Monitoring tools and numerical models for evaluation of monitored natural attenuation of explosives at selected sites

Transcription

1 Monitoring tools and numerical models for evaluation of monitored natural attenuation of explosives at selected sites M. Zakikhani, D. W. Harrelson, J. C. Pemington, J. M. Brannon, M. K. Corcoran, and J. Clark US. Army Engineer Research and Development Center ( ERDC), USA Abstract As part of the natural attenuation study and longterm monitoring of explosives, new types of direct push technologies (DPT) that allow direct detection of contamination, without bringing soil or water samples to the surface for analysis, have been developed in recent years. The DPT has been used to provide data to a multi-objective project conducted at two contaminated sites. The DPT also provided samples for microbiological studies and stratigraphic analysis of Iithologies for the numerical modeling effort. The main objective of the modeling work was to assist field monitoring and data collection for a better demonstration and graphic representation of natural attenuation of explosives. The study has shown that the DPT is an effective tool for sampling both soil and water in varying geologic environments of deposition. DPT technology provided a cost-effective means of providing real-time data concerning the stratigraphy, resistivity and ground water contamination of the sites. The ground water modeling system (GMS) was used to simulate flow and mass transport of contaminants of concern (TNT, RDX) at the sites. Modeling runs for LAAP indicated that the TNT and RDX plumes are static and possibly slowly diminishing 1 Introduction The successful application of monitored natural attenuation reported for other chemicals [1] and the potential for significant cost savings motivated the demonstration project for monitored natural attenuation of explosives.

2 The effectiveness of monitored natural attenuation (MNA) as a remedial alternative depends upon several factors including the type and extent of original or residual contamination, site hydrogeology, and remediation goals. Natural attenuation of explosives may occur through one or more of the following mechanisms: (1) microbial mineralization to very simple, nonhazardous inorganic compounds, (2) microbial transformation to similar or more complex compounds, (3) immobilization due to interactions with the soil or aquifer, and (4) immobilization due to restrictive site geology and/or hydrology. In soil systems, TNT can be reduced to the mono- and di-amino nitro toluenes, which can be incorporated into the soil organic matter. Laboratory evaluations were conducted using [U]-ring- C TNT to measure mineralization. The site selected for this demonstration (Fig. 1) satisfied the criteria that must be considered to build support for natural attenuation. These criteria include the removal and treatment of the contaminant source, historical data that show a decrease of contaminants over time (Fig. 2), and low flow gradient at the site, which allows sufficient time for degradation of residual contaminates before they reach a potential receptor. Figure 1: Louisiana Army Ammunition Plant (LAAP)- Area P and Vicinity

3 Difference ill n o I L El Figure 2: Concentration trends at the selected wells using historical data 1990 and Direct Push Technology Sampling Soil samples were collected tlom 24 locations along 8 transects at LAAP using direct push technologies (DPT). Generally, the penetrations were about 15 m (50 ft)deep andreached theconfming layer. Locations were determined tomeet the following sampling objectives: (1) to refine the vertical and lateral definition of the contamination, (2) to refine site Iithology and hydrogeology, and (3) to provide subsurface soil samples forother tests. Sampling onthe Area Pcap was not permitted due to concern that the integrity of the cap would be compromised. TNT and RDX concentrations from groundwater sampling prior to the date of the CPT event were contoured to identify hot spots prior to establishing transects. The fwst punch on a given transect was used to stratify the site by measuring resistivity to penetration. The strata were defined in terms of lithology. Subsequent punches were used to collect soil samples with a presterilized split spoon, Typically, at least three depths, surface, mid-depth and just above the confining layer were sampled. Well selection criteria included appropriate spatial distribution of wells relative to the original source and relative to the existing conceptualization of the contaminant plume, proximity to potential receptors or site boundaries, and availability of well completion data concerning screen type, depth, and well performance, such as water yields.

4 3 Analytes and analytical methods Due to the several potential attenuation processes affecting explosives, analysis of a battery of parent compounds and transformation/degradation products was necessary. Some of the transformation products of TNT are as environmentally undesirable as the parent compound [2; 3; 4; 5]; therefore, these must be monitored to assure protection of potential receptors. Some geochemical parameters were monitored in the field, while others were assayed in the laboratory. Attenuation processes are typically too slow for accumulation of sufficient degradation products (e.g., nitrates, ammonia, methane, and carbon dioxide) to be distinguished from background. However, the geochemical properties contribute to the weight-of-evidence that conditions in the site are suitable for attenuation processes. 4 Estimates of site capacity for attenuation Adsorption of explosives from groundwater by the site aquifer soils was limited. The measured values of K~ were below 1L kg ]for all soils and contaminants, ranging from no significant adsorption to a high value of 0.84L kg-l. The highest degree of sorption was associated with the soils highest in clay and CEC. The range of sorption coefficients between soils varied over an order of magnitude for explosive compounds. For modeling of contaminant transport at the LAAP site, the use of an average value of KDto represent sorption was appropriate. Although these data indicated limited retardation, low permeability may exert a greater influence on movement of the plume than is exerted by sorption and transformation. These results suggest that mass transport limitations rather than site capacity restrict transport at LAAP. The site exhibited limited capacity for attenuation. 5 Biomarker analyses Three sets of tests were evaluated to determine their usefulness in supporting site evaluation of monitored natural attenuation. These included (1) soil mineralization radioassays, (2) lipid biomarker analyses, and (3) nucleic acid biomarker analyses. These tests were designed to evaluate the microbial degradation potential of the site. Rates of TNT and RDX mineralization were very low in LAAP soils. Rate constants for TNT ranged from less than 1 x 104 to 2.2 x 10-3;for RDX the range was from less than 1 x 104 to 2.0 x Few significant correlations between geochemical parameters and biomarkers were found. Several nucleic acid probes correlated positively with mineralization rate, as did the following parameters determined by lipid biomarkers: biomass, abundance of gram-negative, sulfatereducing and iron-reducing bacteria, and a sulfite reductase. The potential for aerobic degradation of TNT in LAAP soils was suggested by the presence of two

5 catechol oxygenase gene probes. Other observed genes supported potential for both anaerobic and aerobic metabolism of TNT 6 Numerical modeling Numerical modeling provides a mean of quantitatively evaluating multiple natural processes that can be represented as a set of mathematical expressions. The complex and incompletely understood processes involved in the natural attenuation of explosives require the computational power and flexibility of a numerical simulator. Numerical modeling was conducted to compliment the field monitoring and DPT data collection for demonstration and graphic representation of natural attenuation of explosives at the site. The modeling effort focused on conceptualization of the site hydrogeology and reduction of explosives by processes such as immobilization/ degradation, and first order decay. The Department of Defense Groundwater Modeling System (GMS) [7] with its subsurface model, FEMWATER [8], was selected for the modeling element of the demonstration. 6.1 Code Description FEMWATER [8] is a three-dimensional finite element numerical code, which may be used to model flow and mass transport through saturated-unsaturated media. FEMWATER is an enhanced version of two models, 3DFEMWATER [9] and 3DLEWASTE [10]. FEMWATER is integrated into GMS. The flow equations in FEMWATER are based on the continuity and Darcy flow equations. The model application is limited by the assumptions applied to these equations. FEMWATER also can be used for density-dependent problems. FEMWATER simulates the primary processes affecting dissolved-phase contaminant distributions in groundwater including advection, dispersion, sorption, and decay caused by chemical reactions and/or biological transformation. FEMWATER uses the first-order decay as lumped biochemical decay, 6.2 Model limitations Major assumptions and limitations of FEMWATER include the following: (1) the contaminant is transported as a single constituent, thus inter-solute reactions can not be simulated, (2) abiotic and microbial degradation is treated with a tirstorder decay model, (3) adsorption coeftlcient and decay rates can be assigned for different subsurface materials; however, rate constants do not change during simulation time, (4) contaminant sorption is instantaneous and reversible and the adsorbed phase is in local equilibrium. Some of the above assumptions may not be applicable to certain field problems. For this application, the above assumptions were applicable by simplifying some site characteristics required for the modeling without deviating

6 much from actual site conditions. For more details, the reader is referred to FEMWATER model theory documentation [8]. 6,3 Model construction FEMWATER requires basic hydrogeologic and chemical data for simulations. These basic data include hydraulic conductivity, porosity, hydraulic gradient, initial and boundary conditions, distribution (partition) coefficients, and decay rates. The distribution coefficient, Kd, relates the sorbate and solute for linear isotherms. Distribution coefficients for explosives were determined on aquifer soil from LAAP [6]. The retardation factor is a fhnction of both contaminant property, Kd,and soil bulk density, soil porosity, and soil water saturation. The retardation factor provides a general indication of mobility of the contaminant in the soil. The decay rate used in the model were 10-sday- for TNT and 8.13 x 10-6day-l for RDX, which were considerably lower than the batch rates These decay rates were used based on the model calibration, and on the fact that the field decay rates are normally lower than laboratory batch rates. 6.4 Conceptual Model A conceptual model is a powerfid tool for abstracting and simplifying natural phenomena. To develop conceptual models for the site, existing and new hydrogeological data including borehole geologic DPT data, hydraulic conductivity data (measured in DPT samples) and flow boundary conditions were used. Four stratigraphic units were identified at LAAP based on lithologic data, pump test data, and laboratory test results. The subsurface was divided into four hydrogeologic zones; an unsaturated (vadose) soil, an upper terrace aquifer, a semi-confined layer, and a lower terrace aquifer. The soil and the upper terrace aquifer form a shallow unconfined aquifer. The terrace deposits are composed of alternating beds of mixed sands and clay. The distribution of hydraulic conductivityy for each layer was estimated from the conductivity data at the CPT locations. The source of flow recharge at the site was assumed to be from rainfall. Precipitation data collected at the town of Minden, LA near the site were used to estimate average infiltration rate. The yearly averaged rainfall of data was 50.0 in. (127 cm). About i0 of the annual precipitation was estimated to reach the groundwater. 6.5 Initial distributions of contaminants Initial conditions of flow and chemical concentration play a major role in the model outcomes. Different numerical techniques available in GMS were compared to establish a realistic initial flow and mass concentration distributions at the sites. The starting (initial) conditions for the simulation were estimated

7 based ori the first monitoring data. The initial flow and concentration distribution of TNT and RDX were determined using data collected in February 1996, The GMS was used to interpolate/extrapolate the data for all points of the numerical mesh system. 6.6 Model calibration The calibration process is often carried out by manual trial and error, which was the approach employed here. GMS version 3.1 has an option called Map Module, which may be used for the calibration. The differences between the measured and calibrated hydraulic heads were not significant for both sites with complex hydrogeology conditions. The maximum difference between the simulated and measured head was about 3 feet (90 cm) or 1.5 percent of measured heads. The hydraulic conductivity had major effect on the flow calibration compared to the other parameters mentioned earlier. Therefore, more data, which sufficiently provides uniform distribution of hydraulic conductivity, would help the simulated results to be close to the measurements. Both flow and chemical data controlled the transport model of explosives. After the flow model was calibrated, the transport model was calibrated against the chemical input data. The major parameter affected the results of FEMWATER in regard to explosive (transport) was the first-order decay rate. The model had to be calibrated against this rate, The differences between the simulated (calibrated) and measured TNT concentrations were reasonable. The calibrated models were then used for longterm prediction. 6.7 Predictive Simulations The long-term prediction of the plume is needed to demonstrate a clear and meaningfid trend of decreasing contaminant mass and/or concentration over time. The results presented here are based on the following assumptions: 1) no additional source of contamination is added into the site, 2) infiltration rate stays constant throughout the simulations, 3) flow boundary conditions recur every year, 4) no recharge or discharge through pumping occurs during the simulations. The predicted results should be updated and adjusted as new data become available. Jorgensen [11] illustrates an iterative way in which a model prediction may be improved as new information is obtained. A comparison between the simulation for twenty years and the initial conditions of TNT concentration distribution (February 1996) indicated that the TNT plume at LAAP is virtually static to diminishing (Fig. 3 and Fig. 4).

8 : D,mi - m.a, m 700),0+ Wm.m - W.Y. W - +x.1.m - Wd.m. m.m mow O.m : Figure 3: Initial concentration distribution of TNT at the site Y :,,,, $4 Y Lz x Figure 4: Predicted TNT concentration at the site

9 8 Conclusion Monitoring results demonstrated declining contaminated mass over time. Although the site capacity for natural attenuation at the site was low, results of biomarker indirectly demonstrated the potential for microbial attenuation at a slow rate. The model results suggested that this trend would continue. Results of the demonstration suggest that monitored natural attenuation is a viable option that should be among the options considered for remedial of explosivescontaminated sites. References [1] Christensen, T. H., Bjerg, P. L., and P. Kjeldsen, Natural Attenuation: A Feasible Approach to remediation of Ground Water Pollution at Landfills? Ground Water Monitoring and Remediation 20, No. 1:69-77,2000. [2] Honeycutt, M. E., Jarvis, A. S., and McFarland, V. A. (1996). Cytotoxicity and mutagenicity of 2,4,6-trinitrotoluene and its metabolites, Ecotoxicology and Environmental Safety, 35: [3] Jarvis, A. S., McFarland, V, A., and Honeycutt, M. E, Assessment of the effectiveness of comporting for the reduction of toxicity and mutagenicity of explosives-contaminated soil, Ecotoxicology and Environmental Safety 39: , [4] Tan, E. L., Ho, C. -H., Griest, W. H., and Tyndall, R, L. Mutagenicity of trinitrotoluene and its metabolizes formed during comporting, Journal of Toxicology and Environmental Health 36, , [5] Won, W. D,, DiSalvo, L. H., and Ng, J. Toxicity and mutagenicity of 2,4,6- trinitrotoluene and its microbial metabolizes, Applied and Environmental Microbiology 31, , [6] Pennington, J. C., Bowen, R., Brannon, J. M., Zakikhani, M,, Harrelson D., Gunnison, D.W., Mahannah J., Clarke, J. U, Jenkins, Scot Gnewuch, S., Draft Protocol for Evaluating, Selecting, and Implementing Monitored Natural Attenuation at Explosives-Contaminated Sites, Technical Report EL-99-1 O,U, S. Army Engineer Waterways Experiment Station, Vicksburg, MS, [7] Department of Defense Groundwater Modeling System (GMS), Reference Manual.. Version 2.0, Brigham Young University, Engineering Computer Graphic Laboratory, Provo, UT 84602, [8] Lin, H-C. J., Richards, D. R., Cheng Yeh, G-T, Cheng, J-R, Cheng, H-P, and Jones, N. L. FEMWATER A three-dimensional finite-element computer model for simulating density dependent flow and transport, Technical Report CHL-97-12, U. S. Army Engineer Waterways Experiment Station, Vicksburg, MS, [9] Yeh, G.T. 3DFEMWATER: A three-dimensional finite element model of water flow through saturated-unsaturtaed media. ORNL-5567/R 1, Oak Ridge National Laboratory, Oak Ridge, TN, 1987.

10 [10] Yeh, G. T. 3DLEWASTE: A hybrid Lagrangian-Eulerian finite eiement model of waste transport through saturated-unsaturated media. Pennsylvania State University Technical Report, Department of Civil Engineering, Pennsylvania State University, University Park, PA, [11] Jorgensen, D. G. Geohydrologic models of the Houston District, Texas. Ground Water 19, no. 4: , [12] Harrelson, Danny W., Zakikhani M., and Judith C. Pennington, Natural Attenuation of Explosive Chemicals at the Louisiana Army Ammunition Plant, Gulf Coast Association of Geological Societies Transaction, v. 49, p. 129, 1999.