S-SX Contaminant Migration
In support of CH2M Hill Hanford Group, Inc.'s (CHG) preparation of a Field Investigative Report (FIR) for the Hanford Site Single-Shell Tank Waste Management Area (WMA) S-SX, a suite of numerical simulations of flow and solute transport were executed to predict the performance of surface barriers for reducing long-term risks from potential groundwater contamination at the S-SX WMA. The scope and parametric data for these simulations were defined by a modeling data package provided by CHG. This report documents the simulation of eight cases involving two-dimensional cross-sections through the S-SX WMA and one three-dimensional domain of a single tank (quarter symmetry) within the S-SX WMA. The suite of two-dimensional simulations were used to investigate the impact of surface barriers, water-line leaks, clastic dikes, nonuniform inventories, inventories displaced toward the water table, and concentration dependent density and viscosity for the transporting fluid (i.e., water). The three-dimensional simulation was used to investigate the impact of dimensionality on the numerical predictions. Four transported solutes were considered: Tc-99, Cs-137, NO-3, and Cr.
The large quantity of simulation data makes its reproduction in numerical, graphical or visual form impractical. Therefore, selected results are presented in this report, with a majority of the data being archived in electronic form. The two principal objectives of this work were to conduct the simulations and analyses using an open scientific approach and to provide modeling results that could be verified and repeated. In partial fulfillment of these objectives, the source coding for the STOMP simulator, ancillary utilities coding, input files, simulation output files, and converted result files have been archived in electronic form, with sufficient detail to repeat the calculations reported herein.
All simulations comprised steady-flow and transient components, where flow fields developed from the steady-flow component were used to initialize the transient simulation. Steady-flow initial conditions were developed by simulating from a unit hydraulic gradient condition to a steady-flow condition, dictated by the initial meteoric recharge at the surface, water table elevation, water table gradient, no flux vertical boundaries, soil-type zonations and hydrologic properties and location of impermeable tanks. From these starting conditions, transient simulations of solute transport were executed for a 1000-year period (i.e., year 2000 to 3000) that involved changes in the flow fields in response to the application of surface barriers, water-line leaks, or solute-concentration dependent density and viscosity. The physical domains for the two-dimensional simulations were east-west sections across the S-SX WMA boundary. These domains were discretized with grid resolutions of 0.5334 m (1.75 ft) in the horizontal direction and 0.4572 m (1.5 ft) in the vertical direction, yielding 42,900- to 48,516-node grids. The simulations involving a clastic dike used grid refinement to resolve the clastic dike, yielding 50,232-node grids. The physical domain for the three-dimensional simulation was a quarter section of tank SX-108. This domain was modeled at the same grid resolution as the two-dimensional simulations, yielding a 119,422-node grid. Execution times for these simulations varied from 20 to 120 hours, with the longest executions occurring for the three-dimensional simulations and the density and viscosity dependent two-dimensional simulations. Mass balance errors over the 1000-year simulation period for the solute species ranged between 5.38 x 10-5 and 8.80 x 10-2 percent, with the largest errors occurring only for the concentration-dependent density and viscosity simulations.
A principal objective of this investigation was to evaluate the effectiveness of interim barriers to the infiltration of meteoric water (from winter precipitation and snowmelt) on the migration of contaminants from previous leak sources. Sources of contamination at the S-SX WMA include releases from tanks SX-108, SX-115, and S-104 and ancillary equipment, and crib 216-S-25. To assess the impact of surface barriers, water-leaks, clastic dikes, and fluid properties on the migration of known contaminant distributions, eight suites of two-dimensional simulations were executed. The reference suite of simulations (Base Case) considered the migration of contaminants from field estimates of concentration distributions through the vadose zone and groundwater to the S-SX WMA Boundary with no interim barriers but a closure barrier by the year 2040. Contaminant concentrations at the S-SX WMA Boundary were then translated to three compliance points (i.e., 200W Fence, Exclusion Boundary, and Columbia River). Simulation results for the base case scenario predicted arrivals of peak concentrations of the contaminant Tc-99 in the following sequence: S-SX WMA Boundary, year 2048; 200W Fence, year 2186; Exclusion Boundary, year 2348; and Columbia River, year 2549. Arrival times for peak concentrations of the contaminants Cr and NO3 were similar as those for Tc-99, with the variations being primarily due to the initial inventory distributions. The contaminant Cs-137, because of its retardation factor and radioactive decay, was undetected the S-SX WMA Boundary by the year 3000. A summary of the results are provided in Tables 1 through 3, showing the peak time and concentration, for each simulation case and compliance point for the mobile three species (i.e., Tc-99, Cr, NO3), respectively.
The impact of an interim surface barrier was investigated by altering the base-case simulation to include an interim barrier installed by the year 2010. Results from this suite of simulations showed that, whereas, the peak arrival time for the contaminant Tc-99 at the S-SX WMA Boundary was essentially unchanged from the base case simulation, the peak concentration was reduced (17% of the base-case concentration). Translated concentrations to the remote compliance points were similarly reduced.
The impact of a soil-saturating water-line leak was investigated by altering the base-case simulation to include a water-line leak near the dome of tank SX-115, involving 25,000 gallons over a 5-day period. Simulation results showed this quantity of water to be sufficient to saturate the soil between tanks SX-114 and SX-115, but this plume of saturated water rapidly diffused as it migrated down through the vadose zone, having negligible impact below the Plio-Pleistocene layer. Peak concentrations and arrival times at the four compliance points were essentially unchanged from the base-case simulations, demonstrating that the single-event water-line leak had negligible impact on the migration of contaminants from the S-SX WMA.
Clastic dikes, near-vertical geologic features filled with unconsolidated sediments, have been postulated to form a polygonal pattern in the vicinity of the S-SX WMA. The influence of a single clastic dike, situated between tanks SX-108 and SX-109, that extended vertically from the base of the tanks to the top of the Plio-Pleistocene unit was investigated as an alteration to the base-case simulation. This suite of simulations used an altered grid for cross section SX-DD' compared with the base case simulation, where grid refinement was used to resolve the clastic dike. Peak concentrations and arrival times for Tc-99 for the clastic dike simulations at the four compliance points were nearly identical to those for the base case, demonstrating the negligible influence of a single clastic dike situated outside initial contaminant inventory domain.
The influence of initial contaminant inventory distributions were investigated by considered two alterations to the base-case distributions on contaminants. The first alteration involved concentrating and shifting the contaminant mass. For each horizon the contaminant mass beneath the tanks was concentrated in the region between tanks. These simulations showed little change in peak arrival times of the contaminants at the four compliance points, primarily because of the homogenizing affect of the Plio-Pleistocene layer. Peak concentrations for Tc-99, however, were 113% of the base-case values at the S-SX WMA Boundary and 108% of the base-case values at the Columbia River. A second suite of simulations was executed with the horizontally altered initial inventory distributions that included an interim barrier. These simulations showed responses analogous to those for the interim barrier against the base-case simulations. Peak arrival times of Tc-99 at the four compliance points were essentially unchanged, however, the peak concentrations were reduced to 17% of concentrated initial inventory distributions. The second alteration in initial contaminant inventory (i.e., third simulation suite) involved both horizontal concentration of the initial inventory, but also a vertical displacement toward the water table. Results from these simulations showed markedly earlier peak arrival times and concentrations for Tc-99 at the four compliance points. Peak concentrations of Tc-99 at the S-SX WMA Boundary were 348% of those for the base-case initial distributions and 309% of those for the horizontally concentrated initial distributions.
The base-case simulations considered the concentration of transported solutes as having no influence on the properties of the carrier fluid (i.e., the infinitely dilute assumption). At the field-measured concentrations, however, the density and viscosity of water is dependent on the concentration of dissolved solutes, in particular NaNO3. To investigate the influence of concentration dependent fluid properties, a suite of simulations were executed that varied from the base case in that the aqueous-phase density and viscosity were considered to be a function of the NaNO3 concentration, as determined by the concentration of dissolved NO-3 specie. Results from these simulations showed slightly earlier arrival times of the peak concentrations and slightly higher peak concentrations for Tc-99 (e.g., 105% of the base-case concentration at the S-SX WMA Boundary) at the four compliance points. The Cr and NO3 breakthrough curve results showed similar trends. These results indicate that density and viscosity affects have minimal impact on the migration of contaminants through the vadose zone beneath the S-SX WMA, partially due to the influence of the Plio-Pleistocene layer.
A three-dimensional simulation of the SX-108 tank (modeled in quarter symmetry) was executed to provide a quantitative comparison against the two-dimensional cross-section simulations. To isolate the influence of domain dimensionality, the geology of the three-dimensional domain was developed by extrapolating the east-west cross-section geology in the north-south direction. In addition the horizontal and vertical grid spacings were unaltered from the base case simulations. The three-dimensional simulation was executed following the base-case scenario (i.e., no interim barrier, no water-line leaks, and a closure barrier by 2040). The resulting flow fields and solute migration results at the water table show close agreement compared against the two-dimensional base-case simulations. Limited three-dimension simulation results indicate that errors associated with analyzing the S-SX WMA using two-dimensional cross-sections are minimal for the Base Case.
This report documents the simulation process, from converting the modeling data to input format to retrieving archived results. The report is divided into sections that generally follow the overall simulation procedures. First the investigation objectives are summarized, followed by a listing of the numerical simulations that were executed. The next section describes the process of converting the data provided in the modeling data package into input files for the STOMP simulator. Much of this discussion relies on the reader having access to the STOMP guide documents and focuses on the correlations between the modeling data package and STOMP input cards. This section also includes descriptions of converting geologic cross sections into two- and three-dimensional soil distribution maps and converting initial inventory data into distributions of dissolved contaminant concentrations. To meet the modeling specifications four, previously unavailable features (i.e., capabilities) were implemented into the STOMP simulator: 1) solute-concentration-dependent density and viscosity, 2) saturation-dependent permeability anisotropy (i.e., Polmann model), 3) solute-soil-dependent enhanced macrodispersivity, and 4) Courant number limiter (i.e., multiple transport time stepping). Implementation of these capabilities into the STOMP simulator is described. The next section provides a short summary of the code compilation and execution, which included two platforms: 1) workstations operating under UNIX and personal computers operating under Windows NT.
Simulation results are described in summary format, supported with line plots and color-scale images. The result sections begin with descriptions of the techniques and utility programs used to convert the simulation results from the conventional STOMP output format to those forms reported in this document. Results are then presented for each of the cases starting with the coupled vadose-zone and unconfined aquifer simulations and followed with the translation of those results for contaminant transport to the remote compliance points through streamtube modeling. The primary emphasis in reporting results was to provide a straightforward summary of the simulations and streamtube modeling using tables, plots and color-scaled images.
White, M. D., M. Oostrom, and M. D. Williams. July 2001. "FY100 Initial Assessments for S-SX Field Investigation Report(FIR): Simulations of Contaminant Migration with Surface Barriers." Battelle, Pacific Northwest Division. Richland, Washington.