Paper No. 2
Presentation Time: 1:45 PM

CALIBRATION AND PREDICTION OF URANIUM TRANSPORT IN SMALL SCALE FIELD EXPERIMENTS


CURTIS, Gary P., U. S. Geological Survey, 345 Middlefield Road, MS 409, Menlo Park, CA 94025, LU, Dan, Department of Scientific Computing, Florida State University, 400 Dirac Science Library, Tallahassee, FL 32306, YE, Ming, Department of Earth, Ocean, and Atmospheric Science, Florida State University, 303 Carraway Building, Tallahassee, FL 32306 and KOHLER, Matthias, US Geological Survey, 345 Middlefield Road, Menlo Park, CA 94025, gpcurtis@usgs.gov

Prediction of subsurface reactive transport of hexavalent uranium (U(VI)) is challenging because of the inherent complexity of both the subsurface physical structure and the geochemical processes. These systems are inherently open and amenable to multiple plausible conceptual models. Two tracer tests were conducted by extracting contaminated groundwater and then either increasing or decreasing the alkalinity of the pumped groundwater and finally injecting the solution back into the aquifer after the addition of a KBr tracer. In both tests, the observed Br breakthrough at five observation wells was complex and often showed multiple Br peaks indicating physical heterogeneity in the subsurface hydraulic conductivity. Increasing the alkalinity of the injected water from 8 to 23 meq/L initially caused the U(VI) concentration observed 1-2.5m downgradient to increase from 4 µM to 11 µM which was followed by a decrease in U(VI) to as low as 2 µM and then gradually relaxing to the initial concentration. The experiment with the decreased alkalinity followed the opposite trend with U(VI) concentrations initially decreasing followed by an increase above ambient conditions. The complex observed behavior of Br was reproduced with streamtube model. Reactive processes were included in the streamtube model using a set of alternative models that were postulated to account for ion-exchange processes, surface complexation of U(VI) and rate-limited mass transfer processes. Alternative models are being evaluated for their ability to match calibration data and to predict experimental results withheld from calibration.