By Hydrosimulatics INC  


Groundwater Modeling to Assist in the Designof a Remediation Purge Well System in East Bay Township, Michigan

Disclaimer: This project is based on the 1985 U.S. Geological Survey study by F.R. Twenter, T.R. Cummings and N.G. Grannemann.  Some information has been simplified, modified, or omitted to make the problem suitable for educational application. The results and interpretations of this project are not meant to be liable assessments of the real-world contamination problem. 


A subsurface contaminant plume has been identified in East Bay Township, a residential area adjacent to Traverse City, Michigan and home to the Cherry Capital Airport. Prompted by complaints of odors and foaming water in domestic water well supplies in the residential area along Avenue E near East Grand Traverse Bay, the Department of Public Health detected organic substances in the groundwater – including benzene, toluene, xylene, and other compounds characteristic of solvents, cleaning agents, and fuel substances. These chemicals are known to be harmful if ingested by humans.

Subsequent water quality analyses from numerous groundwater monitoring wells indicate that the plume stretches over 4,000 feet from the bay to an area south of Parsons Road, in the vicinity of the US Coast Guard Air Station – a property formerly owned by the U.S. Navy (see Figure 1). A review of old Navy files and records and interviews with long-term residents of the area identified several possible sources of contamination on the Coast Guard property: a former U.S. Navy waste dump on the eastern edge of the property; several former and currently operated fuel dispensers and storage tanks; and storage and disposal of solvents and aviation materials near the Hangar/Administration Building (see Figure 2).


You are tasked with determining if the plume has indeed originated from the former Navy property by conducting a study of the hydrogeologic conditions near the plume and adjacent areas. Furthermore, you will design and analyze a network of hydrologically suitable remediation purge wells to remove the contamination from the groundwater. Your project analysis should also include estimated costs of pumping contaminated groundwater and the removal of contaminated soil at the Air Station.

Specific hydrogeologic questions that should be addressed include:

  • What are the general flow patterns in the region? What controls these patterns? What are the regional sources and sinks of water (e.g., major surface water bodies receiving groundwater, upland areas recharging the water table, etc.)?

  ·     What are the local groundwater flow directions and velocities at the site and downstream from the site? 

  ·    Where does the water pumped by the domestic wells come from? Is the plume in the contribution areas of the wells? Which ones?

  ·       How long does it take for the plume to reach domestic water wells downstream from the site?

  ·       How many purge wells should be used; where should they be located; and at what rates should they pump? 



  1.   Develop a regional model of East Bay Township and the adjacent areas.

 2.  Calibrate the regional model to water level data collected from across the region. This model will     provide boundary conditions for a local model of the site and contaminated area between Parsons     road and East Grand Traverse Bay.

 3.  Develop a local model of the site and the contaminated area between Parsons Road and East Grand   Traverse Bay.

 4. Calibrate the local model to water levels collected on site and in the contaminated area between   Parsons road and East Grand Traverse Bay.

 5. Apply the local calibrated model to determine the well capture areas of potentially contaminated   wells.

 6.  Apply the local calibrated model to predict the 3D movement of contaminants from the Coast Guard   Property (plan and cross-section visualizations).

 7. Design and test multiple groundwater clean-up scenarios (i.e., configurations of the purge well   network);

  8.  Determine an optimal design of the purge well network.

 9. Estimate costs/resources needed for pumping contaminated groundwater and for excavation of   contaminated soils.

10. Organize your findings into a nicely written technical report. Be sure to clearly explain all     assumptions and methods used in your analysis.

The following sections provide more background information and modeling instructions needed to complete the tasks.

Overview of Site Hydrogeology

The Air Station is located on a relatively flat surface that gently slopes to the northeast. The land continues sloping downward north toward East Grand Traverse Bay. South of the site the land surface rises before encountering Mitchell Creek, which drains the area to the south and east of the station. Boardman River and Boardman Lake to the west of the site direct flow to the West Grand Traverse Bay.  The average annual precipitation in the area is about 35 inches.

Underneath the station and adjacent areas are thick glacial deposits of lacustrine origin (i.e., formed through depositional processed associated with lakes). A small exception is the area just south of Cherry Capitol Airport, which is underlain by glacial outwash (deposits of sand and gravel carried by running water from the melting ice of a glacier). The lacustrine deposits consist of an upper sand and gravel unit and an underlying clay unit.

 The upper portion (~20 ft) of the sand and gravel unit consists mostly of fine to medium grained tan sands. Gravel and coarse gray sands are predominant below. Based on wells installed for groundwater chemical analysis and water level monitoring, the sand and gravel unit vary in thickness, from a minimum of about 30 feet to a maximum of about 120 ft (see Figure 4).

 The clay unit is relatively impermeable and its thickness is not known because it has never been fully penetrated within the area. Borehole lithologies suggest that the top of the clay surfaces ranges from 575 to 585 feet near the west side of the site and to about 480 feet on the east. 

Multiscale Approach

      Simulating the plume transport and remediation operations requires detailed information about the groundwater flow at the site. But the site-specific flow conditions are influenced by the regional flow patterns, requiring some understanding of how groundwater head is distributed in the area around (and especially upstream) of the site. The proper way to handle this ‘multiscale’ nature of groundwater systems is to simulate large-scale conditions with a larger, relatively coarse-grid model – the Regional model. This model can then be used to inform a smaller, finer grid model (the Local model).

Regional Model Development

Two modeling areas (domains) have been created for your regional and local-scale modeling (see Figure 5).  The model domains can be added into the MAGNET modeling environment by  uploading the MAGNET file ‘ModelDomains_ForStudents’ available on in the Project Folder on the MAGNET Curriculum Network, or made directly available to you by your instructor.

Alternatively, you can manually add the regional model domain using the ‘DomainRect’ tool (‘DomainDraw’ > ‘DomainRect’). Similarily, the local model area can be added using the ‘Zone Poly’ tool (‘Zones’ > ‘ZonePoly’).

Load the model domains into the MAGNET modeling environment and parametrize the regional model (i.e., assign model inputs) to simulate horizontal (2D) flow in the surficial glacial aquifer (i.e., the sand, gravel and clay aquifer). Run the model in steady-state model to represent long-term average conditions. Use the available spatially-explicit inputs on the Server to represent the aquifer elevations (top and bottom surfaces), hydraulic conductivity, and recharge. The bottom surface follows the bedrock surface, and it is assumed that there is negligible flow to/from the bedrock units.

The exact values of model inputs – hydraulic conductivity and recharge – may require some ‘fine-tuning’ or calibration before the regional model approximates real-world conditions (head values across the aquifer) to a suitable degree. Evaluate the suitability of the model by using Static Water Levels stored on the MAGNET Server to compared simulated heads to measured heads. The ‘cloud’ of data should fall squarely on the 45-degree line of exact match in the calibration plot. Adjust the K and Recharge multipliers and check the calibration plot until this criterion is met.

Local Model Development

A zone feature in the regional model was included to delineate the local model domain. Make this zone ‘active’ by assigning boundary conditions from the regional model. (The next time the model is simulated, only the local model area will be evaluated, with boundary conditions supplied from the regional model. You can also save the results of the regional model as a zipped file that can be uploaded as the Boundary Conditions from the Parent Model.)

Run the local model in steady-state model to represent long-term average conditions. Data collected on and near the site are available with better precision than the SWLs used in regional model calibration. Load the local water level data (see ‘Available Field Data’ section) into the MAGNET data import table and compare simulated water levels to measured water levels. Determine how to adjust K and/or recharge to improve the fit of the local model. Use a new set of K and recharge inputs to re-run the local model. Again, compare the locally measured water levels to the simulated water levels. Repeat this process until there is a satisfactory calibration of the local model.

 For consistency, the local model should apply the same K and recharge inputs as the regional model. The aquifer top surface should still follow the Digital Elevation Model (DEM) available on the Server. The bottom surface, however, is different for the local model. The bottom surface should follow the clay layer surface. In other words, you will only model flow in the sand and gravel layer, assuming negligible flow to/from the clay layer. Scatter points of the elevations at which the clay layer is first encountered are given in the ‘Available Field Data’ section below. These data can be entered directly into the MAGNET scatter point interface for Bottom Elevation of the local model zone feature.

Add particles at the locations of potentially contaminated domestic wells and use reverse particle tracking to determine the well source water areas.

Add a zone feature at the site to represent the source of contamination. The substance occurring in the highest concentrations will be used to simulate the plume migration: toluene (55,500 μg/L). For the purposes of this analysis, you may assume that the concentration of toluene at the site does not change as the plume moves off site (i.e., it is a continuous source). You may want to consider the conservative situation of plume transport (i.e., pure advection of the ‘particles’ only).

 Compare the simulated plume to the plume extent estimated by USGS (see Figure 6). Figure 6 is available as a georeferenced image that can be overlaid to the MAGNET modeling environment for easy comparison. (Figures 2 and 3 are also available as georeferenced images).

Purge Well Network Design

Apply the calibrated local model to evaluate hydrogeologically suitable locations of the purge wells for removing the contaminated groundwater along the length of the plume. Experiment with different locations/configurations and different pumping rates. You may use any number of wells, but you should adhere to the following constraints:

·         pump as little uncontaminated water as possible;

·         minimize the local drawdown (groundwater level decline) around each well to avoid contaminating soils that go from saturated to unsaturated during pumping (‘short-circuiting’ the removal mechanism).

·         Minimize costs associated with well installations and operations.

Also, you must account for the restriction on purge well installations in the residential area between Indian Trails Blvd and the bay.

Decide on a final design and report the locations and pumping rates of the purge wells. Also report expected water levels changes near the wells and estimate the costs of purge well installations and operations. Suggest how the contaminated water should be handled (e.g., how might you treat it and/or where would you dispose of it). You should properly cite any outside resources used in your cost analysis.

Site Remediation

The final step in your analysis is to estimate cost of excavating contaminated soil at the site. Also consider which potential sources should be permanently removed from the site to prevent future contamination. Suggest a monitoring plan for the following 5-10 years.

Available Field Data 

i.                Scatter points, elevations at which clay layer first encountered: 



























SPID=Scatter point ID

X(Lon)=longitude of scatter point

Y(Lat)=latitude of scatter point

SPvalue=scatter point value (elevation of clay surface)

Icolor=-1 … default marker color applied when showing markers on map display


ii.                  Scatter points, water level measurements from monitoring wells


WellID,   Time,   x(Lon),            y(Lat),              Zf,                   Zt,                    LyrIndex, V,                Icolor

K1D,       0,         -85.587227,      44.745162,       170.59656,       169.68216,       0,         182.745888,     -1

K3D,       0,         -85.578638,      44.745722,       172.358304,     171.139104,     0,         182.86476,       -1

K31D,     0,         -85.5807734,    44.7446756,     179.670456,     178.756056,     0,         183.577992,     -1

K33S,     0,         -85.5825351,    44.745477,       182.060088,     181.145688,     0,         183.745632,     -1

K5D,       0,         -85.5790499,    44.7462372,     175.598328,     174.683928,     0,         182.86476,       -1

K8D,       0,         -85.5795202,    44.7469703,     178.725576,     177.506376,     0,         182.8038,         -1

K45S,     0,         -85.5803632,    44.7460995,     180.7464,         179.832,           0,         183.392064,     -1

K13D,     0,         -85.57605,       44.749259,       176.29632,       175.38192,       0,         180.816504,     -1

K32D,     0,         -85.573212,      44.747024,       168.636696,     169.551096,     0,         180.79212,       -1

K15D,     0,         -85.577063,      44.749763,       176.756568,     175.842168,     0,         181.221888,     -1

K11D,     0,         -85.581016,      44.749994,       177.792888,     176.878488,     0,         182.029608,     -1

K16D,     0,         -85.5741764,    44.74964,         172.839888,     171.620688,     0,         180.252624,     -1

K29D,     0,         -85.567555,      44.751773,       164.098224,     163.183824,     0,         177.198528,     -1

K23D,     0,         -85.5748527,    44.7498207,     175.09236,       174.17796,       0,         180.386736,     -1

K9D,       0,         -85.5797745,    44.7472688,     177.582576,     176.668176,     0,         182.843424,     -1

K25S,     0,         -85.5719898,    44.7539879,     173.897544,     172.678344,     0,         176.863248,     -1

KT,         0,         -85.5811013,    44.75005,         178.576224,     177.052224,     0,         182.127144,     -1



WellID=Well ID

Time=0 … time of measurement…use zero when preforming steady-state calibration

X(Lon)=longitude of scatter point

Y(Lat)=latitude of scatter point

Zf=elevation of the well screen top

Zt=elevation of the well screen bottom

LyrIndex-0 … this tells MAGNET to use Zf, Zt info. for assigning well to proper model layer

V=scatter point value (water level)

Icolor=-1 … default marker color applied when showing markers on map display