Mai and Kantzas - 2002 - An Evaluation of The Application of Low Field NMR in The Characterization of Carbonate Reservoirs
Mai and Kantzas - 2002 - An Evaluation of The Application of Low Field NMR in The Characterization of Carbonate Reservoirs
Mai and Kantzas - 2002 - An Evaluation of The Application of Low Field NMR in The Characterization of Carbonate Reservoirs
φ = porosity scanning was also done on dry cores. The cores were then
C = a constant saturated with 2% brine and NMR spectra at 100% brine
FFI = the free fluid index saturation were obtained, along with CAT-scanning. The
BVI = the bound volume of irreducible water cores were then spun at 6000 RPM for 6 hours to achieve
irreducible water (Swi) condition. At this point, NMR and
The porosity value obtained from NMR can be used in the CAT-scanning were obtained again. The NMR spectra were
above equation. C is a constant that might change with obtained by using the CORESPEC 1000TM with TE = 0.3 and
formation. The FFI and BVI are the areas under the T2 NE = 5000. CAT-scanning was performed on a GE9800 with
distribution after the T2cutoff value and before the T2cutoff 140 KVP and 70 mA with a slice every 3 mm.
value, respectively.
Results and Discussion
The mean T2 model is given as5 Various methods of evaluating porosity were employed to
provide a porosity comparison. NMR data were used to
k = aT22gm φ 4 (3) established T2cutoff and Swi values. NMR data were used in
conjunction with CT data to differentiate between the primary
and secondary porosity fraction. This analysis was conducted
Where “a” is a coefficient that depends on formation type and to evaluate the relationship between secondary porosity and
T2gm is the geometric mean of the T2 distribution. permeability. Using the NMR data, permeability models were
Researchers often disagree on the value of the prefactor evaluated and a new model was also developed.
“a” for carbonates. Chang et al. believed that “a” should be
4.75, which is similar to the value used in sandstones7. Porosity. The porosity of each core sample can be calculated
However, Kenyon reported “a” to be 0.112, and Quintero et al. using 5 separate methods: gas expansion, Archimedes
said that it should be 0.413. principle (brine saturation process), NMR (spectra at fully
Chang et al. have also proposed that T2gm should be saturated), CT data at dry condition and CT data at wet stage.
included up to 750 ms only7. They believed that signals with The comparison of porosity between NMR and gas expansion
T2 values greater 750 ms came from vuggy porosity which was given earlier16. The relationship between them was φge =
contribute little to permeability7. 1.09φNMR - 0.01 with an R2 = 0.92. The porosity comparison
between NMR and CT yields a correlatin of φCT = 0.91φge -
Computerized Tomography. CT (or CAT) technology is an 0.007 and a R2 = 0.9217. The main reason for lack of a better
established tool that can be used to aid in reservoir fit was the fact that the cores lost consistently small chips in
characterization. The main principle in CT is that it uses an the centrifuge, making mass balance problematic. In general,
X-ray source that rotates around the sample to obtain one- porosity calculated using NMR spectra at fully saturated
dimensional projections of x-ray attenuation at different condition can yield accurate measurement of porosity. More
angles14. From this data, the cross-sectional slice of the importantly, this implies that the CT fully captures the porous
sample can be reconstructed. The resolution is about 0.5 x 0.5 media information.
mm while the slice thickness varies from 1 mm to 15 mm.
The CT produces 2-D arrays of CT numbers that can be T2cutoff and Swi. T2cutoff is taken as the T2 of the saturated core
translated to density and then to voxel porosity using15: with cumulative amplitude equal to the total amplitude of the
core at Swi16. T2cutoff were plotted against various parameters
ρ g − ρb that were extracted from the NMR data17. It was seen that
φ= (4)
ρg − ρ f T2cutoff is related to T2gm of the fully saturated spectra. This
relationship shows that as T2gm shifts toward the right T2cutoff
Where ρg = grain density increases. It was also seen that as the largest T2 peak (last
peak) in the NMR spectra increases T2cutoff decreases. This
ρb = bulk density
implies that as the porous system has a larger fraction of the
ρf = fluid density
large pores, the amount of trapping is less. T2cutoff is also a
function of the geometric mean of the last peak. As the T2gm
The CT data can be converted into a frequency distribution of
of the last peak increases, the value of T2cutoff increases. This
voxel porosity.
implies that as the T2gm of the last peak increases, the
difference in terms of pore size increases, thus more
Procedure
trapping occurs.
Eighty four samples from 6 different reservoirs in Western
When there is overlap between peaks, it is difficult to
Canada were selected for these experiments. Various methods
correctly estimate the boundaries and the area under each
were used to measure the porosity of the samples. For each
peak. The samples with overlapping peaks in the NMR
sample, porosity was obtained by gas expansion from the dry
spectra were not included in the relationships. A method is
cores. Air permeability was also obtained at this point. CAT-
currently being investigated to overcome this limitation.
4 A. MAI, A. KANTZAS SPE 77401
The same problem is encountered when the size of the The values predicted from this correlation are compared with
fastest T2 peak (first peak) was quantified, however, it was the actual Swi values in Figure 5. It can be seen that the
seen that T2cutoff is a function of the size of the first peak, as estimated values are very similar to the actual values. This
shown in the Figure 1. This figure shows that as the fraction correlation can be used as a first approximation to predict the
of the first peak increases, T2cutoff increases as well. An Swi content rather than using the T2cutoff correlation, which adds
increase in the size of the first peak means that there are more an extra step of summing all the amplitudes at T2 < T2cutoff.
of the smaller pores in the porous medium, which will be left Nevertheless, both methods are available to predict Swi by
saturated at the end of the drainage process. It is also possible using NMR data at fully saturated condition.
that a higher amount of the small pores can lead to higher
trapping of the larger pores as well. This pore topology will Primary and secondary porosity. The primary and
lean to a higher T2cutoff value. secondary porosity fraction can be determined from Ct data
Overall, T2cutoff is a function of the size of the last peak, using a deconvolution of the porosity frequency distribution.
location (T2gm) of the last peak and the size of the first peak. An illustrative example of this procedure is shown in Figure 6.
Various correlations for T2cutoff were developed and the best The method for porosity characterization is verified
correlation found is: qualitatively against the images obtained from CT analysis at
fully saturated condition. A few samples were selected
T2cutoff = 0.06417*(LP)-.74837(T2gm_LP)1.04209 (5) arbitrary to show various degree of secondary porosity. It was
seen that most of the cores with no secondary porosity are
Where LP = amplitude of last peak as a fraction of total similar to that shown in Figure 7. The calculated fraction of
amplitude secondary porosity of this sample from the deconvolution
T2gm_LP = geometric mean of the last peak approach is about 0.05. The images of this sample show a few
visible vugs (dark area). Thus the assumption seems to
This correlation shows that T2cutoff is directly related to the work well.
location (T2gm) of the last peak and inversely related to the size The sample with largest amount of secondary porosity
of the last peak. This is consistent with the observations of the fraction (0.79) in the whole data set is shown in Figure 8.
spectra as discussed above. The comparison between the From the images, this core sample has fairly large porous
estimated and the actual T2cutoff values are shown in Figure 2. spaces with a significant amount of vugs.
It can be seen that a first approximation of the T2cutoff can be A core sample that has an intemerdiate secondary
found using NMR data at fully saturated conditions. As porosity fraction of 0.37 is shown in Figure 9. It can be seen
mentioned before T2cutoff separates the fully saturated NMR that the secondary porosity fraction of this sample is larger
spectrum into the bound and free fluid component. The bound than that shown in Figure 7 and definitely smaller than that of
portion with T2 < T2cutoff is the Swi fraction of the pore space. the sample shown in Figure 8. Thus, it appears that the
Swi could be calculated directly using a correlation. To proposed deconvolution approach is adequate in
find this correlation, Swi was plotted against various distinguishing the primary and secondary porosity fraction.
parameters. Swi was seen to be directly related to the size of It was seen that the secondary porosity fraction is related
the last peak in NMR spectra as shown in Figure 3. This to the last peak and the location (T2gm) of this peak in the
figure shows that as the size of the last peak increases, Swi NMR spectra at fully saturated condition18. This makes sense
decreases. An increase in size of the last peak means that the since the last peak in the NMR spectra corresponds to the
porous medium has a higher number of the large pores, large pores in the porous medium.
meaning that there is less trapping. This relationship is similar
to that of T2cutoff and size of the last peak seen earlier. Swi was Permeability. The two permeability models which utilized
also plotted against T2gm of the last peak, the relationship NMR data were evaluated and a new model is also presented.
between these two parameters is not very good.
Swi is also shown to be a function of the first peak in Evaluation of existing permeability models using NMR data.
Figure 4. The relationship is similar to what was seen before. The two classic permeability models utilizing NMR data were
As the size of the first peak increases, Swi increases, meaning assessed. The free fluid model requires C to be calculated
that more of the pore space is trapped. indepently first. For this data set, C was found to be 0.1008.
A correlation was developed to directly estimate Swi. Free fluid model estimations were compared against the
The correlation obtained is: permeability of air in Figure 10. There is some difference
with the way BVI was calculated. Originally, it was believed
Swi = 0.49249*(FP)0.40745(LP)-1.91963 (6) that T2cutoff for carbonate is 100 ms, thus BVI is the sum of
amplitude at T2 below this point. However, T2cutoff was shown
Where FP is amplitude of the first peak as a fraction of the to vary, thus BVI is the sum of the amplitude upto T2cutoff,
total amplitude. which is not necessary 100 ms16.
This relationship shows that Swi is directly related to the size of The mean T2 model results were compared with the true
the first peak and inversely related to the size of the last peak. permeability values in Figure 11. As mentioned before, there
AN EVALUATION OF THE APPLICATION OF LOW FIELD NMR
SPE 77401 IN THE CHARACTERZATION OF CARBONATE RESERVOIRS 5
was disagreement on the value for the constant “a”. However, agrees well with those obtained by gas expansion method.
a change in “a” only shifts the points but it doesn’t improve T2cutoff values were found to be varying and they relate to
the scattering seen. The modification proposed by Chang et various other NMR parameters. It was seen that T2cutoff is a
al. was also evaluated but it did not improve the scattering. function of the size of the last peak, location (T2gm) of the last
From these figures, it can be seen that another correlation peak and the size of the first peak. The best correlation
is needed to improve the permeability estimation. obtained to estimate T2cutoff was found to be a function of the
size of the last peak and its location.
Development of Permeability Model. Air permeability was Swi correlation was found in terms of the amplitude of the
found to be strongly correlated to the primary porosity first and last peak as a fraction of the total amplitude. The
fraction, as shown in Figure 12. As the primary porosity estimated values of Swi are very similar to the actual values.
fraction increases, permeability decreases. This means that The primary and secondary porosity fraction was found
permeability is directly related to secondary porosity (which is based on an assumption regarding their distributions. These
one minus the primary porosity fraction). This finding values were then compared qualitatively using the CT images.
contrasts those reported by Chang et al., who believed that It was found that in general, the cores with higher secondary
“vuggy porosity contribute only weakly to fluid flow”7. fraction values were seen to have a higher fraction of visible
Permeability of air was plotted against various vugs. Thus this indicates that the assumption is valid.
parameters obtained from the NMR spectra. Even though Permeability models which used NMR data were
permeability does not strongly correlate with one specific evaluated. Both the Free Fluid Index model and the mean T2
variable, many general relationships were seen. Permeability model were shown to be inadequate in estimating
was seen to relate to the size of the first peak, as shown in permeability. Thus a new permeability model was developed.
Figure 13. This figure shows that as the size of the first peak This new model was shown to be a function of the size of the
increases, permeability decreases. This implies as the porous first and last peak. This model was shown to provide a better
media has more of the small pores, which creates restriction to estimation. However, this model requires that the first and the
flow thus decreasing the permeability. last peak has no overlapping, which is not always possible. A
Permeability of air is also correlated with the size of the better way of de-convoluting the peaks is currently being
last peak, as shown in Figure 14. This figure shows that as the developed to improve the deficiency.
size of the last peak increases, permeability increases as well.
An increase in the size of the last peak means that the porous Acknowledgements
medium has more of the larger pores which makes flow The authors wish to thank Michael Benedek of TIPM for
through porous media easier resulting in the increase of CAT-scanning all the cores, and Natalia Mirotchnik of TIPM
permeability. for performing some of the NMR measurements. The authors
Permeability was seen to correlate with Swi as well, as also would like to thank the numerous donors of carbonate
shown in Figure 15. This figure shows that as Swi increases, core plugs, and Suncor Energy Inc., NSERC and COURSE for
permeability decreases. It was noted that Swi itself is a financial support.
function of the size of the first peak and last peak.
By knowing that permeability correlates with the size of Nomenclature
the first peak and the last peak, a correlation was found to be: a = Constant in mean T2 permeability model
BVI = Bound volume index
k = 0.09396(FP)-1.81567(LP)4.55186 (7) C = Constant in the free fluid index permeability model
CAT = Computer Assisted Tomography
This correlation shows that permeability is directly related to CT = Computer Tomography
the size of the last peak and inversely related to the size of the FF = Free fluid index
first peak. This is consistent with the relationships seen FP = Size of first peak as a fraction
earlier. The comparison between the estimated values and the k = Permeability
actual permeability values is shown in Figure 16. This figure LP = Size of last peak as a fraction
shows that the estimated values are fairly similar to the actual NE = Number of echoes
values. It must be noted that this model requires a good rpm = Rotation per minute
estimation of the size of the first and last peak, which proves S/V = Surface to volume ratio
to be difficult when there is overlapping with other peaks. Swi = Fraction of irreducible water
This relationship shows that this correlation can do a better job TE = Inter-echo spacing in CPMG sequence
of estimating the permeability values than the free fluid index T2 = Transverse relaxation time
model and the mean T2 model. T2cutoff = T2 value that separate the bound and free portion
T2gm = Geometric mean of T2
Conclusions T2gm_LP = Geometric mean of the last peak
T2s = T2 relaxation time resulting from surface interactions
It was seen that porosity values calculated from NMR and CT φ = Porosity as a fraction
6 A. MAI, A. KANTZAS SPE 77401
φge = Porosity obtained by gas expansion at the 1999 SPE Annual Technical Conference and Exhibition,
φCT = Porosity obtained by CT Houston, Texas, October 3-6, 1999.
14. Vinegar, H., “X-Ray CT and NMR Imaging of Rocks”, Journal
φNMR = NMR estimate of porosity of Petroleum Technology, pp. 257-259, 1986.
ρ = Surface relaxivity 15. Kantzas, A., “Investigation of Physical Properties of Porous
ρb = Bulk density Rocks and Fluid Flow Phenomena in Porous Media Using
ρg = Grain density Computer Assisted Tomography”, In Situ, Vol. 14 (1), 77-132,
ρf = Fluid density 1990.
16. Mai, A., and Kantzas, A., “On the Characterization of Carbonate
Reservoirs Using Low Field NMR Tools”, SPE paper 75687,
presented at the SPE Gas Technology Sympsium, Calgary,
References Alberta, Canada, April 30 – May 2, 2002.
17. Mai, A., and Kantzas, A., “Characterization of Carbonate
1. Ausbrooks, R., Hurley, N., May, A., and Neese, D., “Pore-Size Reservoirs Using Low Field NMR and X-ray Tomography”,
Distributions in Vuggy Carbonates From Core Images, NMR, presented at the 2002 International Symposium of the Society
and Capillary Pressure”, SPE Paper 56506 presented at the 1999 of Core Analysts, Monterey, California, USA, September 22-
SPE Annual Technical Conference and Exhibition, Houston, 25, 2002.
Texas, October 3-6, 1999. 18. Mai, A., and Kantzas, A., “Porosity Distribution of Carbonate
2. Borgia, G., Bortolotti, V., and Fantazzini, P., “Combined Reservoirs Using Low Field NMR”, presented at the Petroleum
Magnetic Resonance Relaxation and Imaging Quantitaive Society’s Canadian International Petroleum Conference,
Determination of Matrix and Vugular Porosity in Carbonate Calgary, Alberta, Canada, June 11 – 13, 2002.
Cores”, SPE Paper 49294 presented at the 1998 SPE Annual
Technical Conference and Exhibition, New Orleans, Louisiana,
September 27-30, 1998.
3. Asquith, G., and Gibson, C., Basic Well Log Analysis For
Geologists, The American Association of Petroleum Geologists,
USA, 1982.
4. Oraby, M., Chafai, N., Hussing, R., Massengill, D., Clark, J.,
and Pletcher, D., “A New NMR Interpretation Technique Using
Error Minimization With Variable T2 Cutoff”, SPE Paper 38734
presented at the 1997 SPE Annual Technical Conference and
Exhibition, San Antonio, Texas, October 5-8, 1997.
5. Coates, G., Xiao, L., and Prammer, M., NMR Logging
Principles & Applications, Haliburton Energy Services, USA,
1999.
6. Timur, A., “Nuclear Magnetic Resonance Study of Carbonate
Rocks”, The Log Analyst, 13(5), 518-535, 1991.
7. Chang, D., Vinegar, H., Morriss, C., and Straley, C., “Effective
Porosity, Producible Fluid, and Permeability in Carbonates from
NMR Logging”, The Log Analyst, 38(2), 60-72, 1997.
8. Straley, C., Rossini, D., Vinegar, H., Tutunjian, P., and Morriss,
C., “Core Analysis by Low Field NMR”, SCA Paper 9404
presented at the 1994 International Symposium of the Society of
Core Analysts, Stavanger, Norway, September 12-14, 1994.
9. Ohen, H., Enwere, P., and Kier, J., “NMR Relaxivity Grouping
or NMR Facies Identification is Key to Effective Integration of
Core Nuclear Magnetic Resonance Data with Wireline Log”,
SCA Paper 9942 presented at the 1999 International Symposium
of the Society of Core Analysts, Colorado, USA, August 1-4,
1999.
10. Longman, M., Carbonate Diagenesis as a Control on
Strategraphic Traps, presented at the 1981 AAPG Fall
Education Conference, Calgary, Canada, 1981.
11. Marschall, D., Gardner, J., Mardon, D., and Coates, G., “Method
for Correlating NMR Relaxometry and Mercyr Injection Data”,
SCA Paper 9511 presented at the 1995 International Symposium
of the Society of Core Analysts, San Francisco, California,
September 12-14, 1995.
12. Kenyon, W., “Petrophysical Principles of Applications of NMR
Logging”, The Log Analyst, 38(2), 21-43, 1997.
13. Quintero, L., Boyd, A., Gyllensten, A., and El-Wazeer, F.,
“Comparison of Permeability from NMR and Production
Analysis in Carbonate Reservoirs”, SPE Paper 56798 presented
AN EVALUATION OF THE APPLICATION OF LOW FIELD NMR
SPE 77401 IN THE CHARACTERZATION OF CARBONATE RESERVOIRS 7
350 0.7
300 0.6
T2cutoff (ms)
250 0.5
200 0.4
Swi
150
0.3
100
0.2
50
0 0.1
0
0 0.05 0.1
0 0.05 0.1
Amplitude fraction in first peak
Amplitdue fraction in first peak
3 00
y = 1 .01 2 x + 7.7 46 7 0.5
2 50 R 2 = 0 .5 9 88 y = 0.927x + 0.0221
0.4 R 2 = 0.7307
2 00
T2c utoff (m s )
0.3
1 50
Sw i 0.2
1 00
0.1
50
0
0
0 50 1 00 1 50 2 00 2 50 3 00
0 0.1 0.2 0.3 0.4 0.5
100
0.7 90 total
0.6 80
prim
70
0.5 s ec
Frequenc y
60
0.4
Swi
50
0.3 40
0.2 30
20
0.1 10
0 0
10000
1000
100
k (mD)
10
1
0.1
0.01
0.001
0.001 0.1 10 1000 100000
k estimated (mD)
1000
100
10
k (mD)
1
0.1
Figure 8. Sample with largest secondary porosity fraction
0.01
0.001
0.001 0.1 10 1000
k estimated (mD)
10000 1000
1000
100
100
10
k (mD)
10
k (mD)
1 1
0.1 0.1
0.01 0.01
0.001
0.001
0 0.5 1 1.5
0 0.2 0.4 0.6 0.8
Primary porosity fraction Swi
Figure 12. Permeability vs primary porosity fraction Figure 15. Permeability vs Swi
1000
100 y = 0.9159x 0.9556
1000
R2 = 0.4007
100 10
k (mD)
10 1
k (mD)
1 0.1
0.1 0.01
0.01
0.001
0.001
0.001 0.1 10 1000
0 0.02 0.04 0.06 0.08 0.1 k estimated (mD)
Amplitude fraction in first peak
Figure 16. Permeability comparisons
1000
100
10
k (mD)
1
0.1
0.01
0.001
0.5 0.6 0.7 0.8 0.9 1
Amplitude fraction in last peak