|Identification of Relevant Parameters in Multi Fluid Flow Systems|
|Project manager:||Prof. Dr. h.c. Dr.-Ing. E.h. Helmut Kobus, Ph.D.|
|Deputy:||Jürgen Braun, Ph.D.|
|Research assistants:||Jane Allan, Ph.D.|
|Duration:||1.2.1997 - 31.12.1999|
|Funding:||Deutschen Forschungsgemeinschaft, DFG|
This project is part of the research area:
In-Situ Remediation Technologies
Abstract:The behaviour of a DNAPL in the saturated zone was investigated during this research work.
Focussing on the infiltration of TCE in a heterogeneous porous aquifer, experimental and numerical
methods were used and compared to describe the relevant processes and parameters for the
propagation of a DNAPL.
A series of laboratory infiltrations of TCE were carried out in nominally two-dimensional saturated sand packings. The results were used to evaluate the influence of heterogeneities (in the form of coarser and finer sand lenses) on plume development under inherently unstable infiltration conditions. Based on visual observations:
- Migration was dominated by very rapid downward fingering through macroscopically homogeneous portions of the sand packing, while the presence of heterogeneous interfaces arrested the downward migration of a fingering plume.
- At a coarser sand interface lateral spreading occurred followed by the initiation of fingering once the plume has filled the coarser material.
- At a finer sand interface lateral diversion of the plume occurred, however, very little accumulation or “pooling” of the plume over the thickness (85mm) of the experimental model was observed.
- TCE became entrapped (or pooled) at high saturations in the coarse sand lenses, when the capillary pressure was below the entry pressure of the surrounding material
- Because the plume moved through most of the packing as fingers, very little mass was retained, with the exception of the coarser lenses.
The experimental results were compared to simulations using a continuum based numerical model:
- The continuum based model is inappropriate for the homogeneous case because plume spreading was dominated by fingering.
- Predictions of infiltrations in lens packings were reasonable. Predicted plume movement was slightly slower but the main difference was the overestimation of mass retained in macroscopically homogeneous portions of the sand packings where fingering occurred, and therefore underestimation of the extent and depth of the final plume.
The continuum based model predictions of capillary pressure response closely resembled capillary pressures measured throughout the lens sand packings, including portions through which fingering dominated. Differences were generally lower predicted pressures and later predicted arrival of TCE at the tensiometer locations.
Since it is clear that one of the dominant processes controlling flow in this system (fingering) is not addressed specifically by continuum based models, simulations were carried out using a “finger pore space porosity” which is less than the actual porosity. The reduction in the pore space available for flow resulted in increased rates of migration and spreading, which were closer to those observed in the experimental model. More importantly however, the final distribution of TCE at residual or entrapped along the migration path is more accurately represented and therefore the overall extend of contamination in a bottomless system can be more accurately estimated.
Although model predictions of the final TCE distribution improved using the finger pore space model, the model is a very simplistic approach to a much more complicated problem. A more sophisicated approach might be developed which considered other system variable (e.g. front velocity, fluid and porous media properties), which are known to play a role in the development of fingers under unstable conditions.