Does Your Hydraulic Containment System Really Capture Everything?

Hydraulic containment remains one of the most practical options for controlling the spread of large contaminant plumes at great depth (>300 feet below ground surface). Currently, hydraulic containment systems for plumes located in deep and complex hydrogeological settings are designed primarily through numerical groundwater modeling. The ultimate goal of such modeling is often to estimate the groundwater extraction rates required to achieve complete plume capture. Because of time and resources constraints, many modeling practitioners use the following conventional approach to tackle the problem: calibrate the groundwater flow model of interest, then vary the calibrated hydraulic conductivity field over a reasonable range to obtain a range of extraction rates that may be needed for plume capture. Although it is easy to implement, this conventional approach may be inadequate for two primary reasons. First, varying the hydraulic conductivity field alone often makes the model deviate from the calibrated conditions; therefore, extraction rates estimated under such conditions may be misleading. Second, the effects of parameter nonuniqueness are frequently not explored through the conventional modeling approach, potentially resulting in significant additional uncertainty in the estimated extraction rates. While the accuracy of modeled extraction rates do not solely dictate the final success of a hydraulic containment system, a better understanding of the extent of uncertainty can definitely improve decision making for large-scale projects. A classic example of parameter nonuniqueness that has been repeatedly encountered in real world applications is the difficulty in decoupling the effects of recharge and hydraulic conductivity on groundwater head observations. To provide some perspective on this topic from an industry and consulting standpoint, we will present a “hypothetical” case study based on numerous challenging site investigation projects we have worked on. This case includes the following key components: a large contaminant plume in a multi-layered aquifer, a barrier fault that influences plume migration, nearby production well activity, and spatially-varying recharge. We will illustrate the major sources of uncertainty impacting a groundwater flow model, the challenges of collecting useful data to improve model calibration, how parameter uncertainty is translated into uncertainty in estimated extraction rates for a hypothetical hydraulic containment system, and how to communicate issues of uncertainty with decision makers.