Search

Sealing with waterglass is one option of technical improvement of the geological barrier. The process of injection is rather involved, because it depends on the kinetics of reaction of the injected waterglass with salt and brines, on the nature of the fractures and the injection velocity. Generally, up to now only the final result of this impregnation could be tested with injection tests and tomographic methods. We already applied PET process monitoring as laboratory method in a large number of studies of conservative and reactive flow, as well as diffusion experiments, in different geological materials. PET enables to observe the propagation of radiolabelled substances with ultimate sensitivity and with a reasonable spatial resolution of 1 mm. We now developed a method to observe the process of waterglass impregnation into salt rock with PET. Labelling of waterglass is possible by simply adding a small portion of [18F]KF solution, with an activity of around 100 MBq. During the injection of the labelled waterglass into the salt rock we acquire a sequence of PET scans which yield a sequence of the spatial activity distribution in the sample. The activity per voxel is proportional to the volume fraction of waterglass. The method was tested on small volumes of salt grit and a drill core, which was previously damaged with a geomechanical test and characterized with CT-imaging. These first examples were conducted at a low entry pressure, which limited the achievable depth of penetration and thus the achieved permeability reduction. However, PET-monitoring of the flow process before and after the waterglass injection showed significant alterations of the flow field. Generally, this method is applicable also with other impregnation agents and matrices. Our approach, where we combine numerical process simulation based on CT-imaging with direct experimental process observation with PET is suited to improve fundamental process understanding and to verify the underlying assumptions and model codes.

Transport and retardation of chemical species in soils as observed by input-output approaches are commonly interpreted by process simulations and break-through curve (BTC) fitting. Positron emission tomography (PET) provides a direct quantitative spatiotemporal (4D) visualization method for the propagation of compounds labelled with a PET-tracer at intermediate resolution and molecular sensitivity (Kulenkampff et al. 2013). In the framework of SPP 1315, we conducted transport experiments on an artificial soil column with both reactive and conservative tracers, which were monitored with sequential PET imaging. The soil column used (l: 94.5 mm, d: 40 mm; composition: 94% sand, 5% illite, 1% goethite; porosity: 29%) was prepared under CO2-atmosphere and structurally characterized by µCT imaging as widely homogeneous. For the conservative tracer experiment, 5 mL 0.001 M NaNO3 + 0.01 M [18F]KF was flown through the equilibrated column. For the reactive species experiment, 64Cu was produced at the Leipzig cyclotron by the nuclear reaction 64Ni(p,n)64Cu and separation by ion exchange. 5 mL of 0.0008 M [64Cu]Cu(MCPA)2 was produced from 2 mL 64Cu2+ in 0.1 M HNO3, 1 mg Cu(NO3)2•3H2O and 2 mg MCPA in synthetic pore water. The labeled solution was adjusted to pH 5 and flown through the column, which had no former contact with MCPA and had been preconditioned for 4 days with synthetic pore water at pH 5. In both experiments the flow rate was 0.1 ml/min. In the conservative experiment, the break-through occurred after 140 min, and – in spite of the homogeneous packing of the column – the tracer propagation observed with PET showed a preferential flow field towards the rim of the sample. The reactive [64Cu]Cu(MCPA)2 pulse was strongly retarded with a break-through of the activity after 66 h. Fig. 1 shows a snapshot of both experiments after 110 min. Preferential and superficial transport, commonly ignored in input-output approaches, controls the effective volume and reactive internal surface area, and thus impacts interpretation and inverse numerical modelling of BTCs. Such effects can be assessed and quantified with PET process tomography, especially when the pore structure is heterogeneously altered by microbial activity. Reference Kulenkampff, J., Gründig, M., Korn, N., Zakhnini, A., Barth, T., Lippmann-Pipke, J., 2013. Application of high-resolution positron-emission-tomography for quantitative spatiotemporal process monitoring in dense material. http://www.isipt.org/world-congress/7/902.html.

With 1D and axial symmetric 2D finite element simulations of tracer transport in porous media columns, we conducted systematic studies of increasing heterogeneity on the resulting break-through curves (BTC). While preferential flow can still have comparably little effect on the BTC of non-reactive tracer substances, for strongly adsorbing tracer preferential flow can significantly shift the break-through towards earlier arrival times by reducing the reactive surface area. Motivation for conducting this systematic simulation study is our unique GeoPET method (Positron Emission Tomography) that allows the visualization of the spatio-temporal tracer distribution in geologic media (Richter et al., 2000; Gründig et al., 2007; Kulenkampff et al., 2008). The real complexity of thus directly observed flow fields is always surprising. Even for non-reactive tracers quantitative transport simulations typically remain a challenge due to the need for detailed knowledge of hydrodynamic parameter values in 3D with a certain critical spatial resolution. Here we show simulation results obtained by the COMSOL Multiphysics code coupled with PHREEQC (Wissmeier and Barry, 2011) and compare them with measurement results (BTC) from reactive transport (herbicide MCPA on goethite/sand). CD-MUSIC adsorption model parameter were obtained from Kersten et al., (2014). The underlying structural heterogeneity of the simulated column experiments reflects preferential flow along the column boundaries caused by locally elevated permeability in otherwise homogeneous sand packing as verified by the GeoPET images from both conservative and reactive tracer experiments. References Gründig, M., Richter, M., Seese, A. and Sabri, O., 2007. Tomographic radiotracer studies of the spatial distribution of heterogeneous geochemical transport processes. Applied Geochemistry, 22: 2334-2343. Kersten, M., Tunega, D., Georgieva, I. and Vlasova, N., 2014. Surface complexation modeling of herbicide adsorption by goethite: 1. 4-chloro-2-methylphenoxyacetic acid (MCPA). Environmental Science & Technology, submitted. Kulenkampff, J., Gründig, M., Richter, M. and Enzmann, F., 2008. Evaluation of positron emission tomography for visualisation of migration processes in geomaterials. Physics and Chemistry of the Earth, 33: 937-942. Richter, M., Gründig, M. and Butz, T., 2000. Tomographische Radiotracerverfahren zur Untersuchung von Transport- und Sorptionsprozessen in geologischen Schichten. Zeitschrift für Angewandte Geologie, 46(2): 101. Wissmeier, L. and Barry, D.A., 2011. Simulation tool for variably saturated flow with comprehensive geochemical reactions in two- and three-dimensional domains. Environmental Modelling & Software, 26(2): 210-218.

The processes of dissolution and destabilization during and right after the flooding of a salt mine are in principle well understood but in detail still enigmatic. In our attempt to visualize these mechanisms, a combination of structural imaging as well as computer simulations were performed at the pore scale. High resolution positron emission tomography in situ imaging of fluid flow processes in heterogeneously structured drilling cores were matched with high resolution x-ray computer tomography imaging of effective hydraulic microstructures [1]. The microstructures were used as input for lattice Boltzmann based particle tracking simulations [2]. The space and time resolved combined imaging of hydraulic effective structures and fluid flow processes (Figure 1) allows an estimation of the effective volume within the network of pores and fissures and an estimation of the reactive surface area of rock material. Flow path topology concerning, e.g., fingering phenomena, dispersion and molecular diffusion effects can be quantified, and velocity distribution can be observed in 3D [3]. With PET, we observed the propagation of brine labelled with a PET-tracer (124I or 18F) through a mechanically damaged rock salt drill core, which was structurally characterized with CT. The experiments show, that two heterogeneous effects have to be considered with respect to dissolution and destabilization of the material. The first one is a strong localization of fluid paths dominated by fingering phenomena especially along fissures but also in parts in homogeneous porous areas. As a consequence, fluid in motion uses only a small part of the available pathway even under saturated conditions. The fraction of the internal surface of a rock sample which is exposed to the propagating fluid – the effective reactive surface area – decreases with increasing localization of actual transport paths. Therefore, this effect considerably narrows the part of the pore space, where dissolution or other interactions are likely to occur. This effective pore volume and surface area can be quantified experimentally with PET process observation and with CT-based flow field simulations. The second mechanism is a high variability in the streaming velocity patterns along distinct parts of the fractures. Locally enhanced flow velocities increase local dissolution and cause widening of fracture cross sections. This may lead to a self-enhancing effect of increasing flow velocities and flow rates in saline rock. These results may contribute to the understanding of destabilization and dissolution processes and ultimately pylon collapse upon flooding of a salt mine.

Positron-Emission-Tomography (PET) enables direct and quantitative monitoring of the spatiotemporal distributions of dissolved inert and/or reactive PET-nuclides and PET-nuclide-labelled compounds during their passage through decimeter-scaled material samples. We apply our biomedical PET-scanner (ClearPET, Raytest) exclusively to geomaterial samples and reach the physical limit of spatial resolution of nearly 1 mm with pico-molar sensitivity. Artificial soil columns (68.5% fine-medium sand, 25.5% coarse silt, 5% Illit, 1% Goethit) have been prepared in Jena. µCT-images have been produced at the Bundesanstalt für Materialforschung und –prüfung (BAM) and analysed in Mainz. They show layering and inhomogeneous water saturation, yet a roughly homogeneous grain distribution. PET-imaging was conducted during flow-through experiments into the upright column, which is perpendicular to the scanner axis. First flow experiments have been conducted with – presumably – conservative tracers ([18F]KF (decay time 109.77 min) and Eosin). Conventional break-through curves (BTCs) have been recorded both by fluorescence detection of Eosin and with a Gamma Radioactivity Flow-Through Monitor (Gabi Star, Raytest). From the transport process observation by sequential PET-imaging we could derive longitudinal BTCs during the tracer propagation through the column. Further, it could be shown that 18F- was slightly retarded and - more significantly - the effective pore volume was inhomogeneous (preferential transport through the sample). These experiments will be repeated on newly constructed columns with plastic endplates (instead of the original stainless steel ones) and an optimized dead volume. To preliminarily conclude, we demonstrated that the combined tomographic information on pore structure and spatio-temporal transport process observation in a soil model column yields an essential gain in information for describing and parameterizing transport processes, compared to input-output experiments. Such measurements represent the experimental control of ab-initio model simulations of conservative and reactive transport, e.g. [64Cu]MCPA), and of the interpretation of conventional input-output transport experiments, conducted by the partners.

Migration of radionuclides out of a salt-based repository requires transport pathways, where substances propagate by molecular diffusion or even pressure-driven advection. Diffusion coefficients and permeability, being very low in most intact saline rocks, are likely to increase considerably under geomechanical stress, inevitably occurring for instance in the excavation damage zone. Observation and evaluation of transport processes in such heterogeneously structured materials are most demanding, because they likely occur in either extremely localized, macroscopic but rare fractures or in dispersed, mostly indiscernible but still connected microscopic voids. Applying salt solutions labelled with radionuclides we are able to observe such processes directly and - presumably for the first time with due resolution and sensitivity – with our GeoPET method. The same processes are studied with Lattice-Boltzmann flow simulations on µCT-based images of the same samples. This strategy contrasts the common use of computer simulations based on a priori material parameters, and allows for the matching of flow simulation results in “real structures” with real time flow monitoring in the same samples. Here we report on results obtained from a number of drill cores from the ancient Stassfurt salt mine and its overburden, including aged backfill material, which has been analysed in the framework of our joint research project, funded by the Federal Ministry of Education and Research (BMBF) (Wolf et al., 2010).

This thesis is located at the Institute of Radiochemistry, FZD Research Site Leipzig for Interdisciplinary Isotope Research, Reactive Transport Division (FWRT). The main focus of this division is the investigation of transport processes in geosystems by means of radiotracer applications. The main topic of the thesis is the visualization of transport processes in geologic material by means of the in-house development of the GeoPET-method. This work is conducted as part of the scientific joint venture: „Dynamik abgesoffener oder gefluteter Salzbergwerke und ihres Deckgebirgsstockwerks“ („Dynamic of drowned or flooded salt mines and their overburden“), coordinated by the Bundesanstalt für Geowissenschaften und Rohstoffe (BGR). Since the late 19. century drowned salt mines cause severe mining damages in the city of Stassfurt (Saxony-Anhalt). Sink hole depressions and subsidence of the surface below the groundwater table destroyed large parts of the down town. The general causation was lack of experience with salt mines and mining in gypsum karst in the 19. century. The causation in detail why and how exactly salt rock has washed out is more complicated to identify, as streaming mechanisms at the small scale level are partly still unclear; the general fluid dynamics at small scales is partly unknown. To reveal these processes and mechanisms the behaviour of salt brines at the millimetre scale in drilling cores of the different geological units of the salt rock and its surrounding is examined by three-dimensional visualization of the distribution of radioactive labelled water measured with PET. Mechanisms at millimetre scale control mechanisms at the kilometre scale and are of utmost importance for the principal understanding of fluid dynamics. In the laboratory you can have „a look into the rock“. In the field this is not possible this way. Combining PET data with high resolution CT-scans of the samples (conducted by the cooperation partners JGU Mainz and BAM Berlin) allows an alignment of processes of the fluid flow and its associated hydraulic pathway structures. This matching is important for understanding and for generalized conclusion about ongoing processes and is a necessary preparatory work for computer modelling. Lattice-Boltzmann-simulations of velocity fields and streaming patterns based on CT-data are compared with PET-data derived from the same samples. This comparison of the flow patterns is done by means of geostatistic methods that allows scale independant spatial correlation of the patterns and therefore provide scale indpendant parameters like correlation lengths that are a necessity for upscaling. Short term objective is the improvement and validation of parameters and fluid flow concepts derived from small scale simulations. Long term objective is the improvement of upscaling of parameters and concepts to the field scale and a better understanding and prediction of mining damages and groundwater behaviour.

Scaling is a fundamental problem in groundwater hydrology. A typical challenge is the verification that hydrodynamic parameters obtained in laboratory experiments well represent a situation on the field scale. Here were propose that for reactive transport modeling, the prediction of the large scale fluid dynamics and concentration distributions should be based on the characteristics of hydromechanic and geochemical parameter sets on the millimeter and micrometer scale. While it is common sense that chemical reactions take place on the atomic scale, here we show with tomograhic process observation and modeling that also hydrodynamic processes are considerably influenced even by sub-µm scale characteristics of the geomaterial and thus determine the fate and dynamics of the system components in the fluid phase. We applied the spatially highly resolving computer tomography (μXCT) on rock cores for determining the open pore structures. Based on these µm scaled structural information the lattice Boltzmann simulations were conducted. Column experiments were performed on the same samples while applying the process visualization method GeoPET that allows for the non-invasive, quantitative monitoring of e.g. dissolved positron-emitting radio tracers ([18F]KF, [124I]KI) added to the eluent. Visualizing and quantifying transport processes in geological material by means of GeoPET provides a high volume resolution of 1.5 μl (1.3 mm) and extreme high sensitivity for tracer concentrations (10−15 to 10−12 moles/ml). The matching of measured time resolved fluid flow patterns and simulated small scale fluid dynamics is conducted by means of geostatistic methods (variography). It allows validating the structurally related hydromechanical parameters like flow velocities derived from the simulation and offers insights to the linking between ongoing processes on the micrometer scale and its impact on the centimeter scale. The applied scale independent geostatistical tools provide scale independent parameters, like the correlation lengths. Such parameters are a suggested fundamentally important base for valid upscaling to the field scale. We provide results from rock cores with both, relatively simple structured pathways as well as complex ones. In both types only a small part of the available pathway is passed through by the mobile fluid implying that only fractions of the inner surface were available for chemical reactions. Such findings should fertilize the concepts of reactive transport models aiming at larger scales. Therefore we conclude: Microscale information is essential for improving reactive transport models.

Der ehemalige Kali- und Steinsalzbergbau im Raum Staßfurt (Sachsen-Anhalt) führt seit dem 19. Jahrhundert zu stellenweise erheblichen Bergschäden im Stadtgebiet, die auf Subrosion und Konvergenz untertägiger Hohlräume zurück gehen. Gebäude und Infrastruktur sind von großflächigen Absenkungen, Vernässungen und lokalen Tagesbrüchen in Mitleidenschaft gezogen worden. Da man ähnliche Bergbaufolgeschäden in vielen Regionen des Salzbergbaus antrifft, hat sich das aus mehreren Forschungseinrichtungen, Universitäten und Unternehmen bestehende Forschungsverbundvorhaben Dynamik abgesoffener oder gefluteter Salzbergwerke und ihres Deckgebirgsstockwerks unter Federführung der Bundesanstalt für Geowissenschaften und Rohstoffe (BGR) zum Ziel gesetzt, Ursachen, Prozesse und Auswirkungen der Bergschäden exemplarisch und allumfassend zu untersuchen. Um ein möglichst umfassendes Bild der im Untergrund ablaufenden Prozesse und deren Ursachen und Folgen zu erhalten, sind Untersuchung der kleinräumigen Strukturen und Prozesse im Labormaßstab erforderlich. Die großräumige Abschätzung der Grundwasserdynamik im Salinar und im Deckgebirge, also der Ursachen und Folgen der Subrosion, bedarf der Charakterisierung von Wegsamkeiten und der Fluiddynamik bis in den Milli- und Mikrometerbereich, da sich die grundlegenden Prozesse auf der molekularen Ebene in der Mikroskala abspielen und somit die Dynamik in der regionalen Größenordnung bestimmen. Wie auf der Feldskala, so wird auch auf der Laborskala die Verteilung von hydrologischen Parametern, wie z.B. Permeabilitäten, Abstandsgeschwindigkeiten und Dispersionsraten von der heterogenen Struktur des Gesteins kontrolliert, weswegen diese Parameter nicht nur im regionalen sondern auch im Labormaßstab strukturbezogen, also räumlich differenziert, erhoben werden sollten. Daher wurde in enger Zusammenarbeit mit dem Geologischen Institut der JGU Mainz eine Methode zur direkten räumlich aufgelösten und quantitativen Prozessbeobachtung entwickelt und angewendet. Hierfür wurde die Methode der Positronen-Emissions-Tomographie (PET) eingesetzt, ein ursprünglich medizinisches, bildgebendes Verfahren. Dieses Verfahren erlaubt es, zerstörungsfrei, mit höchstmöglicher Empfindlichkeit und geeignetem Auflösungsvermögen, die Bewegung einer mit einem radioaktiven Tracerisotop (hier F-18 und I-124) markierten Wassermenge in Raum und Zeit zu beobachten, wobei die Menge des eingesetzten Radiotracers im Bereich weniger Nanogramm gehalten werden kann. Das System wird somit durch den Tracer auf geringst mögliche Weise beeinflusst und es wird ein sehr realistisches Bild des Fließverhaltens mit einer räumlichen Auflösung von 1 mm und einer zeitlichen Auflösung von 60 s erzeugt. Bislang wird in der Literatur nur von wenigen Ansätzen berichtet, die das Ziel verfolgen, die realistischen Verhältnisse strömender Fluide in heterogen strukturierten Gesteinen dreidimensional abzubilden. Solche Strömungsmuster werden in der Regel mittels stochastischer Modelle durch Computersimulationen erzeugt. Erst ein Abgleich mit gemessenen Strömungsmustern kann zur Validierung dieser Simulationen führen, was auch erst ein valides Upscaling in die Feldskala ermöglicht. In Kooperation mit der Bundesanstalt für Materialforschung- und prüfung Berlin (BAM) und dem Geologischen Institut der Johannes-Gutenberg Universität Mainz (JGU) werden daher die PET-Messungen zur Fluiddynamik mit Lattice-Boltzmann-Simulationen der Fließprozesse verglichen, die auf hochauflösenden computertomographischen µXCT Messungen der internen Struktur der identischen Bohrkerne beruhen. Der Vergleich von PET-Messdaten mit Lattice-Boltzmann - Simulationsdaten bedeutet eine Skalenübertragung um etwa drei Größenordnungen (µm - mm). Daher erfolgt der Abgleich zwischen gemessenen und simulierten Daten auf statistischem Wege. Gemessene und simulierte Strömungsmuster werden mittels Variographie mit einander abgeglichen, da diese Methode skalenübergreifend die Korrelationen zwischen den Fließmustern wider gibt. Ebenso werden räumlich differenzierte Durchbruchkurven und Histogramme für den Vergleich von Messung und Simulation heran gezogen. Diesem Ansatz liegt das Bestreben zugrunde, zu einer Erweiterung des generellen Prozessverständnisses und der Vorhersagbarkeit über das Verhalten von Fluiden in porösen und klüftigen Gesteinen beizutragen und die beobachteten Phänomene und Prozesse von der Labor- auf die Feldskala zu übertragen. Nicht in jedem Fall war dieser Abgleich zwischen Simulation und Experiment jedoch möglich: während in klüftig-porösen Materialien gewöhnlich mit beiden Methoden räumlich stark differenzierte präferentielle Fließwege im Kluftsystem gefunden wurden, konnte in eher mikrostrukturell ausgeprägten Materialien trotz messbarer Permeabilität gelegentlich kein verbundener Porenraum aus den CT-Bildern segmentiert und somit keine Lattice-Boltzmann-Simulation durchgeführt werden. In diesem Fall zeigte die PET-Untersuchung ein diffuses Ausbreitungsverhalten des Tracers.

Positron-Emission-Tomography (PET) enables direct and quantitative monitoring of the spatio-temporal distributions of dissolved inert and/or reactive PET-nuclides and PET-nuclide-labelled compounds during their passage through decimetre-scaled material samples. We apply PET exclusively to geological samples and reach the physical resolution limit of about 1 mm with our small-animal-PET scanner (ClearPET, Raytest). We suggest our GeoPET has unrivalled sensitivity and selectivity for tracer concentrations to some 107 tracer atoms/µl and thus is ideally suited for direct flow and transport process observations in soils. This lower limit of the tracer concentration in the order of about 1 kBq/µl outranges other process observation methods (e.g. NMR or resistivity tomography) by many orders of magnitude. Like in the common medical practice, a combination with µCT for structural imaging would be advantageous for improving the spatial significance. In the past we demonstrated the feasibility of the method, applying in-house developed and medical PET-scanners. The installation of the PET scanner in our controlled area makes possible long-term experiments and the application of non-standard and long-living PET-nuclides (like 124I, decay time 4 days, and 58Co, decay time 71 days). The installation of a new cyclotron will also extend the availability of short-living PET-isotopes for fast process observations (e.g. 11C, decay time 20 min). Application of these nuclides extends the common radiopharmaceutical practice of labelling organic compounds with 18F (decay time 110 min). The density of geomaterials may cause more than 50% of Compton-scattered events, which degrade the image quality. The quantification of the resulting artefacts is under way by Mont-Carlo model-based tools and will significantly empower the scatter-correction procedures. Transport observation studies have been conducted on consolidated, partially fractured rocks and on soils, often showing more or less localized transport pathways. The effective volume and transport network, as well as the effective surface area for specific process conditions can be derived directly and quantitatively from the spatio-temporal tracer distribution in the sample, which can be visualized as 3D-movie, in contrast to indirect observations based on break-through curves at the endpoint of the sample. Comparisons with Lattice-Boltzmann simulations based on structural information obtained by µCT are indicating, that any actual transport field very much depends on elusive boundary conditions and therewith represents just one instance of a variety of possible spatio-temporal distributions with similar effective parameters.

Positron-Emission-Tomography (PET) enables direct and quantitative monitoring of the spatio-temporal distributions of dissolved inert and/or reactive PET-nuclides and PET-nuclide-labeled colloids during their passage through decimeter-scaled material samples. We apply PET exclusively to geomaterial samples and reach the physical resolution limit of about 1 mm with our small-animal-PET scanner (ClearPET, Raytest). We suggest our GeoPET has unrivalled sensitivity and selectivity for our tracer concentrations to some 107 tracer atoms/µl and thus is ideally suited for direct flow and transport process observations in geomaterials. This lower limit of the tracer concentration in the order of about 1 kBq/µl outranges other process observation methods (e.g. NMR or resistivity tomography) by many orders of magnitude. Like in the common medical practice, a combination with µCT for structural imaging would be advantageous for improving the spatial significance. In the past we demonstrated the feasibility of the method, applying in-house developed and medical PET-scanners (Richter et al., 2003, Gründig et al., 2007). The installation of ClearPET in our controlled area made possible long-term experiments, like diffusion of long-living PET-tracers (like 58Co, decay time 71 days) and flow observations over several days (with 124I-labelled compounds, decay time 4 days) (Kulenkampff et al. 2008, Wolf 2010). The installation of a new cyclotrone will extend the availability of short-living PET-isotopes for fast process observations (e.g. 11C, decay time 20 min). The density of geomaterials may cause more than 50% of Compton-scattered events, which degrade image quality mainly by inhomogeneous reconstruction artifacts. These artifacts caused explicitly by geomaterials (and only insignificantly in medical applications) are currently being addressed by model-based scatter-correction procedures. Application examples include the fluid flow visualization in saliniferous rock cores. Different types of flow regimes are identified: in rather dense and low porous rock samples we observe slowly propagating diffuse “clouds of tracer”, whereas in fractured halite and sandstone samples (Fig. 1) we identified networks of extremely localized pathways, each with locally varying propagation velocities (Fig. 2). Quantitative parameterization of pathway and velocity distributions is possible, but still pending. However, with GeoPET-process observations we are capable of evaluating the results of Lattice-Boltzmann simulations, based on structural information from voxel-wise segmented µCT of the same sample. With respect to localized pathways in the fracture system of the halite, the model yields comparable results. Nevertheless, even the high resolution of µCT may be insufficient to visualize the pore-space characteristics in tighter material, and more complex compositions may impair segmentation results from µCT-images. Then, stochastic models which are generated from other information, like pore-size distributions (with missing network information) and integral transport parameters (with missing distribution information), can still be validated by comparing model results with GeoPET-observations.

We apply positron emission tomography (PET) with a high-resolution "small-animal" PET-scanner (ClearPET by Raytest, Straubenhardt) for process observation in rocks. Without affecting its physico-chemical properties, the fluid is labelled with the PET-tracer, a positron-emitting isotope. The annihilation radiation from individual decaying tracer atoms is detected with high sensitivity, and tomographic reconstruction of the recorded events yields a quantitative 3D-image of the tracer concentration. Sequential tomograms during tracer injection are used for the spatiotemporal observation of transport. The raw data have to be corrected, prevalently with respect to background radiation (randoms) and Compton scattering, which is more significant than in common biomedical applications. Although these effects can be considered exactly in principle, we had to develop and apply simplified correction methods for performance reasons. Deficiencies of these correction algorithms generate some artefacts, that cause a lower limit of the tracer concentration in the order of 1 kBq/µl or about 107 atoms/µl, outranging other methods (e.g. nmr or resistivity tomography) by many orders. A number of injection experiments in different rocks have been conducted with PET-process-tomography. New 3D-visualizations of the process-tomograms in fractured rocks showed strongly localized and complex flow paths and some unexpected deviations from the fractures that are deducible from µCT-images. At least, the results demonstrate the large discrepancy between the µCT-derived volume and specific surface area and the hydraulic effective parameters, which also can be analyzed quantitatively with this method. Possibly, these discrepancies and the complexity of the process show the limits of parameter determination methods with model simulations based on structural pore-space models - as long as the simulations are not verified by experimental data.

Anhand der Bergbaufolgeschäden des ehemaligen Kali- und Steinsalzbergbau im Raum Staßfurt (Sachsen-Anhalt) wurden im Rahmen des Forschungsverbundvorhaben Dynamik abgesoffener oder gefluteter Salzbergwerke und ihres Deckgebirgsstockwerks Ursachen, Prozesse und Auswirkungen der Bergschäden exemplarisch und umfassend untersucht. Ein verbessertes Prozessverständnis, das auf kleinskaligen experimentellen Untersuchung der Strukturen und Prozesse im Labormaßstab und damit verbundenen Modellierungen beruht, soll zur Aufklärung des Geschehens und möglicher Folgen beitragen. Zu diesem Zweck wurde in enger Zusammenarbeit mit dem Teilvorhaben zur strukturbezogenen Prozessmodellierung eine Methode zur direkten räumlich aufgelösten und quantitativen Prozessbeobachtung entwickelt und angewendet. Hierfür wurde als bildgebendes Verfahren die Positronen-Emissions-Tomographie (PET) eingesetzt. Hierfür wird ein Teil des injizierten Fluids mit Spuren eines Radiotracers markiert. Die Tracerkonzentration kann zerstörungsfrei, mit höchstmöglicher Empfindlichkeit und einem geeignetem Auflösungsvermögen in Raum und Zeit abgebildet werden. Der Prozess wird durch den Tracer auf geringst mögliche Weise beeinflusst und es wird ein realistisches Bild der Tracerverteilung, bzw. des Fließverhaltens, mit einer räumlichen Auflösung von 1 mm und einer zeitlichen Auflösung von 60s erzeugt. Dieses Fließverhalten lässt sich mit räumlich heterogenen und prozessabhängigen Verteilungen von Parametern, wie effektiven Volumina, Permeabilitäten, Abstandsgeschwindigkeiten und Dispersionsraten beschreiben. Einem besseren Prozessverständnis dient insbesondere der Abgleich mit mit Lattice-Boltzmann-Simulationen der Fließprozesse, die auf hochauflösenden computertomographischen (µXCT) Messungen der internen Struktur der identischen Bohrkerne beruhen. Dieser Vergleich von PET-Messdaten mit Lattice-Boltzmann - Simulationsdaten, der zugleich eine Skalenübertragung um etwa drei Größenordnungen bedeutet, erfolgt mit geostatistischen Methoden. Nicht in jedem Fall war dieser Abgleich zwischen Simulation und Experiment möglich: Während in klüftig-porösen Materialien gewöhnlich mit beiden Methoden räumlich stark differenzierte präferentielle Fließwege im Kluftsystem gefunden wurden, konnte in eher mikrostrukturell ausgeprägten Materialien trotz messbarer Permeabilität gelegentlich kein verbundener Porenraum aus den CT-Bildern segmentiert und somit keine LBM-Simulation durchgeführt werden. In diesem Fall zeigte die PET-Untersuchung ein diffuses Ausbreitungsverhalten des Tracers.