Abstract
Quantification of dispersive mixing is critically important for characterizing and predicting solute transport in porous media. Dispersion is often estimated by fitting to data collected from a nonreactive, conservative tracer test. While this approach may provide quality estimates, the estimate is specific to the site, soil, or experimental conditions in which the test occurred. Currently, there are a limited number of a priori models for estimating dispersivity in fully saturated or air–water systems and no predictive models to account for the presence of nonaqueous phase liquids (NAPL). The overall goal of this study was to critically assess both established and new models to predict dispersivity based on properties of the porous medium and fluid saturation. To accomplish this, we assembled and reviewed existing laboratory scale dispersivity datasets in sandy porous media. Only 2 of the 10 existing model formulations offer predictive capability (as indicated through Nash–Sutcliffe efficiency [NSE]). This article describes the development of new, empirical models that enhance the ability to predict dispersivity in laboratory-scale water-saturated (NSE increases from 0.40 to 0.83) and air–water (NSE increases from −1.1 to 0.75) systems of sandy porous media. Knowledge of dispersivity under water-saturated conditions further improves prediction of dispersivity in the presence of a nonwetting phase (NSE = 0.90). The resulting models have utility for systems with transient water saturation, such as those experienced during infiltration and irrigation events, NAPL source depletion, and delivery of foams and emulsions used in site remediation.