 |
| Titel |
Geostatistical analysis of 3D microCT images of porous media for stochastic upscaling of spatially variable reactive surfaces |
| VerfasserIn |
Marco De Lucia, Michael Kühn |
| Konferenz |
EGU General Assembly 2015
|
| Medientyp |
Artikel
|
| Sprache |
Englisch
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 17 (2015) |
| Datensatznummer |
250113371
|
| Publikation (Nr.) |
 EGU/EGU2015-13565.pdf |
|
|
|
|
|
| Zusammenfassung |
| The 3D imaging of porous media through micro tomography allows the characterization of
porous space and mineral abundances with unprecedented resolution. Such images can be
used to perform computational determination of permeability and to obtain a realistic
measure of the mineral surfaces exposed to fluid flow and thus to chemical interactions.
However, the volume of the plugs that can be analysed with such detail is in the order of 1
cm3, so that their representativity at a larger scale, i.e. as needed for reactive transport
modelling at Darcy scale, is questionable at best. In fact, the fine scale heterogeneity (from
plug to plug at few cm distance within the same core) would originate substantially different
readings of the investigated properties. Therefore, a comprehensive approach including the
spatial variability and heterogeneity at the micro- and plug scale needs to be adopted to gain
full advantage from the high resolution images in view of the upscaling to Darcy
scale.
In the framework of the collaborative project H2STORE, micro-CT imaging of different
core samples from potential H2-storage sites has been performed by partners at TU Clausthal
and Jena University before and after treatment with H2/CO2 mixtures in pressurized
autoclaves.
We present here the workflow which has been implemented to extract the relevant
features from the available data concerning the heterogeneity of the medium at the
microscopic and plug scale and to correlate the observed chemical reactions and changes in
the porous structure with the geometrical features of the medium.
First, a multivariate indicator-based geostatistical model for the microscopic structure of
the plugs has been built and fitted to the available images. This involved the implementation
of exploratory analysis algorithms such as experimental indicator variograms and
cross-variograms. The implemented methods are able to efficiently deal with images
in the order of 10003 voxels making use of parallelization. Sequential Indicator
Simulations are then employed to generate equi-probable realizations of microscopic
structures with varying mineral proportions and porosity but constrained to the spatial
variability observed in the plugs. The statistics computed on the ensemble of realizations
(essentially the distribution of mineral reactive surfaces exposed to porous space) is
integrated at a larger, Darcy scale. In a further step, the analysis of the microscopic
changes in the plugs after exposure to reactive solution establishes the correlations
betweens amount of chemical reactions and changes in the spatial models, thus
deriving some effective correlations which can be injected into the reactive transport
modelling.
In this contribution, we demonstrate the implemented workflow on a series of
images obtained from plugs from a german depleted gas field exposed to H2 and
CO2-charged brines. The geostatistical evaluation of microscale variability of the porous
media contributes to the upscaling of relevant variables and helps estimating - if
not reducing - the uncertainty due to the heterogeneity across scales of the natural
systems. |
|
|
|
|
|
|