). If sufficient natural or contaminant-derived organic carbon is available as a substrate to support the growth of microbial populations, then MNA can be considered as one of the feasible options for managing chlorinated solvent plumes (Wiedemeier et al., 1999; Clement et al., 2000). A detailed protocol is now available for assessing the attenuation processes for applying the technology at field sites (U.S. EPA, 1998, 1999; Lu et al., 1999). According to the protocol, the first task in implementing MNA at a field site involves completion of an initial screening assessment study. Subsequent tasks involve modeling to determine exposures and better quantify fate. The objective of this case study is to illustrate the efforts involved in implementing the MNA screening process, prescribed in U.S. EPA (1998), at hazardous waste sites and the benefits of modeling. The natural attenuation data set collected at the Brooklawn site is used for this purpose. 2. Site characterization data The surficial features of the Brooklawn site and the location of various monitoring wells used in the site characterization effort are shown in Fig. 1. The field site is located to the north of Baton Rouge, LO, USA, approximately 5200 ft (about 1.5 km) away from the

3 T.P. Clement et al. / Journal of Contaminant Hydrology 59 (2002) Fig. 1. Site details and well locations. Mississippi River. The land to the south of the site is largely undeveloped swampy lowlands of the Mississippi River floodplain, known as the Devil s Swamp, and in the north there is a large industrial development site. A small stream known as the Bayou Baton Rouge, which originates about 7 miles north of the site, runs along the western boundary of the site and turns east at the southern site boundary and breaks into several distributaries which discharge into the Swamp. The dissolved contaminant plume at the site extends about 4000 ft in the east west direction and about 1100 ft in the north south direction Geological data The Brooklawn site is located on the interface between ancient Pleistocene sedimentary deposits and the recent alluvial sediments deposited by the Mississippi River. The interface is marked by a topographic bluff line that transverses in the east west direction. The bluff line approximately parallels the Mississippi River, and is about 30 ft higher than the adjacent floodplain. This line is an erosional feature carved by the Mississippi River. To the north of the bluff line, the upland Pleistocene sedimentary deposits predominantly consist of clayey material, and to the south of the line the floodplain alluvium sediments predominantly consist of sandy material. Fig. 2 shows a geologic cross section of the site through the primary contaminant source area. As shown in the figure, the upland Pleistocene deposits consist of clay from the surface (or from the erosional contact point with the alluvium) to a depth of about 160 ft MSL. The conductivity values of the clay

4 136 T.P. Clement et al. / Journal of Contaminant Hydrology 59 (2002) Fig. 2. Representative geological cross section across the source area.

5 T.P. Clement et al. / Journal of Contaminant Hydrology 59 (2002) ranges from 10 6 to 10 4 ft/day. Some interlayers of silt and sand are present within the Pleistocene clay unit. One of the interlayered silt zones, the 40 ft silt layer, seems to be ubiquitously present across the entire site. The hydraulic conductivity of the silt zone is about 0.1 to 1 ft/day. Additionally, an intermediate sand unit, which is a 40-ft-thick sandy silt layer, also runs across the site starting at a depth of about 80 ft MSL. The 400-ft aquifer, which is a highly permeable sand aquifer (with conductivity values in the range 10 to 100 ft/day), underlies the entire site starting at a depth of 160 ft MSL. The alluvial sediments in the floodplain region are generally sandy-to-silty in texture near the surface. With depth, the sediments become interlayered and intermixed with fine sandy silts to medium sand separated by thin discontinuous clay layers which are generally less than 5 ft thick. Clay mixed sands with appreciable amounts of organic material were found in numerous soil borings within the floodplain region. The hydraulic conductivity of the alluvium sand ranges from 1 to 10 ft/day. The thickness of the top alluvial unit increases with distances away from the site. As shown in Fig. 2, along the clay alluvium interface, the alluvial unit pinches out and intersects with the Pleistocene clay unit; and with distances away from the site, towards the river, the alluvial unit become thicker and intersects the 40 ft silt layer and later intersects the intermediate sand layer Background geochemical data Water samples collected from wells P and P , which are located outside the plume (see Fig. 1), were selected to assess the background geochemistry of Brooklawn groundwater. Note that one of the selected wells (P ) is located upstream in the Pleistocene clay unit and the other well (P ) is located downstream in the alluvium unit. The measured geochemical constituents of the Pleistocene and alluvium units are summarized in Table 1. The geochemical data indicate that the alluvium, which is Table 1 Background geochemical characteristics of the groundwater Constituent Alluvium a Pleistocene clay unit b Units Chloride mg/l Specific conductance Amho Dissolved methane not detected not detected mg/l Dissolved oxygen not detected 2 mg/l Dissolved hydrogen nm Inorganic carbon mg/l Iron, ferrous mg/l Nitrite not detected 0.04 mg/l Nitrate mg/l Oxidation/reduction potential mv ph standard units Sulfide not detected 0.1 mg/l Sulfate not detected 8 mg/l Temperature jc Total organic carbon mg/l a Average concentration from wells P and P b Concentration from well P

6 138 T.P. Clement et al. / Journal of Contaminant Hydrology 59 (2002) expected to be the primary transport pathway, can be characterized as an anaerobic system with neutral ph and moderate ionic strength and hardness. The redox characteristics of the Pleistocene clay zone also indicate reducing conditions. Although low levels of dissolved oxygen were measured in P , which is located outside the site boundary, several other wells located within the plume in the Pleistocene clay were devoid of oxygen, thus indicating anaerobic conditions. Measured organic carbon content of the alluvium sediments ranged from 0.39% to 1.13% (Valsaraj et al., 1999) Source locations Based on historic site characterization data, the spatial extents of various DNAPL contamination zones were delineated. The boundaries of the source zones are shown in Figs. 1 and 2. The bulk of the DNAPL mass is present in the eastern portion of the site. However, some minor isolated pits and disposal drains were also present in the western region. Historical data show that most of the waste products were initially disposed into several earthen pits located in the Pleistocene clay unit along the upland portion of the site. During remedial investigations, these upland waste NAPLs were found to be confined with only minimal NAPL mass migration beyond the pits. However, large amounts of NAPL were later disposed into excavated pits in the lower floodplain area, which are surrounded by constructed levees. The NAPL products disposed in this area have migrated beyond the limits of the pits. Further, during a flood event in the 1970s, a portion of the levee failed and NAPL products were directly discharged into the small drainage channels along the southwest corner of the site. This event created a spill area (see Fig. 1), which was delineated based on soil core data collected from the channels. Further site investigations revealed that the presence of NAPL was mostly limited to shallow regions, with isolated deeper occurrences in the northern area of the channel adjacent to the Brooklawn site Contaminant characterization data The contamination currently present at the site consists of large quantities of pooled or trapped NAPL-phase products and various dissolved plumes that emanate from the free phase. The DNAPL present at the site is a complex mixture of several organic compounds. Table 2 shows the groundwater concentrations of various contaminant species measured in two monitoring wells located within the dissolved plume region. The first well, P , is located down gradient from the primary DNAPL source area, and the second well, PBB21-1N, is located near the edge of a secondary spilled source region. The data show that dissolved chlorinated ethene and chlorinated ethane compounds are at concentration levels of one to two orders of magnitude greater than all other compounds. High values of chloroform and 1,2-dichloropropane were also measured in some of the groundwater samples. Chloroform can be biodegraded and is itself an anaerobic biotransformation product of carbon tetrachloride (Beelen and Keulen, 1990; Picardal et al., 1995). Researchers have shown that chloroform can be degraded by anaerobic methanogenic enrichments and by other non-methanogenic anaerobic cultures (Bagley and Gossett, 1995). Literature information also suggests that 1,2-dichloropropane can be completely

7 T.P. Clement et al. / Journal of Contaminant Hydrology 59 (2002) Table 2 Concentration of chlorinated constituents in the alluvium Compound Concentration at well P (Ag/l) Concentration at well PBB21-1N (Ag/l) 1,2,4-Trichlorobenzene a <10 <20 1,2-Dichlorobenzene a,b <5 <5 1,2-Dichloropropane a Bis(2-Chloroisopropyl)ether a Carbon Tetrachloride a <5 <5 Chlorobenzene a,c <5 53 Chloroform a Hexachlorobenzene a <10 <20 Hexachlorobutadiene a <10 <20 Hexachloroethane a <10 <20 1,1,2,2-Tetrachloroethane ,1,2-Trichloroethane ,1-Dichloroethene ,2-Dichloroethane Tetrachloroethene Trichloroethene cis-1,2-dichloroethene trans-1,2-dichloroethene Vinyl chloride a Compound is not included for further study in this initial screening for MNA. b The concentration of all isomers of dichlorobenzene were similar. c Chlorobenzene is included in the table because it is a biotransformation product of other chlorinated benzenes. dechlorinated under anaerobic conditions (Loffler et al., 1997). Since the dissolved levels of chlorinated ethene and ethane compounds were good indicators of the status of overall contamination, further analyses of chloroform and dichloropropane were not included in this screening study. Non-chlorinated contaminants such as the BTEX (benzene, toluene, ethylbenzene and xylene) compounds were also not included because the measured BTEX levels in down gradient wells were much lower than chlorinated compounds. In this initial MNA screening study, we primarily focus on the fate and transport of the following chlorinated ethane and ethene components: 1,1,2,2-tetrachloroethane (TeCA), 1,1,2- trichloroethane (TCA), 1,2-dichloroethane (DCA), chloroethane (CA), tetrachloroethene (PCE), trichloroethene (TCE), dichloroethene (DCE), and vinyl chloride (VC). These components were previously identified as the contaminants of concern for the site based on their chemical, physical, and other transport properties (NPC, 1996). Since chlorinated solvents are prevalent at most DNAPL waste sites, the U.S. EPA (1998) MNA guidelines also primarily focus on analyzing these contaminants. 3. Biodegradation assessment The MNA assessment process involves a six-step screening method (U.S. EPA, 1998). As shown in Fig. 3, the first step in the MNA assessment is to use the site data to answer the important question Is biodegradation occurring at the site? In order to address this

8 140 T.P. Clement et al. / Journal of Contaminant Hydrology 59 (2002) Fig. 3. Monitored natural attenuation screening process flow sheet [adapted from U.S. EPA, 1998]. question, the average concentrations of various geochemical parameters measured in alluvium wells P and P , shown in Table 1, were first used to set the background levels. Groundwater geochemical data collected from four wells including P , P , PBB21-1N, and P , which are expected to represent different types of biochemical and/or hydraulic regimes present at the site, were used to assess the biodegradation conditions. It can be seen from Fig. 1 that the first well, P , is located to the south of the primary NAPL source area. This well is adjacent to the highly contaminated source area and was (prior to pumping, between 1970 and 1995) down gradient from the source. Currently, the well is within the hydraulic capture zone of the

9 T.P. Clement et al. / Journal of Contaminant Hydrology 59 (2002) extraction wells and contains significant concentrations of chlorinated solvents. The second monitoring well, P , is located about 300 ft south of the primary NAPL source area. This well location was also previously (prior to pumping) down gradient from the source, but is now outside the hydraulic capture zone of the pumping wells. The hydraulic gradient at this location changes seasonally due to fluctuations in the Mississippi River stage. The third monitoring well, PBB21-1N, is located at the edge of a NAPL source area in a bayou channel in the spill zone. This well is close to the NAPL source region. The hydraulic gradient in this region has significant seasonal variations and is influenced by the water levels in the bayou channel. The fourth well, P , is located in the swampy area down gradient from the previous monitoring location. The hydraulic gradient at this location also has considerable fluctuations. Using the U.S. EPA (1998) MNA framework, a specified number of points were assigned depending on the concentration of the geochemical indicators observed in the wells. Points were awarded only if the concentration of a geochemical indicator was within the range specified in the screening criteria and if the indicator was not a constituent of the original contaminant source. The points were added and interpreted based on U.S. EPA guidelines to determine whether biodegradation is occurring at the selected location. If the total score was above 15 points, the location was deemed to have a good potential for natural attenuation. Further details of this scoring scheme are discussed in U.S. EPA (1998). The results of the biodegradation assessment for the four selected Brooklawn site wells are summarized in Table 3a and b. As shown in the table, the biodegradation assessment scores for all four locations are found to be greater than 15, indicating a good potential for natural biodegradation. Most of the points assigned to each location were due to the detection of geochemical indicators of the presence of anaerobic environments such environments can support biological activities required for mediating dechlorination processes. A few points were also assigned for the presence of certain degradation products. For example, in wells P and PBB21-1N, cis-dce, a daughter product of chlorinated ethene and chlorinated ethane biotransformation, was present at concentration levels much greater than the concentrations of the other dichloroethene isomers. The concentration of cis-dce was below detectable levels at the other two monitoring locations. Points could not be assigned for the presence of several other biodegradation products (other than cis-dce) even though they were present at very high levels. This is because several of these possible biodegradation products were already present in the NAPL source, although at very low levels. For example, VC, which is a by-product of the reductive dechlorination process, was measured as 25.7 mg/l in well P and 58.2 mg/l in PBB21-IN. However, the measured VC values were consistently low near the source region. Further, the measured weight percentage of VC in the NAPL source was about 0.026%, which yields a maximum effective VC solubility of 0.9 mg/l. Therefore, the high VC concentration levels observed in some of the down gradient wells could have been due to biodegradation. However, because VC is contained within the original NAPL source, points cannot be awarded for the presence of VC according to the U.S. EPA (1998) guidelines. This demonstrates the conservative nature of the scoring strategy, which could result in underestimation of the overall biodegradation potential.

10 Table 3 Bioattenuation screening parameters and scoring Contaminant/geochemical Criteria indicator In NAPL source? PBL (shallow) concentration PBL (shallow) score P (deep) concentration Dissolved oxygen (mg/l) <0.5 N/A Nitrate (mg/l) <1 N/A Iron(II) (mg/l) >1 N/A Sulfate (mg/l) <20 N/A <5 2 <1.0 2 Sulfide (mg/l) >1 N/A Methane >0.5 N/A no data Oxidation/reduction <50 or < 100 N/A 58 1 no data 0 potential (mv) ph 5<pH>9 N/A 6.4 N/A 6.8 N/A Total organic carbon (mg/l) >20 N/A no data Temperature >20 N/A no data 0 Carbon dioxide >2 N/A no data 0 no data 0 Alkalinity >2 N/A Chloride (mg/l) >2 N/A Hydrogen >1 nm N/A < Volatile fatty acids >0.1 N/A no data 0 no data 0 BTEX (mg/l) >0.1 Yes TCE (Ag/l) Yes ,1,2-TCA (Ag/l) Yes < ,2-DCA (Ag/l) Yes < trans-1,2-dce (Ag/l) Yes < cis-1,2-dce (Ag/l) No no data ,1-DCE (Ag/l) No VC (Ag/l) Yes Chloroethane (Ag/l) No <5 0 <5 0 Ethene (mg/l) >0.01 or >0.1 No no data Ethane (mg/l) >0.01 or >0.1 No no data ,1,2,2-Tetrachloroethane (Ag/l) Yes < Tetrachloroethene (Ag/l) Yes < Total P (deep) score 142 T.P. Clement et al. / Journal of Contaminant Hydrology 59 (2002)

11 Contaminant/geochemical indicator Criteria In NAPL source? PBB21-1N concentration PBB21-1N score P concentration Dissolved oxygen (mg/l) <0.5 N/A Nitrate (mg/l) <1 N/A Iron(II) (mg/l) >1 N/A Sulfate (mg/l) <20 N/A Sulfide (mg/l) >1 N/A Methane >0.5 N/A Oxidation/reduction potential (mv) <50 or < 100 N/A no data ph 5<pH>9 N/A 6.15 N/A 7.6 N/A Total organic carbon (mg/l) >20 N/A Temperature >20 N/A no data Carbon dioxide >2 N/A no data 0 no data 0 Alkalinity >2 N/A Chloride (mg/l) >2 N/A Hydrogen >1 nm N/A >8 3 Volatile fatty acids >0.1 N/A no data 0 no data 0 BTEX (mg/l) >0.1 Yes TCE (Ag/l) Yes ,1,2-TCA (Ag/l) Yes ,2-DCA (Ag/l) Yes trans-1,2-dce (Ag/l) Yes < cis-1,2-dce (Ag/l) No no data 0 1,1-DCE (Ag/l) No VC (Ag/l) Yes Chloroethane (Ag/l) No <5 0 <5 0 Ethene (mg/l) >0.01 or >0.1 No Ethane (mg/l) >0.01 or >0.1 No ,1,2,2-Tetrachloroethane (Ag/l) Yes Tetrachloroethene (Ag/l) Yes <5 0 Total P score T.P. Clement et al. / Journal of Contaminant Hydrology 59 (2002)

12 144 T.P. Clement et al. / Journal of Contaminant Hydrology 59 (2002) Development of conceptual models As shown in Fig. 3, in the second stage of the MNA screening process, a conceptual model must be developed to provide the context for understanding the transport processes occurring at the site. Fig. 4 shows the details of the source area, major contaminant transport pathways, and various other surficial features. In our conceptual model development effort, contaminant transport from the major source area where the bulk of DNAPL mass currently resides is only considered. Fig. 4 also shows the location of various monitoring wells and boreholes; data from these wells were used to build the computer simulation model. All model simulations completed in this study focused on predicting the fate-and-transport of contaminants under natural gradient conditions. In order to be consistent with this scenario, the data collected prior to the start of groundwater extraction activities were only used Conceptual model for contaminant transport pathways and exposure points Fig. 5 is a conceptual cross section model of the site, which shows the relative locations of potential exposure points along various transport pathways. The Mississippi River and the 400-ft aquifer were identified as the exposure points of concern. As shown in the figure, the conceptual model considers two distinct transport pathways: (1) a horizontal Fig. 4. Conceptual model for contaminant transport and other site details.

13 T.P. Clement et al. / Journal of Contaminant Hydrology 59 (2002) Fig. 5. Description of source locations and transport paths.

14 146 T.P. Clement et al. / Journal of Contaminant Hydrology 59 (2002) path towards the Mississippi River and (2) a vertical path towards the deep 400-ft aquifer. The total distance along the horizontal transport path, via the alluvium, from the leading edge of the source to the Mississippi River is about 5200 ft. The total distance along the vertical transport path, via the Pleistocene clay, from the bottom of the source to the 400-ft aquifer, is approximately 140 ft. In addition to these two major transport pathways, it is possible that the plumes might reach the 400-ft aquifer through an inclined pathway via the alluvium where the clay layer might be discontinuous. Since little characterization information is available beyond the site boundary, it is difficult to determine the transport characteristics of this inclined pathway, which would include transport via the alluvial zone and some portions of discontinuous clay layers. Preliminary analyses suggest that such a lengthy transport pathway may not be critical because alluvial sediments are naturally less permeable in the vertical direction as compared to the horizontal direction and, hence, are expected to yield more resistance to vertical transport. Also, the observed vertical hydraulic gradient between the alluvium and 400-ft aquifer is very small. Therefore, within the context of this initial screening effort, transport along this inclined pathway was not considered. The model shown in Fig. 5 was conceptualized based on the borehole data shown in Fig. 2 and the time series of groundwater levels shown in Fig. 6. As shown in Fig. 6, the transient groundwater levels in the alluvium and in the 400-ft aquifer were similar, and the trend also closely followed the seasonal variations in the Mississippi River. Water levels in the Pleistocene clay unit do not respond to the river variations. These field observations are consistent with our conceptual model. The water levels observed in the alluvium (P ) and the 40-ft-silt zone (S-UG-1) wells showed an average hydraulic gradient towards the river. The interrelationships between the water level response patterns observed in the alluvium, silt layer, and Fig. 6. Seasonal variations in groundwater levels observed in different hydraulic units.

15 T.P. Clement et al. / Journal of Contaminant Hydrology 59 (2002) Pleistocene clay unit were used to develop a conceptual framework for the transfer of contaminant mass from the source zone to the alluvium unit. Note that the site characterization data, summarized in Fig. 2, reveals that a portion of the DNAPL source disposed in the alluvium pits was in close contact with the silt layer. Therefore, in our conceptual contaminant mass transfer model, it is assumed that the contaminants from the DNAPL source seep into the 40-ft silt layer and later are transported by the horizontal groundwater level gradient between the silt and alluvium zones Estimation of contaminant transport velocity Groundwater levels observed in the silt (S-UG-1) and alluvium (P ) wells were used to estimate a yearly averaged horizontal transport velocity. Table 4 provides the water levels measured in these wells over a 12-month period. As shown in the table, the yearly average of head differences between these two wells is 2.15 ft and the distance between the wells is 800 ft; this yields an average groundwater gradient of Using an estimated porosity value of 0.3 and conductivity value of 5 ft/day for the alluvium, the average transport velocity for the horizontal pathway, towards the river via the alluvium aquifer, is 0.05 ft/day. Similarly, the vertical hydraulic gradient was estimated based on the average head difference between the 40-ft silt well (S-UG-1) and the 400-ft aquifer (well D-UG-1). While the vertical transport path has a much shorter length than the horizontal transport path, the material between the contamination and the 400-ft aquifer is low permeable clay. Based on measured hydraulic conductivity value of ft/day for the clay material, an estimated vertical hydraulic gradient of ft/ft and an estimated porosity of 0.4, the vertical transport velocity is approximated as ft/day. Contaminants Table 4 Calculation of average horizontal hydraulic gradient for the alluvium Month Hydraulic head (ft), Well S-UG-1 (Pleistocene interface) Hydraulic head (ft), Well P (Alluvium) April May a 1.37 June July August September October November December January February March Hydraulic head difference (ft) Average head difference (ft) 2.15 Approximate gradient (based on 800 ft distance between wells) (ft/ft) (toward river) a Estimated value since no data is available for the month.

16 148 T.P. Clement et al. / Journal of Contaminant Hydrology 59 (2002) traveling at this velocity would require over 80,000 years to traverse the 140 ft of clay zone present between the current location of the contaminant source and the top of the 400-ft aquifer. Moreover, monitoring data from the 400-ft aquifer also indicated that the aquifer is uncontaminated. Therefore, further analysis of the vertical pathway was not considered in this initial screening assessment. 5. Biodegradation and reactive transport The U.S. EPA screening model, BIOCHLOR, was used to model the attenuation processes occurring at site. BIOCHLOR is an analytical computer code that is intended for use as a screening-level model to determine if remediation by natural attenuation is feasible at a chlorinated solvent site (Aziz et al., 2000). The code uses a novel analytical solution strategy to solve the multi-species sequential reactive transport problem (Sun et al., 1999; Sun and Clement, 1999; Clement, 2001). BIOCHLOR has the ability to simulate uniform flow with three-dimensional dispersion, linear adsorption, and biodegradation via reductive dechlorination reactions. The model can predict migration patterns of a parent chlorinated solvent species (either TCA or PCE) and its daughter products. BIOCHLOR assumes first-order kinetics to model the biological decay reactions. The use of first-order kinetics is appropriate when the biodegradation rate is primarily a function of the concentration of the contaminant, when the number of microorganisms that can degrade the contaminant is constant over time within the region of interest, and when all other nutrients critical to the biodegradation processes are in abundance. For most fieldscale natural attenuation modeling applications, the first-order assumption may be considered as a reasonable approximation (U.S. EPA, 1998; Aziz et al., 2000) provided the electron donor is not limiting. Moreover, the assumption of first-order kinetics is often acceptable for biodegradation at low pollutant concentration levels (Schmidt et al., 1985), which is typically encountered in most groundwater remediation problems. The disposal operations at the site started between 1968 and 1970, and the field data used for this paper were collected nominally between 1992 and January Therefore, the simulations were completed for 25 years so that direct comparison to the field data is possible. The physical dimensions of the contaminant source were determined based on the scale of the primary source shown in Figs. 4 and 5. The BIOCHLOR model requires several basic transport parameters as input values, which include advection velocity, dispersion coefficients, and retardation factors for the contaminant species. Further, the model requires first-order degradation rate coefficients for the selected chlorinated solvent reductive dechlorination sequence (either chlorinated ethenes or chlorinated ethanes). In the following sections, the methods used for selecting appropriate retardation values, biodegradation rate constants, source zone concentration levels, and other flow and transport parameters are summarized Retardation parameters The effective transport velocity of the contaminants is greatly influenced by the adsorption characteristics of the porous medium. A linear equilibrium-partitioning model

17 T.P. Clement et al. / Journal of Contaminant Hydrology 59 (2002) was assumed to quantify the adsorption characteristics of Brooklawn sediments. Estimates for the partition coefficient K d (and subsequently the retardation factor R) were obtained using an empirical correlation function. Valsaraj et al. (1999) performed sorption experiments using three types of Brooklawn sediments and developed the following function: K d ¼ 10 0:81 K 0:56 ow f oc where K ow is the octanol water coefficient and f oc is the fraction of organic carbon of the soil. Table 5 presents the value of K ow and the calculated values for K d and R values for the chemicals considered in this initial MNA screening study. An f oc value of 0.39%, which was the value measured by Valsaraj et al. (1999) for the sandy alluvial material, was used to compute the retardation parameters summarized in Table Biodegradation rate parameters The Brooklawn site contamination includes a mixture of both chlorinated ethene and chlorinated ethane compounds. The BIOCHLOR model is capable of simulating the degradation of either the chlorinated ethene reaction chain (starting with PCE) or the chlorinated ethane reaction chain (starting with TCA), not the mixture. However, when both ethene and ethane species are present at the site, then it is difficult to interpret ethene daughter products because some of the chlorinated ethenes can be produced from chlorinated ethane decay reactions. Fig. 7 shows possible reaction pathways for degradation of PCE and TeCA and the subsequent chlorinated ethane and chlorinated ethene daughter products (Lorah and Olsen, 1999). In this work, the BIOCHLOR simulations were first completed for each individual ethene and ethane reaction chain. Subsequently, approximations were made to quantify the influence of chlorinated ethenes produced from chlorinated ethane decay reactions. First-order kinetic models are assumed to be sufficient to describe all the reaction steps represented in Fig. 7, which encompasses both chlorinated ethane and chlorinated ethene Table 5 Estimated value of retardation coefficients Compound Log 10 of octanol water a coefficient log K ow Partition coefficient K d (l/kg) b Retardation coefficient (1+qK d /n) c TeCA TCA DCA CA 1 (estimate) PCE TCE DCE d VC a Values from Schwarzenbach et al. (1993). b Estimated from Valsaraj et al. (1999) correlation. c Assuming q=1.6 kg/l and n=0.3. d Value listed is for 1,1-dichloroethene.

18 150 T.P. Clement et al. / Journal of Contaminant Hydrology 59 (2002) Fig. 7. Anaerobic transformation pathways for the chlorinated ethane and ethene compounds. decay reactions. In addition, since many of the biotic reactions need organic carbon as substrate to support the direct dechlorination or co-metabolic reactions, it is assumed that sufficient organic carbon is available at the site. This assumption is reasonable at Brooklawn site because the site is located down gradient from a swamp that has large amounts of decaying material that can supply large amounts of carbon. In addition, carbon sources could also be derived from other waste organic material co-disposed with chlorinated compounds. To assess the biodegradation kinetics of a mixture of chlorinated ethane and chlorinated ethene compounds, the degradation rate coefficients must be considered in conjunction with the fraction of each potential product produced from different parent compounds. As shown in Fig. 7, chlorinated ethane species can degrade and produce less-chlorinated ethane species or chlorinated ethene species. Table 6 presents the estimated fraction of each daughter product that could be expected to be formed from different parent compound. These estimates are based on laboratory data (Lorah and Olsen, 1999; Chen et al., 1996). Table 7 summarizes the assumed values of first-order decay rate coefficients for modeling anaerobic destruction of chlorinated ethane and chlorinated ethene species. These values were used in the BIOCHLOR model to perform the initial fate-and-transport analysis for the Brooklawn site. Note that rate coefficients are only presented for reductive dechlorination along one class of daughter products (e.g., only ethanes or only ethenes) because these are the only sequences that the BIOCHLOR model can describe. Also, to match the required inputs for the BIOCHLOR model, TCA is assumed to be the parent

19 T.P. Clement et al. / Journal of Contaminant Hydrology 59 (2002) Table 6 Estimated fractional percent conversion for chlorinated ethanes and chlorinated ethenes Reaction pathway Estimated yield (mol/mol) Based on data from: PCE!TCE 1.0 Lorah and Olsen (1999) TCE!DCE 1.0 Lorah and Olsen (1999) DCE!VC 1.0 Lorah and Olsen (1999) a TeCA!TCA 0.35 Lorah and Olsen (1999) TeCA!TCE 0.02 Lorah and Olsen (1999) TeCA!DCE 0.63 Lorah and Olsen (1999) a TCA!DCA 0.2 Chen et al. (1996) TCA!VC 0.8 Chen et al. (1996) a DCA!CA 0.7 Chen et al. (1996) DCA!ethane 0.3 Assumed a As expected, sum of all TeCA or TCA or DCA yields is equal to unity. compound of the chlorinated ethane series. Further, all DCE isomers are combined together and assumed to be the daughter product of biodegradation Source concentrations Dissolved concentration levels of chlorinated ethane and chlorinated ethene species measured or estimated to be present near the source zone are listed in Table 8. The first column in the table lists field measured concentration data. These data are averages of dissolved-phase concentrations in the groundwater samples collected from wells W , W , and W , where the water is in direct contact with NAPL products (see Fig. 2 for well locations). The concentrations in the second column were estimated based on a solubility analysis using measured NAPL composition data (Truex et al., 1999). It is interesting to note that in Table 8 the measured concentration values and the calculated effective solubility values compare favorably whenever the field data are above the detection limit. As shown in Fig. 7, chlorinated ethenes can be produced from biodegradation of the chlorinated ethanes present in the NAPL source. The source area concentration of the parent chlorinated ethanes (TeCA and TCA), which can potentially produce chlorinate ethenes, are approximately 60 and 337 mg/l, respectively. Clearly, these high concentrations of ethane parent compounds can yield significant amounts of chlorinated ethenes Table 7 First-order rate coefficients Reaction Value (1/day) Basis TCA!DCA One tenth of the rate estimated for Lorah and Olsen (1999) microcosm data DCA!CA One tenth of the rate estimated for Lorah and Olsen (1999) microcosm data CA!ethane One tenth of the rate estimated for Lorah and Olsen (1999) microcosm data PCE!TCE Set to be equal to TCE conversion rate TCE!DCE One tenth of the rate measured in microcosms using Brooklawn sediments DCE!VC Set to be equal to TCE conversion rate VC!ethene One tenth of the rate measured in microcosms using Brooklawn sediments

20 152 T.P. Clement et al. / Journal of Contaminant Hydrology 59 (2002) Table 8 Contaminant concentrations (in mg/l) near the source zone Constituent Measured values Estimated values a As decay products b TeCA N/A TCA N/A DCA N/A CA <25 0 N/A PCE < N/A TCE < DCE < VC < a Estimated based on solubility analysis. b Calculated using the chlorinated ethane concentrations from the first column and yield data from Table 6. down gradient from the source. The conversion fractional percent data presented in Table 6 can be used to estimate the maximum possible concentration of each chlorinated ethenes produced from the ethane compounds present in the source. These estimates are summarized in the third column of Table 8. Note if the biodegradation rates of ethane reactions are high (when compared to ethene reactions), the mass of ethene produced from ethane may simply be added to the original ethene source levels. This approximation is employed in this study to indirectly simulate the combined ethane ethene reactions Transport parameters The transport parameters used in the model simulations are summarized in Table 9. The transport properties of the alluvium were estimated based on field measurements. Methods used for estimating the hydraulic gradient values were discussed in Section 4. The values of retardation factors for different chlorinated compounds were estimated based on sitespecific sorption data shown in Table 5. The value for porosity was estimated from literature values for similar types of geologic materials. The longitudinal dispersivity value was estimated based on the guidelines presented in Gelhar et al. (1992). The ratio of the longitudinal to transverse dispersivity was assumed to be 0.1, and ratio of the longitudinal to vertical dispersivity was assumed to be Table 9 Parameters used in BIOCHLOR simulations Property Value Hydraulic conductivity (ft/day) 5 Hydraulic gradient Longitudinal dispersivity (ft) 50 Porosity 0.3 Average retardation for ethenes 3.6 Average retardation for ethanes 2.4 Model area (zone-1) length (ft) 5000 Model area width (ft) 2000 Source thickness (ft) 45 Source width (ft) 1000

21 T.P. Clement et al. / Journal of Contaminant Hydrology 59 (2002) BIOCHLOR simulations Three sets of BIOCHLOR simulations were completed to assess whether adequate bioattenuation is occurring at the Brooklawn site, and to evaluate the potential that a steady-state condition may be established such that the contaminant plumes will not reach the receptors. Wherever possible, the simulation results were compared against observed data collected from the test boreholes drilled during a site investigation effort completed in between September 1994 and January These test boreholes, BL , BL , BB , and BB (see Fig. 4), are located approximately 150, 300, 400, and 500 ft, respectively, from the source. Simulation 1 predicted chlorinated ethane transport for a transport period of 25 years so that the model results can be compared to the measured contaminant levels. The sourcezone concentrations used in this simulation were set based on field-measured values (see Table 8, column 1). Both Simulations 2 and 3 predicted chlorinated ethene transport using two types of source conditions. Since measured (above detection limit) source-zone concentrations are unavailable for chlorinated ethenes, the first set of ethene simulation (Simulation 2) was performed using source concentration levels computed based on the concentration estimate from a solubility analysis (see Table 8, column 2). In the second set of ethene simulations (Simulation 3), it was assumed that all chlorinated ethane degradation reactions would occur at a rapid rate that would allow complete ethane degradation close to the source zone. To reflect this condition, the sourcezone concentrations of chlorinated ethenes in Simulation 3 were set equal to the sum of concentrations used in Simulation 2 and the maximum amount of ethene concentrations that could be produced as products of ethane biodegradation (i.e., add ethene concentration levels in column 2 and column 3 of Table 8). Analysis of field data along with model simulations showed that this was a reasonable approximation for the Brooklawn site; this point is discussed in more detail in the following section. The ultimate objective of Simulation 3 was to assess the impact of chlorinated ethane degradation on predicted chlorinated ethene levels. Since BIOCHLOR cannot be used to simulate coupled ethane ethene degradation reactions, this simplified approximation was made to assess the coupled degradation condition Simulation 1 chlorinated ethane reaction chain Fig. 8 presents the model-predicted TCA, DCA, and CA concentration profiles after 25 years of migration. The figure shows the predicted contaminant profiles down gradient from the source zone along the plume centerline under three transport conditions: (1) no attenuation, (2) attenuation by adsorption, and (3) attenuation by adsorption and biodegradation. The results show that the attenuation mechanisms including sorption and degradation considerably retard plume migration. Also, all the biodegradation reactions including production and destruction of intermediate compounds, such as CA, occur within 500 ft down gradient of the source. Field data for TCA and DCA indicate that the concentration of each species is less than 1 mg/l at distances of 150 and 300 ft, respectively, down gradient from the source. With only adsorption-related attenuation, the model predicts the TCA and DCA concentrations

22 154 T.P. Clement et al. / Journal of Contaminant Hydrology 59 (2002) Fig. 8. Comparison of field data against BIOCHLOR simulations TCA and its daughter products (t =25 years).

23 T.P. Clement et al. / Journal of Contaminant Hydrology 59 (2002) as 50 and 450 mg/l, respectively, at these locations. This comparison indicates that the TCA and DCA field data more closely match the concentration profiles predicted by BIOCHLOR under the condition where both biodegradation and adsorption are occurring. Since field data are not available for comparison to predicted CA concentration profiles, these profiles must only be considered as preliminary estimates. The BIOCHLOR model assumes TCA as the parent ethane contaminant and predicts sequential dechlorination of TCA to DCA and later to CA. Thus, TCA produced from its parent ethane compound TeCA, and also the possibility of partial conversion of TCA and DCA compounds to VC (see Fig. 7), which has been measured in microcosm tests (Chen et al., 1996; Lorah and Olsen, 1999), were not considered in this simulation Simulation 2 chlorinated ethene reaction chain Fig. 9 presents model-predicted PCE, TCE, DCE, and VC concentration profiles after 25 years of migration. Similar to the previous analysis, these simulations were also completed under different conditions including no attenuation, attenuation by adsorption, and attenuation by adsorption and biodegradation. Source zone concentration levels used in this simulation are based on the estimates from a solubility analysis (Table 8, column 2). Under the assumed conditions, the model predicted that VC would be produced and degraded within 300 ft down gradient from the source when biodegradation was assumed to occur in the aquifer. With only adsorption-related attenuation, the VC concentration was predicted to continuously reduce toward zero within 300 ft down gradient from the source. The concentration of PCE measured in the field was less than 0.1 mg/l at a distance of 150 ft down gradient from the source. With only adsorption-related attenuation, Simulation 2 predicted 2.5 mg/l of PCE at this location. The predicted PCE profile seems to match the field data only when biodegradation and adsorption were assumed to simultaneously occur in the aquifer. The field data for TCE, DCE, and VC were more difficult to interpret from Simulation 2 because each of the ethene species can also be produced from degradation of chlorinated ethane species. Fig. 9c and d clearly indicates that the fieldmeasured DCE and VC values significantly diverge from the model profiles. In particular, the field-measured VC concentration levels are much higher than those predicted by the model either with or without biodegradation. As illustrated in Fig. 7, chlorinated ethene compounds, particularly VC, can be produced as a by-product of chlorinated ethane degradation. These by-product effects are quantified in the next set of simulations Simulation 3 coupled chlorinated ethane and ethene reactions BIOCHLOR model results for the ethane series (shown in Fig. 8) indicates that TCA concentration levels decrease from over 300 to less than 1 mg/l within 150 ft down gradient of the source. Field data also confirm that the TCA concentration reduces to less than 1 mg/l close to the source zone. Since degradation of TCA can yield ethene compounds, it is possible that significant amounts of chlorinated ethenes are being produced through biological processes within the 300-ft region from the source. The NAPL present at the Brooklawn site contains large amounts of TeCA. As shown in Fig. 7, degradation of TeCA would yield chlorinated ethene compounds. At biologically