Documente Academic
Documente Profesional
Documente Cultură
Summary
Prudhoe Bay is a mixed-wet reservoir where about half the oil recovery is attributable to gravity drainage. Gas/oil relative permeability
data show that gravity-drainage recovery efficiency is poorer for
more fine-grained sandstone and increases as the grain size increases.
Gravity-drainage efficiency also increases with connate-water saturation. Dependence of recovery efficiency on grain size is related to
changes in sorting. An effective grain size, defined by inverting the
Carman-Kozeny relation, provides a useful parameter for correlating
recovery efficiency. This estimate correlates well with visual estimates and direct measurements on disaggregated core. Grain size is
also found to be a more effective parameter for correlating trapped gas
than porosity, a common alternative. Lithology impacts trapped-gas
level with finer-grained, more poorly sorted rock having higher
trapped gas. Trapped gas decreases with increasing microporosity.
Because little gas is trapped in microporosity, a zero-slope generalization of the Land curve better represents trapped-gas data.
Introduction
As a result of the size and economic importance of Prudhoe Bay and
because of the variety of oil recovery methods operating in the reservoir, data for a variety of recovery mechanisms have been collected.
Much of the work to date on understanding relative permeability of
Prudhoe Bay has focused on water/oil1 largely because it is an EOR
target, but gas/oil is at least as important. The oil recovery by gravity
drainage constitutes approximately half the production and potential reserves of the field. Understanding gravity drainage is important for forecasting recovery efficiency in the future and in managing the relative contributions of gravity drainage, waterflooding,
and EOR recovery processes. Because macroscopic recovery efficiency is generally high for the gravity-drainage process, variations
in microscopic efficiency have an even larger impact on overall recovery efficiency than in waterflooding. Gas relative permeability
and trapped-gas measurements are important to predicting miscible
gas usage and recovery efficiency.
Endpoints and Film Drainage
Strictly speaking, a relative permeability endpoint, residual, or irreducible saturation is that saturation at which a phase becomes discontinuous and therefore stops flowing. This definition is meaningful in discussing nonwetting phases, such as gas, but is less
meaningful for wetting phases like oil in the gas/oil system. The
endpoint is controlled by the slow rate of film flow, which depends
on the number of pore volumes (PVs) of throughput or the time allowed for drainage. To understand recovery efficiency, it is more
useful to compare microscopic displacement efficiency,
Ed +1*So /(1*Swi ), at a given small oil relative permeability in a
centrifuge test or a large gas/oil relative permeability ratio in a displacement test. Analysis of gravity drainage at low, stable rates indicates that oil relative permeability is often low enough to control recovery and that gas relative permeability is irrelevant.2 To identify
the effects of relative permeability alone, it is useful to plot recovery
efficiency vs. dimensionless drainage time.2
t5
dkdS
ro
o
*1
DtDgk V
,
Dzfm o
. . . . . . . . . . . . . . . . . . . . . . . (1)
66
Fig. 2Displacement efficiency at krog +0.004 (top) and at a dimensionless time of 40 (bottom) vs. initial water saturation.
Each kind of data point is from a separate sample.
Fig. 3 shows oil relative permeability isoperms from the same set
of experiments. The isoperms show that oil relative permeability is
a function of liquid saturation at higher oil saturations or high oil relative permeability as long as water is not mobile (St0.3, Sl u0.6).
A constant liquid-saturation approximation is nearly adequate even
at relative permeability levels as low as 0.004. At low oil saturations
and sufficiently high water saturations, oil relative permeability depends only on oil saturation. This is consistent with the idea that oil
flow is dominated by the thickness of the oil film between water and
gas at low oil saturations; but when oil films are thick, oil relative
permeability is unaffected by the presence of water. Larger pores
contribute more to hydraulic conductance, according to a relationship of pore size to a power of at least two but probably three or
more. At large oil saturations, the contribution of small pores (i.e.,
those that may be occupied by oil or water depending on the initial
water saturation) is insignificant and the presence or absence of oil
in those pores has negligible impact on oil relative permeability. In
the limit of low water saturations, water is in the microporosity and
does not impact the relative permeability of oil but does impact its
saturation because oil is also in the microporosity. As a result, oil relative permeability depends mainly on total liquid saturation in the
limit of low water and high oil saturations. When the oil saturation
is very low, oil relative permeability becomes nearly independent of
water saturation because oil spreads along the gas/water interface
and in the limit of no oil saturation, the flow must be zero.
Lithology and Effective Grain Size
Studies in the literature indicate that permeability level impacts gas/oil
displacement behavior, with generally higher recovery efficiency at
higher permeabilities. Honoring such trends is an important factor in
obtaining accurate reservoir simulations.16,17 Empirical correlations of
gas/oil relative permeability usually include a permeability dependence, either explicitly or implicitly.16,18,19 Lower-permeability rocks
often have larger capillary pressure exponents, broader pore-size distributions, and poorer sorting, with associated higher Corey exponents
or more unfavorable liquid phase relative permeability curves. The
central idea is that pore-size distribution controls recovery efficiency,
with more broad pore-size distributions leading to lower recovery effiSPE Reservoir Engineering, February 1997
ciency. There are three major causes of low permeability: smaller grain
size, more consolidation or overgrowths, and poorer sorting. Although
grain size itself does not impact relative permeability, the other two
mechanisms lead to broader pore-size distributions20,21 and are often
associated with grain-size changes. These trends can be found in the
Prudhoe data, but the lithologic trends are more complex than those
typically reported in the literature. In particular, the dependence of
permeability and sorting on grain size is nonmonotonic.
Prudhoe Bay consists of sandstones and conglomerates with a
range of grain sizes. A simple way to distinguish lithology is to estimate an average grain size from comparitors, as is standard in geological description. Begg et al.22 did this for a large set of plugs from
Prudhoe Bay. Fig. 4 shows that both porosity and permeability have
a maximum with respect to grain size, which occurs for coarsegrained sands. Because porosity tends to decrease with poorer sorting and large variations in the degree of cementation are not expected for rocks in the same zone, the primary reason for these
variations is that sorting is poor for both very-fine- and very-coarsegrain rocks. Permeability decreases with decreasing porosity, decreasing grain size, and poorer sorting, so the observed permeabilities are consistent with this idea. Because reliable estimates of grain
size are not always available from special core analysis and grain
sizes can change dramatically from foot to foot, it is useful to estimate grain size from permeability and porosity.
The Carman-Kozeny equation estimates permeability from grain
size, porosity and tortuosity. For a uniform sand, the surface to vol-
Fig. 3Oil relative permeability isoperms: kro is a function of liquid saturation at higher So and a function of So at low So .
67
Fig. 5Grain size in phi units based on visual analysis vs. grain
size estimated by Carman-Kozeny relation from permeability and
porosity of the core sample (W). There is good agreement in the
range of coarse to very fine grain but poor agreement otherwise.
Fig. 4Dependence of porosity (top) and log permeability (bottom) on grain size, in phi units, estimated from visual inspection.
1 * f
f
fk , . . . . . . . (2)
and porosity of the plug and the Carman-Kozeny equation. Thus, direct measurements, visual estimates, and estimates based on the
Carman-Kozeny equation all agree.
Fig. 7 shows the dependence of sorting on grain size for disaggregated core. Finer-grained rocks have poorer sorting. The sorting coefficient, defined as the standard deviation in grain size measured
in phi units, increases nearly linearly with the grain size in phi units.
This trend is also apparent in visual inspection. Extrapolating with
respect to grain size in phi units indicates that perfect sorting occurs
around W+0. In this limit, the relative permeability behavior may
approach that of uniform sands. Because poorly sorted rock tends
to have lower recovery efficiency, one can expect that fine-grained
rocks will have poorer recoveries.
Fig. 8 shows recovery efficiency vs. sorting coefficient. As expected, the data show that poorly sorted sandstones have lower recovery efficiency. Recovery efficiency at an oil relative permeability of 0.004 and at a dimensionless time of 40 show the same trend.15
Other studies of Prudhoe Bay relative permeability support and expand on the simple trend with lithology given previously. There are
consistent trends found in studies on other wells. In particular,
SPE Reservoir Engineering, February 1997
Fig. 7Sorting vs. grain size. Data are from laser particle-size
measurements on disaggregated core.
coarse-grained and better-sorted material has consistently more favorable behavior when viewed on a liquid-saturation basis.
Fig. 9 shows displacement gas/oil relative permeability data on composites of different lithologies along with centrifuge oil relative permeability data taken on the plugs from the composites. The conglomeratic
sample has the most unfavorable behavior, the medium-grained sandstone the most favorable behavior. The fine-grained sandstone has less
favorable behavior than the medium-grained sandstone, and the very
fine-grained sandstone slightly less favorable behavior.
Differences in lithology reflect more than differences in microporosity level. Medium to pebbly sandstones have more favorable
relative permeability behavior than fine-grained sandstones both on
a liquid-saturation basis and a displacement-efficiency basis. Con-
Fig. 8Displacement efficiency at krog +0.004 (top) and at a dimensionless drainage time of 40 vs. sorting coefficient. Solid
line is RMA and dashed line is a least-squares fit .
glomeratic samples look more unfavorable on a liquid-saturation basis than on a displacement-efficiency basis largely because they contain more microporous chert, which in these experiments is saturated
with water and does not participate in the flow. While these general
descriptions are useful in exploring differences in behavior, the effective-grain-size concept helps to quantify the impacts and correlate the
data in a more meaningful way. Fig. 10 shows recovery efficiency as
a function of effective grain size for sandstones and conglomerates at
a dimensionless time of 40. The data are derived from centrifuge relative permeability experiments for all available data that contained
connate water. This figure shows that recovery efficiency is lower for
finer-grained samples. The data also show that, for large effective
grain sizes, the conglomerates and sandstones have essentially the
same recovery efficiency, but that conglomerates have a lower recovery efficiency for small effective grain sizes. The difference in recov-
g
gr
1)1cg2
1 * S max
gl
1 ) c g2S g * S gtS max
g
gr
c
c gl
. . . . . . . . . . . . . . . . . . . . . . . (3)
used to fit the data. This equation approaches the Corey equation18
at low gas saturations and has a continuously decreasing slope at
high gas saturations, consistent with the idea that gas enters larger
pores first and then successively smaller pores. The Corey exponent
portion models the way a nonwetting phase becomes connected
(i.e., the dependence of relative permeability on the shape and number of pores in the connected set of pores filled by gas). The latter
portion models the successively smaller contribution of smaller
pores to the relative permeability. The hysteresis behavior implicit
in the choice of reduced saturation is the same as Carlsons24; imbibition relative permeability is the same as secondary-drainage rel70
S max
g
max
1 ) 1S max
gr * 1 S g
1 1*S max
gr
, . . . . . . . . . . . . . . (4)
with S max
gr +0.255.
SPE Reservoir Engineering, February 1997
To understand lithology dependence of trapped gas better, an examination of general trends in the available database was undertaken. Most samples have a characteristic shape of initial/trapped saturation curve similar to the Prudhoe Bay data. To analyze the data
properly, it is important to differentiate between the dependence of
trapped gas on initial gas and the lithology dependence at a given
initial gas level. To do this, the trapped gas corresponding to a maximum gas saturation of unity can be used as a parameter to correlate
with petrophysical properties. Prudhoe data were put on this basis
by fitting a zero-slope Land curve through the data for each sample
available and determining the maximum trapped gas.
Fig. 14 shows that the maximum trapped-gas saturation decreases with increasing porosity level for sandstones. Although this
dependence has been recognized for some time,30,31 it is often overlooked. The Prudhoe data have a similar but weaker correlation with
porosity. Conglomeratic samples have a lower level of trapped gas
at a given porosity level than sandstones. Comparison of Prudhoe
with other data shows a large region of overlap, but Prudhoe has less
trapped gas both for moderate- and low-porosity levels and for conglomeratic samples.
An explanation for the dependence of trapped gas on porosity level
is that lower-porosity samples tend to have larger pore-body to porethroat aspect ratios and lower connectivity. Low porosity caused by
quartz overgrowth decreases pore-throat size more than pore-body
size and leads to higher aspect ratios.32 The major mechanism that
drives gas trapping is disconnection by the choke-off mechanism,
which occurs more readily for higher-aspect-ratio porous media.24,32
If consolidation proceeds to the point where throats are filled, the average number of pore throats connected to each pore body decreases
and there are fewer paths available for gas to escape and trapping increases.31 Another explanation is that low-porosity samples are poorly sorted, which leads to larger levels of trapped gas.
The influence of microporosity explains why Prudhoe Bay sandstones trap low levels of gas. Prudhoe Bay rock has a larger fraction
of microporosity than average sandstone and consequently less
trapped gas for a given porosity level. That microporosity does not
trap gas effectively is apparent in the shape of individual curves relating trapped-gas saturation to maximum gas saturation. Because pores
within microporosity are small, gas enters microporosity only for
larger maximum gas saturations. The change in trapped-gas saturation with increasing maximum gas saturation decreases with increasing maximum gas saturation, ultimately approaching zero, indicating
little trapping in microporosity. Moreover, trapped-gas measurements show a systematically decreasing trapped-gas saturation with
increasing connate-water saturation (Fig. 15). Because connate water
was established in the same way in these experiments, connate-water
saturation is a measure of the level of microporosity, consistent with
other evidence. One explanation for this lack of trapping is that the
SPE Reservoir Engineering, February 1997
Sgt +
S max
gr +
Sj +
Sl +
So +
Sorg +
Sw +
Swc +
Swi +
t+
Dt+
W+
Dz+
f+
D +
mj+
t+
sij +
for the better agreement with the database at higher porosities (e.g.,
see Fontainebleau samples,20 which are virtually microporosity
free) and for the weak trend with porosity, the impact of microporosity on trapped gas partially canceling the impact of porosity level.
Conglomeratic samples often have a larger fraction of pore space in
microporosity,8 accounting for their lower level of gas trapping at
a given porosity level.
Fig. 16 shows trapped gas vs. effective grain size calculated from
the Carman-Kozeny equation. For sandstone samples, there is a
trend of increasing trapped gas with smaller effective grain size,
consistent with the idea that poorer sorting leads to higher trappedgas levels. Moreover, the correlation is more statistically significant
with grain size than porosity.
Conclusions
1. Gravity-drainage oil recovery efficiency decreases with decreasing initial water saturation and poorer sorting. Fine-grained,
low-permeability sandstones tend to have low recovery efficiency
because they have poor sorting.
2. Effective grain size defined by inverting the Carman-Kozeny
relation provides a useful parameter for correlating recovery efficiency. This estimate of grain size correlates well with visual estimates and direct measurements on sandstones.
3. Gas relative permeability depends on rock texture with coarser-grained sandstones and conglomerates having a higher relative
permeability level at a given gas saturation. Much of this difference
between sandstones and conglomerates is because of the impact of
microporosity.
4. Trapped gas depends primarily on porosity or sorting and microporosity level. For sandstones, low porosity and poor sorting lead to
larger trapped-gas levels. Little gas is trapped in microporosity. Because conglomerates contain a larger fraction of microporosity than
sandstones, conglomerates trap less gas at a given porosity level.
Nomenclature
d+ grain diameter, L
Ed + displacement efficiency, fraction
h+ thickness, L
k+ absolute permeability, L2
krg + gas relative permeability
krj + Phase j relative permeability
krog + oil relative permeability
kV + vertical direction permeability, L2
Sg + gas saturation
72
trapped-gas saturation
maximum trapped gas
saturation of Phase j, fraction
liquid saturation
oil saturation
residual oil to gas
water saturation
connate-water saturation
initial water saturation
dimensionless recovery time
drainage time, t
effective grain size
distance from gas/oil contact, L
porosity, fraction of bulk volume
density difference between gas and oil
viscosity of Phase j+(F*t/L2)
dimensionless drainage time
interfacial tension between Phases i and j
Acknowledgments
I thank Arco Alaska Inc. and the working-interest owners of Prudhoe
Bay for permission to publish this paper. A wide variety of data in this
paper were measured by the various companies participating in the
Prudhoe Bay Unit. The results of this work would not be possible
without the dedicated and careful experimental work of many people
in these companies. The interpretations and conclusions presented in
this paper are those of the author and do not necessarily reflect the
opinions of all the Prudhoe Bay working-interest owners.
References
1. Jerauld, G.R. and Rathmell, J.J.: Wettability and Relative Permeability
of Prudhoe Bay: A Case Study In Mixed-Wet Reservoirs, SPERE (February 1997).
2. Richardson, J.G. and Blackwell, R.J.: Use of Simple Mathematical
Models for Predicting Reservoir Behavior, JPT (1971) 1145.
3. Chatzis, I., Kantzas, A., and Dullien, F.A.L.: On the Investigation of
Gravity Assisted Inert Gas Injection Using Micromodels, Long Berea
Sandstone Cores, and Computer Assisted Tomography, paper SPE
18289 presented at the 1988 SPE Annual Technical Conference and Exhibition, Houston, 25 October.
4. Oren, P.E., Billiote, J, and Pinczewski, W.V.: Mobilization of Waterflood Residual Oil by Gas Injection for Water-Wet Conditions, SPEFE
(March 1992) 70.
5. Dullien, F.A.L. et al.: The Effect of Wettability and Heterogeneities on
the Recovery of Waterflood Residual Oil with Low Pressure Inert Gas
Injection, Assisted by Gravity Drainage, paper presented at the 1991
European Symposium on IOR, Stavanger, 2123 May.
6. Erickson, J.W. and Sneider, R.M.: Structural and Hydrocarbon Histories of the Ivishak (Sadlerochit) Reservoir, Prudhoe Bay Field, SPERE
(February 1997).
7. Holstein, E.D. and Warner, H.R. Jr.: Overview of Water Saturation Determination for the Ivishak (Sadlerochit) Reservoir, Prudhoe Bay Field,
paper SPE 28573 presented at the 1994 SPE Annual Technical Conference and Exhibition, New Orleans, 2528 September.
8. Sneider, R.M. and Erickson, J.W.: Rock Types, Depositional History,
and Diagenetic Effects: Sadlerochit Reservoir, Prudhoe Bay Field,
SPERE (February 1997).
9. Narahara, G.M., Pozzi, A.L., and Blackshear, T.H. Jr.: Effect of Connate
Water on Gas/Oil Relative Permeabilities for Water-Wet and Mixed-Wet
Berea Rock, SPE Advanced Technology Series (July 1993) 114.
10. Vizika, O. and Lombard, J.M.: Wettability and Spreading: Two Key
Parameters in Oil Recovery With Three-Phase Gravity Drainage,
SPERE (February 1996) 54.
11. Delclaud, J., Rochon, J., and Nectoux, A.: Investigation of Gas/Oil
Relative Permeabilities: High-Permeability Oil Reservoir Application,
SPE 16966 presented at the 1987 Annual Technical Conference and Exhibition, Dallas, 2730 September.
12. Hagoort, J.: Oil Recovery by Gravity Drainage, SPEJ (June 1980) 139.
13. Dumore, J.M. and Schols, R.S.: Drainage Capillary Pressure Function
and the Influence of Connate Water, SPEJ (October 1974) 437.
14. Owens, W.W., Parrish, D.R., and Lamoreaux, W.E.: An Evaluation of
a Gas Drive Method for Determining Relative Permeability Relationships, Trans., AIME (1956) 207, 275.
SPE Reservoir Engineering, February 1997
30. Katz, D.L. et al.: 1966 How Water Displaces Gas From Porous Media,
Oil & Gas J. (Jan. 10), 5560.
31. Yuan, H.H.: The Influence of Pore Coordination on Petrophysical Parameters, paper SPE 10074 presented at the 1981 SPE Annual Technical Conference and Exhibition, Dallas, 57 October.
32. Wardlaw, N.C. and Cassan, J.P.: Estimation of Recovery Efficiency by
Visual Observation of Pore Systems in Reservoir Rocks, Bulletin of
Canadian Petroleum Geology (1978) 26, 572.
33. Blunt, M.J.: What Determines Residual Oil Saturation in Three-Phase
Flow?, paper SPE 27816 presented at the 1994 SPE/DOE Symposium
of Improved Oil Recovery, Tulsa, Oklahoma, 1720 April.
34. Schneider, F.N. and Owens, W.W.: Sandstone and Carbonate Two- and
Three-Phase Relative Permeability Characteristics, SPEJ (1970) 3, 75.
35. Swanson, B.F. and Hickman, W.B.: Application of Air-Mercury and
Oil-Water Capillary Pressure Data in the Study of Pore Structure and
Fluid Distribution, SPEJ (March 1966) 55.
1.0*
6.894 757
E)00 +mN/m
E)00 +kPa
SPERE
73