Introduction:
For many years, researchers have employed various experimental
techniques to characterize hydrocarbon-bearing porous media. One of the
aims behind characterizing porous media is to be able to extract
hydrocarbons from tight gas and oil reservoirs more efficiently. In this
context, the word ‘tight’ is used to label media with pore size smaller
than 100 µm. Efficient hydrocarbon production is aided through the
understanding of fundamental transport mechanisms and the bridging among
elements of the pore space architecture, namely geometry, connectivity
and transport. In tight media, the latter includes not only
understanding the diverse pore space, but also possible pore-fracture
connectivity. Computed micro-tomography (µCT) and various nuclear
magnetic resonance (NMR) methods have been used for this
characterization. Magnetic resonance imaging (MRI) has been used for
characterization with relative success in recent years.
µCT is one of the most popular techniques used to characterize
geological samples. This technique is nondestructive and can resolve
features at the micro or even nano scale, making this method
advantageous for high resolution. However, µCT utilizes density
contrast, making it a complex process to discern pore space from solid
materials, which requires rigorous and elaborate modeling, image
processing and analysis [1].
Higher-resolution techniques incur in the cost of smaller sample volume.
In contrast, MRI offers direct imaging, namely direct spatial encoding,
not tomography or inversion protocols, of the three-dimensional (3D)
features of porous media. This is possible because MRI draws contrast by
encoding the NMR signals to specific nuclei, e.g. proton spins in
hydrogen-bearing molecules such as water and magnetic spatial gradients
[2]. When compared to µCT, MRI also offers the advantage of
characterizing a more encompassing image of the pore system by imaging a
comparatively larger sample size. The larger size allows resolving
multiscale features in a porous system. MRI and µCT can be complimentary
to each other through digital registration algorithms [1].
NMR methods are not new to geological media characterization, but NMR’s
evolving versatility offers significant advantages. For instance,
Time-Domain NMR (TD-NMR) data can be collected at low-field strength
(0.05 T) [1,3,4]. With this low field, the induced internal
gradients that compromise NMR signals at higher filed strength are
minimized. However, TD-NMR spectra are known to have poor
signal-to-noise ratios (S/N) and it can be challenging to detect signals
with very short (ns) transverse-relaxation times (T2)
[1,3]. Fluid-flow experiments are also absent in the literature
devoted to TD-NMR, because dynamic experiments and the measurement of
rheological quantities cannot be performed simultaneously with TD-NMR
systems [5]. Despite these drawbacks, TD-NMR can be complimentary to
MRI methods in that the former can determine volume of water within the
sample, T2 relaxation distributions, and diffusion
measurements on a relatively fast timescale. Single-point imaging (SPI)
can be performed at low magnetic field, but requires long acquisition
times and is often impractical especially for high resolution images
[3,6].
High-field NMR instruments, e.g. 7 T or 9.4 T, have been used for
imaging in various disciplines, including the geological and medical
fields. A commonly used pulse sequence is rapid acquisition with
relaxation enhancement with inversion recovery (RARE-Inv-Rec), which
uses flip angles of 90° and 180° for an echo recovery [7]. This
pulse sequence works efficiently with high proton density areas, such as
in those found in soft tissue or porous media with pores on the scale of
millimeters. Fluids trapped in these samples have long
T2 values (ms), allowing for a long repetition time of
the applied pulse sequences [8]. With small pores (spanning nm to
µm), which are commonly found in bone or geological samples, the
T2 values are much shorter, (tens of µs), and the
nuclear spins under study become mostly relaxed before commonly used
pulse sequences are fully applied [8]. Other pulse sequences must be
employed in tight porous media that allow for efficient FID acquisition,
along with short repetition times.
Zero Echo Time (ZTE) is a pulse sequence that has been used for image
collection of tight porous media. This pulse sequence uses a smaller
excitation flip angle (5°), constant gradients and radial encoding
techniques [1,3,4,9,10,11]. When compared to the more common pulse
sequences that use 90°-flip angles, the 5°-flip angle used in ZTE allows
for a much shorter pulse time, leading to less relaxation of the
magnetization before the receiver is switched on. Another aspect of the
traditional imaging pulse sequences that causes them to take longer for
full application is the gradient switching process. Additionally,
gradient switching can be very taxing on the instrument hardware and
cause excessive heating. To avoid gradient switching, ZTE uses
nonselective radio frequency (RF) excitation while the gradients are
already acting [12,13]. The strength of the gradient gradually
increases over time, so that spatial encoding is still possible, but the
encoding happens with zero delay. The ramping gradients also reduce eddy
current effects within the magnet [9]. This technique has been used
in multiple fields such as imaging tissues with T2 value
on the µs scale in animal bones [2] and human teeth [10,11]. The
ZTE pulse sequence has also been applied to image porous media such as
dolomite, reef reservoirs, and sandstone [1,3,4,11].
Subsequent to the collection of images of a porous medium, pore-cluster
analysis (PCA) is frequently used to evaluate properties in geological
samples, such as porosity, surface-to-volume ratio, and mass transport
such as single-phase (groundwater) or multiphase flow (oil/gas)
[15-17]. Pore clusters are defined by connected pore voxels that are
in full contact with each other. The criterion commonly referred in
connectivity analysis is that vertex and edges are not considered true
connections [15]. Pore-cluster analysis allows the examination of
pore connectivity through the 3D space and can be used in simulation
exercises such as random walks [17].
The primary goal of the work described herein is to test the ability of
ZTE, combined with PCA, to resolve both pores and fractures in rock
cores with a variety of pore distributions [1,3,4,14], such as
sandstone and carbonate lithologies. We initially use a porous medium
consisting of 4-mm glass beads submerged in water to test the efficiency
and accuracy of a PCA protocol implemented through an in-house developed
Wolfram Mathematica 12.0 notebook. We provide evidence that when ZTE
combined with PCA is applied to tight porous media, connectivity can be
determined, and analyzed producing efficient characterization of tight
porous media. Resolution limits are shown to affect the results of this
analysis.