CBEE Home | enve | igw | pumptreat

Lesson 3 Continued...

3.4. Comparison of IGW to MODFLOW Pump-and-Treat Simulations

Contaminant Transport Under Natural Hydraulic Gradient

Tiedeman and Gorelick (1993) used MODFLOW with a three-dimensional method-of-characteristics model (MOC3D) to simulate the transport of the vinyl chloride plume under the natural hydraulic gradient. Figure 3.44 (A) and (B) show the transport of the plume under the natural hydraulic gradient of the site using MODFLOW and IGW.
Figure 3.44. Comparing (A) MODFLOW vinyl chloride concentration (mg/L) in the contaminant transport model versus (B) IGW under natural gradient conditions

The contaminant concentration legend used for the IGW plume simulation (Figure 3.44(B)) is shown in the pump-and-treat remediation section. Both MODFLOW and IGW show similar temporal results. IGW tends to have smoother concentration contour lines, but the transport velocities and maximum concentration are similar for both simulations. The maximum concentration contour decreases from 800 (red), to 600 (yellow-orange), to 400 (green-blue), to 100 (light blue) from 0 to 15 years. Notice that the three-dimensional MODFLOW model has a similar trend of depth-averaged concentration decreasing throughout the years. The main differences between the two simulations are the initial conditions of the concentration contours used at time zero and the calculations of the concentration contours (i.e., MODFLOW averages concentration in the vertical layer versus the 2D concentration contours given in IGW).

Capture Curve for the 10 Well Optimal Solution

Figure 3.45 (A) and (B) compares the 90% and 50% capture curve determined from MODFLOW and IGW for the 10 well solution. Notice that as the reliability level increases, the size of the capture curve is greater due to the higher extraction rate. The head contours and capture curves compare well for both reliability levels, indicating that IGW did a good job of replicating the 10 well solution from MODFLOW.
Figure 3.45. Comparing (A) MODFLOW versus (B) IGW capture curves for the 10 well solution with a 90% and 50% reliability.

Contaminant Transport under the Ten and Two Well Optimal Extraction Rates

The results for modeling a 5-year remediation time with MODFLOW and IGW are illustrated in Figure 3.46 (A) and (B). For the ten well solution, the size and concentration of the vinyl chloride plumes is similar after 5 years of treatment (relative to their initial concentration). For the two well solution, there is a difference in the remediation of the plume in the two models. In IGW, the stagnation point point that developed between the two wells caused a delay in the remediation time. This probably occured because we had to assign a lower extraction rate to the down-gradient extraction well avoid instability problems within IGW.

(A)

 (B)
Figure 3.46. Contaminant transport using the ten and two extraction well solution with (A) MODFLOW and (B) IGW with a five year remediation time.

3.5. Summary

This lesson used IGW to replicate the Saint Joseph site simulation conducted by Tiedeman and Gorelick (1993). Below is a summary of the modeling steps:

  1. Set Up and Sensitivity Analysis
    1. IGW was set up (Section 3.1) by importing a site map, setting boundary conditions for Lake Michigan and Hickory Creek to constant head values, and settting aquifer properties and characteristics to match data given by Tiedeman and Gorelick (1993).
    2. A sensitivity analysis (Section 3.2) of K1 (aquifer conductivity), K2 (Hickory Creek and lagoon sediment conductivity), and recharge was conducted to calibrate the model to match measured head values and spatial contaminant concentrations reported for the site. The location of the lagoon directly above the groundwater divide made contaminant transport very sensitive to K1, K2, and recharge. Final values were determined by contraining recharge to a rate reasonable for the climate of the site and iteratively adjusting the values of K1 and K2. Final baseline values compared well with the site head values and contaminant characteristics.
    3. The effects of site variability (Section 3.2) on the final baseline values were explored using a random conductivity field for both K1 and K2 using particle tracking and an instantaneous plume. Also, the influence of an unsteady versus steady-state head for Lake Michigan was simulated. There was no significant difference between either the random and uniform field, or the unsteady-state versus steady-state simulations.
  2. Pump and Treat Remediation
    1. The calibrated model was used to simulate the pump-and-treat remediation (Section 3.3) of the vinyl chloride plume heading towards Lake Michigan. Modeling was carried out for the ten and two well optimal solutions given by Tiedeman and Gorelick (1993). The ten well solution had three up-gradient wells that extracted enough flow to change the head gradient so that the plume traveled to the five down-gradient extraction wells. The two well solution required a high pumping rate at the down gradient well and created significant drawdown of the aquifer around that well. This problem was alleviated by turning off pumping at that well after four years. The ten well solution with a reliability level of 50% and 90% remediated the plume within 12 and 10.5 years, respectively. In comparison, the two-well solution with a reliability level of 60% with one well turned off after four years, remediated the plume after 11 years. The capture zone analysis showed that as the reliability level increased, a larger and more dense capture zone developed, which was consistent with Tiedeman and Gorelick (1993).The mass balances for all well solutions gave inconsistent results, which maybe the result of model stability near the extraction wells.
    2. A comparison of the pump-and treat remediation (Section 3.4) results in MODFLOW and IGW showed similar outcomes for the two models.

In conclusion, this exercise suggests that the most reliable and fastest clean-up solution was the 10 well solution with the 90% reliability level. However, the two well solution with the variable extraction rate may be the most economical. The ‘best’ pump-and-treat solution depends on the objectives and criteria developed for the site and may include evaluation of other possible remediation schemes.

This lesson also demonstrates the utility of IGW for illustrating and evaluating groundwater cleanup scenarios. Students can quickly setup simulations similar to the one in this lesson and use the model to visualize the effects of different combinations of wells and flow rates, to monitor the concentration breakthrough curves, to track particle movement and determine capture zones, and to conduct sensitivity analyses on boundary conditions and model parameters. However, IGW does have some limitations with regards to obtaining output information from model simulations (i.e., breakthrough curve, head variation with time, mass balance, etc.), incorporating submodels into model simulation to increase model stability and accuracy;,and saving a simulation that requires a large amount of memory. Despite these limitations, IGW is valuable tool for visualization of groundwater flow and contaminant transport.

3.6. Step-by-Step Tutorial

Click below to link to a pdf document with step-by-step instructions to complete this simulation in IGW.

  • Tutorial 3: Pump and Treat at the Saint Joseph Superfund Site
  • You can also download two bmp images that you will use in the tutorial. To save these files, right click on the file names below and choose the "save target as" option.
  • You can also download the IGW files that will result from successful completion of the tutorial. The IGW files are compressed as a ZIP file. To open it, you will need to save the ZIP file to your local computer and uncompress it with a program such as Winzip or Windows Explorer. To access the unzipped file, first open IGW and then go to file menu and open the file from the saved location (note: you must open the file from within IGW, it will not work correctly if you double click on the file name within Windows Explorer). Please also note that the basemap does not load automatically with the IGW file. You will need to download the file (see above) and import the basemap into the interface as guided in the beginning of the tutorial.

Please note that the lessons and tutorials on our website were designed for IGW version 3.5.6 -- we recommend using this version of the software to ensure compatability between the step-by-step instructions and what you see on your computer screen.

3.7. References

Bedient, Philip B and Wayne C. Huber, 1992, Hydrology and Floodplain Analysis, 2nd ed, Addison-Wesley Publishing Company: Reading, Massachusetts, p. 7-37.

Tiedeman, Claire and Steven M. Gorelick, 1993, Analysis of Uncertainty in Optimal Groundwater Contaminant Capture Design, Water Resources Research, Volume 29, Number 7, pages 2139-2153.

McCarty, Perry L., Semprini, Lewis, et. al, 1990, Evaluation of the In-Situ Methanotrophic Bioremediation for Contaminated Groundwater St. Joseph, Michigan, Western Region Hazardous Substance Research Center: Stanford, California.

Semprini, L., P.K. Kitanidis, D. Kampbell, and J.T. Wilson, 1995, "Anaerobic Transformation of Chlorinated Aliphatic Hydrocarbons in a Sand Aquifer Based on Spatial Chemical Distributions," Water Resource Res. v. 31, no. 4, 1051-1062.

Back to top