Characterization of Anomalous Contaminant Transport via Push-Pull Tracer Tests

Push-pull (single-well-injection-withdrawal) tracer tests are widely used as an economical means of characterizing field-scale solute transport properties such as sorption and dispersion. Typically, these are analyzed by means of analytic solutions that assume transport obeys the radial advection-dispersion equation. We revisit this approach as: (1) Recognition of the ubiquity of anomalous transport and its impact on contaminant remediation necessitates the use of new methods to characterize it, and (2) Improved computational power and numerical methods have rendered reliance on analytical solutions obsolete. Here, we present a technique for characterizing diffusion-driven anomalous transport (i.e., anomalous transport driven by a "trapping" process whose trapping and release statistics are independent of the groundwater flow velocity). Examples include diffusion into low permeability zones, kinetic sorption, and matrix diffusion. Using field observations, we simultaneously calibrate an exponential probability distribution for time spent on a single sojourn in the mobile domain and a truncated power law probability distribution for time spent on a single sojourn in the immobile domain via a stochastic global optimization technique. The calibrated distributions, being independent of the flow regime, are applicable to the same domain under any flow conditions, including linear flow. In the context of the continuous time random walk (CTRW), one may simply define a transition to represent a single trap-and-release cycle, and directly compute the spatiotemporal transition distribution that defines the CTRW from the two calibrated distributions and the local seepage velocity (so that existing CTRW transport theory applies). A test of our methodology against a push-pull test from the MADE site demonstrated fitting performance comparable to that of a 3-D MODFLOW/MT3DMS model with a variety of hydraulic conductivity zones and explicit treatment of mobile-immobile mass transfer. In particular, our model captured the heavy tail portion of the breakthrough curve nearly perfectly. Use of the technique for characterization of dual porosity media and uncertainty analysis at a major remediation site at LANL will also be discussed.