Saltpool Experiment

Introduction

Here the concentration data of two three-dimensional experiments of density-driven flow in a porous medium according to (Oswald and Kinzelbach, 2003) are supplied. They were measured via Nuclear Magnetic Resonance Imaging (MRI). The experiments had a maximum initial salt concentration of 1% salt mass fraction and 10% salt mass fraction and are named saltpool_1 (or salt_l) and saltpool_10 (or salt_d), respectively.

What kind of data is here ?

For each experiment is provided:

• a list of the measurement times including information which temporal phase of the experiment they belong to
• an archive file containing the files of measured concentration values for the 'half" images (at the beginning)
• an archive file containing the files of measured concentration values for the 'full" images
• a file containing the concentration values of the measured breakthrough curve of this experiment (third phase)
• a README.txt file
• four photographic images of a similar experiment to illustrate the experiment

This shall enable any interested person to visualise the measured concentrations according to his/her needs. Using this data in presentations, articles, advertisements, books and similar has to be accompanied with appropriate mention of the origin of this data, e.g. by citation of an appropriate reference as given below. For existing applications of these data as density-driven flow test see also Johannsen et al. (2002) or Diersch and Kolditz (2002).

How are the data arranged in a concentration image ?

The data of each image are provided in sequential slices, which are oriented horizontally in case of the 'half' images (bottom half of porous medium only) and vertically in case of the 'full' images. The concentration values are relative concentrations, i.e. the maximum initial salt concentration has a value of 1.0 . Each slice usually has 80 times 80 values, according to 80 rows and 80 columns in a slice, representing a structured measurement grid with spacing 2.5 mm in both directions. Following the 6400 data points for a slice there is a blank line (to separate different slices). The slices have a thickness of 4.0 mm and therefore the spatial distance between image points is 4.0 mm in the third direction. The images are numbered sequentially. Number '0' indicates an image taken directly before the start of the experiment.

IMPORTANT NOTICE

1. The concentration values have been calibrated using an average signal value for freshwater and an average signal value for maximum saltwater concentration. Therefore concentration values greater than 1.0 and smaller than 0.0 occur. Furthermore, the detection limit of this method is about 5 –10 percent of the maximum concentration used, i.e. relative concentration values below 0.05 could not be distinguished from pure freshwater.
2. The experiments consisted of three temporal phases with different boundary conditions (see Oswald and Kinzelbach, (2003)).
3. The data are provided as they are after calibration. Visualisation has to be done by any user himself. It is strongly recommended to use some kind of smoothing, e.g. a gaussian filter, because there is significant noise in the images. For more advanced methods already applied to this data set you could contact for example Tobias Preusser/Martin Rumpf (University of Duisburg).
4. The images represent the porous medium inside the container. However, there is a small error in spatial location of the images. Therefore the outermost data points could be reduced in signal if effected by the container walls or the very outermost parts of the porous medium could be missing. For details see Oswald et al., (2002) and Oswald and Kinzelbach, (2003).
5. The data were measured in bunches of subsequent slices. Due to the time delay between measurement of different slices there could be already a visible change in the position (and shape) of the saltwater body, if the saltwater is moving relatively fast during this image. This causes jumps of isosurfaces in some of the images (see Oswald et al., (2002)).
6. Simulation runs in terms of benchmarking density-driven flow codes should start with the parameter set of the reference values. However, it is feasible to use any parameter value between the minimum and maximum value of the parameter as given in Oswald and Kinzelbach, (2003). A mathematical benchmark for the experiments is defined for the third phase in Johannsen et al. (2002), giving a grid converged solution using a set of parameters fitted to match the experimental results.
7. Results of numerical simulations should not be interpreted below the scale of a REV, because the physical behaviour could be different on sub-REV scale due the limited validity of physical equations below REV scale. Similar arguments hold for the validity of the common dispersion approach on a scale of at least several dispersion lengths.
8. A benchmarking should use the same parameter set for both experiments, should compare simulated concentrations with measured concentrations (isosurfaces and breakthrough curve), examine grid convergence of results and study parameter sensitivity. A salt mass balance over time could assist the benchmarking additionally (see Oswald and Kinzelbach, (2003)).
9. The measured concentration values of the breakthrough curves also have a measurement error, which is not specified in the file.
10. In case of the salt_d experiment, the 'full' images were corrected for a measuremt error causing a signal loss in side the saltwater body (see Oswald et al., (2002)). This correction could not be applied for the 'half' images at the beginning. Therefore in these images an erroneous concentration loss appears inside the saltwater body. Moreover, in the 'half' images the slices were chosen as 81x81 matrix, because this showed to reflect best the porous medium.

Links to experimental data

References

• Oswald, S.E., and W. Kinzelbach, 2004. "Three-dimensional physical benchmark experiments to test variable-density flow models", Journal of Hydrology, 290(1-2), 22 – 42.
• Diersch & Kolditz, 2002. Variable-density flow and transport in porous media: approaches and challenges, Advances of Water Resources 25(8-12)
• Oswald, S.E., M.B. Scheidegger and W. Kinzelbach, 2002. "Time-dependent measurement of strongly density-dependent flow in a porous medium via Nuclear Magnetic Resonance Imaging", Transport in Porous Media, 47(2), 169-193.
• K. Johannsen, W. Kinzelbach, S. Oswald, and G. Wittum, 2002 "Numerical simulation of density driven flow in porous media", Advances in Water Resources, 25(3), 335-348.

Contact:

Dr. Sascha E. Oswald
Hydrogeology Department
UFZ Centre for Environmental Research
Permoserstrasse 15
D-04318 Leipzig, Germany
Phone: +49 341 235 3985
Fax: +49 341 235 2126
Email: sascha.oswald@ufz.de