The Relevance of Fluid and Porous Media Properties for DNAPL Migration and Entrapment: A Numerical Evaluation of Laboratory Experiments

A large number of sites worldwide are subjected to contamination by dense non-aqueous phase liquids (DNAPLs). This group of typically highly persistent chemicals arise tremendous threats to ecosystems and humankind, especially for groundwater abstraction and usage. In particular, chlorinated solvents have great risk profiles due to their toxic and carcinogenic properties, posing essential needs for appropriate risk assessment and site management strategies. Once released into the subsurface, DNAPLs form so-called source zone geometries (SZGs), i.e., physical shapes containing multiple phases, which represent long-term sources for contamination of downstream groundwater. The complex geometrical and chemical properties of such sources are, together with subsurface characteristics and hydraulic conditions of the aquifer, the most sensitive factors in controlling contaminant plume propagation. As locations of DNAPL sources are widely unknown and subsurface phase exploration methods are limited by technical and financial constraints, in most site assessments, dissolved contaminant plumes are detected only. This fact has led to numerous sites where remediation efforts have been inefficient or even failed, or exceeded economical pre-calculations. Here, improved knowledge on factors controlling source zone formation would lead to better predictions for corresponding SZGs and, therefore, better estimations of contaminant plume evolution and prediction.

A quasi-two-dimensional tank setup formed the basis for generating experimental measurement data of DNAPL migration and entrapment at a Darcy scale under defined laboratory conditions. Three different types of single-size fraction materials (glass beads, filtering glass, and natural sand) were used as homogeneous porous media. DNAPL release into the initially fully water-saturated tank was realized by means of a falling-head boundary condition. Both the aqueous and non-wetting phases were marked for better optical visibility using colorization tracers. All experimental scenarios were conducted under equal ambient conditions (e.g., constant temperature, homogeneous light source). Raw data collection was performed by serial image acquisition from one tank side. A set of customized image analysis and processing approaches was used for the automatized calculation of DNAPL saturation distributions, which served as experimental data for calibrating the base case model scenarios.

For each base case scenario, a representative numerical multiphase flow model was set up using the software codes TMVOC and OpenGeoSys. Starting parameters for calibration were selected based on the tank layout, technical data sheets and hydraulic characterization of the porous media. Each model setup was then run with a specified range of variation for each parameter after successful calibration, whereby parameter ranges were chosen to coincide with physically plausible values at laboratory scale. Through this semi-automatized parameter sensitivity analysis, controlling factors of source zone formation could be identified and ranked along their strength of impact on SZGs. Furthermore, the comparison between results of each software code could identify strengths and weaknesses of each one.