Vadose Zone Contaminant Fate and Transport Analysis for the 216-B-26 Trench
Summary (14 min read)
Introduction
- The BC cribs and trenches in Hanford’s 200 Area are believed to have received about 30 Mgal of scavenged tank waste containing an estimated 400 Ci of 99Tc as well as large quantities of NO3- and 238U.
- The purpose of this study was to develop a conceptual model for contaminant fate and transport at the 216-B-26 Trench site to allow interpretation of the current contaminant distributions and to support identification, development, and evaluation of remediation alternatives.
- For the BC Trench site, such confirmation may possible with a detailed analysis of the features, processes, and events at the 216-B-26 Trench.
1.1 Purpose and Scope
- Accelerated remediation of the BC Cribs and Trenches Feasibility Study (FS) may be viewed as occurring in three phases: the development of alternatives, the screening of the alternatives, and the detailed analysis of alternatives.
- In the following sections, the authors develop containment alternatives for remedial actions and evaluate these alternatives against a no-action alternative to allow elimination of those that would be unable to meet 1.3 preliminary remediation goals.
- The conceptual model describes the important features, events, and processes controlling fluid flow and contaminant transport at the waste site of interest.
- The model is dynamic, becoming more refined as more information about the specific site becomes available.
- The problem of interest here is the fate and transport behavior of contaminants of concern, primarily 99Tc, in the vadose zone at the BC cribs and trenches, with special interest in the 216-B-26 trench.
2.1 Site Features
- Site features include the surface and subsurface physical structure, e.g., hydrostratigraphy, hydraulic, geochemical, and biochemical properties, of the vadose zone that may impact contaminant migration.
- Field and laboratory measurements on undisturbed and repacked samples show the hydraulic and geochemical properties of Hanford sediments to be highly variable in space, even within apparently homogeneous hydrostratigraphic units.
- Thus, the finegrained units, many of which are ≤0.15 m thick, are often not identified.
- There are very few sites that have been subjected to any type of analyses to quantitatively describe the internal structure and heterogeneities in outcrop and core samples.
- Thus, in many cases, there is a lack of site-specific data to support the development of detailed three-dimensional (3-D) geologic models for a given waste site.
2.1.2 Hydraulic Properties
- Accurate predictions of flow and transport in the vadose zone require a characterization of the hydrologic properties.
- No hydraulic properties were measured on the samples from the C4191 borehole and no undisturbed samples were available for characterization.
- Sampling was limited to grab samples, which had undergone considerable mixing, thereby making the interpretation of laboratory measurements a difficult task.
- The pedotransfer functions were used to estimate the saturated hydraulic conductivity, Ks; saturated water content, θs; the Brooks-Corey pore size distribution index, λ,; and the Brooks-Corey bubbling pressure, ψb.
- Calculated parameters are summarized in Section 3.4.
2.1.3 Transport Properties
- Accurate predictions of flow and transport also require detailed characterization of the transport parameters, particularly dispersivity, α. Dispersivity is well known to be scale-dependent, but there are few data for Hanford sediments, especially at scales relevant to the current problem.
- Recent field experiments resulted in the estimation of longitudinal and transverse dispersivities at the Army Loop Road site, but transport distances were limited to a maximum of 1 m (Ward and Gee 2003).
- Thus, αL tended to increase as soil texture became finer and with increasing clay content or as the Brooks-Corey pore-size distribution index, λ, decreased.
- These values, however, are local-scale values and are expected to be smaller than the values needed to predict field-scale behavior.
- The resulting values represent effective values that reflect the total contribution of the various layers.
2.1.4 Geochemical Properties
- Detailed characterization of the geochemical parameters used to describe sorption is also needed for predicting the transport of contaminants in heterogeneous sediments.
- The high values also have high standard deviations and the 1 mL/g has not been substantiated by any other studies.
- Under normal Hanford conditions, zero appears to be the most appropriate value, with a best estimate for the range as 0.0 to 0.1 mL/g (Cantrell et al. 2003).
- Under 2.10 natural Hanford groundwater conditions, U(VI) adsorption is moderate with Kd values ranging from approximately 0.2 to 4 (Cantrell et al. 2003).
- For this conceptual model, a linear sorption isotherm model is assumed.
2.2 Site Events
- Site-specific events are also included in developing the conceptual model because these events affect the choice of initial and boundary conditions as well as the spatial extent of the computational domain.
- Events that are typically included are the natural (e.g., meteoric recharge) and man-made (e.g., accidental and intentional) fluid and contaminant releases.
- Liquid discharges were of two types, 1) unrestricted, which went to cribs, and 2) restricted, which went to specific retention trenches.
- Discharges to most cribs and trenches were completed over relatively short times, ranging from a month to a year.
- These discharges therefore created a strong driving force for vertical transport through the vadose zone.
2.2.1 Release Events
- Release events represent the source terms in the conceptual and numerical models.
- These events must be characterized for quantity and duration of the releases.
- Given the potential for lateral migration of contaminants in these sediments, it is necessary to consider the trenches adjacent to the 216-B-26 trench.
- Six trenches in the immediate vicinity of the 216-B-26 Trench were operational during the same time frame as 216-B-26.
- Median fluid discharge volumes and contaminant inventories, derived from the SIMS model of Simpson et al. (2001), are summarized in Table 2.1 through Table 2.4.
2.2.2 Recharge Events
- 13 trench would have resulted in land clearing.
- It appears that the trench surfaces were kept relatively free of vegetation because a more recent aerial photograph from 2002 show an essentially bare surface similar to that in Figure 2.7 except that there is now a finer-textured overburden layer.
- Backfilling and stabilization would have resulted in a moderate reduction while remediation and closure would bring about a significant reduction, 2.14 depending on the chosen remedy.
- Long-term recharge estimates from natural precipitation (3.5 mm/yr) are based on data reported by Fayer and Walters (1995).
- Recharge rates subsequent to remediation and closure are based on the performance criterion of the prototype Hanford barrier (Ward et al. 1997).
2.3 Fate and Transport
- An understanding of the fate-and-transport processes acting on the contaminants beneath the BC trench site is a pre-requisite for predicting the partitioning of the contaminants of concern (COCs) between the different components of the environment (i.e., soil, water, and air).
- Once this water passes the root zone where evaporation and transpiration can affect the quantity available for transport, it can potentially reach the water table.
- The extent to which these gradients develop is largely controlled by the sediment thermal and hydraulic properties and to a lesser extent by the fluid properties.
- In many respects, the physical and chemical processes prevailing in the vadose zone differ from those that predominate in surface and groundwater (Runnells 1995).
- In the following sections, the authors provide a general description of the fate-and-transport processes and their possible role and importance to the behavior of the major COCs at the 216-B-26 Trench.
2.3.1 Precipitation/Dissolution
- The chemistry of water infiltrating from snowmelt and rainfall generally evolves to different chemical compositions as the water moves through the vadose zone to the water table.
- Thus, the final composition of the vadose zone pore water depends on the mineralogy of the soil as well as the texture and permeability that control the contact time through their effect on permeability (Sparks 1989).
- This is important for the remobilization of existing plumes as there is net dissolution if the vadose zone water is under-saturated with respect to the mineral or chemical species.
- There is much experimental evidence to show that precipitation reactions control the aqueous concentration of many contaminants (Runnells 1995).
- Recent experiments on the migration of strontium conducted under the Vadose Zone Transport Field show that sorption of strontium in Hanford sediments is also strongly influenced by equilibrium 2.15 with calcite (Ward et al. 2003).
2.3.2 Volatilization
- At the BC cribs and trenches, the list of COCs shows several species for which volatilization could be important.
- Mercury in solution can also be volatilized in anaerobic environments and by reaction with dissolved humic acids.
- The importance of volatilization as a transport mechanism depends on the depth to the zone of contamination and the wind velocity at the soil-air interface.
- The COCs at 216-B-26 are at low concentrations (below saturation) and are most likely dissolved in the water phase (i.e., no free non-aqueous phase liquids).
- Volatilization is therefore ignored in this analysis.
2.3.3 Sorption/Desorption
- Sorption/desorption reactions are based on the principle of attraction between aqueous species and the reactive surfaces of the solid phase of soils and sediments.
- At high pH, the surfaces have a net negative charge and therefore attract the positively charged species and repel the negatively charged species.
- Because sorption is a surface phenomenon, the smaller the sediment particle size, the greater the surface area per unit mass of media (specific surface area) and the greater the adsorption.
- Sorbed ions can also be desorbed from the surfaces of soils and sediments when introduced to a solution of different chemistry, especially when the salinity is high.
- This process has been shown to be important in the sorption of 137Cs where high Na+ reduces the normally high Kd of 137Cs (Saiers and Hornberger 1996).
2.3.4 Degradation/Transformation
- In addition to the processes described above, contaminants in the vadose zone can also undergo degradation and transformation over time as a result of chemical and microbiological reactions.
- Degradation and transformation may result in compounds that are less toxic, compounds that are more strongly adsorbed and therefore move more slowly through the soil profile, or compounds that are less soluble.
- Under oxidizing conditions, 99Tc exists as the soluble heptavalent pertechnetate ion, TcO4-, which, owing to its negative charge, is highly mobile in soils and groundwater (Lieser and Bauscher 1987).
- Owing to the long half-lives of the two species of interest, 238U and 99Tc, radiolytic decay and the formation of daughter products are ignored.
- Biochemical transformations of NO3- are also ignored.
2.3.5 Evaporation/Transpiration
- Water can be removed from the vadose zone via losses to the atmosphere.
- These losses consist of evaporation from wet soil and plant surfaces as well as from plants through transpiration.
- Tumble weeds, with roots that can reach down to 20 ft, can take up 90Sr, break off, and blow off waste sites as reported by Marshall (1987).
- While plant uptake can bring dissolved contaminants to the surface where they can become incorporated into plant biomass, water loss from the soil, through passive or active evaporation processes, can lead to a surface accumulation of previously distributed chemicals.
- At this stage of the investigation, a decision was made to ignore these processes.
2.4 Summary of Conceptual Model
- With the features, events, and processes described above, a conceptual model can be developed for the subsurface transport at the 216-B-26 trench.
- It is possible that the absence of any major recharge events or subsequent liquid discharges that could overcome the capillary break at 33.5 m is responsible for the contaminants accumulating in this region above this coarse-textured zone and resulting in higher concentrations in the trailing edge of the plume.
- Lateral spreading can be enhanced by the sloped layers.
- In the first approach, the existing contaminant inventory distribution can be superimposed or convolved with an applicable transport model to predict its transport through the vadose zone under different conditions.
- The system response is based on numerical solutions to the Richards’ flow equation for variably saturated flow and the CDE for multidimensional advection and dispersion.
3.1 Transport Under Steady-Flow Conditions
- An analytical solution to the CDE was used to examine the rate at which a finite amount of a contaminant initially present in the soil either accumulates at the surface or leaches downward under steady-state flow 3.2 conditions.
- Expressions derived by Elrick et al. (1997) were used to describe the spatial and temporal distributions of the resident concentrations of a subsurface pulse input of contaminant.
- These expressions can take into account linear adsorption and first-order decay as well as the effects of a depth-dependent water-content distribution that is assumed to be invariant with time.
- Equations for the equilibrium distribution near the surface during evaporation and for movement toward the water table were used to describe the behavior of finite amounts of the contaminant initially present in the system.
3.1.1 Solution for Contaminant Inventory at Depth, h
- By defining the initial contaminant distribution in terms of f(z) = δ(z–h), the probability density function of the resident concentration for a soil in which the water content is constant with depth is given by Elrick et al. (1997).
- For spatially invariant water content, the spatial components of the hydrodynamic dispersion coefficient include dispersive and diffusive elements and are given by 3.3 mzz DvD +=α (3.3) where αz is the dispersivity in longitudinal direction, and Dm is the molecular diffusion coefficient.
- Spatially invariant water-content profiles are seldom found in field soils.
- For such conditions, Elrick et al. (1997) derived a solution to the CDE for variable θ(z) by transforming the space variable in the CDE to a water-storage term and the time variable to q*t.
3.1.2 Accumulation of Contaminant Under Steady Evaporation
- Under steady evaporation, v is negative, and the upward convective flow counteracts the downward dispersion.
- The resulting distribution is controlled solely by the ratio v/Dz.
3.1.3 Concentration Crossing the Water Table Based on an Arbitrary Initial Distribution
- The concentration crossing a compliance plane at depth z, defined by the water table, is equivalent to a flux concentration and is derived from the resident concentration probability density function using the following relationship (Jury and Roth 1990): z f V Dff RzRF ∂ ∂ −= (3.9) F θθθθθ θθθθ θθθ θθ (3.11).
- Potential impacts to the groundwater can be easily determined for arbitrary inventory distributions and steady-state water-content (dθ(z,t)/dt = 0.profiles.
- The probability density functions given in Equations (3.2) and (3.5) and Equations (3.10) and (3.11) form the basis of analytical solutions for more complex inventory distributions that can be obtained by convolution.
- It should be noted, however, that these relatively simple solutions are based on the assumption of steady flow in very complex natural systems in which flow is typically transient.
- The concentrations predicted at the compliance plane are also leachate concentrations and are therefore not representative of the concentration at a receptor well.
3.1.4 Spatial Moment Analysis
- In the absence of complex state-variable models, a particularly useful method for analyzing spatial data is the method of spatial moments.
- The first moment, m1, measures the mean location of the plume, which when divided by the mean travel time gives the mean pore-water velocity.
- To characterize the 99Tc concentration profile obtained from the C4191 borehole, the zeroth, first, and second moments were determined.
- The analysis of the spatial moments was then extended to )(.
- In addition, the data were fitted to the convolution model by inverse methods to determine the effective dispersion coefficient and pore-water velocity.
3.2 Transport Under Transient Flow Conditions
- Simulating the effects of time-varying surface recharge and operation of the trenches required a transient flow solution to be executed with the solute transport calculations.
- The transient flow-and-transport simulations were initiated using a steady flow solution to the boundary-value problem using the initial boundary values.
- The second stage represented the period after trench operations following backfilling of the trenches.
- These files are used to generate color-scaled plots and animations through Tecplot.(a) A utility program, PlotTo.pl, was used to translate STOMP plot files into Tecplot-formatted input files.
- Surface-flux files are also used to generate rate and integral plots of solutes exiting the computational domain and entering the groundwater.
3.3 Solution Domain
- The physical domain considered for the simulation is a 2-D north-south cross section through trenches 216-B-52 at the north to 216-B-28 to the south.
- Graphical representations of the stratigraphy and trenches were converted to soil zonation maps based on a tilted Cartesian grid.
- The pre-operations condition was simulated as a transient flow problem, starting with a unit-gradient initial condition, which was run out to a time where flow became steady at the recharge rate of interest.
- Nodes representing the trenches were inactive and recharge was applied directly to the trench bottom.
- A number of utilities were developed in-house to convert STOMP plot files into specially formatted input files for graphical presentation using commercial software tools .
3.4 Model Parameterization
- No measurements of flow-and-transport parameters were available for the BC Crib site, and no samples were collected to determine these parameters during the sampling phase.
- Meteoric recharge and parameters for vadose zone flow and transport were developed by PNNL based on published site data and experiments conducted onsite.
- Flow-and-transport parameters were assigned based on the similarity between grain-size statistics of the different soil textures at the site and at previously characterized sites.
- These parameters (Table 3.1 and Table 3.2) and the rationale for their selection are presented below.
3.4.1 Recharge Rates
- Groundwater recharge is representative of the soil water flux through a waste zone to the water table and thus strongly impacts estimated rates of contaminant leaching and remobilization that may occur.
- The first moment, m1, which locates the center of mass, was then used to estimate the travel time and recharge rate according to Equation (3.15).
- The layer sequence in the plume constitutes a capillary break; thus, the plume separation is most likely due to a decrease in suction to a value less than the water entry pressure of the coarse layer after the leading edge passed.
- Data from the Hanford barrier show an average recharge of 27.6 percent of annual precipitation from sparsely vegetated side slopes.
3.4.2 Flow-and-Transport Properties
- Hydraulic properties were estimated based on similarities in grain-size statistics (mean grain size and sorting index) between sediments at the BC Crib site and other characterized sites at Hanford using pedotransfer functions.
- Variable or saturation-dependent anisotropy provides a framework for simulating the effects of saturation on lateral spreading.
- Solute transport parameters include bulk density, diffusivity, sorption coefficients, and macrodispersivity.
- Bulk density for the different sediments was also derived from pedotransfer functions.
- The saturated hydraulic of the aquifer was assumed to be invariant with a value of 1615 m/day in the horizontal direction and 161.5 m/day in the vertical direction.
3.4.3 Bulk Density and Distribution Coefficient
- Both bulk density (ρb) and the distribution coefficient (Kd) estimates are needed to calculate retardation factors for different COCs.
- Under oxidizing conditions, 99Tc exists as the soluble heptavalent pertechnetate ion, TcO4-.
- While these findings provide bounds for Kd values that can be used in a sensitivity analysis of fate and transport, there is no current evidence to support the sorption of 99Tc in Hanford sediments.
- The field-scale estimate of bulk density was calculated as the average of small-scale estimates derived using pedotransfer functions.
3.4.4 Diffusivity and Dispersivity
- The molecular diffusion coefficient for the different species in pore water was derived from literature values (Kemper 1986).
- Owing to the hierarchical nature of heterogeneity that exists in natural soils, it has been suggested that dispersivity is dependent on the scale of observation (Beven et al. 1993).
- There are very few data for Hanford sediments, especially at scales relevant to the current problem.
- An estimate of fieldscale dispersivity was derived from the observed 99Tc distribution by fitting the concentration data to Equation (3.2).
- The resulting values represent effective values that reflect the total contribution of the various layers.
3.5 Input File Generation
- Two types of files were used to drive the STOMP simulator: 1) a simulation control file and material definition and 2) a soil file .
- All input files were written and stored in ASCII text format.
- The simulation control and material definition input files were assembled using a conventional text editor, whereas the zonation file was generated with a utility program.
3.5.2 Zonation File
- The zonation file is an ordered listing (i.e., i,j,k indexing) of integers that identify the rock/soil type for every grid cell in the computational domain.
- The zonation file was generated using information from the geologist’s descriptions of borehole cuttings and geophysical logs.
- Thus, the vertical extent of the computational domain was set at 108.17 m. 3.15 A similar plot for a north-south dissecting the series of trenches is shown in Figure 3.3.
3.5.3 Boundary Conditions
- In all of the simulations, a no-flow boundary was imposed at the bottom of the domain, representing the base of the 5-m thick confined aquifer at 108.17 m.
- Owing to observation of water leaving the monitored domain at the 299-E24-111 test site via the fine-textured layers , the horizontal scale of the modeling domain was increased by 200 m on the side boundaries (north, east, south, and west).
- These boundaries were designated as zero-flux boundaries for water flow and solute transport.
- For the 2-D north-south transect, groundwater was assumed to flow in a southerly direction under a gradient of 1.486⋅10-3 m/m.
- Thus, the south boundary of the aquifer was treated as a hydraulic gradient boundary that allowed water and solutes to flow out.
3.5.4 Source Terms
- The source terms consisted of fluid and contaminant discharges to the series of trenches during trench operations.
- Fluid discharges are reported to have started in late 1956 and ended in early 1957 for most trenches, except for the 216-B-52 Trench, which was operational into 1958.
- All contaminant inventory was assumed to be dissolved in the discharged fluids; thus, the time history is identical to that of the fluid releases.
- Ci while the total nitrate inventory was estimated at 6.7 million kg.
- This rate was then divided by the number of nodes (n = 6) extending over the width of the trench bottom to generate the solute density for each node.
3.6 STOMP Execution
- Owing to the size of the 3-D computational domain (71 × 78 × 1317 =7,293,546 nodes), STOMP transient simulations of field-scale flow on this domain were conducted on the Environmental and Molecular Sciences Laboratory’s (EMSL’s) MPP2 super computer.
- In both cases, the executable form of the code 3.18 was generated from the source code that is under version control by PNNL.
- Next, the source code was compiled and linked using the previously generated parameters definition file and the appropriate libraries to generate the executable form of the code.
- Simulations with the 3-D domain were made using the parallel STOMP90 with the Portable Extensible Toolkit for Scientific Computation solver.
- This solver, which is based on the conjugate gradient method, has a compact storage scheme for the Jacobian matrix and is preferable to the direct band solver for problems over 35,000 nodes.
3.7 Soil-Screening Process
- Simulated concentrations of residual soil contaminants must be evaluated to determine if the contaminants will eventually migrate to the underlying aquifer.
- For such an analysis, the EPA recommends the use of soil-screening levels (SSLs).
- The DAF is essentially the ratio of contaminant concentration in leachate out of the vadose zone to the concentration in ground water at the receptor point.
- In addition, the following assumptions are made: 1) equilibrium between contaminants and soil/water is instantaneous, 2) the adsorption isotherm is linear with concentration, 3) volatilization is negligible, 4) contaminants are homogenous at the source, and 5) the receptor is located at a hypothetical well downstream of the waste 3.19 site, and (6) the system is isotropic.
- Such detailed modeling may ultimately lead to a less restrictive, but still protective, SSL.
3.7.1 Soil-Screening Levels
- Two methods can be used to calculate the receptor point concentration using the leachate radionuclide concentrations to ground water provided by the unsaturated model.
- The first term, in Equation (3.19) estimates the depth of mixing due to vertical dispersivity along the length of ground water travel.
- Aquifer parameters in Table 3.3 were used to estimate the site-specific dilution factors for use with the convolution model.
- Use of the STOMP simulator for SSL calculation requires more site-specific data and a greater modeling effort than using the simple site-specific SSL calculation.
- Inclusion of a confined aquifer at the base of the simulation domain allowed direct simulation of flow and transport in the saturated zone.
3.7.2 Steady-State Convolution Solutions
- Dimensional analysis of the 1-D steady-state convolution solution shows the units of the resident concentration, CR, to be pCi per cubic meter of soil, pCi mb-3.
- The units of the flux concentration, CF, were pCi per cubic meter of soil, pCi mb-3.
- In the analysis, the mean bulk density for the entire simulation domain is used.
- To convert the predicted CR to concentration per unit volume of pore water, the following relationship was used: θ⋅ = 3 * 10 ),( ),( tzC tzC RR (3.19) Uncertainty in the SSL estimates due to uncertainty in the input parameters can be easily accommodated by incorporating probability density functions for the parameters into the solution.
3.7.3 STOMP Simulations
- The STOMP simulator requires fewer assumptions than the simple site-specific SSL calculation based on the convolution solution.
- The STOMP simulator does not perform the required calculations directly.
- Calculation of the breakthrough curves of contaminant concentration at the water table from STOMP simulation results requires some additional processing.
- CERCLA § 121(d), 42 U.S.C. § 9621(d), further specifies that a remedial action must attain a level or standard of control of the hazardous substances, pollutants, and contaminants, which at least attains applicable or relevant and appropriate requirements under federal and state laws, unless a waiver can be justified pursuant to CERCLA § 121(d)(4), 42 U.S.C. § 9621(d)(4).
- The scope of this study is to screen these alternatives to identify those that would minimize the long-term downstream transport of 99TC and NO3- through the vadose zone.
4.1 Detailed Analysis
- The alternatives are analyzed individually against a set of evaluation criteria and then compared against one another to determine their respective strengths and weaknesses and to identify the key trade-offs that must be balanced for the site.
- The results of the detailed analysis are summarized so that an appropriate remedy consistent with CERCLA can be selected.
- The results of the detailed analysis are summarized so that the most effective capping strategy can be pursued.
4.1.1 No-Action Alternative
- The no-action alternative provides an environmental baseline against which impacts of any proposed action and the alternatives can be compared.
- In general, these would include any environmental impacts associated with not satisfying the underlying purpose and need for action.
- At present, the trench surfaces are bare to sparsely vegetated with no infiltration controls.
- The no-action alternative may or may not be a reasonable alternative.
- It is sometimes coupled with monitored natural attenuation, which relies on natural processes to attenuate contaminants in soil and groundwater.
4.1.2 Capping Alternative
- This alternative includes remediation by capping without removing any of the contaminated sediments.
- The fine-textured material minimizes recharge by storing precipitation until it can be recycled to the atmosphere by plants.
- The total area to be remediated would depend on the results of the model analysis and the geophysical investigations.
- At the time of this analysis, capping alternatives had not been identified.
- Table 4.1 summarizes the assumptions about recharge.
4.1.3 Summary of Simulation Cases
- It is assumed to be sandy material with some water-holding capacity that eventually developed some sparse vegetation.
- Following stabilization, the recharge rate was reduced to 25 mm yr-1 and remains fixed until 2012.
- The difference between the five cases results from differences in the drainage criterion for the covers, which are assumed to be of different degrees of robustness.
- For each remedial alternative, model simulations were performed to determine the extent of chemical impacts on soil and groundwater; to identify underlying soil and groundwater flow patterns, and to select the most feasible alternative.
5.1 Steady-State Transport Simulation Results
- The steady-state convolution solutions considered either 1) a depth invariant, θ (z), represented by a mean water content 0.08294 m3 m-3, or 2) a depth-dependent water content, θ (z), derived from the neutron-log measurements.
- Two approaches were used to compare the observed data to the model predictions.
- In the first approach, the plume was treated as a single entity, and the mean recharge rate was estimated from the location of the center of mass and the travel time to the center of mass of the complete plume.
- In the second approach, the plume was decomposed into leading and trailing components, and the analysis was repeated for the two components.
- The results for the depth invariant water content are summarized below.
5.1.1 Fitting the Observed Data
- The field data from the C4191 borehole were fitted to the analytical solution to estimate the mean porewater velocity, from which the recharge rate can be inferred when the mean water content is known as well as the dispersivity.
- Figure 5.1a shows the result when only the longitudinal dispersion coefficient, D, and the pore-water velocity were fitted to the data.
- The fitted profile shows a much larger concentration between the surface and the peak and between the peak and the leading edge of the plume.
- The fitted velocities correspond to mean travel depths of 39.837 m for the leading edge and 29.694 m for the trailing edge.
- Such information, which provides insight into the physical heterogeneity and the relationships between recharge rate and flow anisotropy, can only come for subsurface characterization or field experiments at representative sites.
5.1.2 Soil-Screening Levels Following Remediation
- The convolution models are of greater utility for predicting the potential migration of contaminants initially present in the soil.
- As can be expected, as the mean recharge rate decreases, the pore-water velocity decreases the rate of advance of the plume.
- Simulations with the 1-D model in which measured water content variations with depth taken into consideration resulted in somewhat lower soil concentrations at equivalent times.
- The results for NO3- are summarized in Table 5.2.
5.1.3 Peak Concentrations and Arrival Times at the Water Table
- Figure 5.6 through Figure 5.9 show the temporal variations in 99Tc at the water table predicted by the 1-D convolution model.
- The 900 pCi/L MCL is included on each graph for a reference.
- In should be noted that the 1-D convolution model assumes an average recharge rate and does not take into account episodic highintensity events that could have a greater short-term impact.
- The time to arrival of the MCL decreased from the year 2083, only 78 years from the present, under an assumed no-action recharge rate of 25 mm yr-1, to the year 7792 when a barrier limiting the recharge to 0.5 mm yr-1 was assumed.
- The concentrations reaching the water table can be used to predict 5.13 concentrations at a receptor well using the dilution attenuation factors derived above or by using the results as input to an analytical ground water transport model.
5.2 Transient Transport Simulation Results
- The dispersion of contaminants in the vadose zone and groundwater is an inherently 3-D process.
- Burnett and Frind (1987) found that in vertical-plane simulations, the dimensionality effect became noticeable at a transverse dispersivity of only 1 cm and could result in significant overestimation of a plume length.
- Only transient multidimensional simulations are capable of producing the actual concentration values that drive local processes, including chemical reactions.
- In the sections that follow, results of the transient simulations performed with the STOMP simulator are presented.
- A summary description and comparison of results follows the individual case descriptions summarized in Table 4.1.
5.2.1 Initial Conditions and Saturation Distributions
- In the first phase of the transient simulations, the model was run from time zero to the year 1956 with a recharge rate of 3.5 mm yr-1.
- Simulations with this stratigraphic model resulted in spatially continuous capillary breaks.
- Figure 5.12 shows a simulated moisture-content profile through the center of Trench 216-B-26.
- Furthermore, the range and mean values are quite similar to the observed values.
- The model does tend to over-predict water content relative to the neutron probe, and the discrepancy appears to increase with depth.
5.2.2 Distribution of Water During Trench Operations
- In a typical study of contaminant fate and transport to support the identification of remedial actions, contaminant distributions are needed from multiple locations to establish the initial condition.
- For this part of the simulation, waste-release information was used to generate the necessary boundary conditions for each trench, and the water and dissolved contaminants were injected at the base of each trench at the appropriate rate.
- Figure 5.18 shows the water mass flux across the water table during and after trench operations.
- Thus, these results should be interpreted and treated within the limits of uncertainty of the stratigraphic model and hydraulic parameters.
5.2.3 Distribution of Contaminants During Trench Operations
- The spatial distribution of contaminants and their mass flux across the different boundaries are examined in a fashion similar to that used for water in the previous section.
- The result is a distribution that shows slightly increased mobility in the coarser materials than in the fine-textured lenses.
- The pattern of migration of 99Tc and NO3- across the boundaries should therefore mimic that of water to a large extent.
- Even though saturation-dependent anisotropy predicts an increased tendency for lateral migration, the hydraulic conductivity at low water contents may be too low to significantly affect redistribution at the low potential gradients.
5.2.4 Current Distribution of Contaminants
- Figure 5.24 shows the predicted distribution of 99Tc at the beginning of year 2005.
- When concentration is expressed as the total 99Tc concentration, i.e. concentration per unit volume of soil, the effects of depth variations in water content and bulk density are readily apparent .
- While both of these factors may be responsible for the higher predicted concentrations, the simultaneous prediction of higher apparent dispersion with a higher peak concentration creates a dilemma.
5.2.5 Soil-Screening Levels Following the No-action Alternative
- Figure 5.29 and shows the predicted pore-water concentrations for 99Tc in the year 3000, after the arrival of the peak at the same recharge rate.
- The peak concentration reaching the receptor well is 3147 pCi/L, which exceeds the MCL for 99Tc.
- When compared to the 1D model, a multidimensional model allows redistribution of the plume in the longitudinal and transverse horizontal directions.
- Thus, in all cases of no-action, soil-screening levels are exceeded and the concentration reaching the water table and receptor would likely exceed the MCL.
5.2.6 Arrival Times and Concentrations Under the No-action Alternative
- Figure 5.30 through Figure 5.33 show the mass flux of 99Tc and NO3- at a receptor 100 m down-gradient of the source under the no-action alternatives.
- Nitrate shows similar trends although the first arrival is generally earlier due to the larger concentrations of NO3- and the relatively low MCL.
- In all cases, the concentration reaching the water table increased with a decreasing recharge.
- As shown in Table 5.4, concentrations arriving at the receptor decreased significantly as the recharge rate decreased.
- As pointed out earlier, it would have been impossible to make independent predictions of the current inventory distribution based on 1-D simulations.
5.2.7 Soil-Screening Levels Under the Capping Alternatives
- Figure 5.34 shows the predicted pore-water concentrations for 99Tc in the year 2400 shortly after the arrival of the MCL at the water table under the 0.5 mm yr-1 barrier.
- Figure 5.35 shows a similar plot after the arrival of the peak.
- The values observed at the receptor well are much smaller and are actually below the MCLs.
- To gain further insight into the impact of capping alternative on the contaminant inventory and groundwater quality, the authors calculated the concentrations of 99Tc and NO3entering the water and receptor well for the different capping strategies.
5.2.8 Arrival Times and Concentrations Under the Capping Alternatives
- Undoubtedly, all of the capping alternatives delayed the arrival times of 99Tc and NO3- at the water table significantly beyond those predicted for the no-action alternatives.
- “Dispersion parameters for undisturbed partially saturated soil.” J. Hydrol.
- Handbook of Vadose Zone Characterization and Monitoring, also known as In.
- PNNL 14180, Pacific Northwest National Laboratory, Richland, WA.
Did you find this useful? Give us your feedback
Citations
168 citations
Cites methods from "Vadose Zone Contaminant Fate and Tr..."
...The Tc(VII) aqueous concentrations used (10– 20 lmol L 1) were within the range of those observed in porewaters of Hanford’s BC–crib complex (Serne and Mann, 2004; Ward et al., 2004) where over 400 Ci were released to the vadose zone....
[...]
36 citations
33 citations
26 citations
Cites background from "Vadose Zone Contaminant Fate and Tr..."
...D flow is both qualitatively and quantitatively different, and straightforward extrapolation from lower to higher dimensions in highly heterogeneous formations will lead to significant errors (Dagan 1989; Murray et al. 2001, 2003; Ward et al. 2004)....
[...]
...Recent sampling at the 216-B-26 Trench shows a zone of high 99Tc concentrations between 18 and 53 m (59 and 174 ft) with no reported groundwater contamination (Ward et al. 2004)....
[...]
11 citations
Additional excerpts
...This scale of heterogeneity is typically ignored, but its importance to transport behavior at the Hanford Site has been demonstrated in simulations conditioned on high-resolution (0.076-m) borehole geophysical logs (Ward et al., 2004)....
[...]
References
12,446 citations
4,301 citations
3,173 citations
2,747 citations
"Vadose Zone Contaminant Fate and Tr..." refers background in this paper
...The dispersion process is characterized by a local dispersion tensor, D, which is related to the longitudinal and transverse local dispersivity, αl and αt, respectively (Bear 1972)....
[...]