Phenomenological Modelling of Zero-Valent Iron Deactivation Processes

Michael Finkel1, Irina Kouznetsova1, Markus Ebert2, Peter Bayer1

1 University of Tuebingen, michael.finkel@uni-tuebingen.de, irina.kouznetsova@uni-tuebingen.de,
peter.bayer@uni-tuebingen.de, Tuebingen, Germany
2 University of Kiel, me@gpi.uni-kiel.de, Kiel, Germany

ABSTRACT

Zero-valent iron (ZVI) is widely used as a reactive medium in permeable reactive barriers (PRBs) for treating chlorinated hydrocarbons in groundwater. Numerous researchers have reported on the mechanism and kinetics of contaminant degradation by ZVI. Questions remain, however, concerning the longevity of ZVI that may limit the long-term performance of the treatment technology. Since the complex of processes that are responsible for spatial and temporal ZVI deterioration are not completely understood and appropriate modeling approaches for quantitative predictions are still missing, the longevity estimates of ZVI are based on rules of thumb and general guidance. To fill this gap, we propose a new modeling framework that utilizes a phenomenological modeling approach instead of a processbased approach. We focus on the mathematical description of deterioration phenomena of ZVI, which can quite easily be observed in short- or middle-term (duration in the order of months) accelerated column pre-tests.

The modeling framework consists of (i) a combined zero- and first-order degradation model accounting for the formation of partially dechlorinated products and (ii) a ZVI aging model. The latter describes ZVI aging by lumped parameters that characterize the extent and the rate of ZVI reactivity loss, by means of a timevariant reactivity profile, i.e., a deactivation front propagating through ZVI. Automatic calibration of the model against the concentration measurements in the pre-tests yields degradation rate constants and deterioration parameters. In this way, the calibrated model represents the ZVI performance characteristics for the specific conditions prevailing at the site and is capable of providing site-specific performance predictions over time. The model concept, as well as applications to different accelerated column tests will be presented.