Abstract
Dense, non-aqueous phase liquids, or DNAPLs, represent one common group of groundwater pollutants. The rate of contaminant mass leaving the source zone in the aqueous phase, or mass flux, is introduced as a suitable tool for assessing DNAPL dissolution behaviour, remediation efficiency and future risk assessment predictions.
In all numerical simulations and laboratory tank experiments, aquifer heterogeneity controlled the PCE spill morphologies, resulting in variable vertical DNAPL distribution intervals. A method of mass flux normalisation was developed to consider these contrasting spill morphologies, and for all numerically simulated aquifer realisations, most normalised mass flux values lay within the 0.2 to 0.4 range.
For a range of naturally occurring field conditions, normalised mass flux was essentially independent of changing v or K1, but two traditional mass transfer correlations greatly over predicted normalised mass flux values for the same parameter values. A dual-flow source zone model has been developed to explain normalised mass flux values in relation to DNAPL morphology. The effects of increasing aquifer permeability field heterogeneity were also limited, with many normalised flux values still lying within the range of 0.2 to 0.4.
In a series of six 2-D tank experiments with increasing sand model heterogeneity, the spill behaviour of PCE was controlled by the location of the coarsest sand units. Total DNAPL mass recoveries following a series of surfactant flushes were in the range of 70 to 80%, yet effluent PCE concentrations were greater at the end of each experiment than initial values. Vertical aquifer intervals contaminated by PCE also increased. Normalised mass flux variations were mostly within the range of 0.3 to 0.4, in close agreement with data from the numerical simulations. Increasing the flow velocity of water through each source zone by a factor of two had no effect on effluent PCE concentrations. Thus, macro-scale contaminant morphology, rather than pore-scale mass transfer behaviour, must be considered for accurate source characterisation and mass flux predictions.
In all simulations and tank tests, PITT-based PCE volume estimates showed wide ranges of accuracy, ranging from 70% under prediction to 20% overprediction. However, variations of retardation factor with depth showed high degrees of correlation with upgradient vertical DNAPL distribution intervals, and thus are proposed as a suitable tool for estimating the source zone dimensions required for mass flux normalisation.
Using data from all of the numerical simulations and laboratory tank experiments, a positive correlation has been established between normalised mass flux and average tracer-based source zone DNAPL saturation for low saturation values. Normalised flux remained essentially constant at greater saturations. However, the point where normalised flux decreased corresponded with a very low saturation value, which is unobtainable with current remediation technologies.
This conclusion is of potentially significant benefit for future field application. Assumption of a constant normalised mass flux value enables remediation targets to be identified in terms DNAPL distributions perpendicular to flow. This behaviour can also be used as a tool for assessing the efficiency of remediation operations, and a potential field application is presented.
DNAPL, mass transfer, heterogeneity, flow bypass, mass flux normalisation, PITT