This section briefly describes the conceptual model of the Hanford Site unconfined aquifer system. The conceptual model was developed from information on the hydrogeologic structure of the aquifer, spatial distributions of hydraulic and transport properties, aquifer boundary conditions, and distribution and movement of contaminants. Development of the three-dimensional conceptual model is documented in Thorne and Chamness (1992), Thorne et al. (1993, 1994), and Wurstner et al. (1995).
The Hanford Site lies within the Pasco Basin, a structural depression that has accumulated a relatively thick sequence of fluvial, lacustrine, and glacio-fluvial sediments. The geology and hydrology of the Hanford Site have been studied extensively for more than 50 years and are summarized in documents referenced in Wurstner et al. (1995). The Pasco Basin and nearby anticlines and synclines initially developed in the underlying Columbia River Basalt Group, a sequence of continental flood basalts covering more than 160,000 km2. Overlying the basalt within the Pasco Basin are fluvial and lacustrine sediments of the Ringold Formation and the glacio-fluvial Hanford Formation. Together, these sedimentary deposits comprise the Hanford Site unconfined aquifer system. The saturated thickness of this unconfined aquifer system is greater than 61 m in some areas but pinches out along the flanks of the basalt ridges. Depth to the groundwater ranges from less than 0.3 m near the Columbia River to more than 100 m in the 200-Area plateau. Groundwater in this unconfined aquifer system generally flows from recharge areas to the south and west of the Hanford Site to the Columbia River in the east.
The lateral extent and relationships between the hydrogeologic units of the Ringold and Hanford formations were defined by identifying distinct hydrogeologic units to adequately represent the unconfined aquifer, and by determining geologic contacts between these layers at as many wells as possible. These interpreted areal distributions and thicknesses were integrated into EarthVision,® (c) a three-dimensional visualization software package that was then used to construct a database for formulation in the three-dimensional Site conceptual model. The resulting conceptual model contains nine hydrogeologic units above the uppermost basalt. A brief summary of each of these units, based on descriptions provided in Wurstner et al. (1995), is provided in Table 3.1.
Table 3.1. Major Hydrogeologic Units Used in
the Site-Wide Three-Dimensional Model
| Unit Number | Hydrogeologic Unit | Lithologic Description |
| 1 | Hanford Formation/ Pre-Missoula Gravels |
Fluvial gravels and coarse sands |
| 2 | Palouse Soil | Fine-grained sediments and eolian silts |
| 3 | Plio-Pleistocene Unit | Buried soil horizon containing caliche and basaltic gravels |
| 4 | Upper Ringold Mud | Fine-grained fluvial/lacustrine sediments |
| 5 | Middle Ringold | Semi-indurated coarse-grained fluvial sediments |
| 6 | Middle Ringold | Fine-grained sediments with some interbedded coarse-grained sediments |
| 7 | Middle Ringold | Coarse-grained sediments |
| 8 | Lower Mud Sequence Ringold | Lower blue or green clay or mud sequence |
| 9 | Basal Ringold | Fluvial sand and gravel |
| 10 | Columbia River Basalt | Basalt |
Natural recharge to the unconfined aquifer system occurs from infiltration of 1) runoff from elevated regions along the western boundary of the site, 2) spring discharges originating from the confined basalt aquifer system, and 3) precipitation falling across the Hanford Site. Some recharge also occurs along the Yakima River in the southern portion of the Site. Natural recharge from runoff and irrigation in Cold Creek Valley (upgradient of the Site) provides a source of groundwater inflow to the area of interest. Areal recharge from precipitation falling on the Hanford Site is highly variable, both spatially and temporally, and depends on local climate, soil type, and vegetation. Estimates of recharge based on site-wide variation in these parameters, developed by Fayer and Walters (1995) for 1979 conditions (Figure 3.1), illustrate this range in variability.
The other source of recharge to the unconfined aquifer is artificial recharge from waste water disposal. The large volume of waste water discharged to disposal facilities at the Hanford Site over the past 50 years has significantly impacted groundwater flow and contaminant transport in the unconfined aquifer. The volume of artificial recharge has decreased significantly during the past 10 years and continues to decrease. The major discharge facilities considered in this analysis are summarized in Wurstner et al. (1995).
Boundary conditions shown in Figure 3.2 illustrate that the Hanford Site unconfined aquifer system is bounded by the Columbia River on the north and east and by basalt ridges on the south and west. The Columbia River represents a point of regional discharge for the unconfined aquifer. The amount of groundwater discharging to the river is a function of the local hydraulic properties of the aquifer and of the hydraulic gradient between the groundwater elevation alongside or beneath the river, and the river stage. This hydraulic gradient is variable at any given time, since the river stage is affected by seasonal variations in precipitation and temperatures in other regions of the river drainage system. The river stage is also impacted by weekly and daily changes in river flows at numerous dams on the river, as determined by electric power generation needs, fisheries resources management, and other dam operations. In previous three-dimensional modeling efforts, the entire boundary at the edge of the Columbia River was represented as a constant head over the entire thickness of the unconfined aquifer. The CHARIMA river model (Walters et al. 1994) was used to generate long-term average water surface elevations for the Columbia River based on 1979 conditions. In the current model the Columbia River boundary was extended from the left edge of the river to the middle of the river channel to more accurately reflect the hydraulic interaction of the unconfined aquifer and the river. The surface nodes at the river edge and center were simulated as constant-head boundary conditions reflective of the assumed river stage. The nodes below the surface and along the center of the river were simulated as no-flow boundaries. This design leads to a more accurate approximation of the upward movement of groundwater as the groundwater flow is controlled by the hydraulic gradient between the aquifer and the river.
An areal depiction of the surface finite-element grid and boundary conditions used in the three-dimensional models of the unconfined aquifer is illustrated in Figure 3.2. The finite-element grid is a more regularly spaced grid than has been described in previous reports and used in previous applications. The grid was redesigned to increase the overall effectiveness and efficiency of the three-dimensional model to simulate both flow and transport problems. Most of the interior surface grid spaces are rectangular and are about 750 m on a side. The total number of surface elements used in both the two-dimensional and three-dimensional model is 1606 elements. The three-dimensional model based on this surface grid is made up of a total of 7200 elements (1606 surface and 5594 subsurface elements) and 8465 nodes.
At the Cold Creek and Dry Creek valleys (see Figure 3.2), the unconfined aquifer extends westward beyond the boundary of the Hanford Site groundwater flow model. To approximate the groundwater flux entering the modeled area from these valleys, both constant head and constant flux boundary conditions were defined across these valleys. A constant head boundary condition was specified for Cold Creek Valley for the steady-state model calibration runs. Once calibrated, the steady-state model was used to calculate the flux condition that was then used in the transient simulations. The constant flux boundary was used because it better represents the response of the boundary to a declining water table than does a constant head boundary. Previous versions of the three-dimensional model did not include boundary fluxes where Dry Creek enters the modeled area. In the current model, boundary fluxes are prescribed at the north and east valleys of Dry Creek along the southwestern edge of the model boundary. The addition of these boundary fluxes is consistent with observations of water levels in nearby wells in this area. The unconfined aquifer is also recharged from springs and runoff that infiltrate the aquifer along the northern side of Rattlesnake Hills.
Since the last description of the site-wide model provided in Wurstner et al. (1995), changes have been made to the areal extent of the model, model boundary conditions, and model grid design to reflect the most recent understanding and interpretation of the unconfined aquifer system. The most significant changes incorporated in the current version of the site-wide models were derived from reinterpretation of the 1979 water-table surface of the unconfined aquifer and the top of the basalt, which led to changes in both internal and lateral boundary conditions, including:
Hydraulic properties important to the conceptual model include both horizontal and vertical hydraulic conductivities, storativity, and specific yield. To apply a numerical model, the distribution of these parameters must be specified for each hydrogeologic unit. Hydraulic properties have been measured for the unconfined aquifer (considered as a simple hydrogeologic unit) mainly during aquifer pumping tests and from laboratory permeability tests. The results of these tests have been documented in published and unpublished reports over the past 50 years and in more recent summaries (U.S. DOE 1988; Thorne and Newcomer 1992). As indicated in these documents, the quality of results from aquifer tests at the Site varies widely and is affected by both aquifer conditions and analysis procedures. Thorne and Newcomer (1992) and Wurstner et al. (1995) reanalyzed the aquifer tests, many of which were single-well pumping tests, and they selected the set of aquifer transmissivity calibration data (Figure 3.3) used in the two-dimensional inverse model.
Estimates of model parameters were developed to account for contaminant disperison and adsorption in all transport simulations. Specific model parameters examined included longitudinal and transverse dispersivity (Dl and Dt)) and contaminant retardation factors (Rf). Calculation of effective Rf required estimates of contaminant-specific distribution coefficients as well as estimates of effective bulk density and porosity of the aquifer materials. This section briefly summarizes estimated transport properties.
For this analysis, a longitudinal dispersivity of 90 m was selected to be within the range of recommended grid Peclet numbers (Pe < 4) for acceptable solutions. The 90 m estimate is about one-quarter of the grid spacing in the finest part of the model grid in the 200-Area plateau where the smallest grid spacing is on the order of about 375 m by 375 m. The effective transverse dispersivity was assumed to be one-tenth of the longitudinal dispersivity. Therefore, 9 m was used in all simulations.
For purposes of this analysis, no adsorption was accounted for in simulating the tritium plumes. However, for the iodine-129, technetium-99, uranium, and strontium-90 plumes simulated in this analysis, best-estimate distribution coefficients (Kd) were developed from work summarized in several sources including Rhodes (1956); Nelson (1959); Routson et al. (1978); Serne et al. (1993); Kaplan and Serne (1995), and Kaplan et al. (1996). Table 3.2 provides a summary of these best estimate values, the range of estimates, and the associated reference.
In addition to the estimated distribution coefficient, calculation of contaminant-specific retardation factors used in the model requires estimates of the effective bulk density and porosity. For purposes of these calculations, a bulk density of 1.9 g/cm3 was used for all simulations. The effective porosity was estimated from specific yields obtained from multiple-well aquifer tests. These values range from 0.01 to 0.37. Laboratory measurements of porosity, which range from 0.19 to 0.41, were available for samples from a few Hanford Site wells and were also considered. The few tracer tests conducted indicate effective porosities ranging from 0.1 to 0.25. Based on the ranges of values considered, a best estimate of an effective porosity value for all simulations was assumed to be 0.25.
Table 3.2. Best Estimate Distribution Coefficents Used in Simulations of Iodine-129, Technetium-99, Uranium, and Strontium-90 Plumes
Plume Constituent |
Distribution Coefficient (Kd) (ml/g) | Reference |
|
Best Estimate Value |
Range of Estimates |
||
| Iodine-129 | 0.5 | 0.2 to 15 | Kaplan and Serne (1995), Kaplan et al. (1996) |
| Technetium-99 | 0 | -2.8 to 0.6 | Kaplan and Serne (1995) |
| Uranium | 3 | 0.1 to 79 | Kaplan and Serne (1995) |
| Strontium-90 | 20 | 5 to 173 | Routson et al. (1978); Serne et al. (1993); Rhodes (1956); and Nelson (1959) |
