Evaluating the effectiveness of pumping well configurations for groundwater remediation


By Paul F. Hudak

References listed at bottom of article

Low-cost alternatives for remediating contaminated groundwater are becoming increasingly popular. Permeable reactive barriers can treat contaminants without costly pumping of groundwater (EPA, 2002).

Placed hydraulically downgradient of contaminant plumes, pervious barriers have treated arsenic, metals, organics and other contaminants in groundwater (Guerin et al., 2002; Ludwig et al., 2002; Lai et al., 2006; Gilbert et al., 2010). However, installing permeable reactive barriers, especially deep installations, requires specialized and costly excavation equipment. Additionally, monitoring, removing, disposing of, and replacing filter media can be costly over long operational periods.

Subscribe to our Newsletter!

The latest environmental engineering news direct to your inbox. You can unsubscribe at any time.

At some sites, non-pumping wells equipped with removable porous filters could be used instead of permeable reactive barriers (USGS, 1999). Similar to permeable reactive barriers, filter media in non-pumped wells immobilize or transform contaminants. Conventional drilling rigs can reach much greater depths than trenching equipment; thus, non-pumped wells could be used in deep as well as shallow settings (Hudak, 2009).

However, nearly adjacent placement may be necessary to keep contaminants from migrating between non-pumped wells. Installing, monitoring, and maintaining numerous closely-spaced wells may be cost-prohibitive at many sites (Hudak, 2014).

Alternatively, one extraction well and one accompanying injection well, pumping at very low rates, may create a less costly, hydraulic barrier to contain contaminant plumes in some settings (Cunningham and Reinhard, 2002). A typical installation is collinear, along a transect crossgradient to the prevailing hydraulic gradient, located in front of a contaminant plume. One well extracts contaminated groundwater, which is treated above ground and then injected into the aquifer via the other well.

The injection well also dilutes contaminant concentrations in groundwater. Such dilution complements dilution by fresh groundwater in an aquifer. These processes, combined with the effects of hydrodynamic dispersion, contribute to lowering contaminant concentrations in aquifers. Additionally, low-capacity extraction and injection wells do not require excessive amounts of energy. In some settings, solar energy could power them.

While well pairs may be effective at controlling contaminant plumes, using three wells offers a potential advantage. This involves an extraction well placed directly downgradient of the plume’s leading tip, with peripheral injection wells for containment. Conversely, a three-well pattern may include a central injection well and peripheral extraction wells.

Earlier studies involved flow line distributions for evaluating well pairs (Cunningham and Reinhard, 2002; Wu et al., 2008). Hudak (2015a) used modeled advection and hydrodynamic dispersion to evaluate two-well configurations in simulated homogeneous and heterogeneous aquifers. In a previous study, an injection-extraction well pair outperformed a permeable reactive barrier and non-pumped wells with filter media (Hudak, 2015b).


This study examines a no-action scenario, compared with two-well and three-well alternatives for remediating a contaminated aquifer.

A flow and transport model, MT3DMS (Zheng and Wang, 1999), was used to simulate conditions in a hypothetical unconfined aquifer. MT3DMS involves a finite-difference grid, in this case consisting of 175 rows (oriented east-west), 450 columns (oriented north-south), and one layer. A distance of 0.25 m separated adjacent nodes (centered in cells) along rows and columns. Water table elevation was 5.0000 m and 3.8775 m at nodes in the westernmost and easternmost columns, respectively. The water table sloped eastward, with a regional hydraulic gradient of approximately 0.01.

An elevation of 0 m was set at the base of the aquifer; this was a no-flow boundary, as were the northernmost and southernmost rows of the model. Hydraulic conductivity and effective porosity of aquifer media were set to 0.5 metres/day and 0.25, respectively.

Figure 1 map
Figure 1. Map of initial (top) and residual (bottom) contaminant plume contacting downgradient boundary without intervention; contours in mg/L.

The model generated a flow field and contaminant plume emerging from a 0.56 m2 source area, with a constant concentration of 100 mg/L. This source area was near the western edge of the model domain (Figure 1). The source was active until the plume was 40 m from the eastern edge of the domain. Concentrations at that time were initial conditions for remediation trials (with the source shut off).

In all mass transport simulations, longitudinal dispersivity was 1.0 m, transverse dispersivity was 0.1 m, and the effective molecular diffusion coefficient was 0.00001 m2/d. Plume boundaries coincided with the 1 mg/L concentration contour.

A no-action and three active remediation alternatives were modeled. The active alternatives involved an extraction-injection well pair, with each well pumping at an identical rate, along a linear transect located 5 m downgradient of the initial contaminant plume and oriented perpendicular to the regional hydraulic gradient. The southernmost well injected clean water, considered treated above ground after being extracted from the northernmost well.

Two additional alternatives, each involving three wells along the same transect, were also modeled. One of these alternatives featured a central extraction well and peripheral injection wells. The other involved a central injection well and peripheral extraction wells. The pumping rate at each peripheral well was half the rate at the central well. Thus, the total extraction rate equaled the total injection rate. Each well was screened across the entire aquifer.

Through an iterative process, the model identified the locations of pumping wells, and minimum volume of pumped water necessary to contain and remove the contaminant plume. Flow and transport simulations used the preconditioned conjugate gradient and generalized conjugate gradient solvers, respectively. Mass balance errors were less than 0.01%.


After 1,940 days with an active source, the contaminant plume was 40 m from the eastern model boundary. At this stage, the plume had a maximum width of approximately 14 m. Natural attenuation alone did not contain the contaminant plume on-site. With the source shut off and no intervention, the contaminant plume reached the eastern boundary after 1,660 days. At the time it reached the eastern boundary, the contaminant plume occupied more area, but had lower concentrations, due to the effects of hydrodynamic dispersion and dilution by clean groundwater.

The most efficient well pair capable of containing and removing the contaminant plume had 5 m of separation, with a pumping rate of 0.3 m3/d over 3,690 days. After 3,690 days, concentrations at all model cells were below 1 mg/L, and at no time did concentrations exceed 1 mg/L along the eastern site boundary. By this time, the extraction well had removed 1,107 m3 of water and 5.72 kg of contaminant. The midpoint of the well pair was collinear with the contaminant plume’s long axis.

Attenuation by dilution and hydrodynamic dispersion, in addition to pumping at low rates, effectively reduced contaminant concentrations within the aquifer while containing the plume on-site. Portions of the contaminant plume moved past the well pair, but not off-site.

The most efficient three-well pattern with a central extraction well and peripheral injection wells had 7 m of separation between neighboring wells and pumped 0.3 m3/d over 3,750 days. Over this period, the extraction well removed 1,125 m3 of water and 5.70 kg of contaminant mass. The extraction well lay along an extension of the initial contaminant plume’s long axis.

Figure 2 map
Figure 2. Map of residual contaminant plumes after 2,000 days for two-well and three-well interventions; open circles – extraction wells; dots – injection wells; contours in mg/L.

Finally, the most efficient three-well pattern with a central injection well and peripheral extraction wells had 6 m of separation between neighboring wells. The well pumped 0.4 m3/d for 3,960 days. The extraction well removed 1,584 m3 of water and 5.78 kg of contaminant mass. As in the previous configuration, the central well was collinear with the long axis of the initial contaminant plume. (Figure 2)

Overall, low-capacity pumping well configurations effectively contained and removed the contaminant plume. For each alternative, the most efficient configuration was centered on a downgradient extension of the initial plume’s long axis. Separation between wells varied from 5 m – 7 m, or approximately 36% – 50% of the initial plume’s maximum width.

A well pair was the best alternative, closely followed by a three-well configuration with a central extraction well. These removed a similar volume of water and contaminant mass, over a similar time frame. In addition to being most efficient, the smaller two-well configuration would save on installation, monitoring, and maintenance costs.

Slightly less efficient was the three-well pattern, featuring a central injection well. It pumped a similar amount of contaminant mass as the other alternatives to eventually clean the aquifer. However, it had a 33% higher pumping rate and removed 43% more water which required treatment. With a higher pumping rate, the three-well scheme with centralized injection remediated the aquifer in 7% less time than the two-well scheme.

Overall, results suggest that low-capacity well pairs, and three-well configurations with a central extraction well, may be viable options for remediating some aquifers. They require above-ground treatment of contaminated water, but not the costly excavation and disposal of spent media with permeable reactive barriers. Generally, above-ground treatment systems are easier to monitor and maintain than reactive barriers.

However, the simulated alternatives do have limitations. They require groundwater to effectively carry contaminants to an extraction well(s). If the hydraulic conductivity of an aquifer is very low, and/or contaminants insoluble, alternatives considered here would be ineffective. The simulations also relied upon an on-site buffer zone, within which dilution and hydrodynamic dispersion helped lower contaminant concentrations.

In practice, site-specific conditions dictate the suitability of remediation alternatives.


The objective of this study was to evaluate the relative performance of alternative two – and three-well strategies for containing and remediating a contaminant plume on-site. Results show that a low-capacity well pair, and a three-well scheme with centralized extraction, may be viable alternatives for remediation in some cases, such as the one considered here. For each scheme, the midpoint of the most effective configuration was collinear with the long axis of the initial contaminant plume. Optimal separation between wells was about half the maximum width of the initial contaminant plume.

Paul F. Hudak is with the Department of Geography and the Environment, University of North Texas. This article appears in ES&E Magazine’s February 2018 issue.

References cited in this article

Cunningham, J.A. and Reinhard, M. (2002). Injection-extraction treatment well pairs: An alternative to permeable reactive barriers. Ground Water, 40(6), 599-607.

EPA (U.S. Environmental Protection Agency) (2002). Economic analysis of the implementation of permeable reactive barriers for remediation of contaminated ground water. U.S. Environmental Protection Agency, Washington, D.C.

Gilbert, O., De Pablo, J., Cortina, J-L., Ayora, C. and Cama, J. (2010). In situ removal of arsenic from groundwater by using permeable reactive barriers of organic matter/limestone/zero-valent iron mixtures. Environmental Geochemistry and Health, 32(4), 373-378.

Guerin, T.F., Horner, S., McGovern, T. and Davey, B. (2002). An application of permeable reactive barrier technology to petroleum hydrocarbon contaminated groundwater. Water Research, 36(1), 15-24.

Hudak, P.F. (2009). Internal versus external configurations of passive wells with filter cartridges for cleaning contaminated groundwater. Remediation, 20(1), 133-141.

Hudak, P.F. (2014). Comparison of permeable reactive barrier, funnel and gate, non-pumped wells, and low-capacity wells for groundwater remediation. Journal of Environmental Science and Health, 49(10), 1171-1175.

Hudak, P.F. (2015a). Low-capacity well pairs for treating contaminated groundwater in homogeneous and heterogeneous environments. Remediation, 25(2), 43-53.

Hudak, P.F. (2015b). Low-energy alternatives for removing contaminant plumes in groundwater. The Professional Geologist, 52(4), 23-26.

Lai, K.C.K., Lo, I.M.C., Birkelund, V. and Kjeldsen, P. (2006). Field monitoring of a permeable reactive barrier for removal of chlorinated organics. Journal of Environmental Engineering, 132(2), 199-210.

Ludwig, R.D., McGregor, R.G., Blowes, D.W., Benner, S.G. and Mountjoy, K. (2002). A permeable reactive barrier for treatment of heavy metals. Ground Water, 40(1), 59-66.

USGS (U.S. Geological Survey) (1999). Deep aquifer remediation tools (DARTs): A new technology for ground-water remediation. U.S. Geological Survey Fact Sheet 156-99, Reston, Virginia.

Wu, M.Y., Smits, K.M., Goltz, M.N. and Christ, J.A. (2008). A screening model for injection-extraction treatment well recirculation design. Ground Water, 28(4), 63-71.

Zheng, C. and Wang, P.P. (1999). MT3DMS, a modular three-dimensional multi-species transport model for simulation of advection, dispersion and chemical reactions of contaminants in groundwater systems; documentation and user’s guide. U.S. Army Engineer Research and Development Center Contract Report SERDP-99-1, Vicksburg, Mississippi.

No posts to display


  1. Has this article been published elsewhere or peer reviewed? I would like to cite it for a journal article I intend to submit.


Please enter your comment!
Please enter your name here