CALIBRATED FLOW AND TRANSPORT MODEL FOR TCE IN GROUNDWATER AT BUILDING 460, FORT CHAFFEE, ARKANSAS

Prepared by: Darrel E. Dunn, Fort Lauderdale, Florida

Prepared for: Fort Chaffee Base Transition Team

August 10, 2001

[This is an html version of a report that was submitted in paper copy.  Transcribed by Darrel Dunn, Ph.D., Hydrogeologist, Colorado Springs, Colorado]

PURPOSE

This report describes the development of a groundwater flow and solute transport model at Building 460, Fort Chaffee, Arkansas, and the application of the model to calculating future concentrations of TCE at Little Vache Grasse Creek downgradient from the building.  The TCE was spilled in a northeast-southwest ditch along a road that passes northwest of the building.  The spills occurred sometime after the building was completed in 1942.

MODEL CODE SELECTION

The computer codes selected for the modeling were Modflow, Modpath, and MT3D.  These are widely known and accepted models.  Modflow and Modpath were developed by the U.S. Geological Survey.  MT3D was originally developed at S. S. Papadopulos & Associates, and subsequently documented for the Robert S. Kerr Environmental Research Laboratory of the U.S. Environmental Protection Agency.  MT3D was the preferred solute transport model because the most recent versions include the capability of simulating dual-domain porosity systems, in which immobile domain porosity contains immobile fluid where transport is primarily by molecular diffusion.  It was initially thought that it might be necessary to use dual-domain porosity to simulate the system.  This was not the case, and dual-domain porosity was not used.

GROUNDWATER FLOW MODEL

The groundwater flow model was developed and calibrated to supply groundwater flow velocities for particle tracking and solute transport modeling.

Conceptual Model

The subsurface materials in the vicinity of Building 460 consist of fractured shale overlain by clay.  A weathered zone has developed at the top of the shale.  The water table occurs in all three units at various locations around the site, but is commonly in the shale or the weathered zone.  The shale dips eastward at about one foot per forty feet and subcrops at an angular unconformity beneath the clay.  The dip of the shale is indicated by a dip-slope located northwest of Building 460.

The fractures may be conceptualized as a continuum of fractures with variable attributes.  The attributes that affect the groundwater flow include spacing between fractures, openness of the fractures, orientation, spatial continuity, wall roughness, and fracture fillings.  The depositional environment of shale changes from time to time, causing sedimentary layers with differences in composition and texture that affect the properties of the ultimate lithified shale.  These properties, in turn, affect the fracture attributes.  Consequently, some difference in fracture attributes may be expected to correspond to sedimentary layers, and the spatial distribution of fracture attributes will correlate with layering.  Furthermore, changes in depositional environment occurred laterally at any particular time, so that fracture attributes that affect the hydrologic properties of the shale may be expected to vary laterally within individual layers.  The fracture attributes that affect the hydrologic properties of the shale may be too subtle to identify by ordinary inspection and testing of cores and outcrops, although some, such as fracture spacing and orientation may be observed and described to a limited degree.  Rough estimates of hydraulic conductivity may be generated through in situ and laboratory testing.

The fracture attributes in the weathered zone have been altered during weathering, and the hydrologic properties of this layer may differ significantly from the unweathered shale.  The degree of weathering may vary from place to place due to changes in the parent rock fracturing and lithology.

The flow in the fractured rock and the weathered material derived from it is treated as conforming to Darcy's Law.  This is conventional practice because the equation relating flow velocity between parallel surfaces to separation of the surfaces has the form of the equation relating flow velocity through porous media to the hydraulic conductivity of the media.

Recharge to the groundwater system is through rainfall and snowmelt, and evapotranspiration returns water to the atmosphere.  Evapotranspiration extracts water from the land surface to the base of the zone affected by roots.  The water table conforms roughly to the topography, and the groundwater flows downslope toward Little Vache Grasse Creek.

The flow system is very complex.  Some observations at monitoring sites indicate increasing head with depth and others indicate decreasing head with depth.  Specifically, monitoring well MP03 penetrated a permeable, soft, broken shale at the bottom of the hole, which is 61 feet below the ground surface.  This is a multi-port well and the port in the deep permeable material has a head that is about 5 feet higher than that at the port about seven feet above it.  At MW05, head in the deep well is about 2 feet higher than in the adjacent shallow well.  Conversely, at the MW04 nest of wells, heads are more than a foot higher in a shallower well than an adjacent deeper one with only about 5 vertical feet separating the monitored intervals.  Such vertical head gradients require relatively low vertical hydraulic conductivities in the vicinity of the wells to sustain the head difference, and very heterogeneous lateral hydraulic conductivities.  The higher heads must have relatively high horizontal hydraulic conductivity present in the upgradient direction with low conductivity in the downgradient direction so that the head in the well is influenced more by the higher heads upgradient than by the lower heads downgradient.  The lower heads must have the opposite arrangement of lateral heterogeneity.  In all cases the flow is still downgradient to discharge to Little Vache Grasse Creek, but the head gradients are different in the upgradient and downgradient directions for the wells.  These opposing vertical head gradients can exist with the same upgradient and downgradient boundary  condition heads, but considerable heterogeneity in the hydraulic conductivity of the system is implied.

Discretization

A plan view of the spatial discretization of the groundwater model is shown in Figure 1.  The cell size is a uniform ten foot square, which was chosen to give good solute transport modeling results.  The boundaries were chosen far enough from Building 460 to make boundary effects insignificant in the area of interest.  The grid was oriented with the rows parallel to the ditch on the northwest side of Building 460.  This orientation produced lateral grid boundaries that are approximately parallel to flow lines, and allowed the TCE sources in the ditch to fall in a single row of model cells to facilitate the solute transport modeling.

The grid contains ten layers.  The top layer (number 1) represents the surficial clay.  Layers 2, 3, and 4 represent the weathered zone, and layers 5 through 10 represent the unweathered shale.  The weathered zone was divided into three layers to allow detailed simulation of the vertical distribution of TCE.  The bedrock was divided into six layers to allow detailed simulation of vertical head gradients measured in clustered and multiport monitoring wells.

Time was not discretized in the flow model, which was steady-state.