Abstract
In the past two decades there have been several attempts to compute relative permeability
from high resolution, three-dimensional, X-ray microtomography (micro-CT) images of
the microstructure of a natural porous rock. In these attempts, researchers simulated fluid
flow directly on the imaged 3D pore space to compute relative permeabilities. They then
used laboratory measurements to validate the predictions. Analysis of these works shows
a number of shortcomings in those validations. For example: (i) there has been very
limited direct comparison between the imaged rock and the rock used in the laboratory
tests, i.e. researchers preferred to use the literature data for validation mostly, (iii) there
has been image resolution issues that limited prediction accuracy, (iv) there has been
limited attempt to use high resolution images of multiple fluids in place and (v) the
Lattice-Boltzmann method has instability issues to predict relative permeability at low
phase saturations.
The purpose of this thesis is to test the predictive value of image-based numerical
computations for two-phase, drainage relative permeability using well-defined laboratory
measurements. The experimental data represents a steady-state flow of oil and water in
strongly water-wet, homogeneous outcrop sandstone (Bentheimer) and covers a full
saturation range of both phases. This data is obtained using a standard core sample. Next,
a small sister plug is imaged by micro-CT and the steady-state experiments are repeated
on this plug for three different saturation distributions. These three saturation
distributions are imaged and compared with simulated fluid distributions on the dry
image (using the capillary drainage transform CDT method). The comparison shows
that CDT-based saturation distributions agree with the actual imaged saturation
distributions. Finally, relative permeability computations are made over the CDT-based
saturation distributions. The issues experienced in the previous studies such as the image
resolution and computational capacity are minimized in this study through using an image
of higher resolution and a larger subset.
The thesis demonstrates a good agreement between the image-based computations
made using the CDT method and the laboratory data. The requirements for a successful prediction using the CDT method are strong wetting conditions and capillary-dominated
flow. In order to ensure these conditions, the laboratory tests described in this study
employ the plasma technique for cleaning the core plug and use appropriate flow rates for
controlling the capillary number. The agreement also confirms that steady-state
experimental data is representative for testing image-based predictions.
In this thesis, an attempt is made to use high-resolution micro-CT images of
multiphase distributions in relative permeability computations. It is found that relative
permeabilities are underestimated. This is attributed to snap-off that occurs when the
steady-state experiment is stopped for micro-CT imaging and causes the non-wetting
phase to be disconnected. As a result, the thesis recommends that both steady-state tests
and micro-CT images should be carried out at dynamic conditions for an accurate
validation of image-based methods.