SPE 162265

Variations in Bounding and Scanning Relative Permeability Curves with

Different Carbonate Rock Types
M. R. Dernaika, SPE, Ingrain Inc. - Abu Dhabi, M. A. Basioni, A. Dawoud, M. Z. Kalam, Abu Dhabi Company for
Onshore Oil Operations (ADCO) and S.M. Skjveland, SPE, University of Stavanger

Copyright 2012, Society of Petroleum Engineers

This paper was prepared for presentation at the Abu Dhabi International Petroleum Exhibition & Conference held in Abu Dhabi, UAE, 1114 November 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 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
Relative permeability curves generally exhibit hysteresis between different saturation cycles. This hysteresis is mainly caused
by wettability changes and fluid trapping. Different rock types may experience different hysteresis trends due to variations in
pore geometry. Relative permeability curves may also be a function of the saturation height in the reservoir.

A detailed laboratory study was performed to investigate relative permeability behavior for a major carbonate hydrocarbon
reservoir in the Middle East. Representative core samples covering five reservoir rock types (RRT) were selected based on
whole core and plug X-ray CT, NMR T2, MICP, porosity, permeability and thin-section analyses. Primary drainage and
imbibition water-oil relative permeability (bounding) curves were measured on all the five rock types by the steady state
technique using live fluids at full reservoir conditions with in situ saturation monitoring. Imbibition relative permeability
experiments were also conducted on the main RRT samples to assess the relative permeability (scanning) curves in the
transition zone by varying connate water saturations.

Hysteresis effects were observed between primary drainage and imbibition cycles, and appeared to be influenced by the
sample rock type involved (i.e. wettability and pore geometry). Variations in relative permeability within similar and
different rock types were described and understood from local heterogeneities present in each individual sample. This was
possible from dual energy CT scanning and high resolution imaging. Different imbibition trends from both oil and water
phases were detected from the scanning curves which were explained by different pore level fluid flow scenarios. Relative
permeability scanning curves to both oil and water phases increased with higher connate water. Relative permeability to oil
was explained based on the occupancy of the oil phase at varying connate water saturations. The change in water relative
permeability trend was addressed based on the connectivity of water at the varying connate water saturations. These results
and interpretations introduced improved understanding of the hysteresis phenomena and fluid flow behavior in the transition
zone of a cretaceous carbonate reservoir which can assist in the overall reservoir modeling and well planning.

Introduction
Relative permeability (Kr) is among the most important petrophysical parameters used for reservoir engineering and
management as it provides a great deal of information about the hydrodynamics of fluid flow in reservoir rocks. This
saturation function can be used for estimating productivity, injectivity, hydrocarbons in place, breakthrough time, and
ultimate recovery (Honarpour et al., 1986 and 1995; Heaviside, 1991).

Relative permeability is a function of saturation and saturation history. It depends on the direction of saturation changes as
well as on the maximum and minimum achieved saturations (Masalmeh, 2001 and Jerauld and Salter, 1990). Hysteresis in
relative permeability curves can exist between different saturation cycles. Masalmeh found that water relative permeability
curves exhibited either very little or no hysteresis at all except when considerable part of the pore space became oil-wet. On
the other hand, he found the oil relative permeability curves to show strong hysteresis between the primary drainage and
primary imbibition curves for all wetting states, with very little hysteresis thereafter except for mixed to oil-wet plugs
(Masalmeh, 2001). Actually, most experimental studies in literature have found that hysteresis is large for the nonwetting
phase and either small or nonexistent for the wetting phase.
Much of the hysteresis data in the literature has been obtained with bounding curves; saturations starting at endpoint values
2 SPE 162265

(i.e., irreducible water and residual oil saturations for water/oil systems). Hysteresis is also studied with scanning curves in
which the direction of saturation change is reversed at a number of intermediate saturations. Data such as these are applicable
for modeling reservoir processes in which the water saturation increases or decreases to an intermediate value, then changes
in the opposite direction. Examples include EOR methods that involve movement of an oil bank recovery from coning after a
well is shut in, production of oil from a transition zone, and dissipation of mud filtrate after well completion (Colonna et al,
1972, Land, 1971 and Braun & Holland, 1995).

Hysteresis is mainly caused by contact angle hysteresis, fluid trapping and wettability changes. Contact angle hysteresis is
usually attributed to non-equilibrium effects, contamination or heterogeneity of the surface due to either roughness or
composition. Fluid trapping is a result of pore space geometry and is caused by instabilities in the fluid-fluid interface
configurations. Wettability is the overall tendency of a reservoir rock to prefer one fluid over another, and has been
recognized as an important factor in remaining oil saturation, capillary pressure and relative permeability curves (Anderson,
1987). It depends on rock pore size distribution and rock-fluid interactions. It is influenced by fluid composition, rock surface
mineralogy, temperature, pressure and thickness of water films. Wettability changes (i.e. absence of thin films) can give rise
to considerable hysteresis in the oil-water contact angle (Melrose, 1965). Wettability alteration and contact angle hysteresis
have their origins at the micro-scale. Fluid trapping occurs at a bigger scale and is caused by pore morphology. Both
wettability changes and hysteresis in contact angle largely affect the degree of trapping.

Relative permeability is also a function of pore geometry (Fatt, 1960). There are certain rock properties that affect pore
geometry and can have great influence on reservoir properties such as porosity, permeability, capillary pressure, relative
permeability and resistivity. Table 1 lists some of the main rock properties that are considered important in carbonate
reservoirs. These properties may be different in sandstone reservoirs, which are typically single-porosity systems (i.e., inter-
particle pores) of relative uniform (homogeneous) nature. In carbonate reservoirs the presence of multiple-porosity systems
imparts petrophysical heterogeneity to the gross of the reservoir interval (Mazzullo and Chilingarian, 1992). This
heterogeneity (i.e. variation in rock properties) is the result of subsequent physical and chemical reorganization processes
such as compaction, solution, dolomitization and cementation. These rock properties are interrelated and thus may exert
different effects in different rock types. Therefore, it is not sufficient to characterize a reservoir rock by a single data such as
porosity or permeability. The heterogeneity in rock properties complicates the task of reservoir description where reservoir
properties (e.g. permeability, porosity, saturation and rock types) tend to vary as a function of spatial locations both in
vertical and areal directions. Therefore, proper rock characterization will be necessary to understand variations in rock types
which, in turn, help relate variations in macroscopic measurements (e.g. relative permeability curves) to rock properties and
pore geometry.

In this paper we present experimental hysteresis in relative permeability (bounding) curves between primary drainage and
imbibition on five different carbonate rock types. Imbibition relative permeability experiments were also conducted on the
main RRT samples to assess the relative permeability (scanning) curves in the transition zone by varying connate water
saturations. The rock types were initially characterized by static rock typing which is based on petrophysical data and
geological description. The relative permeability measurements were performed by the steady state technique at full reservoir
conditions using live fluids with in-situ saturation monitoring. These experiments were used to study the effect of different
carbonate rock types on relative permeability and hysteresis trends. Such data is rather scarce in the literature as most of the
available data are either measured on water wet rocks or on limited rock types. The hysteresis data provided in this work is
needed to enrich the available hysteresis models that may lack a complete and consistent description of the hysteresis
phenomena in porous media. In a previous publication (Dernaika et al, 2011) we presented different hysteresis trends (for two
different rock types) that were attributed to the complex relation between the effects of wettability and pore structure. In the
present work we extend our research on the hysteresis patterns to involve five different rock types, and we give further
insight into the variations of rock properties by detailed rock characterization including dual energy CT scanning. We also
investigate the effect of varying connate water saturations on the behavior of relative permeability scanning curves to help
predict fluid flow properties in the transition zone of the reservoir.

Throughout this paper drainage is used to describe oil displacing water. Imbibition describes water displacing oil regardless
of wettability condition. Primary drainage is used to represent oil displacing water from 100% water saturation. Spontaneous
imbibition refers to water displacing oil in water wet pores, and forced imbibition is used to refer to imbibition in mixed wet
pores.
SPE 162265 3

Rock Characterization
The rock (micro) properties that control pore geometry determine many macroscopic properties of the porous medium.
Establishing the relation between micro properties and macroscopic physical properties of a rock sample is an essential
requirement in the understanding of fluid flow behavior and saturation distribution in porous media and in the production of
oil and gas from petroleum bearing reservoirs. If variations in rock types are ignored, laboratory measurements for predicting
fluid flow may be misleading (Morgan and Gordon, 1970).

In this study reservoir rock types (RRT) were initially established based on combined petrophysical properties and geological
description. The petrophyscial properties included measurements from porosity, permeability, NMR T2 distributions and
mercury-derived drainage capillary pressure (Pc) and pore throat size distribution (PSD). Geological descriptions were
obtained from thin-section photomicrographs analysis which aims at defining pore systems, facies and depositional
environment. Mercury injection and thin-section preparation were done on corresponding trims from the plug samples. To
enhance our rock characterization scheme, high definition X-ray Dual Energy CT scanning was acquired on the plugs not
only to investigate the internal structure of the rocks through CT images but also to quantify mineralogy and porosity along
sample lengths through X-ray CT-derived Effective Atomic Number (Zeff) and Bulk Density (Wellington and Vinegar,
1987). Such knowledge of rock fabric and mineralogical distribution is an essential input for reservoir modeling and well
planning (Pranter et al., 2005).

The plugs were CT scanned at two energy levels with calibration material. The images were acquired in helical high-
resolution scanning mode and have an in plane (X-Y) pixel resolution of 0.468 mm and slice thickness (Z) of 0.500 mm. This
Dual-Energy imaging provides two distinct three-dimensional images of the plugs. The high energy images are more
sensitive to bulk density and the low energy images are more sensitive to mineralogy. The method summarized by
Wellington and Vinegar (1987) was used to compute independently the bulk density (BD) and the effective atomic number
(Zeff) for each CT slice position for every plug. The Zeff parameter is related to the photoelectric effect, PEF, commonly
used in wellbore logging. Thus, profiles of BD/Zeff as well as plug averages are generated and used for quality assurance,
rock typing, and heterogeneity assessment.

Figure 1 and Figure 2 provide rock characterization scheme from five different rock types (RRT) found in the reservoir.
Figure 3 gives the poroperm variation for all the samples, and Figure 4 shows crossplot of Zeff versus bulk density and grain
density. Both Zeff and bulk density were derived from the dual energy CT scanning while the grain density values were
measured in a conventional laboratory in combination with Helium porosity. Figure 3 shows an interesting behavior of the
best rock type (RRT1) samples. Those samples have the highest permeability values and yet are at the lower end of the
porosity range within all the rock types. This is a heterogeneity feature in carbonates and, in this case, it is attributed to
dolomitization which caused recrystallization of the limestone mineral resulting in larger pores and pore throats, better
connectivity, and increased pore heterogeneity.

Figure 4 (to the left) plots the average Zeff/BD values from all plugs. All samples lie on the calcite line except RRT1 samples
which come close to the dolomite Zeff line. The bulk density data gives sample porosity by using grain density information
which could be inferred from the effective atomic number or simply from laboratory measurement. Table 2 gives bulk
density and Zeff for three major minerals (i.e. calcite, dolomite and quartz). Dolomite theoretical Zeff value is 13.74, and the
Zeff values for sample #113 and #114 are 14.14 and 14.39, respectively. The higher Zeff values (than 13.74) for these two
samples derived from the dual energy data may be caused due to some low calcite percentages still present in the dolomite-
rich samples. The calcite Zeff is higher than that of dolomite and hence few percentages of calcite may cause a little increase
in the Zeff response. This is also confirmed by the laboratory grain density measurements which gave 2.83g/cc and 2.81g/cc
for sample #113 and #114, respectively. If the samples were made of 100% dolomite we would expect laboratory grain
density measurements at 2.87 g/cc (see Table 2). Interestingly, one may notice that the higher grain density sample (i.e. 113)
gave lower Zeff response than that from sample #114. This is a very consistent set of data which confirms the sensitivity of
the Zeff measurements by the Dual Energy CT scanning. This is an important step towards accurate quantitative
characterization of mineralogy (and porosity) on whole plugs as well as on full diameter cores at early stages of any core
analysis study. The Zeff/GD plot in Figure 4 shows that the calcite samples gave laboratory grain density values at 2.71g/cc
as compared to the dolomite samples which gave higher grain densities.

In Figure 1, individual poroperm data are given above each thin-section photomicrograph for direct comparison. The CT
images are presented in color scale with reference to the color scale bar on top of the CT images in Figure 1(b). Normally, X-
ray CT images come in shades of gray, which are directly influenced by the effective atomic number of the material and its
density. Dense material like calcite appears white and pores appear black. Unresolved pores like micrite would appear as dark
gray. Such gray-scale images may not always show all the detailed internal features and hence we prefer to present the
images in a color scale format where dense materials appear red and pores appear black. Any material density varying in
between would appear as yellow or green based on the color scale bar shown in the figure. It is worth noting that the same
color scale was applied to all images so that it becomes possible to have one-to-one comparison between the different rock
4 SPE 162265

type images and between images in the same rock type to evaluate local heterogeneity and better assess overall quality. This
is actually another important feature of using color scale images.

In Figure 2, two reference lines are shown on the NMR T2 plots to allow direct comparisons with the other rock type curves.
A reference line is also shown on the mercury-derived PSD curves for the same reason. T2 NMR data gives information
about pore body size distribution and can be converted to length unit using surface relaxation parameter. On the other hand,
mercury intrusion data does not provide direct information about pore diameters; it rather assigns the pore body volumes to
their entry throats. Therefore, mercury data contains valuable information about reservoir rocks and rock types because the
resistance offered by the pore structure to fluid flow is controlled by the pore throats.

In Figure 2, we see a general trend in the entry pressure that is increasing as we move from RRT1 to RRT5. This is normally
due to the macro pore sizes decreasing as we move towards the poorer rock types. However, we see that all rock types are
showing similar range of irreducible saturation (at 1000psi mercury pressure, equivalent to 75psi oil/water Pc). This is due to
the fact that all samples do not have many pore throat radii below 0.1 microns as can clearly be seen in the Hg-derived PSD
curves. Nevertheless, the samples would have differences in the intermediate pore throat sizes and this is demonstrated on the
Pc curves for each RRT where the intersections of the curves at 200psi pressure have been shaded to see how this shaded area
is occurring at lower mercury (corresponding to oil in reservoir fluid system) saturations as we move towards RRT5. This
observation proves that the different RRTs have varying intermediate pore throat sizes which can yield differences in
capillary pressures and phase flow.

The importance of this rock characterization phase and the way we present the rock types in Figure 1 and Figure 2 lies on the
local variations identified in each sample and how this variation could play role in producing different macroscopic
measurements like relative permeability functions and hysteresis patterns. Summary of the rock characterization scheme is
presented in Table 3.

Rock Type 1. Thin-section description of RRT1 samples reveals the dolomitic nature of those samples. Sample #113 (with
46mD perm) is fine to medium crystalline dolomite with vugs and intercrystalline meso to macropores. It has a loosely
packed texture and this explains its higher permeability value over sample #114 (with 21mD). Sample #114 is a strongly
dolomitized packstone with intercrystalline micro to macropores. Its matrix microporosity has been reduced in places by
intensive dolomitization. In both thin-section photomicrographs the dolomite grains have grey color, pores are in blue and
(minor) blocky calcite cement can be seen in red. The precursor limestone textures have not been preserved in these two
samples. The corresponding CT images in Figure 1(b) confirm the heterogeneity seen in the poroperm data and clearly show
the large connected pore channels along plug lengths. Those channels appear as black to blue in color representing low pixel
values with reference to the color scale bar on top of the CT images. Low pixels represent low density with big pores while
high pixels in red would represent dense materials like calcite or dolomite with low porosity.

The X-ray CT-derived Dual Energy (DE) data along sample lengths (i.e. bulk density and Zeff) are given in Figure 1(c). The
Zeff (mineralogy) distributions reveal uniform profiles, whereas bulk density curves show high pore heterogeneity along
sample lengths. This is in line with the heterogeneity detected from poroperm data, and indeed consistent with the color
distributions seen in the CT images. Such heterogeneity would lead to high pore channels with lower porosity on the overall
plug. Sample #113 and #114 show porosity variations along sample lengths as 5.3% and 3.6%, respectively. This is an
important piece of information to consider when explaining relative permeability curves and in-situ saturation profiles that
are usually influenced by capillary end effects.

Figure 2 shows low entry capillary pressure characteristics for this rock type with very low irreducible saturations at 1000psi
mercury pressure. This pressure corresponds to 75psi oil/water capillary pressure when scaled using mercury and reservoir
fluid interfacial tension and contact angle data (Morrow and Melrose, 1991). The pore throat size distribution curves show
that sample #113 has larger pore throats compared to sample #114. This is indeed in line with the plug permeability
measurements. Both curves present most of the pore throats above the reference line at 1 micron. The NMR T2 distribution
curves give data on the plug scale while the samples are 100% brine saturated. Again, we see larger T2 values for the sample
#113 confirming presence of larger pore bodies than in the other sample.

Rock Type 2. In Figure 1(a) thin-section description of RRT2 samples reveals the calcitic nature of those samples. Sample
#9 (with 11mD perm) is grainstone with intraparticle porosity in the micrite and interparticle intercrystalline porosity. It
reveals diverse pore types with concavo-convex grain contacts. Sample #15 (1.75mD) is a grainstone to rudstone showing
shell debris and heterogeneous texture with intraparticle porosity in the micrite. The photomicrographs agree with the
measured plug permeabilities, where the higher permeability plug (i.e. sample #9) clearly shows larger connected pore
channels as opposed to sample #15. The corresponding CT images on the whole plugs in Figure 1(b) present consistent
observations to the thin-sections and show that sample #9 has large pore channels as revealed in the large area of low pixels
(i.e. blue color) in the relevant image. On the other hand, sample #15 has larger proportion of the higher pixels (green/yellow
SPE 162265 5

colors) distributed along the sample length which hinders the flow in the large pore channels.

The dual energy data plot confirms these observations and yields lower bulk density (i.e. higher porosity) distribution for
sample #9. Both samples give uniform Zeff distribution around 15.7 confirming the calcitic nature of those samples. The
fracture seen at the top of the CT image of sample #9 was induced by implementation of improper procedures which were
done after the relative permeability experiments were completed. The reason we are showing the image is to confirm that
indeed sample #9 has got different pixel distributions at the plug scale and to confirm it has lower bulk density which could
be inferred from the bottom part of the plug that was not damaged in the laboratory. This is also in line with the Helium
porosity data which gave higher porosity value for sample #9 (0.279 frac) compared to sample #15 (0.236 frac).

These variations in the local heterogeneity between those samples in RRT2 are not very clearly identified in the mercury
derived data and NMR T2 distributions from Figure 2. These variations though would appear in the mercury and NMR
distributions if the scales are blown out and adjusted to show the minor changes. For instance, the peak in the mercury-
derived pore throat size distribution curve lies at 1.15 microns for sample #9 as compared to sample #15 which shows the
peak at 0.87 microns. This is a sufficient change in the pore throat sizes to yield large variations in permeability and
connectivity which was not initially perceived by the mercury and NMR data. This analysis shows the importance of the high
definition dual energy CT scanning that gives bulk density (i.e. porosity) variation along sample length as well as produces
continuous image log and makes it possible to see the internal structure of the rock along flow paths. The entry pressure Pc in
RRT2 is higher than the pressure in RRT1 simply because the macro pores in RRT2 samples are smaller than in RRT1
samples. This is also seen in the PSD curves. Although there is large difference in permeability between the RRT2 samples
this was not reflected in the entry pressures in the mercury data. This must be related to the sample sizes between plug and
trim and the location from which the mercury trim was taken (heterogeneity). The permeability and porosity variations
between these samples in RRT2 and the earlier observations from CT data may suggest that sample #15 could better be
classified as RRT3!

Rock Types 3&4. These two rock types are calcitic and almost similar in nature and distributions, and may actually be
classified as one rock type. They are packstone with intraparticle porosity in the micrite. Similar to the above investigations
we see interesting consistency between thin-section photomicrographs, CT images, dual energy data, mercury and NMR T2
distributions. The main point to emphasize here is the variation in poroperm data that can clearly be linked to the CT color
scale images. For instance, sample #4 has higher poroperm data than sample #6 in the RRT4 plugs, and this is reflected in the
CT images, where there is larger area of the low pixel values (blue color) distributed throughout sample #4. This was also
confirmed by the photomicrograph descriptions which revealed the fact that all samples experienced reduction in the
interparticle pores by intercrystalline cement except sample #4.

Rock Type 5. The plugs in RRT5 are amazingly similar in every single data generated or measurement obtained from them.
They are calcitic in nature, bioturbated packstone with intercrystalline microporosity between microcrystalline calcite
cements. Interparticle pore space is reduced by micritic matrix and pore-filling calcite and dolomite cements. They gave the
least poroperm data and the most homogeneous bulk density distributions. One can hardly differentiate between these two
samples and it will certainly be interesting to see how similar the multi-phase flow behavior will be!

Relative Permeability Bounding Curves

Before the start of the relative permeability program all samples were thoroughly cleaned by flow through techniques using
repeated cycles of several hot solvents to render the rocks water wet (presumed wettability condition before oil entered the
reservoir). The samples were then saturated with 100% simulated formation water that is representative to the reservoir water
composition. Primary drainage and imbibition water-oil relative permeability (bounding) curves were measured on all the
five rock types by the steady state equilibrium fractional flow technique using live fluids at full reservoir conditions with
gamma ray as in situ saturation monitoring (ISSM). All the samples underwent an aging period of minimum 4 weeks at the
end of primary drainage at full reservoir conditions without changing the experimental setup. This is believed to render the
rocks mixed wet for representative wettability condition during the imbibition process. Table 4 gives flooding rates and
capillary numbers for RRT1 and RRT5 samples. The capillary number is given as viscous forces to capillary forces (Nc =
v/!), where stands for viscosity, ! interfacial tension and v fluid velocity. The same flow rates were used in drainage and
imbibition, and the capillary numbers were comparable in both saturation cycles for each RRT. The capillary numbers given
in Table 4 are calculated from the imbibition experiments. The flooding rate used for RRT2 to RRT5 samples was 20ml/hr.
Higher flooding rate (i.e. 30ml/hr) was used for the RRT1 samples due to their higher permeability values.

Figure 5 gives the primary drainage and imbibition relative permeability for all RRT samples. It also gives the hysteresis
behavior between the primary drainage and imbibition cycles. Figure 5(a) presents the primary drainage relative permeability
curves, whereas Figure 5(b) presents the imbibition relative permeability curves. Figure 5(c) plots the hysteresis curves
between primary drainage and imbibition. From each RRT classification there are two samples. In each plot of Figure 5(a)
the two samples from the same RRT are plotted together. Similarly, the imbibition curves from the two samples in each RRT
6 SPE 162265

are plotted together in each of the plots in Figure 5(b). On the other hand, Figure 5(c) gives each sample data in a different
plot for each RRT classification to emphasize the hysteresis trend for each plug sample. Saturation cycle (i.e. primary
drainage or imbibition) together with sample number and RRT are indicated in the title chart of each plot. This is a heavy set
of data which aims at comparing relative permeability behavior between different carbonate rock types found in the reservoir
under study. Each row in the figure represents a single rock type data as classified earlier in Figure 1 and Figure 2. Actually,
Figure 5 presents the different RRT Kr data in the same fashion as the RRT classification in Figure 1 and Figure 2 so that
direct comparisons can be made between the RRT and the relative permeability behavior.

In the same fashion Figure 6 presents the primary drainage and imbibition in-situ (water) saturation monitoring profiles
(ISSM) along sample lengths for all RRT samples. Each saturation profile represents equilibrium water saturation at the end
of each fractional flow rate that was used during the steady state relative permeability experiment. These ISSM curves are
necessary to check quality of the relative permeability curves. Two main features are normally detected from such curves:
heterogeneity and capillary end effect. A bump flood was designed at the end of each saturation cycle to establish more
uniform water saturation (Sw) along sample lengths. The Sw profiles after the bump floods are monitored through the ISSM
curves. The saturation cycle (i.e. primary drainage or imbibition) together with sample number and RRT are indicated in the
title charts.

Primary Drainage. In Figure 5(a) we see little differences between the primary drainage (PD) relative permeability (Kr)
curves between the samples in the same RRT. These differences seem to become smaller as we move towards the poorer
RRT samples. The differences are obviously caused by local heterogeneity which is seen more in higher permeability
carbonate samples that are characterized by vugs, moldic porosity and diversified ranges of micro to macro-pores. This is
actually in line with the observations from Figure 1 and Figure 2 which showed less rock property variations (between the
samples in the same RRT) as we moved towards the poorer RRT5 samples. Larger variations in rock properties (see Table 1)
clearly yield variations in macro physical properties such as relative permeability.

There is no clear relation between the Krw-Kro intersection (saturation) point and rock type in Figure 5(a). However, all the
intersection points for all the rock type samples seem to range between 0.55 to 0.65 water saturation. This is basically
suggesting that the effective oil permeability becomes equal to the effective water permeability before oil invades 50% of the
pore space in primary drainage. This is rather an expected result due to the fact that water leaves most of the macro pores and
oil occupies only the macro and meso pores as oil advances into the pore system in the primary drainage cycle. At around
50% water saturation oil permeability would be higher than that of water as oil invades more of the pore space and water
flows in the smaller pore space and in water layers.

In Figure 5(a) sample #113 (in the RRT1 plot) suffered from large capillary end effect which caused the Kro curve to flatten
at lower water saturation (Sw). This can clearly be seen in the corresponding ISSM curves in Figure 6. The bump flood at the
end of drainage decreased the water saturation further and gave more uniform Sw profile. All samples were subject to a bump
flood to produce more representative saturations. This was necessary due to the capillary end effect, which is basically the
saturation gradient along sample lengths as can be seen in Figure 6. The capillary end effect is defined as the accumulation of
preferentially wetting phase (i.e. water in primary drainage) at the outlet when displacing the wetting phase by the non-
wetting phase (i.e. oil). This observation confirms that the primary drainage experiment was conducted at preferentially
water-wet condition. This, in turn, suggests that the solvent cleaning procedures implemented at the restoration stage of the
samples were efficient.

Without the bump floods the water saturations at the end of primary drainage range from 0.2 to 0.3 fractions. After the bump
floods, those saturations decreased to a range of 0.05 to 0.23. Similarly, the Kro increased after the bump flood from a range
of 0.2-0.5 to a range of 0.65-0.85. These changes (i.e. increase in Kro and decrease in Sw) emphasize the importance of the
bump flood design and demonstrate the impact of the capillary end effect on relative permeability curves. Different
samples/rock types may experience different magnitudes of capillary end effects due to variations in rock properties. This is
especially true for water wet states where the significance of the pore geometry is most apparent (Wardlaw, 1980).

Imbibition. In Figure 5(b) the imbibition Kr curves are presented in the same fashion as the PD Kr curves in Figure 5(a). The
imbibition Krw-Kro intersection points occur at lower Sw values than the intersection points of drainage. The imbibition
Krw-Kro intersection points decreased to a range of 0.45 to 0.6. This might be an indication of wettability change to less
water wet conditions (Honarpour et al., 1986) but conclusions about wettability alterations from Kr curves can be risky
especially for intermediate and/or mixed wet conditions (Cuiec, 1991).

Without bump floods the water saturations at the end of imbibition range from 0.75 to 0.85 fractions. After the bump floods
these saturations increase to a range of 0.80 to 0.95. Similarly, the Krw increased after the bump flood from a range of 0.2-
0.3 to a range of 0.2-0.55. These changes (i.e. increase in Krw and Sw) emphasize the importance of the bump flood design
and demonstrate the impact of rock properties and wettability on the capillary end effect and remaining oil saturations. Figure
SPE 162265 7

6 does not show severe end effects in imbibition except for sample #9 from RRT2 and sample #22 from RRT3. This can be
seen in the relevant ISSM curves which show lower Sw values towards the outlets of the samples. These effects were
minimized by the bump floods as can also be seen in Figure 6 in the relevant plots.

The ISSM curves for RRT1 samples in Figure 6 show fluctuations in both drainage and imbibition as compared to the other
rock types. This is not noise in the saturation profile data; it is caused by the local heterogeneity along the lengths of the
samples which become very apparent in the presence of vugs.

The imbibition Kr curves in Figure 5(b) for each of the two samples in each RRT show similar behavior except for minor
variations due to local heterogeneity which is basically related to the specific types and relative percentages of pores present,
and their distribution within the rock. The similar imbibition behavior of the different samples in the same RRT confirms that
the static rock typing scheme established from petrophysical measurements and geological descriptions is valid and yields
similar imbibition dynamic data for the samples under study. This is an interesting examination of the effect of static rock
characterization on the dynamic data and rather important input information for an ultimate dynamic reservoir modeling
(Masalmeh, 2007). The only violation to this link between static and dynamic rock typing is RRT2 samples which had
already been shown to give variations in the static rock characterization between sample #9 and #15. Although these two
samples were initially selected to be RRT2 but the reality is that sample #15 is better fitted into RRT3 as explained above.
This again confirms the consistency established between static and dynamic rock typing. There may not be a guarantee to
establish a good link between the static and dynamic rock types due to local heterogeneity and wettability but proper static
rock typing is certainly the way to reduce discrepancies that may arise between the two rock typing schemes.

The above observations made between primary drainage and imbibition relative permeability characteristics indicate that
there were some changes in the wettability conditions of the samples after aging. It is actually expected to observe mixed wet
behavior in such carbonate samples during the imbibition process. This has been indicated by the low remaining oil
saturations obtained at the end of imbibition together with the increase in water saturations (from drainage to imbibition) at
the Krw-Kro intersection points. In addition, most of the Krw values attained at the end of imbibition suggest mixed wet
behavior (Honarpour et al., 1986 and Masalmeh, 2001). Nevertheless, the effect of wettability on the relative permeability
end points is still subject for discussions, and these indications are only rules of thumb that are questioned by several
researchers (e.g. Morrow, 1990), especially for mixed wettability systems.

Figure 7A presents (in semi-log scale) relative permeability curves for different rock type samples to assess the effect of rock
types on the relative permeability behavior in primary drainage and imbibition. Plot (a) and plot (b) give the primary drainage
Krw curves and the primary drainage Kro curves, respectively, while plot (c) and plot (d) give the imbibition Krw curves and
the imbibition Kro curves, respectively. Figure 7B plots Figure 7A in linear scale. In primary drainage the curves did not
show major variations between the different RRT samples. At the high water saturation end, Kro tends to be larger for large
pore sized samples at the start of oil invasion. At the lower Sw end (i.e. towards end of drainage), the curves seem to be
influenced by capillary end effects. The imbibition curves in plot (c) and plot (d) for both Krw and Kro, respectively, show
large variations between the different rock types. One can also detect larger variations in the imbibition water relative
permeability than in the imbibition oil relative permeability curves.This result highlights fundamental differences between
drainage and imbibition fluid flow mechanisms at the pore level. It also indicates potential differences in the flow pattern
between oil and water.

In plot (c), the larger pore sized samples (i.e. RRT1 and RRT2) show higher Krw than the smaller pore sized samples (i.e.
RRT4 and RRT5) at a given water saturation. In this perspective RRT1 and RRT2 samples show more mixed wet
characteristics than the poorer RRT4 and RRT5 samples. This is also supported by the imbibition Kro trends in plot (d) for
RRT1 and RRT2 samples, which tend to show more rapid reduction upon water invasion into the large oil wet pores.
Opposite to the Krw behavior the large pore sized samples (i.e. RRT1 and RRT2) give lower Kro values at a given Sw. All
those samples were exposed to the same fluids at the same conditions and there is no clear reasoning behind possible
variations in the wettability conditions between the samples under study. The main differences between those samples are the
rock geometrical properties (see Table 1) that tend to give rise to different fluid mechanisms in imbibition more than in
primary drainage. This should be the result of more complex fluid flow regime in imbibition as will be explained below.

Hysteresis. Figure 5(c) shows the hysteresis behavior between primary drainage and imbibition. Each plot presents the Kr
curves from one rock sample to compare between the different saturation cycles. RRT1 samples gave a different hysteresis
pattern to all other rock types. In RRT1 the imbibition Kro is lower than the primary drainage Kro curve at the same water
saturation, and the imbibition Krw is higher than the primary drainage Krw curve at the same water saturation. For all the
other rock types (i.e. RRT2 to RRT5 samples) the imbibition Kro curve is lower than the primary drainage Kro curve (this is
a similar behavior to RRT1) but this time the imbibition Krw curve is lower than the primary drainage Krw curve at the same
water saturation. The hysteresis trend from RRT1 samples can be interpreted as mixed wet to oil wet behavior and the
hysteresis trend from the other RRT samples as water wet behavior (Jerauld and Salter, 1990 and Masalmeh, 2001). This
8 SPE 162265

behavior was explained in details in a conceptual hysteresis model by Masalmeh (Masalmeh, 2001) and was later represented
in schematic diagrams in our previous work (Dernaika et al, 2011) due to the importance and scarcity of such experimental
curves, and because it was related to our experimental data presented in that paper as is the case in this work as well.

The hysteresis model suggests that the samples showing similar Kr hysteresis pattern to our RRT1 samples are mixed wet to
oil wet and that the other samples showing the Kr hysteresis pattern as in our RRT2-RRT5 samples are water wet. This is
explained on the basis of fluid flow mechanisms and sequence of invasions. The model suggests that for the water wet case
both imbibition relative permeability to oil (Kro) and to water (Krw) are shifted to lower water saturations due to oil trapping
which causes lower mobile oil saturation. This causes the imbibition Kro to be lower than the primary drainage Kro curve,
and will cause the imbibition Krw to be higher than the primary drainage Krw. However, oil trapping may hinder water flow
and thus may lead to lower imbibition Krw than the primary drainage Krw. Hence, there will be two opposing effects on the
imbibition Krw and, which in many cases can cause no major hysteresis in the Krw curve between primary drainage and
imbibition in water wet systems. For the mixed wet case, imbibition will start displacing big oil wet pores and small water
wet pores. Water will occupy more big pores during imbibition than during primary drainage, and this may lead to higher
imbibition Krw than primary drainage Krw curve. Imbibition Kro will be lower than the primary drainage Kro since there
will be fewer large pores occupied with oil.

That was a brief explanation of the primary drainage and imbibition hysteresis model based on different wettability scenarios.
This explanation seems to be generally consistent with previously published hysteresis trends. However, this model - as many
other introduced models for fluid flow in porous media - needs to be utilized carefully when judging wettability states out of
relative permeability macro measurements. The reason for this is the lack of sufficient number of relative permeability
experiemnts performed at full reservoir conditions with live fluids on many different rock types to support this model. To
quickly jump to the conclusion that our RRT1 samples are mixed wet to oil wet and that our other RRT samples are water
wet may not be true since all those samples were saturated with the same fluids at the same experimental conditions.
Moreover, previous capillary pressure curves measured on similar rock types at reservoir conditions did not show any
spontaneuous imbibition and indeed showed mixed wet characteristics from forced imbibition data (Dernaika et al, 2011).
Therefore, the different relative permeability hysteresis trends reported in this work must be related to differences in pore
geometry between the different rock type samples in addition to the wettability considerations. There must be combined
effects from both pore geometry and wettability that may result in such variations in hysteresis trends.

When it comes to wettability it is not easily determined in natural complex porous media since it depends on rock pore sizes
and their distribution as well as on rock-fluid interactions. It is influenced by fluid composition, rock surface mineralogy,
temperature, pressure and thickness of water films. Upon oil invasion by capillary action a thin film of water remains
adsorbed on the pore surface and protects the solid from the oil. When capillary pressure becomes higher, at a certain critical
capillary pressure level, the water film ruptures and oil gets access to the surface. Stability of the water film depends on
prevailing capillary pressure and disjoining pressure isotherm (Hirasaki, 1991). The disjoining pressure isotherm is dependent
on the surface mineralogy, fluid composition and pore shape. When oil gets in contact with the surface wettability alteration
occurs and the degree of alteration depends on the fluid and surface compositions (i.e. rock-fluid interactions). The pore
corners and crevices where water is still present remain water-wet. This leads to development of heterogeneous forms of
wettability, where some parts of the rock surface are altered to oil-wet, while other parts remain water-wet ( Kovscek et al,
1993). In our case, what might induce different wettability states between RRT1 and the other rock types are: differences in
mineralogy (dolomite versus calcite), pore size, pore size distribution and pore shape. All other conditions such as fluid
composition, temperature and aging time were fixed for all rock types. Mineralogy is a major factor in the rock-fluid
intercations but there is no evidence about various wettability alterations between dolomite and calcite minerals. Actually,
contact angle experiments indicate similar wettability conditions (i.e. contact angles) are attained in the presence of oil and
water for both dolomite and calcite minerals (Vijapurapu, 2002). Pore size/distribution and pore shape may influence which
pore size will be converted to oil wet: concave versus convex pores (Kovscek et al, 1993) but again there is no clear evidence
this might be ocuuring in our system! In this perspective, at the end of primary drainage, the Augmented Laplace Equation
may predict random distribution of mixed wet pores, large mixed wet pores or small mixed pores.

What seems to be more obvious difference between the two systems (RRT1 versus other RRTs), and which may have an
effect on the fluid behavior is the pore geometry of the porous medium. The dolomite rock type (i.e. RRT1) has loosly
packed texture with large open pore throat sizes, whereas the calcite rock types in this study are mainly cemented with
reduced pore throat sizes. Such a difference in the pore system may yield differences in the ratio of pore body to pore throat
size, which is commonly referred to as the aspect ratio. One would normally expect the more loosly packed rock type with
large pores and pore throats to have lower aspect ratio than the calcitic well cemented samples with small pores. The aspect
ratio is one of the most important rock properties of porous media because it controls the surface energy of the reservoir that
is available to drive imbibition. It is responsible for the inefficient conversion of capillary work to surface energy due to
capillary instabilities that accompany the displacement (Morrow et al, 2008). The higher the aspect ratio is the less efficient
the drainage capillary work is and hence less surface energy will be available to drive imbibition. Therefore, higher aspect
SPE 162265 9

ratios would lead to higher degrees of hysteresis. In addition, the aspect ratio was found to be the most important structural
determinant of the relative permeability hysteresis behavior (Jerauld and Salter, 1990).

In primary drainage, oil invasion occurs by piston like displacement, where oil enters the pore from one of the ends,
displacing water originally present. This displacement occurs when the capillary pressure is increased to the capillary entry
pressure required for invasion of the interface. In imbibition, both piston-like invasion and snap off mechanisms are possible.
Piston-like invasion occurs when the capillary pressure is decreased to the capillary entry pressure required for the
displacement. The capillary entry pressures for both drainage and imbibition are governed by the Laplace equation. Snap-off
occurs as a result of increased water content from the corners or by swelling of films in water wet pores. Water may invade
the entire cross-section and cut off oil, resulting in trapped oil. The amount of trapped oil is strongly dependent on the shape
of the fluid interface. The interface shape is a function of the contact angle and the geometry of the pore. Breakage of the
continuity of the oil phase occurs at the junction of the pore with the exit throat. This capillary instability in the exit throat is
caused by the interface moving to maintain equal curvature at all points. Snap-off and trapping in single pores appear to be a
function of the aspect ratio (Wardlaw, 1982). Snap-off only occurs if piston-like invasion is topologically impossible since
the capillary pressure for piston-like invasion is favourable compared to the capillary pressure for snap-off (Lenormand,
1983). Snap-off can be more favourable over piston-like displacement in spontaneous imbibition if the pore is surrounded by
several oil-filled throats, when advancing contact angle is small and when the pore system is characterized by large aspect
ratio.

The above details about wettability and pore geometry impose a challenge in explaining the different hysteresis patterns
observed between different rock type samples and clearly confirm that the type of displacement that occurs at the pore level
depends upon whether the process is drainage or imbibition, surface wettability and geometry of the pore space.

The different hysteresis trends obtained in this research study may be explained by the type of the pore-level displacement
mechanism occurring in each rock type. In the dolomite RRT1 samples the controlling fluid mechanism in imbibition must
be the piston-like invasion. This is supported by the loose packed texture of the samples which are characterized by large
pore throats as evidenced from thin-section photomicrographs, CT images and mercury-derived pore throat size distribution
(PSD). The wettability change that must have happened during aging at the end of primary drainage certainly altered some of
the pore surfaces to oil wet which normally renders pores to a mixed wet state. The degree of wettability alteration may not
be assessed from the current measurements but this may be related to initial oil saturation, fluid and surface compositons and
aging time. Such pore geometry with the assumed mixed wet state would yield the measured hysteresis trend in RRT1 and
would be in line with the proposed hysteresis model (Masalmeh, 2001), where imbibition starts displacing big mixed wet
pores and small water wet pores. Water occupies more big pores during imbibition than during primary drainage, and this
would lead to higher imbibition Krw than primary drainage Krw curve. Imbibition Kro will be lower than the primary
drainage Kro since there will be fewer large pores occupied with oil.

For the calcite cemented samples (i.e. RRT2-RRT5) the fluid displacement in imbibition must be governed by the snap-off
mechanism at the pore scale, where it is known to be favoured in large aspect ratio systems with pores of more water wet
behavior. Those rock types are characterized by smaller pores than the RRT1 samples as revealed in thin-sections, CT images
and PSD curves. In addition to the aspect ratio, the accessibility of individual pores contributes to the flow behavior and to
the magnitude of permeability. Fluids do not seem to invade individual pores as set by the entry pressures of the
corresponding pore throats. Invasion often occurs in large clusters (Yuan, 1991), and therefore the size of the pore throats and
the number of pore throats emanating from a pore body (often called coordination number) largely affect the accessibile
interconnections in the pore system. Smaller pore sized systems may reduce the accessibility of individual pores and hence
may yield higher invasion pressures that may result in smaller magnitudes of permeability. In imbibition, lower accessibility
of individual pores may promote trapping of oil clusters by the snap-off mechnism which eventually may hinder water flow.
During primary drainage oil invades the pore system and alters wettability in certain pore surface locations as explained
earlier. The pore corners and crevices where water is still present remain water-wet. The area of the water wet surface would
be comparable to all pores invaded by the oil (Helland and Skjveland, 2005). This would mean that the established water-
wet surface area is expected to be almost the same for all invaded pores between RRT1 and RRT2-RRT5 samples. Thus, the
smaller pores in the calcite rocks would have larger fraction of water wet surface, and should therefore exhibit a more water
wet behavior than the larger pores in RRT1 during imbibition. It is actually obvious from Figure 5 that RRT5 samples with
the smallest pore and pore throst sizes have shown the most water wet imbibition Kr behavior (see Figure 7A and Figure 7B
as well). This does not necessarily mean that this RRT5 is water wet! It shows water wet behavior due to combined effects of
wettability and pore geometry. To emphasize this argument similar RRT samples have been measured in porous plate at
reservoir conditions and did not show any positive capillary pressure imbibition. Similar RRT samples were also tested in
wettability index experiments and gave large negative values that ranged from -0.55 to -0.66, indicating mixed wet states.
Previous researchers had commented on the risk of evaluating the wettability of a reservoir solely from relative permeability
curves because of the heterogeneous nature of wettability and the uncertain degree of wettability alterations (Cuiec, 1991 and
Torsaeter, 1988). From the NMR T2 curve and mercury PSD one would be tempted to infer the aspect ratio from peak values
10 SPE 162265

of the T2 curve and mercury PSD. This can be easily calculated from the given values in Table 3. For RRT1 samples (i.e. 113
and 114) the T2 by PSD peak ratios are 373/2.7(138) and 295/1.53(193), respectively. Sample #9 from RRT2 and sample
#138 from RRT5 would give T2 by PSD peak ratios as 257 and 345, respectively. This clearly shows that the aspect ratio
would increase as we move from RRT1 to RRT5. This is not a calculation of the aspect ratio but it can certainly help
understand the change of the aspect ratio with rock type. T2 value was used to infer pore body size. This analysis of pore
geometry and wettability behavior has been very useful in explaining the variations seen in the hysteresis patterns among the
various rock types involved. This same analysis helps explain the variations of the imbibition Kro and Krw curves in Figure
7A and Figure 7B. The combined effects of wettability and pore geometry from RRT2 to RRT5 samples suggest that these
RRTs are characterized by higher aspect ratio and more water wet behavior which can explain the observed hysteresis trend
by the water wet hysteresis model presented earlier (Masalmeh, 2001). Such a behavior would lead to fluid (oil) entrapment
and yield reduced imbibition Krw compared to primary drainage Krw curve.

The differences between these dolomite and calcite samples are not specific to the type of minerals; they are related to the
pore geometry. Sample #9 from RRT2 is calcite-rich and yet tends to show similar hysteresis behavior to the dolomite RRT1
samples. The imbibition Krw curve of sample #9 in Figure 5 shows no hysteresis with the primary drainage Krw at the low
Sw range. At higher Sw the imbibition Krw curve tends to decrease below the primary drainage curve due to the large
capillary end effect depicted in the corresponding in-situ saturation monitoring plot in Figure 6. This effect caused the
imbibition Krw curve to flatten and to give lower permeability values at high Sw. However, the bump flood at the end of
imbibition reduced the capillary end effect as can be seen in the ISSM plot and the imbibition Krw end point almost
approached the primary drainage Krw curve. This is not a water wet behavior. This is clearly influenced by the type of pores
in the calcite sample and the established wettability condition at the end of primary drainage. In Figure 7A sample #9 from
RRT2 shows similar behavior to the RRT1 sample. The pore geometry is more in favour of piston-like displacement and the
large pore throats enhances the accessibilities of the pore throats in the porous rock sample. One would tend to believe that
those samples were subject to a range of mild to moderate wettability alteration degrees because the pore geometry effects
could still be seen in the imbibition behavior. If the samples were severly aged by a kind of chemical for instance then the
pore geometry effects would be different and the strong oil wet characteristics of the pores would fully govern the flow and
perhaps promote piston-like displacement regime.

As mentioned above, Figure 7A and Figure 7B present very interesting effects of the pore geometry in primary drainage and
in imbibition. Although the samples have different pore geometries the primary drainage relative permeability curves did not
show major variations. However, the imbibition Kr behavior was different. The reason for this is the different displacement
mechanisms that govern flow in each saturation cycle. In primary drainage flow advances with piston-like displacement
where oil invades successively smaller pores. Primary drainage occurs in the pristine state of the rock, and hence should
always represent water wet behavior. In here the flow is mainly controlled by available pore throat sizes. In imbibition flow
can be controlled by piston-like invasion or snap-off mechanism. Both displacement mechanisms are actually possible in the
porous medium depending on the wettability state of the pore and the pore geometry. In here the flow is controlled by the
aspect ratio which may yield snap-off and fluid trapping, by pores which could behave as water wet or mixed wet depending
on size, degree of wettability alteration and individual pore accessibility. This can be very complex set of flow regimes
occurring at the pore level which makes the prediction of the macro flow very difficult. Figure 7A and Figure 7B represent a
good example to emphasize the importance of sample selection and rock typing which may lead to measurement
discrepancies if not done properly.

Relative Permeability Scanning Curves

Three plug samples from the main reservoir rock type #3 were selected to undergo fractional flow steady state (SS) relative
permeability experiments to investigate the imbibition Kr behavior in the transition zone of the reservoir under study. The
experiments were performed at representative conditions to simulate reservoir behavior. Each sample was prepared to
different target initial oil saturation (Soi) by primary drainage SS Kr experiment. At the end of drainage, and before the start
of imbibition, the samples were aged in crude oil at reservoir temperature for four weeks.

At the top of the transition zone big pores and small pores are invaded with oil, while at the bottom of the transition zone
only the big pores may be invaded with oil; the smaller pores would stay filled with water and would remain water wet. This
would yield variations in the wettability state (along the length of the transition zone), which are governed by the increase in
water saturation with depth. One would expect more oil wet rocks at the top of the transition zone (i.e. low Swi) as compared
to the base of the zone where the rocks should be more water wet. Such variations in saturations and wettability can have
large impact on the production characteristics along the depth of the transition zone (e.g. Masalmeh, 2000).

Figure 8 shows the poroperm data for the three transition zone (TZ) samples together with the poroperm data from the Kr
bounding curve program. The figure also presents the corresponding CT images and sample numbers (SN) from the TZ rock
samples. The CT images confirm the poroperm variations seen in the TZ samples. In Figure 9 plug #10 started with the
lowest initial oil saturation (Soi) of 0.32frac., followed by plug #25 (Soi= 0.52frac.), and finally sample #14 (Soi=0.68frac.).
SPE 162265 11

The bounding imbibition fractional flow Kro curve (plug #22) is also shown in the figure for reference and to investigate the
imbibition Kro scanning curve behavior in a wider saturation range. It would certainly be more representative to carry out the
scanning curve experiemnts (at varying Soi) on one plug but this would lead to extensive time delays in the project timing.

The tests appear to show trends of increasing kro with decreasing initial oil saturation (Soi). The imbibition Krw scanning
curves are plotted in Figure 10, and they also show increasing krw with decreasing initial oil saturation. This type of data is
of great importance in the understanding of the flow behavior in transition zones. The impact of wettability variations within
hydrocarbon transition zones is not fully understood, and such data is rather scarce in the litersature especially the water
relative permeability variation with Swi, where it cannot be derived from imbibition centrifuge experiemnts. Oil relative
permeability (Kro) curves were obtained in the literature by imbibition single speed centrifuge experiments at different initial
water saturations (Masalmeh, 2000). The Kro increased with Swi at a given Sw. That is, Kro increased with increasing water
wetness in the samples. Such a behavior would lead to the conclusion that the mobility of oil in transition zone can be higher
than conventionally assumed. Therefore, large volumes of oil can be recovered (Masalmeh, 2000). To confirm this
conclusion, however, variation in water relative permeability with Swi (i.e. wettability state) is needed (as in this research
study) due to its significance and profound influence on waterflood efficiency (Jackson et al, 2003).

Figure 9 and Figure 10, respectively, present increasing Kro and Krw with increasing Swi (i.e. decreasing Soi) at a given Sw.
For the larger Kro curves with higher Swi, this is due to the fact that at low Swi oil flows in both big and small pores whereas
at high Swi oil flows in large pores only. This variation of Kro with Swi agrees with similar results obtained from imbibition
single speed centrifuge experiemnts (Masalmeh, 2000). Figure 9, however, shows that for sample #25 (Soi=0.52frac), Kro
starts large but decreases quickly to very low Kro values after 0.69frac Sw. This behavior occurs for low Soi samples, where
oil phase tends to lose its connectivity after certain amount of pore space has been water flooded. Because the oil at low Soi
only exists in the bigger pores water flood can easily sweep those pores leaving very little oil in the pore space which causes
the oil to lose its connectivity and the oil permeability to drop dramatically. For even lower Soi of 0.32frac (plug #10) and
upon water flood, imbibition Kro drops quickly since there are fewer large pores occupied with oil. This behavior was absent
with the larger Soi samples because oil would be present in wider range of pore sizes and hence oil production occurs with
steady reduction in Kro from different pore sizes and hence we do not see sharp decrease in Kro. The oil phase can maintain
good connectivity until lower oil saturations are attained. This is of course directly related to the oil phase saturation
distribution at the pore level. On the other hand, Krw in Figure 10 shows even more increases with decreasing Soi. Hysteresis
in both Kro and Krw may indicate non-water wet conditions. At low Swi there are less connected water-filled pores and
hence water can initially only flow in large oil wet pores and small water wet pores. The percentage of water wet pores here
should be critical because it will govern the fraction of water flow in the oil-wet and the water-wet pores! This will in turn
have direct impact on the value of Krw. Therefore, at very low Swi, and depending on pore sizes and wettability distribution
water can flow in large oil wet pores. Krw increases gradually and slowly as water saturation increases rapidly due to the
invasion of the large pores. The increase of Krw may change slope and increases rapidly when water connectivity increases
until water spans the whole pore space. The start of water invasion from a higher Swi can give rise to higher Krw because
water would invade the larger pores only and due to the large water saturation which helps speed up water connectivity and
yield larger Krw. Sample #22 (Soi=0.79frac), sample #14 (Soi=0.68frac) and sample #25 (Soi=0.52) all show change in the
slope of Krw increase with Swi. The change of slope of the Krw curve occurs at around 0.67frac Sw for sample #22 and #14,
while it occurs at 0.61frac Sw for sample #25. The linear scale in Figure 11(b) clearly shows the rapid increase in Krw at a
certain Sw which must be related to the improved water connectivity as water invades the pore space.

Figure 12(a) shows scanning curve for sample #25 (Soi=0.52frac) in between bounding drainage and imbibition curves of
sample #22 which is believed to be of the same RRT but permeability compariosns show that sample #25 is at higher
permeability value than sample #22. This could be the reason for this or could simply be related to pore geometry and
wettability considerations. For sample #14 (Soi=0.68frac) its scanning curve is in between the bounding drainage and
imbibition curves. It is worth noting that sample #14 is at lower permeability value than sample #22. From the earlier pore
geometry investigations in this paper, permeability changes may indicate variations in pore geometries including pore sizes
which may heavily affect the imbibition Kr behavior. Sample #14 gives higher Krw than sample #22 at low Sw although
sample #22 has higher permeability. The higher Krw of sample #14 should be the effect of Swi as explained above. At higher
Sw sample #22 becomes at higher Krw than sample #14. This is believed to be caused by the higher absolute permeability of
sample #22. From this analysis hysteresis from Krw seems to be more evidenced and influenced by pore structure and
wettability than hysteresis of Kro. It is clear that Kr is largely affected by Swi which might be related to wettability and pore
sizes. Such variation of Kr with Swi could be significant to affect production and flow behavior at the reserovir scale.
12 SPE 162265

Conclusions and Observations

The following conclusions and observations can be summarized from this reaserch study:
1. Dual Energy CT scanning derives accurate porosity and mineralogy distributions along sample lengths that can enhance
sample selection and improve static rock typing. Longitudinal color-scale CT images have been essential in detecting
local heterogeneity among various rock type samples.
2. The initially established static rock typing scheme based on petrophysical properties and geological description was
validated by imbibition relative permeability and hysteresis trends. This validation is essential when assigning saturation
functions into dynamic reservoir modeling.
3. Mercury data from smaller trims can be misleading if not supported by plug measurements. This may lead to
discrepancies in the static rock typing scheme.
4. Samples within the same rock type may show little variations in relative permeability curves due to local heterogeneity.
5. Different rock types may not show major variations in primary drainage Kr. This must be due to the less complex fluid
flow mechanism governed by piston-like displacement.
6. Imbibition Kr curves at representative reservoir conditions show large variations among different rock types. This effect
was attributed to the combined effects of pore geometry and wettability in the complex imbibition flow.
7. Imbibition Krw curves show more variations with rock types than imbibition Kro.
8. Primary drainage and imbibition relative permeability curves showed different hysteresis patterns on different rock type
samples which could be attributed to the combined effects of wettability and pore geometry.
9. Imbibition Kr scanning curves by the steady state fractional flow expriemnts at reservoir conditions showed trends of
increasing kro with decreasing initial oil saturation. This was in agreement with previously measured Kro scanning curves
from centrifuge.
10. Imbibition Krw scanning curves were also measured and showed increasing krw with decreasing initial oil saturation.
This data is not usually obtained in imbibition centrifuge programs. Such data is of great significance in affecting
production and flow behavior at reservoir scale.

Nomenclature
Symbols
BD Bulk Density
GD Grain Density
! Interfacial Tension
KL Klinkenburg Corrected Gas Permeability
Kr Relative Permeability
Kro Oil Relative Permeability
Krw Water Relative Permeability
Nc Capillary Number
Sw Water Saturation
Swi Initial Water Saturation
Soi Initial Oil Saturation
Sor Residual Oil Saturation
SS Steady State
v Velocity
Viscosity
Zeff Effective Atomic Number

Abbreviations
Frac Fractional
Imb Imbibition
ISSM In-Situ Saturation Monitoring
PD Primary Drainage
PSD Mercury-Derived Pore Throat Size Distribution
RRT Reservoir Rock Type
SN Sample Number
TZ Transition Zone

Acknowledgment
The authors wish to acknowledge ADNOC and ADCO Management for permission to publish the results of this research
study. Ingrain Inc.-Abu Dhabi are acknowledged for providing the Dual-Energy CT scanning data and images, with special
thanks to Dominic McGurk.
SPE 162265 13

References
Anderson, W.G., Wettability Literature Survey Part 4: Effects of Wettability on Capillary Pressure, JPT (Oct. 1987) 1283-1300.
Anderson, W.G., Wettability Literature Survey Part 5: The Effects of Wettability on Relative Permeability, JPT (Nov. 1987) 1453-68.
Braun, E.M., and Holland, R.F.: Relative Permeability Hysteresis: Laboratory Measurements and a Conceptual Model (SPE 28615) SPE
Reservoir Engineering (August 1995).
Cuiec, L.E., Evaluation of Reservoir Wettability and Its Effects on Oil Recovery, Interfacial Phenomena in Oil Recovery, N.R. Morrow
(ed.), Marcell Dekker, New York, 319-75, 1991.
Colonna, J., Brissaud, F., and Millet, J.L.: Evolution of Capillarity and Relative Permeability Hysteresis SPEJ (Feb. 1972) 28; Trans.,
AIME, 253.
Dernaika, M.R., Kalam, M.Z., Basioni, M., Dawoud, A. and Skjveland, S.M.: Hysteresis of Capillary Pressure, Resistivity Index and
Relative Permeability in Carbonate Rock Types SCA 2011-14 presented at the SCA 2011 conference, Austin, Texas, USA, September
(2011).
Fatt, I.: Microscopic Heterogeneity in Reservoir Rock and Its Influence on Reservoir Behavior, J. Institute of Petroleum (July 1966) 231.
Heaviside, J., Measurement of Relative Permeability, in Interracial Phenomena in Oil Recovery by N. R.Morrow (Ed.), Marcel Dekker,
New York, 377 (1991).
Helland, J.O., Skjveland, S.M., (2005): Physically-based capillary pressure correlation for mixed-wet reservoirs from a bundle-of-tubes
model. SPEJ (SPE 89428), presented at the 2004 SPE Symposium on Improved Oil Recovery, Tulsa, OK, Apr. 1721.
Hirasaki, G.J., Thermodynamics of Thin Films and Three-Phase Contact Regioins, Interfacial Phenomena in Oil Recovery, N.R. Morrow
(ed.), Marcell Dekker, New York, 23-76, 1991
Honarpour, M.M., Huang, D.D. and Al-Hussainy, R.: Simultaneous Measurements of Relative Permeability, Capillary Pressure, and
Electrical Resistivity with Microwave System for Saturation Monitoring, SPE 30540 presented at the SPE Annual Technical
Conference & Exhibition held in Dallas, USA., 22-25 October, 1995.
Honarpour, M., Koederitz, L. and Harvey, A.H., Relative Permeability of Petroleum Reservoirs, CRC Press, Boca Raton, (1986).
Jackson, M.D., P.H. Valvatne, and M.J. Blunt,: Prediction of wettability variation and its impact on waterflooding using pore-to reservoir-
scale simulation, J. Pet. Sci. and Eng. (2003) 39, 231. Presented at the 2002 SPE Annual Technical Conference and Exhibition, SPE
77543.
Jerauld, G. R., and S. J. Salter, Effect of pore-structure on hysteresis in relative permeability and capillary pressure: Pore-level modeling,
Transport in Porous Media, 5, 103-151, 1990.
Kovscek, A.R., Wong, H., and Radke, C.J.: A pore-level scenario for the development of mixed wettability in oil reservoirs, Am. Inst.
Chem. Eng. J. (1993) 39, No. 6, 10721085.
Land, C.S.: Comparison of Calculated with Experimental Imbibition Relative Permeability, SPEJ (Dec. 1971) 419; Trans., AIME, 251.
Lenormand, R., C. Zarcone, and A. Sarr, Mechanisms of the Displacement of One Fluid by Another in a Network of Capillary Ducts,
Journal of Fluid Mechanics, 135, 337-353, 1983.
Masalmeh, S.K.: High Oil Recoveries From Transition Zones, SPE paper 87291 presented at the 2000 SPE Abu Dhabi International
Petroleum Exhibition and Conference, Abu Dhabi, 13-15 October.
Masalmeh, S.K.: Experimental Measurements of Capillary Pressure and Relative Permeability Hysteresis SCA 2001-23 presented at the
SCA 2001 conference, Edinburgh, September (2001).
Masalmeh, S.K. and Jing, X.D., Improved Characterization and Modelling of Carbonate Reservoirs for Predicting Waterflooding
Performance, SPE paper 11722_PP, presented at the International Petroleum Technology Conference held in Dubai, U.A.E., 4-6
December (2007b).
Melrose, J.C, Wettability as Related to Capillary Action in Porous Media, paper SPE 1085, SPEJ (1965), paper presented at SPE-AlChE
Joint Symposium on Wetting and Capillarity in Fluid Displacement Processes, 1959.
Mazzullo, S.J., and Chilingarian, G.V., 1992, Diagenesis and origin of porosity, in G.V. Chilingarian, S.J. Mazzullo, and H.H. Rieke, eds.,
Carbonate Reservoir Characterization: A Geologic-Engineering Analysis, Part I: Elsevier Publ. Co., Amsterdam, Developments in
Petroleum Science 30, p. 199-270.
Morgan, J.T. and Gordon, D.T.: Influence of Pore Geometry on Water-Oil Relative Permeability, Journal of Petroleum Technology
(1970), SPE 2588 presented at the SPE 44th Annual Meeting in Denver, Colo., September, 1969.
Morrow, N.R., Wettability and Its Effect on Oil Recovery, JPT, 1476-84, Dec. 1990.
Morrow, N.R. and Melrose, J.C.: Application of Capillary Pressure Measurements to the Determination of Connate Water Saturation,
Interfacial Phenomena in Petroleum Recovery, N.R. Morrow (ed.), Marcel Dekker, New York City (1991) 257-287.
Morrow N.R., Fischer H., Li Y., Mason, G., Ruth, D., Seth, S., Tong, Z., Unsal, E., Wickramathilaka, S., Wo, S., Yin, P.: (2008)
Fundamentals of reservoir surface energy as related to surface properties, wettability, capillary action, and oil recovery from fractured
reservoirs by spontaneous imbibition Final Technical Report, Submitted by: University of Wyoming Laramie, WY 82071 DOE Award
No.: DE-FC26- 03NT15408.
Pranter, M.J., Hirstius, C.B., Budd, D.A., 2005. Scales of lateral petrophysical heterogeneity in dolomite lithofacies as determined from
outcrop analogs: implications for 3-D reservoir modeling. AAPG Bulletin 89, 645662.
Torsaeter, O, A Comparative Study of Wettability Test Methods Based on Experimental Results from North Sea Reservoir Rocks, paper
SPE 18281 presented at the 1988 SPE Annual Technical Conference and Exhibition, Houston, Oct. 2-5, 1988.
Vijapurapu, C.S.: The Effects of Rock and Fluids Characteristics on Reservoir Wettability, A Thesis submitted to the Graduate Faculty
of the Louisiana State University and Agricultural and Mechanical College in partial fulfillment of the requirements for the degree of
Master of Science in Petroleum Engineering in the Department of Petroleum Engineering, December 2002
Wardlaw, N.C, The Effects of Pore Structure on Displacement Efficiency in Reservoir Rocks and in Glass Micromodels, paper SPE 8843
presented at the First Joint SPE/DOE Symposium on Enhanced Oil Recovery at Tulsa, Oklahoma, April 20-23, 1980.
Wardlaw, N.C, The Effects of Geometry, Wettability, Viscosity and Interfacial Tension on Trapping in Single Pore-Throat Pairs, JCPT,
May-June 1982, Montreal.
14 SPE 162265

Wellington, S.L. and Vinegar, H.J.(1987): X-Ray Computerized Tomography, J. Pet Tech 39 (8): 885-898. SPE-16983-PA.
Yuan, H.H.,Pore-Scale Heterogeneity from Mercury Porosimetry Data, SPEFE, 1991.6, 2, 233240; Trans., AIME, 291. SPE-14892-PA.

Table 1 Main rock properties that affect pore geometry in carbonates

Pore System (intergranular, intercrystalline, vuggy, fractured)
Pore (Throat) Size (relative presence and distribution)
Aspect Ratio (Pore body to pore throat)
Pore Configuration (surface area, tortuosity)
Particulate System (sorting, shape, packing)

Table 2 Bulk Density (BD) and Effective Atomic Number (Zeff) for certain pure minerals
Mineral BD (g/cc) Zeff
Calcite 2.71 15.71
Dolomite 2.87 13.74
Quartz 2.65 11.78

Table 3 Summary of the rock characterization scheme

Hg Micron T2 at
Sample Thin-section GD BD Porosity Perm saturation at the NMR
RRT Mineral Zeff
# description (g/cc) profile (frac) (mD) at 200psi peak of peak
(frac) PSD (ms)
Vugs and
intercrystalline meso
Non-
113 1 to macropores Dolomite 14.14 2.83 0.174 46 0.97 2.7 373
uniform
(loosely packed
texture)
intercrystalline micro Non-
114 1 Dolomite 14.39 2.81 0.195 21 0.95 1.53 295
to macropores uniform
Non-
Interparticle porosity, uniform
9 2 intraparticle porosity Calcite 15.59 2.70 (from 0.279 11 0.82 1.15 295
in micrite (grainstone) CT
image)
intraparticle porosity
15 2 in micrite (grainstone Calcite 15.66 2.70 Uniform 0.236 1.75 0.82 0.87 294
to rudstone)
Non-
22 3 Packstone Calcite 15.76 2.70 0.212 2.08 0.65 0.72 184
uniform
Non-
72 3 Packstone Calcite 15.67 2.70 0.214 2.06 0.65 0.60 184
uniform
Non-
4 4 Packstone Calcite 15.69 2.70 0.227 5.34 0.52 0.72 233
uniform
Packstone (pores Non-
6 4 Calcite 15.78 2.71 0.213 3.02 0.62 0.50 233
reduced by cement) uniform
Bioturbated
138 5 packstone(microporo Calcite 15.51 2.73 Uniform 0.184 0.98 0.23 0.42 145
sity between cement)
Bioturbated
packstone
139 5 Calcite 15.47 2.73 Uniform 0.184 1.01 0.23 0.42 145
(microporosity
between cement)

Table 4 Flooding rates and capillary numbers in the Kr experiments

RRT1 RRT5
Total flooding rate 30 ml/hr 20 ml/hr
-8 -8
Capillary number (Nc = v/!) 6.8x10 4.7x10
SPE 162265 15

(a) Thin-section microphotographs (b) CT images (c) CT-derived DE data

Figure 1 (a) Thin-section photomicrographs. (b) Color scale CT images at 500 microns/pixel (color scale bar in pixels is shown on
top of the images). The middle longitudinal CT slice is shown here. (c) X-ray CT-derived dual energy (DE) bulk density (BD) and
effective atomic number (Zeff) data along sample lengths. Two plugs are selected for each rock type. There are five different rock
types identified in the reservoir. Sample number and poroperm data are shown above each thin-section photomicrograph. Sample
number is also identified above each CT image.
16 SPE 162265

Figure 2 Two plots to the left, respectively: Mercury-derived drainage capillary pressure (Pc) and pore throat size distribution (PSD).
The Pc curves are shown with shaded squares at 200psi to emphasize the variation of rock properties within the different rock
types. The plot to the right: NMR T2 distributions on 100% brine saturated plugs. Reference lines are added on the distribution plots
to follow property changes with rock types. Two plugs are selected for each rock type. There are five different rock types identified
in the reservoir.
SPE 162265 17

Figure 3 Permeability versus porosity for the five different rock types

Figure 4 Plot to the left: Zeff versus BD for all rock types. Zeff values are indicated for pure Calcite, Dolomite and Quartz. The CT-
derived Zeff values show that RRT1 samples are mostly dolomitic while the other RRT samples are calcitic.
Plot to the right: Zeff versus GD for all rock types. The laboratory measured GD values for RRT1 samples are in the range of 2.8-
2.85g/cc. This confirms the dolomitic nature of these samples inferred from the X-ray CT-derived Zeff values. Similarly, the other
RRT samples give consistent Zeff and GD values confirming the calcitic nature of the samples.
18 SPE 162265

(a) PD Kr (b) Imb Kr (c) PD-Imb Kr Hysteresis

Figure 5 (a) Primary drainage Kr curves for all RRTs. (b) Imbibition Kr curves for all RRTs. (c) Hysteresis in the Kr curves between
PD and Imb for all the RRTs. Every plot row presents one rock type and the RRT number is indicated in the chart title. RRT 1 curves
gave different hysteresis trend from the other RRTs.
SPE 162265 19

Figure 6 Primary drainage and imbibition in-situ (water) saturation profiles (ISSM) along sample lengths for all RRT samples.
Saturation cycle (i.e. primary drainage or imbibition) together with sample number and RRT are indicated in title charts. These ISSM
curves are necessary to check quality of the relative permeability curves reported earlier. Two main features are normally detected
from such curves: heterogeneity and capillary end effect.
20 SPE 162265

Figure 7A (Semilog plot) Relative permeability curves for different rock types to assess effect of rock types on Kr behavior in
primary drainage and imbibition. a) Primary drainage Krw curves. b) Primary drainage Kro curves. c) Imbibition Krw curves. d)
Imbibition Kro curves.
SPE 162265 21

Figure 7B (Linear scale of Figure 7A) Relative permeability curves for different rock types to assess effect of rock types on Kr
behavior in primary drainage and imbibition. a) Primary drainage Krw curves. b) Primary drainage Kro curves. c) Imbibition Krw
curves. d) Imbibition Kro curves.
22 SPE 162265

Figure 8 Permeability versus porosity for the five different rock types in the Kr bounding curve program including three transition
zone (TZ) samples in the Kr scanning curve program with their corresponding CT images and sample numbers (SN). The CT images
are good representation of the poroperm variations seen in the TZ samples.
SPE 162265 23

Figure 9 Oil relative permeability imbibition scanning curves for the transition zone samples from RRT#3. Additional imbibition Kro
curve from RRT3 (i.e. SN 22) bounding curve is plotted to compare with wider range of Soi values. Each imbibition curve starts from
different initial oil saturation (Soi) to investigate fluid flow behavior in the reservoir transition zone.

Figure 10 Water relative permeability imbibition scanning curves for the transition zone samples from RRT#3.
24 SPE 162265

(a) Kro Scanning Curves (b) Krw Scanning Curves

Figure 11 (a) Linear scale plot for Kro scanning curves from Figure 9. (b) Linear scale plot for Krw scanning curves from Figure 10.

(a) Scanning curve (Soi=0.52frac.) with Bounding curves (b) Scanning curve (Soi=0.68frac.) with Bounding curves

Figure 12 (a) Scanning curve at Soi=0.52frac plotted with drainage and imbibition bounding curves. (b) Scanning curve at
Soi=0.68frac plotted with drainage and imbibition bounding curves.