Revisiting the Cape Cod Bacteria Injection Experiment Using a Stochastic Modeling Approach

Colloid filtration models developed in the 1970s for application to homogeneous sand filters used in water treatment have been coupled with the advection-dispersion equation and successfully applied to transport of microorganisms through lab-scale sand columns. Application of this approach to field-scale problems requires adjustment of the model formulation to address the realistic complexity posed by spatial variability of geologic materials known to affect flow and transport processes in aquifers. Representation of geologic heterogeneity as a three-dimensional random permeability field has been incorporated analytically and numerically into mathematical models of colloid transport. In both cases applications to hypothetical field-scale scenarios have shown significant effects of aquifer heterogeneity on the colloid transport process compared to equivalent homogeneous systems. We have applied a particle-tracking technique implementing several different colloid filtration models to a previously-published data set from Cape Cod, Massachusetts, where non-growing, non-motile stained bacteria were injected into an aquifer and recovered 7 meters away in multiple observation wells. The parameters of the heterogeneous hydraulic-conductivity field (mean, variance, and correlation scales of the ln K field) were calibrated based on bromide breakthrough curves from the bacteria/bromide injection test, in addition to independent hydraulic conductivity measurements from a nearby plot. Application of the particle-tracking model to the bacteria data illustrates the robustness of the modeling approach and the sensitivity of transport parameters to the physical heterogeneity representation and the colloid filtration model chosen. In addition, because the distribution diameter sizes of the injected bacteria was known in this case, we show that utilizing a size distribution rather than a mean bacteria diameter as input gives a much more realistic prediction of the composite bacteria breakthrough curve. Results of the simulations also point to needs for quantifying the correlation of local-scale colloid transport parameters (single collector efficiency factor, collision efficiency, detachment) to hydraulic conductivity variability, as well as mathematical incorporation of the effects of geochemical heterogeneity on the three-dimensional transport process.