You are on page 1of 15

SPE 160870

Correlations Between NMR Relaxation Response and Relative Permeability


From Tomographic Reservoir Rock Images
Tariq M AlGhamdi, C. H. Arns, SPE, University of New South Wales, and R. Y. Eyvazzadeh, SPE, Saudi Aramco

Copyright 2012, Society of Petroleum Engineers

This paper was prepared for presentation at the SPE Saudi Arabia Section Technical Symposium and Exhibition held in Al-Khobar, Saudi Arabia, 8–11 April 2012.

This paper was selected for presentation by an SPE program committee following review of information contained in an abstract submitted by the author(s). Contents of the paper have not been
reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material, as presented, does not necessarily reflect any position of the Society of Petroleum
Engineers, its officers, or members. Papers presented at the SPE meetings are subject to publication review by Editorial Committee of Society of Petroleum Engineers. Electronic reproduction,
distribution, or storage of any part of this paper without the written consent of the Society of Petroleum Engineers is prohibited. Permission to reproduce in print is restricted to an abstract of not
more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of where and whom the paper was presented. Write Librarian, SPE, P.O. Box
833836, Richardson, TX 75083-3836, U.S.A., fax 01-972-952-9435.

Abstract

NMR is typically used in the petroleum industry to characterize pore size and identify fluids in fully and partially saturated
reservoir samples. While the NMR relaxation response can be used to estimate the permeability of the rock, it may also
provide information about fluid distribution for multi-phase systems which would lead to the estimation of effective
permeability of fluids at partial saturations and derivation of relative permeability to assess hydrocarbon recovery. Using a
random walk method, we simulate the NMR response as function of saturation on tomographic images of Bentheimer and
Berea sandstone as well as a Ferroan Dolomite samples. Fluid distributions are simulated using the capillary pressure curves of
these samples via capillary drainage transform (CDT) allowing the calculations of the saturations directly on the images. The
magnetic susceptibility of minerals and fluids is used to calculate the internal magnetic fields from the material distributions of
solids and fluids. We show that the logarithmic mean of the NMR T2 distribution is a robust measure of permeability and
results in strong correlations between NMR response and relative permeability of both fluids. The observed relative
permeability from NMR in our work is in excellent agreement with relative permeability calculations on direct image based
using the lattice Boltzmann method (LBM). We have also compared our NMR results for the wetting phase to published
experimental results on Bentheimer and Berea sandstone samples and observed strong agreement. Using NMR numerical
calculations, we demonstrate that internal gradients aid the establishment of relative permeability correlations for the non-
wetting phase.

Introduction

Nuclear magnetic resonance (NMR) is increasingly used to estimate pore size distribution and to identify fluids [Kleinberg
(1996); Strange (1996); Akkurt (1998) and Sun (2004)]. One of the main advantages of NMR is its ability to provide an
estimate of permeability [Curwen (1995); Dunn (1999) and Hidajat (2002)]. An essential tool in reservoir description is the
estimation of permeability from NMR responses by empirical formulae. The NMR T2 relaxation time response for CPMG
sequence is a function of parameters such as saturating fluid bulk relaxation time T2b, surface relaxivity ρ, surface area to pore
volume S/V, diffusion coefficient D, the gyro-magnetic ratio of the proton γ, the local magnetic field "G" and the CPMG echo-
spacing time tE [Bloembergen (1948)]. For a single pore and fast diffusion the NMR T2 relaxation is described as the
following [Straley (1994) and Akkurt (1995)]:

1 1 S D
=( ) + (ρ ) + (g γG t E ) 2 ………………................………………………………………....……. (1)
T2 T2b V 12

In the weak coupling regime the process of NMR relaxation will generate a magnetization decay as a function of relaxation
times of the individual pores. The magnetization decay resulting from spin-spin relaxation can then be expressed as
−t
M ( t ) = ∑ (a i ) M 0 exp( ) ……………………………...........….. . . . . . . . . . . . . . . . . . . . ........... . . ..... . . . .. . . . . (2)
T2
2 SPE 160870

Correlations between NMR responses and permeability were first established by Seevers [Seevers (1966)] using T1 relaxation
times, both for laboratory measurements and in the borehole. NMR T2 correlations with permeability have been introduced by
Kenyon [Kenyon (1988)] and superseded spin-lattice relaxation measurements because transverse relaxation responses gave
better results for sandstones and can be measured much faster. A wide range of relationships between NMR spin-lattice
relaxation responses and physical properties including permeability are given by [Sen (1990); Banavar and Schwartz (1987);
Dunn (1999) and Arns (2005)]. Kenyon's NMR permeability model using transverse relaxation time T2 became much
recognized [Straley (1994) and Westphal (2005)] even for carbonates due to the advantage of faster acquisition to obtain
relaxation measurements.

Jayanth Banavar and Lawrence Schwartz [Schwartz (1987)] were the first to measure NMR relaxation responses at partial
saturations but suggested that industry should perform NMR measurements in mixtures of oil and water to match the reality of
hydrocarbon reservoirs, which frequently are partially saturated. Their partial saturation measurements depicted the behavior
of ratios of spin-lattice relaxation times at multiple saturations to the NMR response at 100% saturations. Straley [Straley
(1991)] was among the first to address partial saturations directly. Using a 10 MHz NMR instrument, he performed drainage
centrifuge experiments on clay-rich low permeability sandstone samples by air and kerosene so that only the water phase
contributes to the NMR spin-lattice relaxation. He noted during water drainage that the long peak of T1 disappears as
saturation decreases and the short T1 remains the same. These observations led to the derivation of free fluid index by NMR in
agreement centrifuge experiments.

More recently, NMR responses of partially saturated rocks have been used to give information about fluid flow [Chen (1993);
Tessier (1998), and Toumelin (2003)]. Chen [Chen (1993)] conducted drainage experiments on Bentheimer sandstone using
nitrogen to displace water from the sample and measure NMR spin-lattice relaxation (T1) at different saturations. He
introduced a power law model to describe the decrease in T1 relaxation time with decreasing water saturations. While not
directly deriving a relative permeability curve for the wetting phase, he suggested that this is feasible.

In 1994, Chen [Chen (1994)] reported the wetting phase relative permeability from water-nitrogen experiments at a magnetic
field strength of 2 Tesla and room temperature conditions on Bentheimer and Berea sandstone in addition to a limestone
carbonate reservoir sample. From the experiments, profiles of spin-lattice relaxation time at multiple saturations were
generated. The relationship by Katz and Thompson relating T1 to morphological length scales [Katz and Thompson (1987)]
was used in combination with Archie’s law [Archie (1942)] to derive the relative permeability for the water phase from NMR
T1 measurements.

Further progress in the assessment of dynamic flow through reservoir cores was made with NMR saturation imaging
[Rothwell (1985), Chen (1992) and Kulkarni (1998)]. Kulkarni used NMR imaging to map the saturation profiles of oil and
water for a limestone carbonate sample. This led to an estimate of two-phase flow functions like relative permeability and
capillary pressure. The experiment showed the possibility of deriving relative permeability via NMR measurements, but the
experiment was limited to water saturations above 40 percent.

Xue [Xue (2004)] used a 50 MHz NMR imaging setup to determine fluid saturation and used Chen's model [Chen (1994)] to
derive the wetting phase relative permeability on a sandstone sample. Spin-lattice NMR T1 measurements in full and partial
saturations were gathered by displacing water by nitrogen from the sample and resolving T1 spectra locally. She concluded
that her analysis is not validated and might be subject to flow stability conditions governing relative permeability. Ioannidis
[Ioannidis (2006)] repeated the NMR experiments of Chen and Xue on glass beads for transverse relaxation and predicted
relative permeability successfully using high field at 500 MHz frequency. He modified Chen’s model by replacing spin-lattice
time T1 with transverse relation time T2 in the permeability correlations.

The interpretation of NMR relaxation responses is complicated by the nature of the Laplace inversion needed. Simulations of
NMR responses assist these interpretations and can provide additional insights. The most accurate NMR response simulations
are carried out on high resolution CT images, mainly utilizing random walk techniques [Kim (2000); Valckenborg (2002);
and Arns (2011)]. The work presented by [Arns (2011); Toumelin (2003); and Talabi (2008)] highlighted advancement in
NMR numerical modeling for porous media on high resolution tomographic images and 3D pore network models at multiple
saturations. Earlier simulation methods did not account for internal gradients, which would lead to erroneous results if iron
minerals are present [LaTorraca (1995) and Keating (2010)]. Newer simulation techniques include internal gradient effects by
explicitly calculating the internal field distributions, e.g. [Valckenborg (2002); Arns (2011); and Chen (2011)].
SPE 160870 3

[a] [b] [c]

Fig. 1-Segmented tomographic images of [a] Bentheimer sandstone [b] Berea sandstone and [c] Ferroan-Dolomite.

In this work, we use digital images of a set of samples (Bentheimer, Berea and Ferroan Dolomite) and calculate the
petrophysical properties porosity, permeability, and relative permeability directly on the tomographic images. NMR relaxation
responses are simulated at low field (2 MHz) as a function of saturation and internal gradients accounted for. First, NMR
relaxation responses for both fluid phases at fully saturation states of each phase (100% water and 100% oil) are simulated to
calculate absolute permeability of both phases. Then the NMR response of each phase at different saturations is simulated.
Following the extraction of spin-spin relaxation times behavior with saturation, we adopt and modify the Schlumberger-Doll
research (SDR) NMR permeability equation to accommodate the relation of NMR T2 relaxation response as a function of
saturation. Excellent results are observed matching NMR relative permeability to image based (LBM) relative permeability in
particular the non-wetting phase. Finally, we compare our wetting phase relative permeability results for Bentheimer and
Berea sandstone with published results.

Methodology

Sample characterization. In this work, we consider three samples: Berea and Bentheimer sandstone and a Ferroan Dolomite
from the Middle East. XRD results on these samples revealed that Berea comprises of 86% quartz and a mixture of different
clay minerals (6% Kaolinite, 3.5% Ankerite, 2.9% Illite and 1% feldspar). The Bentheimer sandstone exhibits 2-3% of
Kaolinite clay mineral. The Ferroan Dolomite contains 90% dolomite and around 10% Ankerite; this type of dolomite is called
Ferroan due to the existence of iron-bearing minerals. All three samples display a relatively homogeneous micro-structure with
high porosity and permeability. The samples (Fig 1) were imaged utilizing a high resolution X-ray computed tomography
facility at a field of view of 20483 voxels [Sakellariou (2007)] with an image resolution of around 3 microns.

Image processing and analysis. The sample processing of high resolution tomographic images is a critical step [Sheppard
(2004)] for the accurate definition of pore and solid volumes respectively. Raw tomograms exhibit a wide range of different
noise types and image artifacts and a set of sophisticated filters is utilized to accurately quantify the solid, pore and
intermediate phases. The final process is the segmentation of the sample into two or three phases by defining the volumes of
pore, solid and clay or micro-porosity from intermediate phase. Sample dimensions and characteristics are given in Table 1.
The reported porosity corresponds to the porosity derived from the tomographic images after segmentation, assuming that the
clay regions have a porosity of 50%.

TABLE 1- SUMMARY OF PETROPHYSICAL PROPERTIES OF THE SAMPELS

Sample Size (voxels) Resolution (micron) Porosity Permeability mD


Bentheimer 800 2.9 0.23 2777
Berea 1080 2.83 0.19 523

Ferroan
1020 2.18 0.18 642
Dolomite

Numerical simulation

Image based fluid saturations. The fluid distributions in the samples are numerically simulated directly on the voxelated
tomographic images via a capillary drainage transform [Hilpert (2001)]. The simulation performs a fixed capillary pressure
4 SPE 160870

[a] [b] [c]

[d] [e] [f]

[g] [h] [i]

Fig. 2-Simulated drainage and imbibition using capillary drainage mechanism. [a-c] Bentheimer sandstone ([a]100% [b] 50% and [c]
25% Sw), [d-f] Berea sandstone ([d] 100% [e] 80% and [f] 44% Sw), [g-i Ferroan-Dolomite Carbonate ( [g] 100%, [h] 75% and [i] 25%
Sw), (white is the invading non-wetting phase, black is the defending wetting-phase, gray is the solid-phase, red is the clay region).

corresponding to a pore entry, mimicking the standard mercury intrusion experiment. Partitioning fluids utilizing numerical
analysis provides information that is not accessible by conventional core analysis. (Fig 2) shows different saturations profiles
simulated on the tomographic images for the three reservoir samples used in this work.

Permeability and relative permeability

Absolute and relative permeability are calculated using the lattice Boltzmann method [Gunstensen (1991) and Shan (1993)],
which is well suited to deal with multiphase flow in complex geometries. Using the static fluid distribution with no slip
boundary conditions, permeability for each phase is calculated at different saturations. This allows relative permeability for
each phase to be calculated according to

K w, eff
K rw = . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ….... . (3)
K abs
SPE 160870 5

K o , eff
K ro = . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ……. . (4)
K abs

This direct image based calculation of relative permeability using numerically derived fluid distributions was validated
experimentally on Bentheimer sandstone [Hussain (2011)]. The absolute permeability values for the samples are given in
Table 1 and agree well with experimental data for Bentheimer and Berea available in the literature.

NMR Simulation

Surface relaxation: The spin relaxation time of a saturated porous system is numerically calculated using a lattice random
walk method [Mendelson (1990) and Bergman (1995)]. Initially the walkers are placed randomly in the 3D pore space. At
each time step i the walkers are moved from their initial position to a neighboring site and the clock of the walker advanced by
τi= ε2/(6D0), where D0 is the bulk diffusion constant of the relevant fluid, reflecting Brownian dynamics and ε is a small
fraction of the voxel size. We treat each random walk as the movement of a spin packet with initial strength Mw (t = 0). At
each time step (i) of length (τi), the strength of the walker is reduced by the survival probability Si, with the strength of the
walker at time t = ∑i τi given by

M w ( t ) = M w (0 )∏ Si . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . ............ . . .... . . . . . . (5)


i
Here Si = SbSs, where Sb = exp (−ti/Tb) for bulk relaxation and Ss = 1−ν for surface relaxation. For steps within the same fluid
Ss = 1. The killing probability ν is related to the surface relaxivity ρ via
ρε ρε
Aν = ( ) + O ( ) 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . .... . . . . . . . . . . . . . . ...... . . . . .. . (6)
D0 D
Here A is a correction factor of order 1 (we take A = 3/2) accounting for the details of the random walk implementation
[Mendelson (1990); Bergman (1995)]. This leads to
6ρτi
SS = 1 − . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..... . . . . . . (7)
εA
Dephasing due to internal gradients: To capture the dephasing of random walkers caused by internal fields, we model the
phase accumulation of the random walkers for the CPMG sequence explicitly. For this, one needs to derive the local magnetic
field strength. For the cases considered here the internal field, which results from susceptibility contrast between minerals and
fluids, is accurately described by a dipole approximation, since the magnetic susceptibilities of all components are small
compared to one. The internal magnetic field is numerically calculated on the tomographic image by assigning an effective
magnetic susceptibility to each voxel and convoluting the dipole field around the susceptibility field [Arns. (2011)].We
calculate the dipole field when the distance from the dipole center (r) is larger than the radius of the dipole (a) is given as
μ 0 ⎛⎜ 3(mr)r − r 3 ⎞⎟
B dipole =
4π ⎜ 5 ⎟ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . (8)
⎝ r ⎠
For the dipole field inside the sphere when r is less than a
2
B dipole = μ 0 m ….. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (9)
3
Here μ0 is the magnetic permeability of the vacuum and m is the dipole magnetic field for a unit volume of the lattice spacing
(resolution of tomographic image). When we apply an external magnetic field to (H) to the sample, the susceptibility field will
result in an induced internal magnetic field response (Binternal ). We use this to model the internal gradient distribution on high
resolution tomographic images at different fluid saturations for a given applied static magnetic field that is oriented in the z-
direction. The phase evolution of a spin with reference to the Lamor frequency at the starting position ω0 = g Bz(0) is given by

[ ]
N j
φ D = φ − φ 0 = ∑ γτ i B z ( t j ) − B z ( 0 ) , t j = ∑ τ i . . . . . . . . . . . . . . . ................... . .. . . . . . . . . (10)
j=1 i =1

The total magnetization decay including dephasing for an individual spin is then given by Mw (t) = Mw (t) * cos (φd) and
6 SPE 160870

recorded for the echo positions (maximal coherence) of the CPMG sequence. An element-wise sum over the magnetization
decays of the individual walkers finally results in the total magnetization decay.

Modeling parameters: The important material parameters for NMR modeling are surface relaxation time and the
susceptibilities of the individual components. For surface relaxivities we use the values published by [Talabi (2008)]. We use
as susceptibilities of the minerals and fluids the results of a recent literature survey [Potter (2008)]. The values for the different
parameters are reported in Tables 2-4. To derive the volume susceptibility of the minerals and in particular the intermediate
clay phase, we carried out XRD analysis for all three sample. Since most reservoir fluids are diamagnetic [Potter 2004], we
used as analogues for the simulation water and dodecane. We obtained further fluid properties (see Table-3) hydrogen Index
(HI) and diffusion coefficients and bulk fluid properties from the Halliburton NMR catalog [Coates (1999)].

For our simulations, we define the lattice spacing as being a fraction of the resolution of our sample. The resolution is about 3
microns and we typically fine-grained the system by a factor of 10 to achieve a good time resolution of the CPMG pulse
sequence. We use between 60000-100000 walkers per each simulation. The NMR T2 pore size distribution can be obtained by
inverting the magnetization decay profile [Arns (2005)].

NMR Permeability correlations

NMR is a practical petrophysical tool especially in determining permeability and has been widely used [Timur (1969); Sen
(1990); and Prammer (1994)]. Expanding the approach to partial saturations enables the calculation of effective permeability
for the relevant fluids. The Schlumberger-Doll-Research (SDR) NMR permeability correlation found by Kenyon [Kenyon
(1988)] is described as

K NMR = a ∗ T 2 ∗ φ 4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . ..... . . . . . . . (11)


2lm
We modify the NMR permeability equation of Schlumberger-Doll-Research (SDR) to account for partial saturated samples by
calculating the logarithmic mean T2lm from NMR T2 responses at different saturations. The modified version of the SDR
equation becomes

K NMR (S w ) = a ∗ T 2 (Sw ) ∗ φ 2 4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . … . . . . . . . . . (12)


2lm
Here the porosity in (Eq. 11) denotes the fractional porosity of the effective phases (water/oil) and T2lm is a calculated as
function of saturation. The final version of modified (SDR) for relative permeability is then described
2
⎛T (S ) ⎞
K r, NMR = ⎜⎜ 2lm w ⎟⎟ ∗ (S w )4 … . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ……... . . . ….. . .. . . . . . (13)
⎝ T2lm ⎠
T2lm is calculated as the logarithmic mean of the transverse relaxation time T2 and the volume fraction of pores at different
sizes
∑ (a i * log(T2i ))
T2lm = exp . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ............ . . . . . . . . . . . . . . . . . . . . . . (14)
∑ ai
Results and Analysis

NMR response of samples saturated with single fluid. We first simulate the response on Bentheimer, Berea and Ferroan-
Dolomite samples fully saturated with water or oil respectively. The response of the water phase includes the NMR response
of the clay regions, which are considered 50% water saturated. Consequently, the NMR response of water includes a fast
relaxing component in the clay region and the region around it, which can exchange. The NMR water response is dominated
by surface relaxation, while the oil phase response is mainly due to bulk relaxation [Akkurt (2005)]. (Fig 3) illustrates the
NMR responses of the saturated rocks. Bentheimer sandstone exhibits only a small amount of clay and at the same time shows
a weaker susceptibility contrast. Consequently, the short relaxation time peak attributed to water saturated clays is weak. For
Berea sandstone both susceptibility contrast and clay fraction are higher, leading to a much more noticeable clay peak. For
Ferroan dolomite the short relaxation time peak is caused by a combination of internal gradients and increased surface
relaxivity, making the peak position dependant on echo spacing.

NMR simulation response at partial saturations. NMR responses for partially saturated samples are derived using the
saturation methods (capillary drainage) presented before. For water and oil phases, the simulation parameters are presented in
Tables 2, 3, and 4. The wetting phase (water) exhibits a non-zero surface relaxivity with the solid. However, for the non-
wetting fluid there are neither interactions with the solid surface nor the wetting fluid. Thus, the non-wetting fluid can only
SPE 160870 7

relax by internal gradient effects or bulk relaxation. We report the NMR response of both phases at different saturations in
Figure 4. For the water phase starting at fully water saturated state, the relaxation time decreases as drainage takes place until
the samples are fully drained. This is due to large pores being drained first then followed by the next larger. The decrease in
relaxation time is governed by the surface area to pore volume ratio that water phase signal is dependent on.

Diffusional coupling effects between clay regions and resolved pore space were examined in all three samples. As samples
desaturate, the clay regions stay coupled with the free water, since diffusion coupling effects increase. The shift to smaller
relaxation times is explained by the increase surface to volume ratio. As a result of this diffusional coupling, larger amplitudes
of short transverse relaxation times were visible at lower water saturations. The diffusional coupling effects varied here, as
both Berea and Ferroan Dolomite exhibited larger peaks than Bentheimer and this could be due to more paramagnetic clays
being present compared to Bentheimer which contains diamagnetic Kaolinite. These remarks are very crucial in the
consideration of cut-off analysis defining the bound fluid index (BVI) and thus the accuracy of calculating the irreducible
water saturation. Considering a cut-off value leads to more accurate permeability estimation in the free fluid region.

Consider now the NMR relaxation response of the oil phase. The Bentheimer sandstone illustrated marginal increase in
relaxation time as oil saturation decreases. In the Berea and Ferroan Dolomite samples, incremental increase and wider
separation in relaxation time of oil phase is seen as oil saturation decreases. In addition to bulk fluid relaxation in the oil phase,
internal field gradients play a significant role in the relaxation of the oil signal. The magnitude of the internal fields based
relaxation was observed to be weaker in Bentheimer and stronger in Berea and Ferroan Dolomite. The increase in relaxation
time for Berea and the Ferroan dolomite with decreasing saturation can be explained by loosing access to smaller pores and
crevices, where internal gradients are stronger. For both phases, we calculated the logarithmic mean of relaxation time values
(Eq. 14) as function of saturation in all the samples and generated the profiles for each phase (Fig 5).

Absolute and relative permeability from NMR relaxation measurements. Both phases’ absolute permeabilities are
calculated using (Eq. 11). To account for the influence of clay on the absolute permeability of the water phase, we apply cutoff
analysis on the logarithmic average mean of relaxation time T2lm the same way as one would typically for experimental data.
The cut-off value used in Bentheimer and Berea sandstone samples is 33 msec, which is shorter than the cut off value used for
the Ferroan Dolomite sample 67 msec. Relative permeability correlations from numerical NMR relaxation responses are
calculated from the effective NMR permeability of each fluid at the different saturations from the NMR response at partial
saturations from the profiles of logarithmic mean of relaxation times of each phase (Fig 5) using (Eq. 12). Relative
permeability is calculated via (Eq. 13) for both water and oil phases. Strong correlations are observed in all three samples (Fig
6) between the relaxation responses of both phases against the relative permeability calculated directly on the tomographic
images of samples by lattice-Boltzmann. This suggests that internal gradients played a major role in establishing the
correlations for both phases but powerfully the non-wetting phase that was modeled with zero surface relaxation. We validate
the simulation results on Berea and Bentheimer sandstone samples with those measured experimentally by Songhua Chen
[Chen et al. (1994)]. Our results are in excellent agreement to his results and observed relative permeability of the wetting
phase (Fig 7). This shows that numerical simulations can be accurate in modeling fluid flow dynamics by incorporating all
important elements including the presence of internal magnetic fields.

Conclusions

A random walk method is used to simulate NMR response as function of saturation on Berea, Bentheimer and Ferroan
Dolomite reservoir samples. We perform XRD analysis on all the samples to calculate magnetic susceptibility using the
reported literature values of magnetic response to minerals and fluids. Careful modeling of susceptibility of minerals forming
the sample including the clay is essential as internal gradients from susceptibility contrast between fluid and surrounding solid
interfaces enhances the NMR spin-spin relaxation measurements for both wetting and non-wetting phases. Strong correlations
between relative permeability and NMR spin-spin relaxation response are observed on all three samples. The analysis of
diffusional coupling effects of clay micro-pores should be considered as it would distinguish bound fluid from free fluid
regions. As a result, use of cut-off analysis to calibrate the logarithmic mean value leads to more accurate permeability
estimates which results in enhancements in relative permeability correlations for the wetting phase. As oil phase does not relax
via surface relaxation, internal gradients from susceptibility contrast assisted in establishing a surface related/weighted
relaxation mechanism. Our results for Bentheimer and Berea sandstone are in excellent agreement also to previously published
work by Songhua Chen from the relation of NMR spin-lattice and saturation for the water phase.

The work presented here comprise of a systematic approach that shall adopt NMR experimental procedure. Mineralogy here
plays a major role in the simulation of surface relaxation and mainly the volume susceptibility of the samples at which the
magnitude of internal gradients shall be assessed. The result from this work might provide hints for estimating dynamic flow
relative permeability much faster and with non-invasive methods like nuclear magnetic resonance.

This work demonstrates that NMR is capable of estimating accurately relative permeability in partial saturated samples.
8 SPE 160870

Relative permeability experiments are very expensive and time consuming. It would be highly desirable to extend this study to
more complex rock, accompanied by experimental measurements.

Acknowledgments

The authors acknowledge the University of New South Wales and Australian National university for their support through
providing the samples for this work and facilitating the computations through access to the National Computing Infrastructure
(NCI). TA acknowledges Saudi Aramco for their financial support through a PhD scholarship at the University of New South
Wales. CHA acknowledges the Australian Research Council for an Australian Research Fellowship (DP0881112).

References
Akkurt, R. et al.: “Enhance Diffusion: Expanding the range of NMR direct hydrocarbon-typing applications,” SPWLA 39th Annual
Logging Symposium, SPWLA (May 1998).
Akkurt, R.: “NMR logging of natural gas reservoirs,” SPWLA 36th Annual Logging Symposium, SPWLA (June 1995).
Akkurt, R.: Practical NMR Logging (October 2005).
Archie, G.E. “The Electrical resistivity log as an aid in determining some reservoir characteristics,” Trans. AIME (1942) 146, 54.
Arns, C.H. et al.: “Comparison of experimental NMR measurements with simulated responses on digitized images of mono-mineralic
rocks using Xray-CT,” Improved Core Analysis for Unconventional Fields, Society of Core Analysts, Noordwijk, number SCA2009-44
in The 23rd International Symposium of the Society of Core Analysts (September 27-30 2009) .
Arns, C.H. et al.: “Analysis of T2-D relaxation-diffusion NMR measurements for partially saturated media at different field strength,”
Society of Core Analysts, Halifax, number SCA2010-17 in The 24th International Symposium of the Society of Core Analysts (October
4-7 2010).
Arns, C.H., AlGhamdi, T. and Arns, J.Y.: “Numerical analysis of nuclear magnetic resonance relaxation-diffusion responses of
sedimentary rock,” New Journal of Physics (2011) 13, 015004.
Arns, C.H., Knackstedt, M.A. and Martys, N.: “Cross-property correlations and permeability estimation in sandstone,” Phys. Rev. E
(2005) 72, 046304.
Arns, C.H.: “An analysis of NMR-permeability scaling rules by numerical MRI,” 48th Annual Logging Symposium, Society of
Petrophysicists & Well Log Analyst, Austin, Texas (June 2007).
Banavar, J.R. and Schwartz, L.M.: “Magnetic Resonance as a Probe of Permeability in Porous Media,” Phys. Rev. Lett. (1987) 58,
1411.
Bergman, D.J. et al.: “Self-diffusion in a periodic porous medium: A comparison of different approaches,” Phys. Rev. E (1995) 51,
3393.
Bloembergen, N., Purcell, E.M. and Pound, R.V.: “Relaxation Effects in Nuclear Magnetic Resonance Absorption,” Phys. Rev. (1948)
7, 679.
Carr, H.Y. and Purcell, E.M.: “Effects of diffusion on free precession in nuclear magnetic resonance problems,” Phys. Rev. (1954) 94,
630.
Chen S, Kim KH, Qin F, Watson AT. “Quantitative NMR imaging of multiphase flow in porous media”, Magnetic . Resonance
Imaging (1992).
Chen, S., Hsie-Keng and Watson, A.T.: “Fluid saturation-dependent nuclear magnetic resonance spin-lattice relaxation in porous media
and pore structure analysis,” J. Appl. Phys. (1993) 74, 1473.
Chen, S., Liaw, H.K. and Watson, A.: “Measurements and analysis of fluid saturation-dependent NMR relaxation and linebroadening in
porous media,” Magnetic Resonance Imaging (1994) 12, 201.
Chen, S., Lilong Li, Gigi Zhang and Jason Chen, "Magnetic resonance for downhole complex-lithology earth formation evaluation",
New J. Phys. (2011) 13, 085015.
Coates, G.R., Xiao, L. and Prammer, M.G.: NMR Logging Principles and Applications, Halliburton Energy Services (1999).
Kim, I. C., Cule, I.C.K.D. and Torquato, S.: “Comment on "Walker diffusion method for calculation of transport properties of
composite materials",” Phys. Rev. E (2000) 61, 4659.
Curwen, D.W. and Molaro, C. “Permeability from Magnetic Resonance Imaging Logs,” PWLA 36th Annual Logging Symposium,
SPWLA (June 1995).
Dunn, K.J., LaTorraca, G.A. and Bergman, D.J.: “Permeability relation with other petrophysical parameters for periodic porous media,”
Geophysics (1999) 64, 470.
Gunstensen, A.K. and Rothman, D.H.: “Lattice Boltzmann model of immiscible fluids,” Phys. Rev. A (1991) 43, 4320.
Hahn, E.L.: “Spin echoes,” Phys. Rev. (1950) 80, 580.
Hidajat, I. et al.: “Permeability of porous media from simulated NMR response,” Transport in Porous Media (2002) 48, 225.
Hilpert, M. and Miller, C.T.: “Pore-morphology based simulation of drainage in totally wetting porous media,” Advances in Water
Resources (2001) 24, 243.
Hürlimann, M.D., Helmer, K.G. and Sotak, C.H.: “Dephasing of Hahn echo in rocks by diffusion in susceptibility-induced field
inhomogeneities,” Magnetic Resonance Imaging (1998) 16, 535.
Hussain, Furqan: “Experimental validation of image based drainage relative permeability”, Ph.D. thesis, University of New South
Wales (2011).
Ioannidis, M.A., Ghatzis, I. and an R. Perunarkilli, C.L.: “Unsaturated Hydraulic Conductivity from Nuclear Magnetic Resonance
Measurements,” Water Resources Research (2006) 42, 6 PP, W07201.
Katz, A.J. and Thompson, A.H.: “Prediction of rock electrical conductivity from mercury injection experiments,” J. Geophys. Res.
(1987) 92, 599.
SPE 160870 9

Keating, K. and Knight, R.: “A Laboratory study of the effect of Fe(II)-bearing minerals on nuclear magnetic resonance (NMR)
relaxation measurements,” Geophysics (2010) 75, F72.
Kenyon,W.E. et al.: “A three part study of NMR longitudinal relaxation properties of water saturated sandstones,” SPE formation
evaluation (1988) 3, 626.
Kleinberg, R.L.: “Utility of NMR T2 distributions, connection with capillary pressure, clay effect, and determination of the surface
relaxivity parameter r2,” Magnetic Resonance Imaging (1996) 14,761.
Knackstedt, M.A. et al.: “Digital Core Laboratory: Properties of reservoir core derived from 3D images,” Asia Pacific Conference on
Integrated Modelling for Asset Management, SPE 87009, Kuala Lumpur (March 2004).
Kulkarni, R., Watson, A.T. and Nordtvedt, J.E.: “Estimation of Porous Media Flow Functions using NMR Imaging Data,” Magnetic
Resonance Imaging (1998) 16, 707.
LaTorraca, G. and Dunn, K.J.: “Magnetic susceptibility contrast effect on NMR T2 Logging,” SPWLA 36th Annual Logging
Symposium, SPWLA (June 1995).
Mendelson, K.S.: “Percolation model of nuclear magnetic relaxation in porous media,” Phys. Rev. B (1990) 41, 562.
Potter, D.K., AlGhamdi, T.M. and Ivakhnenko, O.P.: “Sensitive carbonate reservoir rock characterization frommagnetic susceptibility:
mineral quantification, correlation with petrophysical properties, and anisotropy,” International Symposium of the Society of Core
Analysts, Society of Core Analysts, Abu Dhabi (2008).
Prammer, M.G.: “NMR Pore Size Distributions and Permeability at the Well Site,” Presented at the SPE 69th Annual Technical
Conference and Exhibition, Soc. Pet. Eng., New Orleans (1994) .
Rothwell, William P. and H. J. Vinegar, "Petrophysical applications of NMR imaging," Appl. Opt. 24, 3969-3972 (1985)
Rueslatten, H. et al.: “NMR Studies of an Iron-Rich Sandstone Oil Reservoir,” International Symposium of the Society of Core
Analysts, Society of Core Analysts, The Hague (1998).
Sakellariou A. , C. H. Arns, A. P. Sheppard, R. M. Sok, H. Averdunk, A. Limaye, A. C. Jones, T. J. Senden, and M. A. Knackstedt,
Materialstoday (2007) v10 no. 12, p44-51 .
Seevers, D..: “A Nuclear Magnetic Method for Determining The Permeability of Sandstones,” SPWLA 7th Annual Logging
Symposium, SPWLA (1966).
Sen, P.N. et al.: “Surface-to-volume ratio, charge density, nuclear magnetic relaxation, and permeability in clay-bearing sandstones,”
Geophysics (1990) 55, 61.
Shan, X. and Chen, H.: “Lattice Boltzmann model for simulating flows with multiple phases and components,” Phys. Rev. E (1993) 47,
1815.
Sheppard, A.P., Sok, R.M. and Averdunk, H.: “Techniques for Image Enhancement and Segmentation of Tomographic Images of
Porous Materials,” Physica A (2004) 339, 145.
Straley, C. et al.: “NMR in partially saturated rocks: laboratory insights on free fluid index and comparison with borehole logs,”
SPWLA 32th Annual Logging Symposium, SPWLA (June 1991).
Straley, C., Vinegar, H. and Morriss, C.: “Core Analysis by Low Field NMR,” (1994), SCA1994.
Strange, J.H., Webber, J.B.W. and Schmidt, S.D.: “Pore size distribution mapping,” Magnetic Resonance Imaging (1996) 14, 803.
Sun, B. and Dunn, K.J.: “Methods and limitations of NMR data inversion for fluid typing,” J. Magn. Res. (2004) 169, 118.
Talabi, O.: Pore Scale simulation of NMR response in porous media, Ph.D. thesis, London Imperial College (2008).
Tessier, J.J. and Packer, K.J.: “The characterization of multiphase fluid transport in a porous solid by pulsed gradient stimulated echo
nuclear magnetic resonance,” Phys. Fluids (1998) 10, 75.
Timur, A.: “Effective porosity and permeability of sandstones investigated through nuclear magnetic resonance principles,” The Log
Analyst (1969) 10, 3.
Toumelin, E. and Torres-Verdín, C.: “Modeling of Multiple Echo-Time NMR Measurements for Complex Pore Geometries and
Multiphase Saturations,” SPE Reservoir Evaluation & Engineering (2003) SPE 85635.
Toumelin, E. et al.: “Reconciling NMR Measurements and Numerical Simulations: Assessment of Temperature and Diffusive Coupling
Effects on Two-Phase Carbonate Samples,” Petrophysics (2003) 44, 91.
Valckenborg, R.M.E. et al.: “Random-walk simulations of NMR dephasing effects due to uniform magnetic-field gradients in a pore,”
Phys. Rev. E (2002) 65, 21306.
Westphal, H. et al.: “NMR Measurements in Carbonate Rocks: Problems and an Approach to a Solution,” Pure and Applied Geophysics
(2005) 162, 549.
Xue, S.: Towards improved methods for determining porous media multiphase flow functions, Ph.D. thesis, Texas AM University
(2004).
10 SPE 160870

TABLE 2- NMR FLUIDS PROPERTIES INPUT PARAMETERS

Parameter Water Clay Oil

HI 0.94 0.47 1.04


D0 [ cm2/ s] 2.30E-05 4.10E-06 8.00E-06

ρ[μm/s] 3 1 0

T2b [sec] 1 0.1 1

TABLE 3- SURFACE RELAXIVITY MATRIX FOR MODELING


INTRACTIONS BETWEEN PHASES

ρ[μm/s] Water Clay Solid Oil

Water 0 10 3 0

Clay 10 10 3 0

Solid 3 3 0 0

Oil 0 0 0 0

TABLE 4-VOLUME SUSCEPTIBILITY OF MINERALS AND FLUIDS USED TO


CALCULATE EFFECTIVE SUSCEPTIBILITY OF THE SAMPLES

-5
Material Vol. susceptibility [SI , x10 ]

Quartz -1.641
Calcite -1.311
Dolomite -1.37
Kaolinite -1.68
Feldspar -0.1695
Ankerite 36.1
Illite 4.16
Water, H2O -0.9035
Dodecane, C12H26 -1.26
50kppm NaCl brine -0.935
Bentheimer Clay region, 50% brine -1.31
Berea Clay Region, 50% brine 4.71
Ferroan-Dolomite clay region,50% clay 2.38
SPE 160870 11

a]

[b]

[c]

Fig. 3- NMR response of full saturations of oil and water [a] Bentheimer [b] Berea and [c] Ferroan Dolomite
12 SPE 160870

[a] [b]

[c] [d]

[e] [f]

Fig. 4- NMR relaxation response at partial saturations of both phases water [a, c, e] oil [b , d ,f[ incorporating internal gradients from
susceptibility contrast [a-b] Bentheimer [c-d] Berea and [e-f] Ferroan Dolomite
SPE 160870 13

[a]

[b]

[c]

Fig. 5- Transverse relaxation logarithmic mean time profiles of each phase at partial saturations [a] Bentheimer [b] Berea and [c]
Ferroan Dolomite
14 SPE 160870

[a]

[b]

[c]

Fig. 6- Relative permeability from NMR response of partially saturated samples compared with direct image based relative
permeability [a] Bentheimer [b] Berea and [c] Ferroan-Dolomite
SPE 160870 15

[a]

[b]

Fig. 7- Experimental validation of NMR relative permeability of wetting phase from numerical simulations with published results for
wetting phase relative permeability on [a] Bentheimer and [b] Berea sandstone samples.

You might also like