Strategies for Cost-Effective Chemical Delivery and Mixing for Bioremediation

In situ bioremediation plans often call for injection and extraction wells or other technologies that will ensure delivery and removal of the reactants and products of the cleanup reactions. However, it can be difficult to anticipate their effectiveness in the heterogeneous conditions of an aquifer. WRHSRC researcher Peter Kitanidis and his research team are working on this problem. They are are developing computer models that help practitioners visualize aquifer conditions and design treatment systems that enhance aquifer mixing.

Click on highlighted words and illustrations or links will pop up.

Dr. Kitanidis, a professor of Environmental Engineering at Stanford University, comments that mixing is an understudied part of bioremediation, “Laboratory studies often focus at the bench scale and do not consider the difficulties involved in scaling up.” Several factors make mixing a difficult process to anticipate. First, aquifers are heterogeneous. Variation in the geology of the aquifer will influence characteristics such as dispersion and sorption and it can be prohibitively expensive to determine geologic characteristics in detail. Second, Kitanidis points out that there is a lack of process understanding of mixing in aquifers. Most mixing studies have focused on surface conditions where flow is turbulent. However, in the porous media of a geologic formation viscosity dominates and suppresses turbulence. The slow rate of mixing in non-turbulent systems also means that it is likely to be the limiting factor for chemical reactions. Third, mass transport in a formation is influenced by external factors that may be hard to predict. For example, the regional flow field, may respond to changes in nearby river stage.

A Case Study at Oak Ridge National Laboratory

In recent years, Kitanidis and his team have informed their mixing studies with work at a field site at Oak Ridge National Laboratory. The site is a good place to study mixing because the aquifer geology, chemistry, and flow field are extremely complex. The goal at the site is for chemical reduction of Uranium (VI) to less mobile Uranium (IV). The aquifer is highly weathered saprolite and, in addition to the high uranium concentration, the pH is exceptionally low (~3.5) and nitrates are exceptionally high (~10 g/L). Nitrates need to be removed and pH needs to be raised in a way that permits biological activity but prevents metals from precipitating and clogging the porous medium. In addition, the regional flow field is influence by a nearby stream so that the flow rates and direction change with rainfall events.

A team of researchers from Oak Ridge National Laboratory and Stanford University (under the direction of Craig Criddle and Philip Jardine) has developed a cleanup process involving both above ground and in situ treatment steps. Injection and extraction wells are set up to create two recirculating loops (Figure 1 and Figure 2). The outer loop captures contaminated water and brings it to the surface for removal of metals and nitrogen. Once it is reinjected, it creates a “protected” zone around the inner loop where the aquifer chemistry is appropriate for microbial uranium reduction.

Kitanidis and his team developed mathematical models of the flow, transport and biogeochemistry of the system. The models enable the team to visualize injection, extraction, and recirculation patterns and compare predictions with the results of experiments and field tests. They also enable determination of breakthrough curves and residence times for the reactants and products involved in the cleanup reactions.

Software

Stanford University graduate students Mike Fienen and Jian Luo designed the models as functions that run in the mathematical software MATLAB ®. They describe the first module, called Complex Capture Zone Analysis Routine or ComCZAR, in a forthcoming article in the Journal of Hydrology. A public use version of the model is available online ( http://www.stanford.edu/~fienen/software/ ). ComCZAR is a two-dimensional, analytical model that creates an image of streamlines showing capture zones for wells in a layered homogenous aquifer with anisotropic transmissivity and a regional flow field.

Fienen, Luo, and Kitanidis developed a second module for delineation of capture, release, and recirculation zones. This series of MATLAB algorithms also enables the user to create isochrons (contours of equal travel time) and breakthrough curves for extraction wells. The authors are preparing a paper on this module and a version is available online at (http://www.stanford.edu/~jianluo/software/ ).

Kitanidis and his team hope these models can be applied to other field sites where injection and extraction wells will be implemented. They believe that a better understanding of aquifer mixing a suite of effective computer models can help practitioners develop more cost-effective bioremediation systems. Over the next year, they plan to develop a manual on chemical delivery and mixing and the use of their modeling programs.