SPE 149042

Sensitivity of Elastic Properties to Fabric Heterogeneity in

Carbonate Rocks
R Sharma, SPE, Colorado School of Mines, M Prasad, Colorado School of Mines, M Batzle, SPE,
Colorado School of Mines, S Vega, Petroleum Institute

Copyright 2011, Society of Petroleum Engineers

This paper was prepared for presentation at the SPE/DGS Saudi Arabia section Technical Symposium and Exhibition held in Al-Khobar, Saudi Arabia, 15–18 May 2011.

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 does not necessarily reflect any position of
the Society of Petroleum Engineers, its officers, or members. 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 SPE copyright.

Abstract
Carbonate reservoirs hold more than 60% of the world’s total remaining oil and 40% of gas reserves
(Schlumberger, 2007). With recovery factor averaging around 25-30%, these reservoirs still have substantial
amount of trapped hydrocarbons. The primary reason for this poor exploitation of reserves in carbonate stems
from the fact that it is challenging to characterize storage and flow properties in these reservoir. Carbonate rocks
are known to exhibit depositional and developmental processes (EOR flooding) related heterogeneity.
Diagenetic processes like dissolution and precipitation in presence of certain fluid types such as: brine, may
have drastic effects on the matrix properties (Eberli, 2003). Diagenesis can trigger heterogeneous distribution of
petrophysical properties that may lead to heterogeneous saturation and elastic property distribution. We present
our analysis on sensitivity of heterogeneity in carbonate fabric to time-lapse flow and elastic properties
variation. We found that at full (end point) saturation, the most heterogeneous sample imbibed 20% less brine
than the least heterogeneous sample. Also, at full saturation, the least heterogeneous sample showed 45% more
increase in normalized bulk modulus and 10% more decrease in normalized shear modulus than the most
heterogeneous sample. Finally comparing our measurements and Gassmann predictions, it seems that these
carbonates mainly exhibit patchy saturation.

Introduction
Lithology, rock texture, saturation (uniform versus patchy), and saturants are known to affect the measured
elastic and flow properties in carbonate rocks (Knight et al., 1998 & Adam et al., 2006). Interdependence of
porosity, pore-types with acoustic velocity is well catalogued in the literature. However, not many studies are
out to establish interdependence between permeability and velocity, quantitative heterogeneity and pore size
distribution in carbonates.

In this work, we aim to better understand the differences in elastic properties of least heterogeneous and most
heterogeneous type carbonates in response to fluid saturation experiments. Because these carbonate samples
belong to reservoirs that have heterogeneous pore distribution and therefore they might contain complex flow
channels. It is important to understand how the fluid will flow in these flow channels, how much of it will be left
behind, and the impact it will have on the reservoir elastic properties. Rafavich et al. (1984) and Wilkens et al.
(1984) mentioned that porosity is the most important factor controlling velocity, and pore-fluid type has no
statistical relevance. On the other hand, Japsen et al., (2000) and Assefa et al., (2003) suggested that pore type,
pore fluid compressibility, and saturation affect on velocities and elastic properties in carbonate rocks are the
most important factors.
SPE-SAS-1049

Data, Methods and Measurements

CT scan images, Scanning Electron Microscope (ESEM), and acoustic techniques were employed to
qualitatively and quantitatively characterize heterogeneity in these carbonate samples and to relate them with the
changing trends in saturation and elastic property variations. We used three carbonate core plugs, which were
very similar in mineral (99% calcite) composition but markedly different in fabric composition. Figure 1, shows
the optical and the CT scan images of the samples arranged in order of increasing heterogeneity. Figure 2 shows
Scanning Electron Microscope (ESEM) results for the most heterogeneous (D) and least heterogeneous (J)
sample. Sample D shows poor grain size sorting and pore size distribution and possibly, a non corresponding
porosity and permeability relationship whereas, sample J shows uniform grain size sorting which indicate
towards a uniformly corresponding porosity and permeability relation.

Optical Images CT Scan Images

φ = 0.29 SAMPLE D
D mD
k = 23.4 Heterogeneous

φ = 0.30 SAMPLE K
K
k = 15.2 mD

φ = 0.20 SAMPLE J
J
k = 2.6 mD

Homogeneous

a) b)
Figure 1: Sample heterogeneity characterization using (a) optical and (b) CT scan images

J-Homogeneous

D- Heterogeneous

Figure 2: Grain size and pore size distribution using Environmental Scanning Acoustic Microscope (ESEM) in (a)
heterogeneous plug, (b) homogeneous plug
SPE-SAS-1049 3

To confirm for uniform or anisotropic permeability in samples, we decided to use shear wave propagation
through the samples. We used ultrasonic (1 MHz) wave through five propagation planes in the same sample.
Figure 3a displays four panels detailing the least heterogeneous (J) sample. The displays are; 1) CT scan image
and the sampling wavelength of the ultrasonic wave through this sample, 2) histogram with density number
distribution in terms of Houndsfield Units, 3) optical image with porosity, and 4) shear-wave propagation along
five planes in the sample. The shear-wave pattern in the five planes showed almost similar behavior and the
shear onset was picked within a narrow band, hence negligible velocity anisotropy in the sample. On the other
hand, in the most heterogeneous sample (D), the shear waveform pattern was very distinct in the five
orientations and shear onset was picked from a wide band (Figure 3b). This shows that heterogeneity in fabric
can introduce velocity anisotropy in the rock. However, if this velocity anisotropy can lead to permeability
anisotropy needed to be explored.

5% cutoff 5% cutoff

1.75 mm

Porosity = 19%

5
4 Maximum CT Number ( > 5%) − Minimum CT Number (> 5%)
Heterogeneity Number (%) = 0.5 * 100
ShearxWave
3
CT Number Highest Bin Percentage = 17%
J-Homogeneous

2
1

a)

5% cutoff 5% cutoff

2.2 mm

Porosity = 29%

DV_S1_dry2
0.15
55
4MaximumCT Number(> 5%) − MinimumCT Number(> 5%)
ity Number(%) = 0.5 *
Heterogene
0.10
4

CT Number Highest Bin Percentage

x100 = 44%
3
3

0.05 2
2
Amp (mV)

1
0.00 1

-0.05
D-Heterogeneous
-0.10
Shear Wave

-0.15

b) 0 10 20 30 40 50
Time (us)
60 70 80 90 100

Figure 3. Integrated charaterization of samples using shear wave profiling, Imaging and density distribution in a)
homogeneous sample, b) heterogeneous sample
SPE-SAS-1049

From the argument that porosity to velocity has an inverse correlation and porosity to permeability has direct
correlation, it follows that velocity and permeability will also have an inverse relation. We measured
permeability using a Pressure Decay Profile Permeameter (PDPK) along the same five orientations as used for
the shear-wave velocity measurements. We found that the least heterogeneous sample does not show any shear-
wave velocity anisotropy and so does no permeability anisotropy (Figure 4a) as was expected from the uniform
distribution of pores and grains in the sample (Figure 2b). On the other hand, the most heterogeneous sample
(Figure 4b) shows that it has very prominent shear-wave velocity anisotropy and a suggestive inverse relation
with permeability in the shear- plane. Such permeability anisotropy can lead to preferential flow in matrix.

Normalized by Minimum Perm.-Vel. Correlation in J Normalized by Minimum Perm.-Vel. Correlation in D

1.20 1.20
Homogeneous Permeability Heterogeneous Permeability
S1 Velocity S1 Velocity
101 S2 Velocity 101 S2 Velocity

P erm eab ility

76 76
P e rm e ab ility

V elocity
V elo c ity
1.10 1.10
51 51

26 26

1 1.00 1 1.00
0 1 2 3 4 5 6 0 1 2 3 4 5 6
a) Orientation b) Orientation

Figure 4: Polarized shear-wave velocity and permeability anisotropy in a) homogeneous sample, b) heterogeneous
sample.

We conducted saturation experiments to saturate our samples with 8000 ppm brine. Saturation experiment was
carried out in two stages. In the first stage samples were saturated under natural imbibition condition after
drawing vacuum on the samples. We then weighed the samples using Archimedes principle to calculate the
volume of brine imbibed by the samples. We called this stage as partial saturation stage because the volume of
water imbibed by the samples was less than the pore volume of the samples. We measured ultrasonic
compressive and shear wave velocity in the samples at this stage. In the second stage, the samples were
saturated at pressure (1000 psi) under undrained conditions. This pressure of 1000 psi was reached in gradual
steps of 50 psi and under controlled flow rate of 0.05 mL/min. After allowing sufficient time for fluid to reach
equilibrium, the samples were reweighed to calculate the additional imbibed brine at elevated pressure
condition. We called this stage as full or end-point saturation stage because we found that at this stage the least
heterogeneous sample was already oversaturated (exerted pressure more than maximum capillary pressure in the
sample) with brine. We again measured ultrasonic compressive and shear-wave velocities in the sample at this
end-point saturation stage. Table 1 presents brine filled volume at partial and end-point saturation stage.

In all, least heterogeneous sample imbibed 97% of the pore volume in comparison to 80% by most
heterogeneous sample at full saturation. This saturation experiment signifies the amount of hydrocarbon that is
likely to remain trapped in more heterogeneous formations, for example 20% in this work.

Table1. Brine filled porosity at partial and full saturation

Sample Brine Filled Porosity, Brine Filled Porosity, Total Porosity, CMS-300 Permeability, CMS-300
PS (%) FS (%) (%) (mD)
D 44 80 29 23.4
K 51 90 30 15.2
J 90 97 20 2.6
FS-Full Saturation, PS- Partial Saturation

Figure 5 is the graphical representation of the points mentioned in Table 1. The horizontal axis represents
saturation (Sw). Residual brine saturation is assumed to be constant to start with. Dotted curves represent the
natural imbibition and bold curves the forced imbibition. In all, homogeneous sample imbibed 97% of the pore
SPE-SAS-1049 5

volume in comparison to 80% by heterogeneous sample at full saturation. This saturation experiment signifies
the amount of hydrocarbon that is likely to remain trapped in more heterogeneous formations, for example 20%
in this work.

6.00 J-Homogeneous

4.00 Natural Imbibition

Trapped Gas
C apillary P ressure, P cwg

+VE
2.00 Heterogeneous Homogeneous
Swr Sgrw

0.00
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

-2.00 Heterogeneous

-VE Forced Imbibition

-4.00 D- Heterogeneous

Remaining Gas
-6.00
Brine
Water Saturation, Sw

Figure 5: Flow heterogeneity in our samples using partial and end-point saturation values at natural and forced
imbibitions. More heterogeneous sample shows more residual gas in the sample

We did not monitor the capillary pressure data in our experiments however; our results are very much in
agreement with Special Core Analysis (SCAL) experiment carried out by Masalmeh and Jing (2004) for residual
saturation. The authors described that the samples used in their experiment had almost similar porosities but had
different permeability values. Figure 6, shows imbibition capillary curves and photomicrographs of the samples
that are farthest apart in terms of the residual water. Sample (s17) that has more residual water has more
heterogeneous distribution of pores than the sample (s18) that has the least residual water. This is very much in
agreement with the flow results we obtained on our set of homogeneous and heterogeneous samples.

Figure 6: Dynamic petrophysical and petrographic correlation using (a) Imbibition capillary curves showing
significant difference in residual water saturation for (b) less heterogeneous samples #18, and (c) more heterogeneous
sample #17, having very similar porosity (27-30%) (Masalmeh and Jing, 2004).
SPE-SAS-1049

We then calculated bulk and shear modulus from the measured ultrasonic velocities at partial and full saturation
(Table 2) by accounting for the changes in bulk density of the samples at two stages. Figure 7 presents partial
and fully saturated (normalized to dry state) bulk and shear modulus. The horizontal axis represents saturation
increase which is equivalent of time in a reservoir life. Point value 1.00 on the vertical axis is the dry state
values of moduli. Above this line, we plotted bulk modulus and below the line, shear modulus. The two
heterogeneous samples are plotted in dotted lines. These two samples have very similar storage but different
flow characteristics and therefore ended up having different saturation after pressure saturation stage. This
indicates towards the role of pressure maintenance in different part of the heterogeneous reservoir.

Partial and full saturation stages are marked as PS and FS respectively. Under similar saturation conditions, they
imbibed different amount of brine and ended up having different residual gas saturation. At partial saturation,
homogeneous sample show a 22% added increase in bulk modulus and a 17% more decrease in shear modulus
than the heterogeneous sample. At full saturation, homogeneous sample show 45% extra increase in bulk
modulus and 10% extra decrease in shear modulus than the heterogeneous sample. This signifies that under any
EOR scheme, homogeneous and heterogeneous sections in the reservoir may end up having different saturation
and a very heterogeneous distribution of elastic properties with time.

Elastic Moduli versus Brine Saturation

2.00

30%
1.75
Elastic Modulii Normalized to Dry State

45%

1.50 49%

22%
7%
1.25 27%

1.00 - 4%
3% 10%
17% 10%
0.75 -21%
Shear Mod
Bulk Mod
0.50
0.00 0.20 0.40 0.60 0.80 1.00
Brine Saturation
Blk Mod Homogeneous Blk Mod M.Heterogeneous Blk Mod Heterogeneous
Shear Mod Homogeneous Shear Mod M.Heterogeneous Shear Mod Heterogeneous

Figure 7: Bulk and shear modulus measured at partial and fully saturated conditions normalized to their respective
dry state values. The two way arrows show the difference in relative changes in bulk and shear modulus between
homogeneous and heterogeneous at partial and fully saturated states.

Elastic Property Modeling

We used Gassmann equation to model the fluid substitution changes in the elastic properties using accurate in-
place fluid information. Since Gassmann is a uniform saturation model, a comparison of the measured
compressive-wave velocities with the modeled will give a good idea about the saturation type happening in the
samples. Cadoret et al. (1992) presented the first x-ray image evidence of patchy and uniform saturation in
carbonates (Figure 8). Figure 9 shows compressive-wave velocity values measured in homogeneous and
heterogeneous samples. The first three bars are measurement at dry, partial and fully saturated stages. The next
two bars are Gassmann calculated results corresponding to partial and fully saturated stages. The last bar is the
Gassmann velocity at full brine saturation.
SPE-SAS-1049 7

Table 2: Measured acoustic velocity at partial and full saturation

Vp at PS Vs at PS Vp at FS Vs at FS
Sample
(m/s) (m/s) (m/s) (m/s)
D 3750 2040 3820 2000
K 3100 1750 3200 1600
J 3050 1550 3390 1590
Vp, Vs - compressional and shear-wave velocity; PS, FS- Partial and Full Saturation

To start with, the heterogeneous sample had strongest dry frame and hence show the least saturation and
substitution effects. Measured values are always more than the corresponding Gassmann predicted values. This
observation can be tied down to reasons such as; 1) data acquisition using ultrasonic waveform, known for pore
fluid non-equilibrium and over estimated measurements, 2) a patchy saturation in the samples that will again
have higher estimated measurements than when the sample would have uniform saturation (Sengupta et al.,
1999). A non-Gassmann like saturation could indicate towards the patchy saturation in carbonate reservoirs as
displayed in Figure 9.

a) b)

Figure 8: Micrographs of the saturation maps of Brauvilliers limestone showing (a) patchy and (b) uniform saturation.
(Cadoret et al., 1992)

Modeled and Measured Vp versus Porosity

4000
Heterogeneous Mildly Homogeneous
Heterogeneous
Compressive-wave Velocity (m/s)

3500

3000

2500
29 30 20
Porosity (%)
Measured_Dry Measured_Partial Measured_Full
Gassmann_Partial Gassmann_Full Gassmann_Brine (100%)

Figure 8: Comparison of measured and Gassmann predicted bulk modulus in partial and fully saturated conditions.
Partial and full saturation prediction using Gassmann assumes uniform saturation (Wood’s average) of realistic fluid
in place.
SPE-SAS-1049

We will present a comparison of the measured velocities at partial and end-point saturation with results from
other fluid substitution models like Biot’s high frequency approximation of saturated properties, Squirt flow
model addressing local pore fluid non-equilibrium mechanism, and ‘patchy’ saturation model addressing partial
saturations in the rocks. Figure 8 also suggests that the samples with similar dry frame strength (irrespective of
the difference in porosity) have similar patterns of comparison between measured and Gassmann predicted
velocity values.

Discussion and Conclusion

Heterogeneous fabric can set in preferential flow and therefore can set in heterogeneity in fluid saturation and
elastic properties distribution. Partial to end-point saturation effects on elastic properties can be significant. This
situation is analogous to the time-lapse saturation and elastic property changes in the reservoir, when subjected
to pressure and saturation changes under EOR regimes.

We witness shear modulus weakening upon brine saturation much more than can be accounted for change in
bulk density upon saturation. These shear change can have geo-mechanical implication in reservoir for time-
lapse observations.

We also found that samples with similar dry frame strength instead of similar porosities show similar patterns of
comparison between measured and Gassmann predicted velocity results. The heterogeneity will have a role to
play in occurrence of these patterns.

The comparison of measured and modeled velocity values indicate that carbonates usually undergo patchy
saturation and Gassmann’s model in its original form is not sufficient to predict 4D saturation changes in these
reservoirs.

Acknowledgement
We express out sincere thanks to DHI consortium at UH and CSM and the Petroleum Institute, Abu Dhabi for
their financial support. We also thank Center for Rock Abuse at CSM for lab facilities and valuable technical
discussion.