PTRL3001 Course Manual
PTRL3001 Course Manual
PTRL3001 Course Manual
RESERVOIR ENGINEERING II
by
Val Pinczewski
School of Petroleum Engineering
University of New South Wales
Sydney NSW 2052.
AUSTRALIA
March 2013
Prepared for
IMPORTANT NOTICE
2012 University of New South Wales. All rights are reserved.
This copy of the manual and accompanying software was prepared in accordance
with copyright laws for the sole use of students enrolled in a course at the University of New South Wales. It is illegal to reproduce any of this material or to
use it for any other purpose.
Contents
1 INTRODUCTION
1.2 Factors Influencing the Design of Water and Gas Flooding Projects
11
15
2.1.1
Interfacial Tension . . . . . . . . . . . . . . . . . . . . . .
15
2.1.2
Adhesion Tension . . . . . . . . . . . . . . . . . . . . . . .
15
2.1.3
17
2.1.4
18
2.1.5
Capillary Pressure . . . . . . . . . . . . . . . . . . . . . .
20
2.1.6
Laplace Equation . . . . . . . . . . . . . . . . . . . . . . .
22
2.1.7
23
2.1.8
29
29
35
2.1.9
38
38
41
41
42
2.2.1
Capillary-gravity equilibrium . . . . . . . . . . . . . . . .
44
2.2.2
45
2.2.3
51
55
57
59
60
2.6.1
61
2.6.2
Cory Curves . . . . . . . . . . . . . . . . . . . . . . . . . .
64
2.6.3
66
2.6.4
67
70
2.7.1
Darcys Law . . . . . . . . . . . . . . . . . . . . . . . . . .
71
2.7.2
72
2.7.3
Stones Model I . . . . . . . . . . . . . . . . . . . . . . . .
75
2.7.4
Stones Model II . . . . . . . . . . . . . . . . . . . . . . .
76
2.7.5
77
79
80
3 RECOVERY FACTORS
104
3.4.2
3.4.3
3.4.4
126
3.6.2
3.8.2
3.8.3
3.8.4
3.8.5
176
4.1.1
4.2.2
4.2.3
215
226
6.1.2
. . . . . . . . . . . . . . . 229
6.3
6.2.1
6.2.2
6.4.2
250
7.1.1
7.1.2
7.1.3
7.1.4
Chapter 1
INTRODUCTION
Hydrocarbon recovery from naturally occurring reservoirs depends on the reservoir drive mechanism and the complex interaction between rock and fluid in the
complex interconnected pore spaces of the reservoir rock as the hydrocarbon is
displaced from the pore space. The efficiency of the displacement process is central to the estimation of recovery factors and rates of recovery. The most common
displacement processes involve the displacement of oil by water or gas - water and
gas flooding.
Water and gas flooding technologies have been developed over the last 50 years or
so. Water and gas flooding are referred to as secondary recovery recovery process
and typically follow primary recovery operation which use the reservoirs natural
energy to produce oil. The purpose of water and gas flooding is to enhance and
accelerate oil recovery. The table below shows typical oil recovery factors for
different reservoir drive mechanisms.
Rock and Fluid Expansion
Solution gas drive
Gas cap drive
Water drive
Gravity drainage
Combination drive
5-30%
3-7%
20-40%
25-75%
up to 80%
30-60%
Water and gas injection is usually applied to improve recovery efficiencies in rock
and fluid expansion and solution gas drive reservoirs by maintaining or increasing
reservoir pressure. The presence of an oil-water contact or the identification of
an extensive aquifer in the vicinity of the field does not guarentee an effective
water drive. Faulting close to the field may effectively isolate the reservoir from
the aquifer.
6
Figure 1.1: Faulting which may isolate the reservoir from the aquifer
The main engineering tool used to design and evaluate water and gas flooding
projects is numerical reservoir simulation. In order for the engineer to effectively
use this tool it is necessary for him to be familiar with the basic mechanisms of
flooding processes. A key to understanding any displacement process is fractional
flow theory.
The design of a water or gas flood project requires the engineer to estimate the
likely recovery factor for the flood. The initial estimate is made using approximate
analytical methods based on fractional flow theory. This is later refined using
reservoir simulation studies. The overall recovery factor or recovery efficiency,
ER , for any oil recovery process may be expressed as the product of four individual
efficiencies:
ER = ED EM EA EV
where,
ED
EM
EA
EV
is
is
is
is
the
the
the
the
displacement efficiency
mobilization efficiency
areal sweep efficiency
vertical sweep efficiency
the volume of the reservoir swept or contacted by the injection fluid. The product
of the areal and vertical sweep efficiencies, EA EV , is referred to as the volumetric
sweep efficiency which represents the fraction of the reservoir volume swept or
contacted by the injection fluid. All of the above individual efficiencies depend,
to some extent, on the volume of displacing fluid injected into the reservoir cumulative injection - and the overall recovery factor increases with increasing
injection volume.
In evaluating flooding processes the engineer is working with information at two
length scales - macroscale or field scale and microscale or core plug scale. At the
macroscale the factors affecting oil recovery efficiency are;
1. mobility ratio
2. heterogeneity and reservoir geometry
3. gravity.
These are the key factors affecting overall or volumetric sweep efficiency - the
fraction of the reservoir volume swept by the injected fluid.
At the microscale the factors affecting oil recovery are;
1. wettability
2. fluid properties - fluid viscosities and densities
3. rock microstructure - pore geometry and connectivity
4. capillary pressure
5. relative permeability
6. initial water and gas saturations and residual oil saturation
The above factors determine the overall recovery from the swept zones of the
reservoir. Since relative permeabilities are are measured on core plugs at length
scales much smaller than the length scales associated with heterogeneity it is
necessary to up-scale laboratory measured relative permeabilities and residual
saturations. The up-scaled values are usually referred to as thickness averaged or
pseudo properties. Pseudo-relative permeabilities are a common example of such
properties.
1.1
1.2
The following factors are important in identifying the suitability of reservoirs for
water and gas flooding;
1. Reservoir geometry
2. Heterogeneity and permeability
3. Fluid properties
4. Fluid saturations and wettability
5. Reservoir depth
Reservoir geometry
Reservoir geometry will influence the location of wells. Reservoir dip angle and
permeability will largely determine the stability of the displacement front. If the
dip angle is high and the permeability is also high it may be possible to achieve
a gravity stable displacement leading to high oil recoveries.
Heterogeneity and permeability
Heterogeneity occurs at all scales in reservoirs. On the macroscale or reservoir
scales layering and faulting are the most serious heterogeneities. In fractured and
carbonate reservoirs containing significant amounts of micro-porosity, the macroporosity (fractures and vugs) may be effectively swept by the injection fluids but
the micro-porosity (matrix and microporosity) may be largely by-passed. If the
formation contains a thin layer with very high permeability commonlr referred
to as a thief zone, rapid channeling and bypassing will develop resulting in early
breakthrough of the injected fluid and low oil recovery.
Heterogeneity also determines the continuity of sands and pay zones. Sand continuity is one of the main factors determining the success of a flooding project.
Thin reservoirs with low permeability will exhibit low water injectivity requiring
large numbers of injection wells or excessively high injection pressures which may
exceed the the formation fracture pressure. THese reservoirs, particularly if they
have high dip angles, may be better candidates for gas flooding.
Fluid properties
Oil viscosity is the most important fluid property influencing the performance
of a flooding project. Oil viscosity is important in determining the shape of the
fractional flow curve and the mobility ratio for the displacement bot of which
determine overall oil recovery.
Fluid saturations and wettability
The higher the initial oil saturation the greater the chance of a successful flooding
project. The higher the initial oil saturation the larger the target for the flooding
project and the greater the liklyhood of recovering a significant volume of oil.
Reservoir wettability determines the distribution of the fluids in the pore spaces of
the reservoir rock. The distribution of fluids determines the relative permeabilities
for the rock and the residual oil saturation. Both have a profound effect on oil
recovery.
Reservoir depth
Reservoir depth has a significant impact on flooding project costs. Injection costs
increase with depth because injection pressures increase with depth. Lifting costs
also increase with depth. The cost of lifting oil from deep wellslimit maximum
economic water-oil and gas-oil ratios and reduce oil recovery.
In shallow reservoirs injection pressures may exceed formation fracture pressures.
10
1.3
The following factors are important in determining the optimum reservoir pressure or time to initiate waterflooding;
1. Oil viscosity
2. Free gas saturation
3. Well productivity
4. Costs
5. Initial oil saturation
Oil viscosity
Oil viscosity is lowest at the bubble point pressure. Lower oil viscosities result in
higher oil recoveries.
Initial gas saturation
In waterflooding projects initial gas saturations in the oil zone of the order of
10% result in residual oil saturations which are lower than when an initial gas
saturation is not present. The folowing figure shows a correlation presented by
Craig (1971) which accounts for the reduction in residual oil saturation as a
result of the presence of gas in the oil zone. The correlations are based on data
for water-wet rocks.
I a three phase oil, water and gas system under water-wet conditions gas is the
more strongly nonwetting phase. The presence of a gas saturation forces oil out
of the larger pores into smaller pores. This increases the connectivity of the oil
phase and makes it easier to recover by waterflooding where water being the most
strongly wetting phase tends to go to the smaller pores.
11
Figure 1.2: Effect of gas saturation on waterflood residual oil saturation (Society
of Petroleum Engineers)
12
Example
An oil reservoir is being evaluated for a secondary waterflood. Indications are
that the reservoir is water-wet. Oil-water relative permeability data shows that
the residual oil saturation is 35%. This would be the residual oil saturation for a
waterflood conducted at a pressure above the bubble point pressure where there
is no initial gas saturation in the oil zone. What would be the the residual oil
saturation if the waterflood was conducted at a pressure below the bubble point
pressure where the initial flowing gas saturation is 14% ?
Well productivity
High reservoir pressures result in higher well production rates which is desirable
in any flooding project.
Costs
Capital and operating costs are important in determining optimum flooding time.
These costs are closely related to reservoir pressure. Injection costs increase with
increasing reservoir pressure. Optimising the timing of a flooding project requires
an economic analysis of a range of development scenarios related to reservoir
pressure.
Initial oil saturation
This is the most crucial factor in determining the overall success of a secondary
flooding project. The reservoir must contain a sufficient volume of oil at the end
of the primary phase of production to make the secondary flood economically
viable. This requires a high initial oil saturstion for the project. High initial
oil saturations also mean high oil relative permeability and higher oil production
rates.
13
Chapter 2
WETTABILITY, CAPILLARY
PRESSURE AND RELATIVE
PERMEABILITY
Wettability controls the distribution and flow of fluids through the pore spaces
of a reservoir. It is a measure of the preference that the rock has for either the
water or hydrocarbon phase in a water-hydrocarbon-rock system. It is defined
as the tendency of one of the fluids to spread on or adhere to the solid surface.
Wettability affects everything.
In a water-wet system the water occupies the smaller pore spaces and contacts
most of the rock surfaces. As the oil is located in the larger pores and not in
contact with the rock surface waterflood oil recoveries may be high. In contrast,
in a oil-wet system the oil is located in the smaller pores and in contact with the
rock surface making oil recovery more difficult. The reverse may be true when
gasflooding.
Wettability effects also make it possible to reduce waterflood residual oil saturations by gasflooding and gasflood residual saturations by waterflooding. In a
water-wet system waterflood residual oil is located in the larger pore spaces. The
injection of a non-wetting phase such as gas allows some of this oil to be recovered
because the gas prefers to flow through the larger pore spaces where the residual
oil resides ie., the gas goes where the residual oil is.
The wettability of a reservoir is very difficult to determine or measure because
the original wettability of reservoir cores is often altered by contamination from
drilling and coring fluids. For this reason cores should always be cut with low
invasion drilling and coring bits.
14
2.1
2.1.1
WETTABILITY
PRESSURE
AND
CAPILLARY
Interfacial Tension
When two immiscible fluids are in contact, the fluids are separated by a well
defined interface which is a few molecular diameters thick. Molecules in the bulk
phases (away from the interface) are surrounded by the same molecules and the
resultant or net attractive force acting on a molecule is zero. At the interface
the resultant net forces on molecules is not zero because the molecules on both
sides of the interface are different. The inequality of the intermolecular forces on
each side of the interface gives rise to two very important characteristics of the
interface:
1. The non-zero forces on molecules at the interface places the interface in
tension. Since the interface between two fluids is flexible the tension at the
interface causes the interface to deform ie., the interface is curved.
2. Since the forces on the two sides of the interface are different and the interface has an area, the pressure (force per unit area) is different on each
side of the interface. The difference between the pressures on each side of
the interface is called the capillary pressure.
The difference between the net forces on molecules in the bulk fluid and the
interface means that work must be done to bring a molecule from the bulk fluid
to the interface, i.e., work is required to create new interfacial area. The work
is referred to as free interfacial surface energy or the free energy of the surface.
Since a system will always seek an equilibrium state which minimizes the free
energy of the system, the result of interfacial tension is to pull the interface into
the shape of a sphere - minimum surface area per unit volume.
The interfacial tension between two fluids (say fluids 1 and 2), 12, is the force
per unit length required to create new surface and is usually expressed in units
of dynes/cm. This is equivalent to surface energy per unit area (erg/cm2 ). Interfacial tension is a measure of specific surface energy.
2.1.2
Adhesion Tension
When two immiscible fluids (say oil and water) are brought into contact with a
solid surface, the equilibrium configuration is shown schematically in the attached
15
Figure 2.1: Simplified schematic of molecules at an interface between two immiscible fluids
16
figure. In analyzing this system we must consider a total of three surface tensions
or surface forces - water-oil (wo ), solid-water (sw ) and solid-oil (so ).
At equilibrium (a mechanically stable configuration)), the resultant horizontal
force at any point on the three-phase contact line is zero. Since the interfacial
tensions are forces per unit length, the force balance may be written as,
so = sw + wo cos
where is called the contact angle, or,
cos =
so sw
wo
2.1.3
AT
wo
The ratio of the areas of the solid wetted by water and oil depend on the value
of the contact angle, , which depends on the adhesion tension (see the attached
figure). The smaller the contact angle, the greater the area wetted by the water.
The wetting preference of the solid, referred to as the wettability of the solid, is
therefore related to the value of the contact angle.
When the contact angle is zero, the adhesion tension is AT = wo and the water
spreads as a thin film to cover the entire solid surface. The solid is said to be
completely water wet. When the contact angle is 180o , the adhesion tension is
AT = wo, and the water takes the form of a sphere which contacts the surface
at a single point and the solid is completely covered by oil. The solid is said to
be completely oil wet. When the contact angle is 90o , the adhesion tension is
AT = 0, and there is no tendency for the water or oil to spread. The solid has no
wetting preference is referred to as being of neutral wettability. When the contact
angle is < 90o , the solid is said to be water-wet. When the contact angle is >
90o , the solid is said to be oil-wet. Terms like weakly oil wet and weakly water
wet refer to contact angles close to 90o .
17
Figure 2.3: Interfacial tensions for hydrocarbons at their bubble point temperature
2.1.4
Since interfacial tensions arise from molecular interactions at the solid-fluid and
fluid-fluid interfaces, the chemical properties of the solid surface (mineralogy) and
the fluids (oil and water) will influence the magnitude of the contact angle and
therefore wettability. Polar organic compounds commonly found in all crude oils
(asphaltenes) may react with the solid surface (adsorb) and cover it to form a
preferentially oil wet surface. This requires that the crude oil comes into direct
contact with the solid surface. If this requires wetting water films to rupture,
it may produce mixed wettability conditions where some parts of the surface are
oil-wet (parts of the pore space where films are sufficiently thin to rupture) while
other parts remain water-wet.
Interfacially active compounds (naturally occurring surfactants) which tend to
accumulate at interfaces can lower water-oil interfacial tensions and affect wettability. Cation exchange between formation brine and water in clays can alter the
electrical forces at solid-fluid interfaces and therefore affect wettability by altering
the adhesion tension. Electrical forces are also affected by acid-base interactions
between the rock surface and crude oil. As a consequence, the chemistry and pH
of the reservoir brine have an important effect on wettability.
The salinity and pH of the connate and injected water can have a profound effect
on wettability. Chemicals and additives in drilling and coring fluids may contaminate core samples as a result of fluid invasion during the drilling and coring
18
operations and therefore change the wettability state of the core. Restoring reservoir wettability in contaminated core plugs is the major problem in special core
analysis. Failure to restore the reservoir wettability state will produce erroneous
laboratory test results.
Wettability of reservoir rocks
Reservoir rocks are commonly classified as being water-wet, oil-wet or
intermediate-wet. The attached figure shows the range of contact angles usually associated with the different wettability states. Strongly oil-wet states are
not normally encountered in actual reservoirs. Intermediate or mixed wettability
states are usually termed as dalmation wetting with the rock having both oil-wet
and water-wet patches.
The attached figure shows a schematic of how a reservoir rock may change wettability from a strongly water-wet state to dalmation wettability. The figure shows
the pore-scale distribution of oil and water as a function of capillary pressure. At
low capillary pressure (low oil saturation) the rock surface is completely water-wet
with a thick water film separating the oil from the rock surface. As capillary pressure increases (oil saturation increases) the film separating the oil from the rock
surface decreases in thickness. At high capillary pressure the water film becomes
sufficiently thin for intermolecular forces to become a factor in determining its
stability. If the film becomes unstable and ruptures oil directly contacts the rock
surface. Surface active molecules in the oil (asphaltenes) adsorbe on the exposed
rock surface and render it oil wet (lower row of figures). If the film remains stable
the rock remains strongly water wet (upper row of figures).
The above discussion illustrates the complexity of describing reservoir wettabil19
2.1.5
Capillary Pressure
Consider the case of a vertical capillary tube placed in a large beaker containing
water as shown in the attached figure. The glass is preferentially water-wet and
the adhesion tension is positive causing the water to be drawn up in the tube to
a height, h, above the free water level in the beaker.
If the air pressure at the free water level (h = 0) in the beaker is pa , the water
pressure in the capillary at the free water level is also pa - pressures in a static
system vary only with depth. The pressure on the water side of the interface in
the capillary at h = h, is
pw = pa w gh
where w is the density of water and g is the gravitational constant.
20
Figure 2.6: Mechanism for wettability change from strongly water-wet to oil-wet
(intermediate wettability) or dalmation wettability
The pressure on the air side of the interface in the capillary at h = h, is
pg = pa g gh
where g is the density of air.
The pressure difference across the interface is
pg pw = (w g )gh
Since g << w , this is approximately,
pg pw = w gh
The pressure change or jump across a curved (stressed) interface is called the
capillary pressure, Pc . It is written in general terms as,
Pc = pn pw = (w g )gh
where pn is the non-wetting fluid pressure and pw is the wetting fluid pressure.
Both pressures being measured at the height of the interface.
The attached figure shows different possible menisci configurations in a capillary
depending on the system wettability. The figure clearly illustrates the relationship
between contact angle, capillary pressure and the distribution of fluids in the tube.
21
2.1.6
Laplace Equation
The height of the capillary rise in a tube of radius R, is determined by the balance
between the adhesion force drawing the interface up and the gravitational force
pulling the interface down. Remembering that interfacial tension and adhesion
tension have units of force per unit length and that force is the product of area
and pressure, we can write the force balance at the interface in a capillary as,
(2R)AT = (R2 h)(w n )g
or
2AT
R
We have previously shown that AT = wn cos wn . Substituting for AT gives,
Pc =
Pc =
2wn cos wn
R
Pc
R
=
=
=
To convert capillary pressure to more convenient pressure units use the following
conversion factors:
dyne/cm2
dyne/cm2
dyne/cm2
1106
1.02106
1.45105
=
=
=
Pa
atm
psi
Laplaces equation is of central importance to the understanding of fluid distribution in porous media, two and three phase flow and to the understanding of
reservoir seals.
2.1.7
Figure 2.10: Schematic of a drainage displacement in a pore and threshold capillary pressure
24
2wn cos wn
rt
25
26
r (cm)
102
103
104
105
106
107
Water-oil system.
= 27 dyne/cm.
cos =0.8 ( = 37o )
dynes/cm 1.45 105 = psi.
Solution hints
Use the equation
Pc =
2cos
r
27
Figure 2.12: Schematic of the restored state method for measuring drainage capillary pressure in core plugs
28
2.1.8
The bundle of capillary tubes model for a porous medium can be used to illustrate
the relationship between drainage capillary pressure, pore size distribution and
saturation in a porous medium. Consider the bundle of capillary tubes model
for a reservoir core plug in a core holder as shown in the attached figure. The
capillaries are initially all full of water. Oil is injected at the inlet by increasing the
oil pressure in small steps. The outlet is fitted with a water-wet semi-permeable
membrane which allows water to freely leave the system but prevents oil from
leaving. Since water is hydraulically connected to the outlet, the water pressure is
always atmospheric and the capillary pressure is simply equal to the oil pressure
minus a constant (atmospheric pressure).
Since each tube in the bundle has a different radius and therefore a different
threshold pressure for the entry of the hydrocarbon phase (non-wetting fluid),
the largest capillary which has the lowest threshold capillary pressure will be
filled first.
When the largest capillary is full, the capillary pressure will increase, with no
further change in saturation, until the capillary pressure reaches the threshold
pressure for the next largest tube.
This continues until the smallest capillary is filled or until the capillary pressure
reaches a maximum value which is insufficient to enter the next smallest tube.
The displacement process then stops.
As additional tubes are filled, the saturation of the non-wetting fluid (hydrocarbon) increases and the water saturation decreases. This produces the stepped
capillary pressure-water saturation relationship shown in the figure.
2.1.9
Actual reservoir rocks have more complex pore structure than the bundle of capillary tubes model used above. Moreover, variations in mineralogy mean that the
contact angle or wettability may be different from point to point in the medium.
However, as for the simple capillary bundle the pore filling process is still controlled by capillary pressure. Laboratory measurements on actual rocks therefore
provide a mean description of rock wettability and capillary pressure. The three
most commonly used laboratory methods for measuring capillary pressure are;
29
Figure 2.13: Restored state or porous plate (membrane) method for measuring
drainage capillary pressure
30
Figure 2.14: Restored state method for measuring drainage capillary pressure at
reservoir conditions
1. de-saturation through a porous disk or membrane,
2. centrifugal method,
3. constant pressure mercury injection.
Desaturation through a porous disk or membrane (restored state
method)
This method simulates the drainage displacement process by which accumulating
hydrocarbon displaced water from the porous medium. The laboratory experiment to measure capillary pressure is similar to that discussed for the idealized
bundle of capillary tubes model discussed above. A schematic of the actual test
apparatus is shown in the attached figure.
31
The sample of reservoir rock is first cleaned, dried and saturated with wetting fluid (brine). Because brine composition will affect wettability and clay
swelling, it is best to use actual reservoir brine in the test.
(ii)
The brine saturated core is enclosed in a cell filled with hydrocarbon and
fitted with a brine wet semi-permeable membrane (a ceramic disk with very
small capillaries or a thin polymer membrane) which is impermeable to the
non-wetting phase (hydrocarbon). The semi-permeable disk or membrane
acts as a barrier to oil flow because the radius of the pores in the disk or
membrane is so small that the threshold capillary pressure for oil to enter
the water-wet pores is too high.
(iii)
(iv)
(v).
When the oil pressure has increased to a value just sufficient for oil to enter
the largest pore in the sample, the corresponding capillary pressure is called
the threshold capillary pressure.
(vi)
As the capillary pressure increases and more hydrocarbon enters the pore
space the water saturation in the core decreases. The actual water saturation in the core at any step is determined by material balance from
measurements of the displaced brine volume.
(viii)
If the test is conducted over a period of several weeks or so, the rock may
change wettability as water wetting films which prevent oil from directly
contacting the rock surface thin and rupture with increasing capillary pressure.
Typical drainage capillary pressure curves for actual reservoir rocks are shown in
the attached figure.
Imbibition displacements may also be measured by starting the test with an
oil saturated core, filling the chamber with brine and replacing the water-wet
membrane with an oil-wet membrane.
32
Figure 2.15: Drainage capillary pressure curves for different rock types
33
maintained until liquid ceases to flow. The cumulative liquid produced is recorded
for each speed. The test is terminated when an increase in speed fails to produce
additional fluid or when the maximum possible centrifuge speed is reached.
The centrifuge method produces similar results to the porous plate or semipermeable membrane method for similar wettability conditions. The major advantage of the method is that it is quick and capable of achieving much higher
capillary pressures than the porous plate method.
The centrifuge method can be used to measure both drainage and imbibition capillary pressures. For drainage displacements, the core is saturated with brine and
spun in a cup filled with oil. For imbibition displacements, the core is saturated
with oil and spun in a cup filled with brine.
A recently developed procedure allows imbibition capillary pressure tests to be
run at pseudo-reservoir conditions in the centrifuge. The core must be cut and
preserved under conditions that retain true wettability. Reservoir temperature is
maintained in the centrifuge and the core is saturated with water and gas-free
crude oil from the reservoir. Gas must be removed from the crude because the
centrifuge environment is at atmospheric pressure.
Typical drainage and imbibition capillary pressure curves are shown in the attached figure. The area between the drainage and imbibition capillary pressure
curves, which is called the hysteresis loop, results because oil forced into rock pores
cannot be so easily forced out. In fact part of the oil, the irreducible oil satura-
34
Figure 2.17: Typical drainage and imbibition capillary pressure curves showing
capillari hysterysis
tion, cannot be forced out no matter how high a water phase pressure is used.
A negative capillary pressure occurs when the wetting phase (water) pressure is
greater than the non-wetting phase pressure. In the centrifuge imbibition test,
only negative capillary pressures are possible. However, this test defines the low
oil saturation region of the curve which is important in determining waterflood
residual oil saturation and water drive recovery efficiency.
2.1.10
J(Sw )-function
From the previous analysis of the capillary bundle model for a porous medium
we derived a relationship between permeability and capillary radius,
k=
r2
8
2 cos
r
35
cos
q
Dividing both sides of the above equation by the RHS makes Pc a dimensionless
function which we call the J (Sw )-function (pronounced j-function),
Pc
cos
= J (Sw )
The J (Sw )-function is a universal dimensionless capillary pressure curve. Knowing this function allows us to calculate the capillary pressure curve for any porous
medium from a knowledge of k, , and cos . This actually works - sometimes!
Most times it fails because it implies a constant irreducible water saturation. Actual reservoir rocks are far more complex than a simple bundle of capillaries and
display a wide range of irreducible saturations (see the attached figure). However, the J (Sw )-function is sometimes useful in interpolating laboratory capillary
pressure data - particularly if the lab data is grouped on the basis of irreducible
water saturation.
36
Figure 2.19: J -function correlation for drainage capillary pressure data - most
data cannot be correlated using the simple J -function correlation
37
2.1.11
It is not convenient to measure capillary pressure in the laboratory at reservoir conditions using reservoir fluids. Laboratories usually use refined oil-water,
water-air or mercury-air fluid systems because working with these systems is
more convenient and less expensive than working with reservoir fluids at reservoir conditions. It is therefore necessary to convert laboratory measured capillary
pressure data to reservoir conditions.
Capillary pressure is given by,
Pc =
2 cos
r
2( cos )Lab
r
(Pc )Res =
2( cos )Res
r
( cos )Res
( cos )Lab
Note that the use of this equation requires that the laboratory actually measure
both ( cos )Lab and ( cos )Res . The latter requires a measurement at high
pressure and temperature and is therefore expensive.
2.1.12
Measurement of Wettability
2.
3.
Wettability can also be inferred from relative permeability tests. This will be
considered under the section of the course notes dealing with relative permeability.
2.1.13
The wettability of most water-oil-rock systems becomes progressively more waterwet as temperature increases. Although the trend of wettability with temperature
is based on laboratory measurements at relatively low temperatures, we may
expect hotter reservoirs to be more strongly water-wet than cooler reservoirs, all
other things being equal.
2.1.14
Reservoir Wettability
Intermediatewet
75o 105o
Oilwet
105o 180o
2 (7%)
2 (8%)
4 (7%)
15 (50%)
21 (84%)
36 (66%)
A < 62o
A > 133o ;R = A
12 (40%)
2 (8%)
14 (26%)
10 (33%)
16 (64%)
26 (47%)
8 (27%)
7 (28%)
15 (27%)
41
2.2
2.
Gravity gravity acting on oil and water (fluids of different density) produces a pressure differences between the oil and water phases which increases with height above the oil-water contact (OWC).
Equilibrium between capillary and gravity produces an initial distribution of fluids in the reservoir in which water saturation decreases with increasing height
above the OWC. Figure 1.21 shows a simple capillary tube bundle model for a
porous medium in which the water saturation decreases with depth above the
OWC which is defined as the depth at which the water saturation is 100%.
42
Figure 1.21 also shows the hydrostatic pressure-depth plot for the reservoir. Note
that the capillary pressure at the OWC is not zero. Since the OWC is in a porous
medium the water-oil interface is in a pore of finite size so the capillary is non-zero.
The capillary pressure at the contact corresponds to the the threshold capillary
for the porous medium as measured for the rock drainage capillary pressure curve
- the capillary pressure required to enter the largest pore in the system.
Since the capillary pressure at the OWC is not zero the oil and water gradient
lines intersect at a depth below the OWC. This depth is called the free water
level. The free water level has no physical significance but is a useful reference
depth for defining the oil and water gradient lines.
The major terms used to describe the vertical distribution of fluids in a reservoir
which is in static capillary-gravity equilibrium are summarised below:
Parameter
Symbol
Definition
FWL
Water-oil contact
WOC
Water-oil Transition
zone
Connate water
saturation
Swc
Irreducible water
saturation
Swi
Threshold capillary
pressure
Pct
43
2.2.1
Capillary-gravity equilibrium
Consider the case of a single vertical capillary as shown in the attached figure.
The capillary rise analysis is similar to that for the air-water system considered
earlier. For the oil-water system shown in the figure, the capillary pressure for
any point in the reservoir at vertical height, h, above the free-water level, Pc,h ,
2wo coswo
= po,h pw,h
r
where po,h and pw,h are the oil and water pressures at the interface in the capillary.
Pc,h =
Since Pc,0 = 0 at the free water level (h = 0), po,0 = pw,0 = p0 , and we can write
that,
pw,h = p0 w gh
and
po,h = p0 o gh
where w and o are the respective fluid densities and g is the gravitational
constant.
Substituting for the above pressures gives,
Pc,h = po,h pw,h = (p0 o gh) (p0 w gh)
Pc,h = (w o )gh
or,
h=
Pc,h
g
44
In field units,
h=
Pc,h
0.433
where,
Pc,h
h
=
=
=
The above equations show that the capillary rise in a capillary tube is directly
proportional to capillary pressure - the larger the capillary pressure (the smaller
the capillary) the greater the capillary rise.
2.2.2
The capillary pressure curve for a reservoir rock shows that water saturation
decreases with increasing capillary pressure.
Capillary pressure at the free water level is zero:
Pc F W L = po F W L pw F W L = 0
The oil and water pressures at the free water level are therefore equal,
po F W L = pw F W L = pF W L
The pressures in the oil and water phases above the free water level decrease
with height from the pressure pF W L by the respective gravitational heads,
po h = pF W L o gh
pw h = pF W L w gh
where h is the height above the free water level.
45
2.
3.
The following steps are used to compute the initial water saturation
distribution in a reservoir from laboratory capillary pressure data:
1.
Since laboratory capillary pressure data is uasually measured using nonreservoir fluids we first convert laboratory Pc data to reservoir conditions.
This requires laboratory measured values of interfacial tensions and contact
angles for both the laboratory and reservoir fluid systems.
2.
Log data, core data, formation testing and/or production testing are used
to establish the depth of the water-oil contact (DW OC ). Note that the depth
of the WOC is the highest depth at which the water saturation is 100% - the
depth or capillary pressure at which the oil can pass through the largest pore
throat in the rock. This capillary pressure is called the threshold capillary
pressure for the formation, Pct . It is read from the laboratory measured
capillary pressure curve and corresponds to the value of capillary pressure
at 100% water saturation.
3.
The threshold capillary pressure, Pct , is used to calculate the depth of the
free water level below the WOC, hPct ,
hPct =
Pct
(w o )
DF W L = DW OC + hPct
46
4.
The capillary pressure at every height in the reservoir between the top of
the reservoir and the free water level, h, is computed using,
hPc =
Pc
(w o )
and the corresponding water saturation is read from the capillary pressure
for the rock (converted to reservoir conditions).
5.
The height above the free water level is converted to depth using,
D = DF W L h
47
Pc (lab)(psi)
10.00
6.00
4.00
3.00
2.00
1.20
0.70
0.47
0.40
0.35
The laboratory tests were conducted with test fluids for which cos = 70
dynes/cm. The reservoir oil-brine system has cos = 28 dynes/cm with
o = 54.7 lb/ft3 and w = 63.8 lb/ft3.
Construct a graph showing the variation of water saturation with depth. What is
the average connate water saturation (initial water saturation) in the reservoir?.
[Answer: 0.30]
The following solution is presented in the spreadsheet [VSAT.mcd] and you
should carefully study this spreadsheet. In a spreadsheet calculation it is more
convenient to fit an analytical equation to the capillary pressure data and then
work with the equation rather than the tabular data itself. The equation which
we will use to fit drainage capillary pressure data is described in Section 1.3.
The equation requires three parameters to be estimated - Swc , Pct , and n. The
first two parameters can usually be estimated from the capillary pressure data
itself. The last parameter controls the curvature of the fitted curve. The curve
fit is performed using PCFIT.mcd located in the folder PcFit. You fit the curve
by entering values for the three parameters. I could have automated the curve
fitting procedure - but than there would be nothing for you to do!
48
Solution outline
1.
(cos)f ield
(cos)lab
28
= 0.4Pc lab
70
This gives the following field capillary pressure curve at reservoir conditions:
Sw
.186
.205
.236
.280
.378
.489
.642
.789
.878
1.000
2.
Pc f ield (psi)
4.0
2.4
1.6
1.2
0.8
0.48
0.28
0.19
0.16
0.14
144Pct
(w o )
144 0.14
= 2.22feet
(63.8 54.7)
This distance is usually small and in hydrostatic pressure calculations it is
usually assumed to be zero.
hPct =
3.
4.
144Pc
(w o )
144Pc
= 15.824Pc
(63.8 54.7)
49
Sw
.186
.205
.236
.280
.378
.489
.642
.789
.878
1.000
5.
Pc (field) (psi)
4.0
2.4
1.6
1.2
0.8
0.48
0.28
0.19
0.16
0.14
h( feet)
63
38
25
19
13
8
4
3
3
2
50
2.2.3
The attached figure shows a layered reservoir system with different capillary pressure curves for each layer. If the layers are all in hydraulic communication (no
flow barriers between layers), the oil and water pressures are continuous across
layers as is the difference between the fluid pressures - capillary pressure. Since
the water saturation in a layer follows the shape of the capillary pressure curve
for that layer, the water saturation-depth relationship is different in the different
layers and water saturation displays discontinuities at the boundaries between
layers. The more complex the layering, the greater the number of discontinuities
in the vertical water saturation distribution.
51
Depth
(ft.ss)
39984007
40074019
40194031
40314037
40374046
40464055
52
TABLE II
Irreducible Water Saturations
(From single point centrifuge capillary pressure tests at 150 psi)
Zone
Plug No.
1,3
2
6
5
1
2
3
4
Depth
(ft.ss)
4005
4012
4050
4038
Irreducible Water
Saturation (% PV)
15.0
21.0
24.0
31.0
TABLE III
Drainage Capillary Pressure Data
(Laboratory Measurements IFT=47.5 dynes/cm)
Saturation
(% PV)
100
90
80
70
60
50
40
30
20
Pc (psi)
Pc (psi)
Pc (psi)
Pc (psi)
Plug No.1 Plug No.2 Plug No.3 Plug No.4
0.25
0.38
0.46
0.88
0.27
0.41
0.49
0.93
0.29
0.44
0.53
1.00
0.32
0.49
0.59
1.10
0.36
0.55
0.66
1.24
0.42
0.64
0.78
1.45
0.51
0.81
1.00
2.00
0.69
1.19
1.63
1.32
Solution hint
Begin by fitting curves to each of the capillary pressure data sets (folder PcFit)
and then enter all the necessary data in VSATL.mcd. Note the equation for the
capillary pressure in a shale (Sw = 1 for all Pc ).
53
54
2.3
Laboratory capillary pressure data are usually limited to a small number of measured points because of the time and expense involved in making larger numbers
of measurements. We therefore need to interpolate between measured values to
obtain intermediate values.
Capillary pressure measurements are also made on a small number of plugs spanning the range of permeabilities and porosities characteristic of the major sand
bodies in the reservoir. If curves for different permeabilities and porosities are
required, they must be estimated or interpolated from the available capillary
pressure curves.
The J -function is often used to interpolate and extend available capillary pressure
data. A problem with the J -function is that connate water saturation is usually
different for rocks having different permeability and/or porosity.
Another method, known as the permeability method, is also commonly used. In
this method the capillary pressure curves for a number of plugs having different
permeabilities are read to make a plot of water saturation against the logarithm
of permeability for constant values of capillary pressure. The resulting plots may
usually be approximated by straight lines which have the following general form:
Sw = a log k + C for constant Pc
This plot can be used to estimate capillary pressure curves for plugs having
intermediate values of permeability.
We will use a combination of both of the above methods. First we fit the following
general curve to each set of measured capillary pressure data.
log Pc = m log SwD + log b
where, SwD is a normalized water saturation given by
SwD =
Sw Swc
1 Swc
Since the capillary pressure at Sw = 1 is the threshold capillary, Pct , we can write
the above equation as (b = Pct ),
log Pc = m log SwD + log Pct
It is convenient to re-define m as
m=
55
1
n
where n is a number related to the curvature of the capillary pressure curve. The
larger the value of n, the more steeply curved the capillary pressure curve.
With the above curve-fitting procedure, the measured capillary pressure curve for
each plug is represented by three values Swc , Pct and n. If a number of capillary
pressure curves are available for different plugs, we can plot Swc , Pct and n against
permeability or, from the J -function, against k/, and use the resulting plots to
estimate Swc , Pct and n for plugs having different permeability and porosity.
56
Figure 2.27: Residual oil formation by snap-off and capillary pressure hysterysis
2.4
The attached figure shows a drainage displacement (LHS) followed by an imbibition displacement (RHS). This simulates the accumulation and production phases
of a reservoir.
Drainage - a non-wetting phase displacing a wetting phase, say, oil displacing water (accumulation).
Imbibition - a wetting phase displacing a non- wetting phase, water displacing oil (production).
The sequence shows that snap-off of oil in pore throats results in a history dependent residual oil saturation or hysteresis. The creation of residual oil saturation
in this manner is referred to as capillary trapping because snap-off results from
a capillary driven instability of the oil filament in the throat. In actual reservoir
rock the pore structure is more chaotic than that shown in the figure and residual
oil is isolated in clusters as shown in the attached figure.
57
Figure 2.28: Residual oil blobs in a glass bead pack (top left hand side) and Berea
sandstone
Drainage processes are controlled by pore throats because a threshold pressure
must be overcome for the oil-water interface to enter a throat i.e., throats impede
the advance of the oil-water interface in drainage displacements. Imbibition processes are controlled by pore bodies because the wetting fluid prefers to be in the
smaller pore throats i.e., pore bodies impede the advancing water-oil interface.
Since the correlation between throat and pore sizes in real reservoir rocks is not
perfect i.e., not all large pores are connected by large throats, the paths taken by
oil and water are different in drainage and imbibition. This, together with the
snap-off mechanism discussed above, results in capillary pressure hysteresis and
different drainage and imbibition capillary pressure curves for the same rock.
58
2.5
The attached figure shows a trapped oil blob in a porous medium. The oil blob can
be mobilized (moved) by increasing the water velocity or flooding rate. Darcys
law for the flowing water phase is
v=
k P
w L
vw L
= Pw,U Pw,D
k
where Pw,U and Pw,D are the water upstream and downstream pressures, respectively.
The oil blob is trapped (static) and the pressure in the oil is everywhere constant
and equal to Po (assuming negligible gravity forces). Since the upstream part of
the blob reside in pore bodies, large pores where the capillary pressure is low, we
can neglect capillary pressure and approximate,
Pw,U = Po
59
vw
v
=
kow
w ow
2.6
RELATIVE PERMEABILITY
2.6.1
ko A Po
o L
qw =
kw A Pw
w L
where qo and qw are the oil and water rates, respectively. ko and kw are the effective permeabilities of the oil and water phases, respectively. These permeabilities
are measured in laboratory core flood experiments.
61
ko
k
kw
k
where k is the single fluid rock permeability which is known from previous measurements.
krw =
kkro Po
A
o
L
62
qw =
kkrw Pw
A
w
L
The groups, (kkro /o ) and (kkrw /w ), are called the phase mobilities.
A typical set of relative permeability curves is shown in the attached figure. The
relative permeability end-points are the irreducible (connate) water saturation
and the residual oil or non-wetting fluid saturation.
Relative permeability is;
- a strong function of fluid saturation.
- a function of rock properties ( and k) i.e., pore space micro-structure.
- a strong function of wettability.
- generally a weak function of fluid properties, however, large changes in
interfacial tensions can affect relative permeability.
A typical set of relative permeability curves for water-wet and oil-wet conditions
for the same rock sample are shown in the attached figure. Typically, water-wet
curves display an end-point water relative permeability of less than 0.2 and a
cross-over of the water and oil relative permeability curves at water saturations
greater than 50%. Oil-wet curves display an end-point water relative permeability
greater than 0.5 and a cross-over of water and oil relative permeability curves at
63
water saturations lower than 50%. The shapes of relative permeability curves
provide an indication of rock wettability.
The difference in end-point relative permeabilities between water-wet and oil-wet
conditions may be explained by the differences in pore-scale fluid distribution for
water-wet and oil-wet systems. As shown in the attached figure, in a water-wet
system oil occupies the larger pores and the interconnecting pore throats while
water resides in thin wetting films. When water displaces oil from a pore body,
oil snaps-off leaving a residual oil blob which effectively blocks the pore to water
flow. This results low water relative permeability at residual oil saturation - a low
warer end-point relative permeability. On the other hand, for an oil-wet system
there is no snap-off of oil when water displaces oil from a pore body. The residual
oil is in the form of thin wetting films which do not block the flow of water.
End-point water relative permeability is therefore high.
2.6.2
Cory Curves
Since relative permeability is an empirical concept, there exist no general theoretical expressions for the dependence of relative permeability on phase saturation,
wettability or fluid and rock properties. However, experimentally measured relative permeability data are generally smooth and can usually be represented by
the following analytical expressions,
o
kro = kro
So Sor
1 Sor Swi
no
nw
Sw Swi
1 Sor Swi
o
o
kro and krw are the respective fluid relative permeabilities when the other phase
is at its residual or irreducible saturation - end-point relative permeabilities. The
exponents no and nw are determined from a least-squares fit to the experimental
data.
o
krw = krw
The above equations are known as Cory curves and are often written using an
alternative nomenclature as,
1 Sw Sor
1 Sor Swi
eo
Sw Swi
1 Sor Swi
ew
kro = Eo
krw = Ew
We will use the above equations to smooth laboratory measured relative permeability data in a later exercise to estimate the displacement efficiency of a linear
waterflood.
64
65
2.6.3
In the limit of zero interfacial tension, a two fluid system reduces to;
- a mixture of two components in a single phase system (no interfaces exist),
- phase saturations correspond to the single phase mixture compositions (volume basis),
- since there is no capillary pressure, the trapped fluid saturations are zero,
For these conditions the effective relative permeabilities are straight line functions
of saturation (composition).
We therefore expect that as the interfacial tension decreases;
- relative permeability curves tend to straight line functions of saturation,
- both trapped or irreducible fluid saturations decrease,
66
2.6.4
67
68
69
2.7
THREE-PHASE FLOW
In the most general case we are interested in the simultaneous flow of three fluids,
oil - o, water - w and gas - g. Examples of three-phase flow abound - gas injection
into a transition zone, production of oil from a transition zone when the pressure
is below the bubble point pressure, and simultaneous water and gas coning are a
few commonly encountered examples.
For simplicity, we will begin by assuming that the phases are immiscible (not
a good assumption for most reservoirs). We will relax this assumption in the
next section. When the phases are immiscible we simply extend the previously
discussed two-phase development to three phases. The three fluid saturations
must sum to one.
So + Sw + Sg = 1
(2.1)
(2.2)
Pcgo = Pg Po
(2.3)
The third capillary pressure may be obtained from a simple combination of the
other two i.e.,
Pcgw = Pg Pw = Pcgo + Pcow
(2.4)
(2.5)
Laboratory measurements show that three-phase capillary pressures are well approximated by two-phase oil-water and gas-oil data,
Pcow = Pcow (Sw )
(2.6)
(2.7)
70
2.7.1
Darcys Law
Following the previously described treatment of two-phase flow we write for each
of the three phases,
vg =
kkrg
(Pg g gD)
g
(2.8)
vo =
kkro
(Po o gD)
o
(2.9)
kkrw
(Pw w gD)
w
(2.10)
vw =
in a water-oil system
in a gas-oil system
71
Conservation Equations
The conservation equations for each phase may be written as,
(g Sg )
t
(o So )
(o vo ) + qo =
t
(w Sw )
(w vw ) + qw =
t
(g vg ) + qg =
(2.11)
(2.12)
(2.13)
Combining the above equations with Darcys law yields the governing equations
for three-phase flow,
"
(2.14)
"
(2.15)
(2.16)
g kkrg
(g Sg )
(Pg g gD) + qg =
g
t
o kkro
(oSo )
(Po o gD) + qo =
o
t
"
w kkrw
(w Sw )
(Pw w gD) + qw =
w
t
2.7.2
Three-phase relative permeabilities are very difficult to measure with any degree
of certainty in laboratory tests. They are almost always estimated from measured
two-phase data. The estimates are largely empirical in nature and are all based on
the arguments originally proposed by Stone(1970). These arguments are outlined
below together with comments based on more recent fundamental studies of threephase relative permeability which tend to support Stones original arguments.
Models for determining three-phase relative permeabilities from two-phase experimental data are usually based on the following assumptions:
1.
water is the wetting phase and with respect to this wetting phase, the hydrocarbon phases (oil and gas) behaves as a lumped non-wetting phase.
Visualization experiments in glass micro models (see the papers at the end
of this section) show that this assumption is reasonable - see the attached
figure for both negative and positive spreading systems.
72
2.
gas is the non-wetting phase and with respect to this non-wetting phase, the
liquid phases (oil and water) behaves as a lumped wetting phase. Again,
this assumption appears reasonable on the basis of glass micro model experiments.
The above assumptions suggest that the behavior of the complex threephase system may be approximated by the behavior of two simpler pseudo
two-phase systems i.e., a hydrocarbon-water system and a gas-liquid system.
3.
The above assumptions imply that the wetting phase (water) in a three-phase
system will behave in a similar manner to the wetting phase (water) in a twophase water-oil system when both are at the same water saturation. Similarly, the
non-wetting phase (gas) in a three-phase system will behave in a similar manner
to the non-wetting phase (gas) in a two-phase gas-oil system when both are at the
same gas saturation. This means that the water and gas relative permeabilities
in the three-phase system (krw and krg ) at saturation Sw and Sg are the same as
the relative permeabilities of water in the two-phase water-oil system (krwo ) at
73
the same water saturation Sw and the same as the relative permeability for gas
in the two-phase gas-oil system (krgo ) at the same gas saturation Sg , respectively.
The treatment of the intermediate phase, oil, is a little more complex. Stone
(1970) argued that in a three-phase system at saturation Sw and Sg (So = 1
Sw Sg ) it should be more difficult for oil to move than in the corresponding
two-phase water-oil (So = 1 Sw ) and oil-gas (So = 1 Sg ) systems. This is
a reasonable expectation because the oil saturation in the three-phase system is
lower than the oil saturation in either of the two-phase systems at the same Sw
and Sg . If the oil saturation is lower, the oil phase occupies less of the pore-space
and fewer of the possible flow paths. Since it will be more difficult for oil to
move in a three-phase system than in the corresponding two-phase systems, the
relative permeability for oil in the three-phase system, kro , should be smaller than
the oil relative permeability in the two-phase water-oil system at the same water
saturation (krwo ) and the oil relative permeability in the two-phase gas-oil system
( krog ) at the same gas saturation. In the absence of anything better we could
simply take, kro = krow krog . This guarantees that the three-phase oil relative
permeability is always lower than either of the corresponding two-phase values.
This is essentially what Stones models and their various derivatives boil down
to.
74
Figure 2.39: Oil-water distribution in a strongly water-wet glass micromodel distribution of wetting (water) and non-wetting phases are similar to those in
three-phase flow
2.7.3
Stones Model I
Stone (1970), on the basis of statistical arguments, proposed that the oil saturation in a three-phase system may be estimated using the following equation,
kro = Sow g
where w and g are given by,
w =
krow
1 Sw
g =
krog
1 Sg
So Som
1 Swi Som
Sw =
Sw Swi
1 Swi Som
75
Figure 2.40: Two-phase oil-water and gas-oil relative permeability used with
Stones model to estimate three-phase relative permeabilities. Swmax is the maximum possible water saturation (Swmax = 1 Sorm Sgc ), Sgc is the critical gas
saturation (the saturation at which gas becomes mobile or the irreducible gas saturation), Sgmax is the maximum possible gas saturation (Sgmax = 1 Swc Sorm ).
Sg =
Sg
1 Swi Som
Som is the minimum oil saturation achieved in a three-phase system and must be
measured in laboratory tests.
Three-phase relative permeabilities computed with this model are shown in the
attached figure. They are in good qualitative agreement with limited measured
three-phase data.
2.7.4
Stones Model II
krow
where krow
is the oil relative permeability in the two phase oil-water system at
the irreducible water saturation.
Note that this equation does not require the specification of Som . This is actually a major disadvantage with tespect to Model I because it can result in the
76
2.7.5
(2.17)
If the end-point relative permeabilities are not equal to 1 then the models predict
oil phase relative permeabilities which only approximate the two-phase curves in
the limits of two-phase flow. Aziz and Settari have shown that this inconsistency
is removed if it is assumed that the two-phase gas-oil data is measured in the
presence of irreducible water.
The oil-water system at Swi is identical to the oil-gas system at Sg = 0. This is
equivalent to defining the absolute permeability as the effective permeability of
the oil phase at Swi . This amounts to defining,
krow (Swi ) = krog (SL = 1) = krocw
(2.18)
where,
SL = 1 Sg = So + Swi
With this modification Stones models may be written as,
Stones Model I
kro = Sow g
(2.19)
where,
w =
(2.20)
g =
(2.21)
Stones Model II
kro = krocw [(krow /krocw + krw )(krog /krocw + krg ) (krw + krg )]
(2.22)
2.8
The relatively recent research on three-phase flow has indicated that gravity
drainage of oil spreading films may contribute significantly to additional oil recovery from a previously water flooded reservoir. This has stimulated interest in
using immiscible gas flooding as an enhanced oil recovery process. In general,
waterflood residual oil saturation may be reduced (residual oil mobilized) by:
- Increasing viscous forces - a very limited option because very large increases
in injection rates are required to increase the capillary number for the flood
sufficiently to significantly reduce residual oil saturation.
- Reducing interfacial tension by injection of chemicals or using a miscible
injection gas (EOR processes).
- Introducing a third, non-wetting phase, gas.
Gas, being a non-wetting phase, will tend to;
- go to where the oil (also a non-wetting phase) is trapped.
- displace disconnected oil from the larger pore spaces which contain the bulk
of the trapped oil.
- reconnect isolated or residual oil blobs by the formation of oil spreading
films for systems where the interfacial tensions result in a positive value for
the oil-gas spreading coefficient.
The above mechanisms can result in oil recoveries which are significantly higher
than those achieved with waterflooding.
Gas flooding a core at waterflood
residual oil can reduce the residual oil saturation from approximately 30% to well
below 10%. The same is true when a core at gasflood residual oil saturation is
flooded with water.
79
Figure 2.42: Micro-CT units at the ANU/UNSW Digital Core Research Laboratories.
2.9
Digital Core Analysis (DCA) is a new core analysis technique which allows important petrophysical rock properties to be determined from small rock fragments
such as sidewall cores and drill cuttings which are unsuitable for conventional laboratory analysis. In addition to the obvious cost savings resulting from greatly
reduced core cutting costs DCA makes it possible to generate larger volumes
of data which reduces the level of uncertainty associated with conventional and
special core analysis data.
DCA is based on very high resolution three-dimensional imaging of the porespaces in reservoir rocks. Pore-scale physics coupled with the imaged pore-space
allow direct computation of a wide range of rock properties including permeability,
porosity, capillary pressure, relative permeability and residual saturations. The
following figure shows a typical X-ray micro-CT (computer tomography) unit.
The units shown were built by the Australian National University (ANU) for the
ANU/UNSW (University of New South Wales) Digital Core Research Group. The
units employ a cone-beam geometry and have a resolution down to one micron.
80
Figure 2.43: Comparison between sample size for micro-CT DCA and a conventional core plug.
81
Figure 2.45: Three-dimensional image of a high porosity carbonate outcrop sample (Mount Gambier).
82
Mineralogy
X-ray images measure the density of materials - the attenuation of the X-ray
in passing through a material increases with increasing densiry. This property
can be used to differentiate between different minerals in the sample where the
contrast in densities is high. The figure below shows a section through a sandstone
sample with dolomitic intrusions - the brighter the colour the higher the density.
The following figure shows fractures and partial mineralization in a coal sample.
83
84
Figure 2.48: Partitioning the imaged pore-space into pores and throats.
85
Figure 2.50: Network representations for a well sorted sandstone and a complex
reservoir carbonate.
Networks for different rock types are very different reflecting the strong differences
in rock microstructures. The following figure shows equivalent network representations for a well sorted sandstone and a highly heterogeneous reservoir carbonate
sample. The differences in geometry (pore and throat sizes) and topology (how
pores and throats are interconnected) are clearly evident. These differences may
be anticipated to have a profound effect on the petrophysical properties of the
individual rocks.
Large data sets from small rock fragments
Heterogeneity occurs at all length scales. A typical core plug measurement in a
laboratory produces one value of porosity and the corresponding permeability. If
the core plug was cut into a hundred smaller pieces and it was possible to make
measurements on the small pieces we would have a hundred values of porosity
and the corresponding permeability. Because of heterogeneity each porositypermeability value would be different and the overall result may be similar to
measurements on a hundred different core plugs cut from the same sand unit.
86
Although cutting a core plug into a hundred smaller pieces and making measurements on each piece separately is impractical using conventional laboratory
measurement methods it is possible using DCA. The above figure shows how a
small irregular rock fragment is imaged and the image then sub-sampled into
many smaller sub-sets. The porosity and permeability of each sub-set is then
computed.
The figure below shows a comparison between the porosity-permeability relationships determined for 55 core plugs using conventional laboratory tests and
the data obtained from fragments cut from 4 of the core plugs using DCA. The
correspondence between the two methods of analysis is very good.
The following figure shows porosity-permeability data for 10 different sand units
using DCA. The data was obtained using small fragments of rock cut from 10
core plugs one from each sand unit. This cannot be done using conventional core
measurement methods.
The porosity and permeability values for the DCA samples are calculated directly
from the images of the pore space. Porosity is computed by a simple voxel
count of solid-void voxels. Permeability is calculated by solving the Navier-Stokes
equation for fluid flow in the pore space. The following figure shows a small part
of the imaged volume used to make the calculation. The figure shows computed
streamlines through the interconnected pore space.
87
88
89
Figure 2.54: Computed streamlines for single phase flow in the pore space of a
sandstone rock fragment.
The following figure shows a comparison between image based permeabilities and
conventional permeability correlations. The imaged based data follows the same
correlations as those empirically determined for conventional core data.
90
91
Relative Permeability
Relative permeability may be calculated directly on the images, however, it is
usually more convenient to make this calculation on the network representation
using network models or pore-scale simulators. The following figure shows such a
simulator based on fundamental pore-scale physics for multiphase flow in porous
media. The simulator has a number of advantages over conventional SCAL the
major one being that it is possible to remove capillary end effects.
The simulator determins the interconnected channels occupied by each fluid as
a function of capillary pressure and simple hydraulic calculations determine the
corresponding relative permeabilities.
92
93
Another important advantage of the network model based calculations is that the
effect of displacement rate on residual oil saturation and relative permeability can
be incorporated into the calculations. The above figure shows the computed displacement patterns for an imbibition displacement as a function of displacement
rate measured in terms of the capillary bumber - the ratio of viscous to capillary
forces.
The major pore-scale mechanism for the creation of residual oil in imbibition
displacements at water-wet conditions is snap-off in pore throats. As capillary
pressure decreases and water saturation increases, wetting films in pore throats
swell and may snap-off or disconnect oil producing residual oil. The snap-off
process requires flow of wetting fluid through wetting films which is a slow process.
If the displacement rate is high the displacemeny front moves faster than film
flow can supplu wetting fluid to the displacement site and snap-off is suppressed
resulting in lower residuals and a more efficient displacement.
94
Figure 2.59: Laboratory measured residual saturation as a function of displacement rate for carbonate and sandstone rocks.
The above figure shows that for low displacement rates (low capillary numbers)
the displacement is very diffuse with residual oil created by snap-off spanning
the sample resultimg in low oil recoveries. As the displacement rate increases the
displacement becomes more frontal with a clearly defined displacement front and
the region ahead of the front affected by snap-off reducing in size. At the highest
rate the displacement is almosr purely frontal with a little snap-off. Oil recoveries
at these rates are higher than at lower displacement rates.
Also shown above is laboratory measured data for residual oil saturation as a
function of displacement rate for a variety of carbonate and sandstone rocks. The
data clearly shows that residual saturation decreases with increasing displacement
rate and that carbonates are far more displacement rate sensitive than sandstones.
95
Figure 2.60: Comparison between measured residual saturations for Berea sandstone as a function of displacement rate and DCA network predictions for the
same rock (green circles are measured data blue crosses and line are DCA predictions).
The residual saturation data for sandstones was mainly based on measurements
for Berea sandstone. The above figure shows a comparison between measured
residual saturations for Berea sandstone as a function of displacement rate and
DCA network predictions for a Berea sample which has similar porosity and
permeability which was imaged and the image subjected to DCA. The agreement
between the inaged based calculations and the measured data is very good.
96
Figure 2.61: Comparison between laboratory measured oil-water relative permeability curves for Berea sandstone and DCA network predictions for the same
rock (grey boxes are measured data and the coloured lines are DCA network
model predictions).
The above figure shows a comparison between laboratory measured oil-water relative permeability curves for Berea sandstone and DCA network predictions for the
same rock (grey boxes are measured data and the coloured lines are DCA network
model predictions). The data was measured at strongly water-wet conditions and
a displacement rate corresponding to a capillary number Ca = 3.5 106 . The
computed relative permeabilities used networks extracted from images of Berea
cores having similar porosity and permeability and the computations were made
at displacement rates significantly higher and lower than the measured rate. The
computations show that relative permeability for strongly wetting Berea is insensitive to rate for the range of displacement rates shown. This is thought to be
due to the low pore-throat aspect ratio (1.7) which tends to suppress snap-off.
97
Figure 2.62: Effect of pore-throat aspect ratio on network model computed residual oil saturations.
The above figure shows computed residual oil saturation as a function of displacement rate for the Berea network with an aspect ratio of 1.7 and the same
network with the aspect ratio increased to 6.8. The computations show that
residual saturations for networks with higher aspect ratio are much more sensitive to displacement rate than lower aspect ratio networks. The aspect ratio
of 6.8 is typical of the Mount Gambier carbonate reported by Tie and Morrow
(2005).
The following figure shows a comparison between measured and computed gaswater residual saturations as a function of porosity. The measured data was
reported by Hamon (2001) of Total. The computed residuals shown as the large
red circles are for three representative sandstone core fragments provided by Total.
The agreement between the measured and DCA computed residuals is excellent.
Capillary pressure
Capillary pressure is computed directly on the images using simple invasion algorithms based on LaPlaces equation relating capillary pressure to pore and throat
sizes. The following figure shows a comparison between MICP measured capillary
pressure and image computed capillary pressure for a sandstone rock fragnent.
The agreement between the measured and computed capillary pressure curves is
excellent.
98
Figure 2.64: Comparison between MICP measured capillary pressure and image
computed capillary pressure for a sandstone rock fragnent.
99
Figure 2.65: Comparison between MICP measured capillary pressure and image
computed capillary pressure for a sandstone rock fragnent.
100
Figure 2.66: Comparison between MICP measured capillary pressure and image
computed capillary pressure for a sandstone rock fragnent.
101
102
103
Chapter 3
RECOVERY FACTORS
3.1
Recovery factors are usually calculated using reservoir simulation models. However, the techniques described in this chapter are useful in determining first-order
estimates of recovery factors quickly. First-order estimates of this type should
always be determined prior to undertaking a reservoir simulation study. This
provides a quality check on the data which will be used in the simulation study
and identifies the most important factors affecting recovery. Knowledge of these
factors will influence the design of the simulation model and the overall course
of the simulation study. A useful rule-of-thumb - if you cannot or have not done
the calculations outlined in this chapter, you should not be doing a reservoir
simulation study.
3.2
RECOVERY FACTORS
The overall recovery factor, ER , for any oil recovery process may be expressed as
the product of four individual factors:
ER = ED EM EA EV
where,
ER
ED
EM
EA
EV
is
is
is
is
is
the
the
the
the
the
Figure 3.1: Oil-water relative permeability curves for water wet conditions
For the most accurate estimates the above factors should be determined by detailed reservoir simulation studies. However, realistic quick estimates of recovery
factors can be made by assigning values between 0 and 1 for each of the four
effeciency factors. For example, a reservoir for which each of the four efficiency
factors are 0.8 the overall recover factor is 0.41. In other words, only 41% of the
original oil in place would be recovered.
All of the above individual efficiencies depend, to some extent, on the volume of
displacing fluid injected into the reservoir - cumulative injection. In some texts
and courses the areal and vertical displacement efficiencies are usually grouped
and expressed as a single volumetric sweep efficiency, Ev = EA EV . This is
almost always the case when a numerical reservoir simulator is used to determine
recovery efficiency.
3.3
MOBILIZATION EFFICIENCY, EM
Soi Sorp
Soi
105
where Soi (= 1 Swi ) is the initial oil saturation, Swi is the irreducible water
saturation and Sorp , or simply Sor , is the residual oil saturation for the recovery
process. It is a measure of the maximum amount of oil that can be recovered by
a particular flooding process.
EM is defined as the maximum possible oil recovery for a process as a fraction
of the initial oil in-place. It represents the moveable oil volume in the reservoir.
Miscible displacement processes may have Sorp close to or equal to zero. For a
process with Sorp = 0, EM = 1.
The calculation of EM for an immiscible displacement only requires a knowledge
of Soi and Sorp . These are usually measured in special core analysis tests - relative
permeability and capillary pressure tests discussed in the previous sections. Soi
and Sorp are the end-point saturations on the relative permeability curves. The
most important point to remember is that a realistic estimate of EM requires
a representative sample of the reservoir rock and laboratory testing at restored
state conditions.
As discussed previously, the initial or connate water saturation depends on rock
microstructure and the prevailing reservoir capillary pressure. The residual oil
saturation saturation depends on trapping on the pore-scale or micro-scale. The
main mechanism responsible for the creation of residual oil is snap-off in pore
throats. This is a capillary pressure driven instability which is favoured by high
pore-throat aspect ratios.
106
3.4
DISPLACEMENT EFFICIENCY
107
Figure 3.4: Saturation profile in a one-dimensional immiscible displacement showing shock front and shock front saturation
3.4.1
In order to evaluate ED we must determine the average oil saturation in the swept
zone, So .
Consider the case of a constant rate linear (one-dimensional) water flood in a
homogeneous isotropic horizontal reservoir. Assume that porosity, permeability
and fluid viscosities are constant, the oil and water are incompressible and that
capillary pressure is sufficiently small for it to be neglected or that the displacement rate is sufficiently high for viscous forces to dominate capillary forces. These
assumptions allow us to study the essential features of immiscible displacements
in porous media. The displacement with these assumptions is usually referred to
as the Buckley-Leverett displacement problem.
Water is injected at x = 0 and production occurs at x = L. We can define a
dimensionless distance xD = x/L.
The cumulative water injection volume, Wi , at any time, t, is related to the
constant injection rate, qT , by,
Wi = qT t
Time and cumulative water injection volume are therefore both equivalent measures of the progress of the flood.
We can define a dimensionless cumulative water injection volume, WiD , by divid-
108
ing the cumulative water injection volume by the reservoir pore volume,
WiD =
Wi
AL
3.4.2
Buckley-Leverett Displacement
A material balance for the water phase over a volume element, Ax, where A is
the cross-sectional area and x is the distance along the length of the flood, gives
w qw |x w qw |x+x =
(Axw Sw )
t
109
Figure 3.6: Material balance over a volume element in a one-dimensional immiscible displacement
110
Sw
qw
= A
x
t
Sw
Sw
dx +
dt
x
t
Sw
Sw
dx +
dt
x
t
Sw
Sw dx
=
t
x dt
Sw dx
qw Sw
= A
Sw x
x dt
or,
1 qw
dx
=
dt
A Sw
The above equation gives the velocity with which a constant saturation, Sw , is
convected in the direction of the displacement. We write the equation as,
dx
1 qw
=
dt Sw
A Sw Sw
to emphasise that the velocity and partial derivative are evaluated at the specified
saturation Sw . In order to evaluate the velocity we need to determine the value
of the partial derivative of qw with respect to Sw at saturation Sw . To do this it
is convenient to introduce a new variable.
Fractional water flow, fw
We define the fractional flow of water, fw , as,
fw =
qw
qw + qo
where qo is the oil flow rate. The more commonly used oil-field term for fractional
flow of water is water-cut. For an incompressible system qw + qo is constant and
equal to the injection rate, qT . We can therefore write,
fw =
qw
qT
or,
qw = qT fw
and substituting for qw in the equation for the velocity of a constant saturation,
gives
dx
qT dfw
=
dt Sw
A dSw Sw
Since fw = fw (Sw ), dfw /dSw |Sw is independent of time, and we can integrate the
above equation to determine the distance traveled by a constant saturation as a
function of time.
112
qT dfw
dx =
A dSw Sw
dt
which gives,
x|Sw
qT t dfw
=
A dSw Sw
x|Sw
Wi dfw
=
A dSw Sw
The distance travelled by saturation Sw at time, t, is proportional to the cumulative water volume injected and the slope of the fractional flow curve at that
saturation.
If we divide both sides of the equation by L, the length of the reservoir, we have,
x
Wi dfw
=
L Sw
AL dSw Sw
qT t
= WiD
AL
(dimensionless distence)
xD |Sw = WiD
113
dfw
dSw Sw
qw
qw + qo
For horizontal flow we may write Darcys law for each fluid as,
qw =
kkrw A pw
w x
qo =
kkro A po
o x
kkrw A p
w x
qo =
kkro A p
o x
Substituting for qw and qw in the expression for fw and cancelling common terms
gives,
fw (Sw ) =
1
krw /w
=
)
krw /w + kro /o
1 + (w /o )(kro /krw
For typical relative permeability curves, the above equation results in an S-shaped
fractional flow curve as shown in the attached figure.
For a dipping reservoir gravity must be included and Darcys law is written as,
!
kkrw A
qw =
w
p
+ w g sin
x
kkro A
qo =
o
p
+ o g sin
x
114
Figure 3.7: Relative permeability curves and fractional flow curve for a viscosity
ratio of 2
where is the angle of inclination or the reservoir dip angle. The fractional flow
equation, after a little manipulation becomes,
fw =
The above equation shows that gravity modifies the course of a displacement by
modifying the shape of the fractional flow curve.
Defining the constant terms introduced by the gravity term as,
kA
(w o )g sin
qT o
we can rewrite the fractional flow curve as,
G=
fw (Sw ) =
1 Gkro
)
1 + (w /o )(kro /krw
For horizontal displacements ( = 0) or high rate displacements in dipping reservoirs where viscous forces dominate gravity forces, G = 0 and the equation reduces to the previously derived expression for fw . For viscosity dominated displacements the fractional curve is typically S-shaped as shown in the attached
figure. For gravity dominated displacements the fractional flow can take on negative values and result in very efficient displacements.
115
Figure 3.9: Fractional flow curve and slope of the fractional flow curve
116
Figure 3.10: Non-physical saturation profile (two saturation values at the same
location).
Welge construction
The equation,
xD |Sw = WiD
dfw
dSw Sw
can be used to construct the saturation profile for any cumulative injection volume
or time. This can be done by selecting an arbitrary number of specific saturations
in the range Swi Sw (1 Sor ). At t = 0 (WiD = 0) all the saturations are
at xD = 0. The above equation can be used to calculate the position of any
saturation Sw at time t or WiD = 0 by determining the slope of the fractional
flow curve for each saturation of interest. Figure 3.7 shows that the slope of
the fractional flow is a bell shaped curve. Since the velocity of a particular
saturation is proportional to the slope of the fractional curve at that saturation,
the velocitities of small and large water saturations are small with the velocity
having a maximum at intermediate saturations.
Figure 3.8 shows that the resulting saturation profile is not physically realistic
since it implies the existance of two values of saturation at the same location in
117
Using the previous equation written at the shock front saturation, Swf ,
xf D = WiD
dfw
dSw Swf
Sw Swi
dfw
= 1/
dSw Swf
This result shows that the average saturation behind the shock front, Sw , is
directly related to the velocity of the shock front saturation.
We can also evaluate Sw by integrating the Sw (= Sw (x))-profile,
1 Z xf
Sw dx
Sw =
xf 0
Since we have that,
x|Sw
we can write,
Wi dfw
=
A dSw Sw
dfw
Wi
d
dx =
A
dSw
and
Sw =
R xf
0
Sw d
dfw
dSw
dfw
dSw Swf
Sw =
dfw
Sw dS
w
or
Sw =
Sw =
dfw
Swf dS
w
Swf
i xf
0
R xf
0
dfw
dSw
dSw
dfw
dSw Swf
ixf
Rx
0f
0
dfw
dSw Swf
dfw
Sw dS
w
dfw
(1 Sor ) dS
w
dfw
(1Sor )
( fw |Swf 1)
dfw
dSw Swf
Since the slope of the fractional curve at (1 Sor ) is zero, we have that,
Sw =
dfw
Swf dS
w
Swf
( fw |Swf 1)
dfw
dSw Swf
119
1
dfw
=
dSw Swf
Sw Swi
Combining the above equations gives,
1 fw |Swf
dfw
1
=
=
dSw Swf
Sw Swi
Sw Swf
This result has a simple graphical interpretation which is known as the Welge
construction. If we construct the tangent to the fractional flow curve from the
point (Sw = Swi , fw = 0), as shown in Figure 3.10, then to satisfy the above
equation the point of tangency must have coodinates (Sw = Swf , fw = fwf ) and
the tangent line must extrapolate to the point (Sw = Sw , fw = 1). The Welge
construction therefore allows us to determine Sw , Swf , and fw |Swf .
The construction shows that the saturation distribution remains single valued
(physically meaningful) if all saturations smaller than the shock front saturation,
Swf , are removed. Further analysis shows that this is equivalent to making the
areas ahead and behind the shock front equal (Figure 3.9) i.e., satisfying the
overall mmaterial balance for the water phase.
The above equation may be rearranged to provide an expression for the average
water saturation behind the shock front,
Sw = Swf + 1 fw |Swf
dfw
1/
dSw Swf
Note that Sw is constant with a value determined by values of the Swf , fw |Swf
and dfw /dSw |Swf .
120
121
3.4.3
Np
AL
WiD =
Wi
AL
Since the saturation behind the shock front is constant, the displacement
efficiency for the swept part of the displacement (behind the shock front) is
also constant.
122
Figure 3.14: Fractional flow curves and tangent constructions for gasflooding of
a 5cp oil in a horizontal and highly dipping reservoir.
2. At breakthrough .
The shock front is at the production face, xf = L.
The saturation at breakthrough is the shock saturation Sw,bt = Swf .
The average saturation in the reservoir is the average saturation behind the
shock, Sw
A simple material balance gives,
Np,btD = Wi,btD = (Swf Swi )
Effect of gravity on displacement efficiency
The displacement effeciency at breakthrough or the average saturation in
the reservoir behind the shock front, Sw is a measure of the relative efficiency
of the displacement. The higher the value of Sw the better the displacement.
Since Sw is read directly from the tangent to the fractional flow curve it
follows that the shape of the fractional flow curve determines the efficiency
of the displacement. The attached figure shows the effect of gravity or dip
angle on the shape of the fractional flow curve for the gasflood of an oil
having a viscosity of 5cp. The large density difference between gas and oil
can result in a marked change in the shape of the fractional flow curve and
a large increase in the value of Sw .
123
3. After breakthrough .
After breakthrough x = L and the saturation and fractional flow at the exit
face (x = L), Swe and fwe , increase with cumulative injection (time). The
Welge equation may be re-written at the exit face, x = L, as,
Sw = Swe + 1 fw |Swe
dfw
1/
dSw Swe
dfw
xD = WiD
dSw Sw
dfw
1
=
dSw Swe
WiD
Soi So
Sw Swi
=
Soi Sor
1 Swi Sor
Figures 3.12 1nd 3.13 show typical fractional flow curves and the resulting displacement efficiencies. The figures illustrate the following important points,
- displacement efficiency increases with increasing injection volume (approaches 1 for very large injection volumes).
124
125
3.4.4
The use of fractional flow curves requires us to determine the slopes or derivatives of these curve. Laboratory measured relative permeability curves are rarely
sufficiently smooth for them to produce smooth fractional flow curves and well
behaved derivatives. The usual practice is to smooth relative permeability curves
before using them.
A good way of smoothing laboratory relative permeability data is to fit an analytical function to the data. A useful function of this type is the Corey equation
introduced in the previous chapter. This is which is written as,
kro = kro |Swi (So)eo
krw = krw |Sor (Sw )ew
where So and Sw are normalized saturations defined as,
So =
So Sor
1 Swi Sor
126
So =
Sw Swi
1 Swi Sor
kro |Swi and krw |Sor are the end-point relative permeabilities and eo and ew are
phase exponents. The equations are sometimes expressed as,
kro = Eo (So )
eo
krw = Ew (Sw )
ew
127
krw
0.000
0.002
0.009
0.020
0.033
0.051
0.075
0.100
0.132
0.170
0.208
0.251
0.300
kro
0.800
0.610
0.470
0.370
0.285
0.220
0.150
0.095
0.060
0.030
0.015
0.010
0
o = 50 cp
o = 5 cp
o = 0.5 cp
What is the mobility ratio for each displacement and what is the effect of mobility
ratio on displacement performance?
128
[ Answers:
at breakthrough
Case-1
Case-2
Case-3
Where there is no answer entered the reservoir is still producing clean oil i.e.,
before breakthrough.
Case-1
Case-2
Case-3
M=18.8
M=1.9
M=0.19 ]
Solution hints
This calculation, as well as those which follow, are tedious to do by hand and are
complex enough to require considerable time to program in Excel or any other
spread-sheet. My suggestion is that you become familiar with [BLD.mcd] as soon
as possible. This will also help you in following the consequences of the theory
which you should now have studied and understood.
Follow the following steps to solve the problem:
1. The first step is to fit analytical functions to the relative permeability data.
Smooth relative permeability curves result is smooth fractional flow curves
and this, in-turn, results in smooth saturation and recovery profiles. Smooth
relative permeability curves also produce faster convergence and smoother
results in reservoir simulators - it is well worth the effort to smooth laboratory relative permeability curves.
We will fit Cory-curves to the laboratory relative permeability data. Make
sure that you have read the section in your course notes describing these
129
curves. The parameters Eo and Ew are read from the end-points in the
relative permeability table. The parameters eo and ew control the curvature
of the curves and can be determined by eye in [BLD.mcd] or by least-squares
in [KRFIT-1.mcd] in the folder [Cory Fit]. The relative permeability data
is contained in the data file [KR1-1.prn]. Your answers may differ a little
from those in the course notes because your Cory fitting parameters may
be a little different.
2. Open [BLD.mcd] and input the basic reservoir and rate data. Double check
this to be sure that you are solving the correct problem. Enter the bestfit values for the Cory-curve parameters. [BLD.mcd] allows you to change
the Cory-curves without affecting the laboratory data. This is useful for
performing sensitivity studies. Note that all the calculations are performed
using the Cory-curves and not the actual laboratory data.
3. [BLD.mcd] performs the fractional flow and Welge tangent construction
automatically. Check to see how this is done. If you discover a quicker way
of doing this, let me know. The results of this calculation are summarized
under the heading Displacement Summary at Breakthrough. Note that the
fractional flow is equivalent to water-cut.
4. Check that the equations used to calculate recoveries after breakthrough
correspond to those in the course notes. The best way to read flood performance results after breakthrough is to scroll the table by entering different
values of Nx . Note that there are a total of 1000 points for the calculations.
For points before breakthrough the reservoir is producing clean oil - the
displacement front has not yet reached the outlet face. You can read the
value of the breakthrough point directly above the summary table.
5. Check the saturation profile calculation - it follows the procedure outlined
in the course notes to remove non-physical saturations by introducing the
shock front. Enter different values for pore volumes injected (EP V.inj ) and
see [BLD.mcd] calculate the resulting saturation profile. Look at the effect
of viscosity ratio on the shape of the saturation profile at similar pore
volumes injected.
6. Use [BLD.mcd] when revising the theory of immiscible displacement. It
should help you put the calculation and theory in context.
130
kv
o
g
o
g
=
=
=
=
=
=
0.35
500 md
0.5 cp
0.013 cp
0.75 gm/cm3
0.12 gm/cm3
kro
1
0.77
0.66
0.59
0.44
0.32
0.22
0.15
0.10
0.06
0.037
0.020
0.010
0.004
0.0012
0.0004
0.00002
0
131
krg
0
0
0
0.00003
0.0013
0.0062
0.0172
0.0359
0.0639
0.01
0.15
0.213
0.29
0.37
0.47
0.57
0.68
0.80
(a) Calculate the critical displacement velocity in ft/day and the critical displacement rate in RB/day using a basis of 1 acre for the displacement area. Calculate
the time for gas breakthrough the displacement efficiency at gas breakthrough for
the following flooding rates (expressed in terms of the critical flooding velocity;
vc , 0.5vc , 0.1vc and 0.01vc .
(b) Maintaining the same flooding rates (the same injection rates in RB/day) as
in (a) calculate the time for gas breakthrough the displacement efficiency at gas
breakthrough if the dip angle was 5o . What is the critical displacement rate for
this dip angle?
(c) Repeat the calculation in (b) for a horizontal reservoir where the dip anfle is
zero.
Note: For gas displacements well life effectively terminates at gas breakthrough.
The combination of very high gas mobility and well hydraulic problems when
producing gas with significant volumes of liquid result in the well gassing out
quickly after breakthrough.
[Answer: (a) qc =13849 RB/day-acre or 1.785 ft/day, (i) 15 days, 0.48, (ii) 35
days, 0.564, (iii) 238 days, 0.758, (iv) 7.7 years, 0.893.
(b) qc =1207 RB/day-acre or 0.156 ft/day, (i) 12 days, 0.394, (ii) 25 days, 0.402,
(iii) 147 days, 0.468, (iv) 6.4 years, 0.745.
(c) qc =0 STB/day-acre or 0 ft/day, (i) 12 days, 0.387, (ii) 24 days, 0.387, (iii)
121 days, 0.387, (iv) 3.3 years, 0.387.]
Your answers may be a little different depending on the values of the parameters
you have used to fit the relative permeability curves.
Solution hints
This problem is designed to show you the effect of flooding rate on displacement
efficiency when gravity effects are important. The derivation of the equation
used to calculate the critical displacement velocity is given in Section 2.7.2. Note
the changes in the shape of the fractional flow curve and the Welge tangent
construction as the flooding rate decreases and gravity begins to dominate viscous
forces.
When gas flooding a thin sand with little or zero dip the best strategy is to flood
at the highest rate possible. Can you tell me why - bonus points if you can.
132
Figure 3.17: Typical field waterflood development - areal sweep efficiency is proportional the developed area
3.5
The areal sweep efficiency is the fraction of total reservoir area swept by injected
fluid. EA depends on the fraction of the reservoir area which has been developed and on the sweep efficiencies for a typical development pattern. EA can be
increased by,
- increasing the developed area by drilling more wells infill or development
drilling and by use of horizontal wells to contact more of the reservoir.
- improving pattern sweep efficiency - using a more efficient development
pattern.
Areal sweep efficiency is primarily a function of;
- displacement mobility ratio, M, - defined later in the section,
- injected fluid volume - pattern sweep efficiency increases with increasing
cumulative injection volume,
- reservoir heterogeneity - increasing heterogeneity results in decreasing sweep
efficiency as a result of areal by-passing.
133
A number of commonly used flooding patterns are shown in the attached figure.
These are briefly described below.
Direct Line Drive
Alternating lines of injection and production wells are arranged so that injection
and production wells are located opposite each other. The injection pattern is
specified by the distance between wells and the distance between lines of injection
and production wells.
Staggered Line Drive
This pattern is similar to the direct line drive. Lines of injection and production
wells are arranged so that injection and production wells are no longer opposite
each other but displaced by half of the distance between wells.
Five-Spot
This is a special case of a staggered line drive where the distances between lines
of injection and production wells and the distances between wells is the same.
With this arrangement four production wells are always located at the corners of
a square with a production well at the centre.
Seven-Spot
Injection wells are located at the corners of a hexagon with a production well in
the centre.
Nine-Spot
This pattern is similar to a five-spot with an additional injection well at the middle
of each side of the square. For this pattern each production well is associated
with eight injectors.
134
135
3.6
Mobility Ratio
kkri
i
Oil is usually the displaced phase. Water is usually the displacing phase. However,
in a gas flood to recover oil, gas may be the displacing phase. In waterflooding
a gas reservoir gas is the displaced phase. For simplicity we will consider the
displaced phase to be oil and to allow the displacing phase to be either water or
gas. We will simply refer to the displacing phase as (i = d). The mobility ratio
136
is therefore be written as
(kr /)d
(kr /)o
M=
Sw
0.670
0.525
0.380
krw (Sw )
0.260
0.110
0.025
kro (Swi )
1.0
1.0
1.0
M
0.26
1.10
2.50
The above table shows that for immiscible displacements, the mobility ratio is
usually small or close to unity irrespective of the viscosity ratio. A 100 fold
increase in the viscosity ratio results in only a 10 fold increase in mobility ratio. This is usually referred to as the relative permeability effect in immiscible
displacements.
End-point relative permeability based mobilities do not require a fractional flow
displacement analysis for their evaluation. The end-point relative permeabilities
137
can be read directly from the relative permeability curves. For this reason endpoint relative permeability based mobilities are the most commonly used. Thay
are almost always used in correlations for sweep efficiencies.
3.6.1
The attached figure shows the progress of a water flood in a horizontal laboratory
scaled five-spot pattern displacement (Craig et al. JPT Jan. 1955). Caudle and
Witte (JPT Dec., 1959) developed a correlation for areal sweep efficiency for a
five-point pattern based on laboratory sand pack experiments similar to those
shown in the attached figure. The correlation, shown in the attached figure, is
based on data for both immiscible and miscible displacements. The correlation
is useful for estimating areal sweep efficiency as a function of mobility ratio and
the fractional flow of the displacing fluid.
The figures show that the low areal sweep efficiencies associated with the higher
mobility ratios are due to the development of viscous fingering which results in
increasing levels of by-passing with increasing mobility ratio. Viscous fingering
138
M EAbt
0.26 0.89
1.10 0.68
2.50 0.55
EA
fw =0.98
1.0
0.99
0.98
Figure 3.22: Areal sweep efficiency for a five spot pattern flood (Caudle and Witte
(1959))
For a miscible displacement the end-point relative permeabilities are unity (no
residual saturations - the relative permeabilities are straight line functions of saturation or concentration). The mobility ratio is equal to the viscosity ratio. The
following table shows that areal displacement efficiencies for miscible displacements may be considerably lower than for immiscible displacements for the same
viscosity ratios.
Pattern Areal Sweep Efficiencies for Miscible Displacements
o /w
1
10
100
M
1
10
100
EAbt
0.69
0.43
0.41
EA
fw =0.98
0.99
0.91
0.68
3.6.2
Willhite (1968) presents a useful correlation for Craigs (1955) graphical relationships for areal sweep efficiency. The relationship for areal sweep efficiency at
breakthrough, EA,bt =, is
EA,bt = 0.546 + 0.0317/M + 0.302/eM 0.0051
(3.1)
Figure 3.24: Craigs (1955) areal sweep efficiency data for mobility ratios in the
range 1-10 which are typical of what is normally encountered in waterflooding
inverse of mobility ratio as a function of for mobility ratios in the range 1-10
which are typical of what is normally encountered in waterflooding.
Willhite (1968) presents a useful correlation for Dyes et al. (1954) graphical
relationships. The areal sweep efficiency at breakthrough, EA,bt =, is given by,
EA,bt = 0.546 + 0.0317/M + 0.302/eM 0.0051
(3.2)
1
1+X
(3.3)
where,
X = [a1 ln(M + a2 ) + a3]fw + a4 ln(M + a5) + a6
142
(3.4)
a1
a2
a3
a4
a5
a6
The above algebraic correlations are useful in computer and spreadsheet calculations.
143
3.7
For the case of a homogeneous formation the low areal displacement efficiencies
at high mobility ratios are caused by viscous fingering (see the attached figures).
If the reservoir or sand being flooded is inclined gravity forces can dampen the
development of viscous fingers and increase the displacement sweep efficiency.
When this occurs the displacement is said to be gravity stabilized. It is therefore
important to be able to calculate the displacement rate at which gravity effects
dominate viscous forces - the forces responsible for viscous fingering.
Consider the attached figure which shows an immiscible displacement in a dipping
reservoir (dip angle ). We assume that capillary pressures are sufficiently small
for them to be neglected and that the displacing fluid is heavier than the displaced
fluid. The otherwise horizontal interface between the fluids is perturbed slightly
- the interface experiences a small disturbance or perturbation. The perturbation
length, x, is small and represents the site at which a potential viscous instability
may initiate.
Since fluid flows from high pressure to low pressure, the interface is stable if
P2 > P1 i.e., the perturbation on the interface moves in a direction opposite
to the overall displacement and therefore does not finger ahead of the front viscous fingers collapse back to the displacement front. The interface is unstable
144
if P2 < P1 i.e., the perturbation on the interface moves ahead of the displacement
front. If P2 = P1 the interface is neutrally stable because the perturbation neither
grows nor decays.
Darcys law may be written for each phase as,
e
kkr1
v1 =
1
dP1
+ 1 g sin
dx
kk e
v2 = r2
2
dP2
+ 2 g sin
dx
!
!
e
e
where kr1
and kr2
are the end-point relative permeabilities.
The above equations may be rearranged to give the pressure gradient in each
phase as,
dP1
v11
= e 1 g sin
dx
kkr1
v22
dP2
= e 2 g sin
dx
kkr2
If the pressure on the unperturbed interface is, P0 , we can write for the pressure
in each phase, at the head of the perturbation,
dP1
P1 = P0 +
dx
xf +x
dP2
P2 = P0 +
dx
xf +x
dP2
dx
dP1
x
dx
xf +x
x
xf +x
We define a critical displacement velocity, vc , corresponding to the neutrally stable condition P2 = P1 . At velocities above the critical rate P2 > P1 and the
perturbation grows producing an unstable displacement. At velocities lower or
equal to the critical rate small perturbations on the interface do not grow and
the displacement is said to be gravity stable. Grabity stable displacements result
in very high sweep efficiencies. At the critical displacement rate,
dP2
dx
=
xf +x
145
dP1
dx
xf +x
vc 2
vc 1
2g sin = e 1g sin
e
kkr2
kkr1
k(1 2 )g sin
e
e
(2 /kr2
1 /kr1
)
e
kkr1
(1 2 )g sin
1 (M 1)
M is the end-point mobility ratio. Note that fluid-1 is the displacing fluid usually water or gas when oil is the displaced fluid (fluid-2).
The significance of the critical velocity is that:
for v > vc
for v = vc
for v < vc
3.7.1
In crestal injection the injection well is located at the top of the structure. In
basal injection the injection is at the base of the structure. Gas being lighter
than oil is usually injected at the top of the structure. When this is done gravity
segregation assists in improving the effectiveness of the displacement. Water is
usually injected at the base of the structure for similar reasons.
If a displacement is carried out at a rate below the critical displacement rate,
i.e., the displacement is gravity stable, the areal sweep efficiency is close to one
(EA = 1). For a horizontal reservoir, zero dip angle, vc = 0 and the displacement
is always unstable with the extent of viscous fingering being determined only
by the mobility ratio. Note that the displacement rate also has an effect on
the displacement efficiency, (ED ) in a dipping reservoir. This was illustrated in
Problem 3.2.
146
Figure 3.26: Water and gas injection well locations in dipping reservoirs
[Answers:
(i) o = 50 cp, M=18.8, EAbt = 0.54; EA90%= 0.75; EA99%= 0.86
(ii) o = 5 cp, M=1.9, EA bt = 0.60; EA 90% = 0.91; EA 99% = 1
(iii) o = 0.5 cp, M=0.19, EA bt = 0.96; EA 90% = 0.96; EA 99% = 1 ]
147
=
=
=
=
=
0.9 gm/cc
0.65 gm/cc
5 cp
0.05 cp
0.18 cp
148
Figure 3.27: Vertical displacement flow regimes and vertical sweep efficiency at
breakthrough in two-dimensional linear floods
3.8
Vertical sweep efficiency is the fraction of the vertical section of the reservoir area
which is swept by injected fluid. Maximum vertical sweep efficiency is achieved
when the displacement front proceeds uniformly in a plane perpendicular to the
flooding direction.
Any factors which distort the displacement front reduce vertical sweep efficiency.
Any factors which restore a distorted displacement front increase vertical sweep
efficiency.
Areal sweep efficiency is affected by;
- gravity segregation,
- reservoir heterogeneity - reservoir stratification or vertical heterogeneity.
149
3.8.1
The distance between an injector and a producer in almost any flooding process
is large compared with the thickness of the reservoir undergoing the flood. This
means that in most practical cases gravity forces dominate viscous forces in all
parts of the reservoir other than the region in the immediate vicinity of the wells.
Dumore (1964) showed that for displacements in a homogeneous medium where
the injected fluid is lighter than the displaced fluid (gas-flooding)the injection
may occur at rates above the critical displacement rate, at the critical rate and
below the critical rate.
Rates above the critical rate (Region IV)
- the displacement is completely unstable and vertical sweep is dominated by
viscous fingering (viscous forces dominant). Although the displacement is
unstable, vertical sweep efficiencies may actually be quite high because the
detrimental effect of viscous fingers is reduced by the displacement velocity
150
being greater than the growth rate of viscous fingers ie., the displacement
front overtakes the viscous fingers resulting in a largely uniform displacement front.
High reservoir displacement rates can in reality only be achieved when fluid
viscosities are low, permeabilities are high and cross-sectional areas are
small (v = q/A). These conditions are usually met when flooding light oils
with gas in high permeability thin sands. Such floods are therefore usually
conducted at the highest practical displacement rate.
Intermediate rates (Region III)
- both viscous and gravity forces are important. Gravity forces distort the
displacement front decreasing sweep efficiency but still help in reducing the
detrimental effects of viscous fingering. The displacement is partially stable
with viscous finger growth dampened by gravity forces.
Rates below the critical rate (Region I)
- there are no viscous fingers and the displacement is completely dominated
by gravity forces. Gravity segregation causes the lighter injection fluid to
completely over-ride the displaced fluid. This can result in very low vertical
sweep efficiencies.
The three flow regimes discussed above are shown schematically in the attached
figure for the case of a horizontal miscible displacement. Vertical displacement
efficiency is clearly determined by the relative magnitudes of the gravitational
and viscous forces.
The gravitational (vertical) force is given by,
PV = gh
where is the density difference between the fluids, g is the acceleration due to
gravity and h is the sand thickness.
The viscous (horizontal ) force is given by Darcys law as,
PH =
uo L
k
where u is the displacement velocity, o is the oil phase viscosity, L is the distance
between injector and producer and k is the sand permeability.
The viscous (horizontal ) to gravity (vertical) force ratio is given by,
PH
uo L
=
PV
kgh
151
Figure 3.29: Vertical sweep efficiency at breakthrough for horizontal linear floods
(u in bbl/day-ft2 ; o in cp; L in feet; k in md; in gm/cc; h in feet)
Horizontal beds
Craig (1980)studied scaled laboratory displacements in homogeous porous media
and developed a correlation between vertical sweep efficiency at breakthrough and
the horizontal to vertical force ratio. The correlation is shown in the following
figure. The correlation shows that sweep efficiency at breakthrough is a strong
function of the mobility ratio for the displacement.
The attached figures shows vertical displacement efficiency at breakthrough for
linear horizontal flow in a uniform system. These figures are useful in estimating
the breakthrough vertical displacement efficiency for horizontal displacements
in uniform porous media. The figures shows that Unfavorable mobility ratio
accelerates the formation of a gravity tongue which results in by-passing and low
sweep efficiency.
152
0.5
50
200
1000
0.50
0.75
0.04
1.20
ft/D
ft
md
ft
g/cc
g/cc
cp
cp
Solution:
Convert displacement rate to the required units:
u
=
0.5
=
0.5
= 0.5/5.615
=
0.089
ft/D
ft3 /ft2 D
bbl/ft2 D
bbl/ft2 D
o
1.2
=
= 30
CO2
0.04
Figure 3.30: Stable (a) and (b) and unstable (c) displacements in a dipping
reservoir. (a) stable G > M 1, M > 1, < ; (b) stable G > M 1, M < 1,
> ; (c) unstable G < M 1.
Dipping beds
As seen in the previous section the displacement in a dipping reservoir or reservoir
unit is stabilized by gravity if the flooding rate is below the critical rate, qc given
by,
qc =
The following figure shows that although the interface between the displacing and
displaced fluids is stable the interface itself is not necessarily flat, however, it is
stable with slope , given by the following equation,
dx
= tan =
dy
M 1G
tan
G
where,
154
Figure 3.31: Typical porosity and permeability variations with depth in a waterflood section
G=
3.8.2
The permeability of reservoir rocks can vary by many orders of magnitude over
short vertical distances. In contrast, porosity may change by only a few percentage points over the same scale. Reservoirs are usually considered to be homogeneous with respect to porosity but heterogeneous with respect to vertical
variations in permeability.
155
3.8.3
Measures of Heterogeneity
n
X
hj
is
The average porosity, ,
n
1X
=
j hj
h 1
To allow a simple analysis we assume that the flow in each layer is horizontal ie.,
there is no flow across layers (no cross-flow), the displacement in each layer is
piston-like (zero capillary pressure) and the displacing and displaced fluids have
equal viscosity.
Since there is no cross-flow the arrangement of the layers is unimportant and we
can rearrange the layers in order of decreasing permeability. The flow in each
layer is given by Darcys law,
qj =
kj hj W p
where W is the width of the system. The flow in each layer is therefore proportional to
qj (kj hj )
Total flow rate, q, is proportional to
q (k h)
157
158
Figure 3.34: Waterflooding in the presence of heterogeneity and gravity segregation effects
Figure 3.35: Vertical layering with layers arranged in order of increasing permeability
159
Figure 3.36: Flow-storage capacity plots for different formations (Paul et al. 1982)
n
n
1X
1 X
qj =
kj hj
q 1
hk 1
160
F dC 1
The integral in the above equation is the area under the flow-storage capacity
(F C) curve. The total area of the F C plane is unity. With the above
definition the Lorenz coefficient for a homogeneous reservoir is 0, and for an
infinitely heterogeneous reservoir it is 1.
Solution hint
Count squares to estimate the area under the F C curve (I did it this way) or
read-off the points, enter into a spreadsheet and calculate the area numerically
or by fitting a curve and integrating.
161
C=0.5
C=0.841
(F 0 )C=0.5
k)k
n hn
dF
F
Fn Fn1
(1/h
kn
=
=
=
=
=
n
dC n
C n Cn Cn1
k
(1/h)h
where it is assumed that the porosity for all the layers is approximately constant.
Using this result we can write VDP as,
VDP =
k50% k84.1%
k50%
2.
3.
4.
From the log-normal probability plot determine k50% (the log mean permeability) and k84.1% permeability values corresponding to the indicated
probability levels.
5.
k50% k84.1%
k50%
The attached figure shows a typical log permeability-probability plot and the
calculation of the permeability variation coefficient.
163
164
165
Part-2
Another sand in the same reservoir was also examined in a similar manner. The
data is given below. What can you say about this sand?
INTERVAL
PERMEABILITY
NUMBER (ft)
(md)
1
162
2
126
3
70
4
79
5
380
6
212
7
714
8
28
9
31
10
36
11
63
12
48
13
45
14
63
15
32
[Answer:VDP =0.472 - heterogeneous and need to use 4 sub-sands]
solution hints
Go to the problem folder, open [VDP.mcd], read-in the correct data file and see
the answer.
3.8.4
This is the simplest system with all layers acting independently and each layer
is continuous from injector to producer. Gravity segregation effects within layers
is neglected. The simplicity makes the system amenable to basic analysis. The
most sophisticated method for estimating vertical sweep efficiency is based on the
work of Dykstra and Parsons. Their work is based on more than 200 flood tests
on more than 40 reservoir core samples combined with simulation studies.
166
Dykstra-Parsons Method
Dykstra and Parsons performed laboratory linear floods in horizontal layered
systems. They developed graphical correlations for vertical sweep efficiency as a
function of vertical permeability variation coefficient VDP , mobility ratio M and
water-oil ratio W OR. The correlations were presented for different W OR ranging
from 0.1-100.The attached figures show their correlation for terminal water-cuts
of 50%, 83%, 96% and 99%. These correlations may be used to estimate EV .
de Souza and Brigham (1981) developed a mathematical correlation for the
Dykstra-Parsons graphical data valid for mobility ratios M = 0 10 and
VDP = 0.3 0.8. The correlation is given by,
Y =
(W OR + 0.4)(18.948 2.499VDP )
(M 0.8094VDP + 1.137)10X
(3.6)
where,
2
X = 1.6543VDP
+ 0.935VDP 0.6891
(3.7)
a2 = 0.181
a4 = 0.00462
a6 = 0.000277
167
a5
+ a6 Y
ln(Y )
(3.8)
Figure 3.40: Vertical sweep efficiency as a function of Dykstra-Parsons coefficient of vertical heterogeneity - OWR=1 fw =50%, OWR=5 fw =83%, OWR=25
fw =96%, OWR=100 fw =99%.
168
Figure 3.41: Vertical sweep efficiency as a function of the de Souza and Brigham
(1981) correlating parameter Y
169
[Answers:
(i) o = 50 cp, M=18.8, EV bt = 0.02, EV 80%= 0.13, EV 90%= 0.26, EV 99%= 0.72
(ii) o = 5 cp, M=1.9, EV bt = 0.53, EV 80% = 0.53, EV 90% = 0.67, EV 99% = 0.97
(iii) o = 0.5 cp, M=0.19, EV bt = 0.86, EV 80% = -, EV 90% = 0.86, EV 99% = 0.99 ]
3.8.5
As for areal sweep efficiency, gravity can overcome the detrimental effects of vertical heterogeneity and result in a highly efficient plug-type displacement. This
occurs when the flood is dominated by gravity i.e., when the displacement velocity is below the critical velocity. As shown previously, the critical velocity
equation requires the permeability in the direction of the displacement. For a
heterogeneous or layered reservoir the displacement rate in each layer must be
below the critical rate for the layer. When a flood is occurs at rates below the
critical rate, both EA and EV are close to 1.
If the overall flooding rate is q, a good estimate of the rate in each layer, qi , is
obtained by assuming that the total rate is allocated to each layer according to
the layer permeability-thickness product ki hi i.e.,
ki hi
qi = P
q
ki hi
(3.9)
qi
Ai
(3.10)
3.9
This section contains two problems which will allow you to use the methods
discussed in this chapter to obtain first-order estimates of overall recovery factors.
These estimates are often used in screening studies of possible field developments
and in preliminary studies prior to field reservoir simulation studies. If you cannot
do these example problems you should not be doing, or contemplate doing, a
reservoir simulation study.
Problem 3.8 - Comparing recovery factors for waterflooding and polymer flooding [BLD.mcd]
A waterflood is being conducted in a flat (horizontal), layered reservoir on a
direct line drive. We are requested to examine the feasibility of converting the
waterflood to a polymer flood. We will estimate the recoveries in the five-spot
to assess the potential of the polymer flood by comparing the recovery factors
for a waterflood and a polymer flood . A large difference in recovery factor
would suggest good potential for the polymer flood and justify a more detailed
investigation involving reservoir simulation or a field pilot test. A small difference
in recovery factor would kill the proposal for conversion to a polymer flood.
The distance between lines of injection and production wells is 2000 feet. The
distance between wells on a line is 1000 feet. Wells are designed to inject 500
RB/day of water or water containing polymer.
The relevant reservoir-fluid properties are given below;
k50%
k84.1%
h
Dip angle,
o
w
Bo
polymer
0.22
260
130
25
0o
5.0
0.5
1.1
5.0
md
md
ft
cp
cp
RB/STB
cp
171
Relative permeability curves for the oil-water system are given in the following
table.
Sw
0.20
0.25
0.30
0.35
0.40
0.45
0.50
0.55
0.60
0.65
kro
1
0.60
0.38
0.25
0.17
0.10
0.06
0.03
0.01
0
krw
0
0.03
0.05
0.08
0.13
0.16
0.23
0.30
0.35
0.45
The relative permeability curves for a polymer flood are expected to be similar.
The economic limit for production is a water-cut of 98% (fw = 0.98). Compare
the performance (recovery factor and cumulative injection) of a waterflood and
a polymer flood at breakthrough and the economic production limit. Summarize
your findings and prepare a recommendation for future action.
[Answer:
EM
ED
EA
EV
ER
waterflood
bt
fw = 0.98
0.56 0.46
0.56 0.75
0.55 0.51
0.93 0.96
Winj
(PV)
0.07 0.21
0.38 2.1
polymer
bt
fw = 0.98
0.56 0.78
0.56 0.88
0.80 0.93
1
0.98
0.33
0.49
0.35
1.25
Time
(years)
2.2
22.5
3.8
13.5 ]
Solution hints
1. You may get slightly different answers from those given above. The answers
depend to some degree on the parameters used to fit the relative permeability
data. However, your conclusions should remain the same.
2. Although we are calculating on a five-spot flooding pattern, in the absence of
data for this pattern (the case here for EV ), we can use the curves for a linear
flood to estimate EV . This is because displacement and sweep efficiencies are
relatively insensitive to the type of pattern.
172
173
k
h
VDP
Dip angle,
o
CO2
o
CO2
Bo
0.22
10,000 md
10 ft
0.3
45o
5.0
0.05
0.90
0.65
1.1
cp
cp
gm/cc
gm/cc
RB/STB
kro
1
0.77
0.66
0.59
0.44
0.32
0.22
0.15
0.10
0.06
0.037
0.020
0.010
0.004
0.0012
0.0004
0.00002
0
174
krg
0
0
0
0.00003
0.0013
0.0062
0.0172
0.0359
0.0639
0.1
0.15
0.213
0.29
0.37
0.47
0.57
0.68
0.80
The reservoir will be flooded on a 1000 ft. line-drive with 600 ft. between wells
on a line. The flood will be conducted at the critical rate to ensure a gravity
dominated flood. What is the recovery factor at gas breakthrough? What would
the recovery factor at gas breakthrough be if the dip angle was zero ie., the
reservoir was horizontal?
Answer:
bt
EM
0.8
ED
0.43
zero dip bt
0.8
0.31 0.41
45o dip
EA
1
EV
1
ER
0.34
0.09 0.01
Solution hints
1. If the displacement rate is equal to or less than the critical rate you can assume
that areal and vertical sweep efficiencies are 1.
2. The results of the calculations clearly show something which is now well
understood by the hydrocarbon production industry - to be effective a gas flood
must be gravity stabilized.
3. For a gas injection process, breakthrough corresponds to the end of the injection process. The viscosity of gas is very low compared with that of oil (and
water). When gas breaks-through at a well the oil rate declines and the gas rate
and producing gas-oil-ratio (GOR) increase rapidly. The well gasses-out and gas
injection ceases because large volumes gas are now simply being cycled through
the reservoir. On the other hand, since oil and water have viscosities which are
usually of the same order and large produced water volumes can be handled
economically, waterflood projects can be produced to very high water cuts or
fractional flows.
175
Chapter 4
PSEUDO-FUNCTIONS
4.1
TYPES OF PSEUDO-FUNCTIONS
Modeling the effects of heterogeneity on the recovery and performance of reservoirs is difficult. Simple one-dimensional fractional flow analyses require estimates of vertical sweep efficiency which is strongly affected by layering - a twodimensional problem. Reservoir simulation studies require large finely gridded
three-dimensional models to adequately represent heterogeneity. These models
will often contain many more grid blocks than can be accommodated in practical
full-field studies where the huge volume of data and processing time necessitate
the use of simpler models.
The dimensionality and number of grid blocks in a simulation study can be reduced by averaging the properties of groups of smaller blocks. The number of
layers in an in an analytical layered reservoir model can be reduced by averaging
the properties of the individual layers. The attached figure shows a x-section from
a highly detailed three-dimensional simulation model overlayed by a much coarser
grid. The properties of the coarser grid are obtained by averaging the properties
of the finer grid blocks. All averaging processes result in a loss of information. In
this case the loss of information is the description of the finer-scale heterogeneity
or sub-grid block heterogeneity.
The lost description of the finer scale heterogeneity will affect the flow of phases in
the coarser grid. We can reintroduce this information, in an approximate sense, by
inputting modified relative permeability curves or pseudo relative permeabilities
in the coarser grid model. The rock relative permeability curves used in the finegrid model are modified to make the flows in the coarse grid block model match
those in the finer grid block model.
176
Figure 4.1: Detailed geostatistical model and corresponding simulation grid which
has lost information on sub-grid block scale heterogeneity
177
178
4.1.1
Selection of layering
The layering is usually selected on the basis of a permeability log constructed from
a porosity log using a suitable porosity-permeability transform. Similar permeability values are grouped into layers. The layers will almost always contain a
range of different porosity and permeability values, however, the Dykstra-Parsons
coefficient of permeability variability should be less than 0.3 for each layer. These
are averaged to obtain the layer values. The appropriate averaging method is
thickness weighed.
P
hi i
i hi
hi ki
i hi
n = Pi
kn = Pi
where the subscript i refers to the individual values of the rock property within
layer n.
179
Figure 4.3: Pseudo-functions - making a column of fine grid blocks behave like a
single coarse grid block - the coarse grid block contains more than one rock type.
Figure 4.4: Fine grid X-sectional model used to generate dynamic pseudo functions. Different colors represent different permeability levels (32md to 2840md)
180
4.2
4.2.1
DETERMINING PSEUDO-FUNCTIONS
Dynamic pseudo-functions
181
Figure 4.5: Saturation map for the fine grid X-sectional model - all saturation
functions are therefore known. The simulator also provides the pressure field.
182
K
X
zk
k=1
The length, L, and width, W , of the of both the fine and coarse grid blocks are
the same, L = x and W = y.
Equating the total pore volume in any column of fine grid blocks with the pore
volume in the corresponding coarse grid block gives,
K
X
(zk xy)k =
k=1
K
X
(zk xy)
k=1
where is the porosity of the coarse grid block. Solving for the coarse grid
porosity gives,
P
zk k
= P
zk
where it is understood that the summations are over a column of fine grid blocks
k = 1, K.
Average permeability
The coarse grid block permeability k is determined from the requirement that
for a pressure difference P across the column of fine grid blocks and the same
183
pressure difference across the coarse grid block the total flows are the same. Using
Darcys law and equating the total flow for the column of fine grid blocks to the
flow for the coarse grid block gives,
k(hy)
p
kk (zk y) p
=
x
k=1
K
X
gives
Solving for the average permeability, k,
zk kk
k = P
zk
P
where kj is the fine-grid block permeability. This is the average horizontal permeability where the flow through the column of fine grid blocks is in parallel. A
similar result is easily obtained for average vertical permeability where the flow
through the column of fine grid blocks is now in series.
The horizontal permeability of the course block is equal to the thickness weighed
average permeability of the column of fine grid blocks.
Average saturation
The coarse grid block must contain the same volume of water or wetting saturation as the column of fine grid blocks. Equating the water volumes in the column
of fine grid blocks with that in the coarse grid block gives,
K
X
(xyzk )k Sw k = (hy)Sw
k=1
Solving for average water saturation, Sw , and substituting the previously derived
gives
expression for ,
P
zk k Swk
Sw = P
zk k
Pseudo relative permeability
It is more convenient to write the flow equations (Darcys law) in terms of phase
potentials rather than pressures. Phase potential is defined as,
= P gD
184
where is the phase potential, P is the phase pressure, is the phase density,
D is the depth and g is the gravitational acceleration. The difference in phase
potential is,
= P gD
For linear flow, Darcys law may be written for the oil rate in each fine grid block
as,
qo k =
o k
kk kro k
yzk
o
x
where kk is the permeability of the fine grid block, krok is the corresponding
oil relative permeability, o is the oil viscosity and ok is the difference in oil
potential driving the flow. The difference in oil potential is given by,
o j = o(i+1,k) o(i,k)
where the subscript i + 1 refers to the upstream column of blocks and irefers to
the column of grid blocks being averaged. This corresponds to the commonly
used up-stream weighing.
In an analogous manner we write for the water rate,
qw k =
kk krw k
w k
yzk
w
x
where krw k is the fine grid block water relative permeability, w is the water
viscosity and w k is the difference in water potential driving the flow and is
given by,
w k = w(i+1,k) w(i,k)
The total oil rate for the column of fine grid blocks is given by the sum of the
individual block rates,
qo =
qo k =
kk krok
ok
yzk
o
x
qw =
qw k =
w k
kk krw k
yzk
w
x
qo =
o
k kro (hy)
o
x
qw =
w
kkrw (hy)
w
x
and
where kro and krw are the oil and water pseudo relative permeabilities, respectively.
The corresponding fluid potentials are,
o =
o(i+1)
o(i)
and
w =
w(i+1)
w(i)
Equating the expressions for phase rates for the column of fine grid blocks and
for the coarse grid block and solving for the pseudo relative permeabilities gives,
kro =
kk kro k zk o k
o
kh
krw =
kk krw k zk w k
w
kh
and
All the terms on the RHS of the above equations are known with the exception
o and
w . In order to evaluate the
of the average coarse block phase potentials
pseudo relative permeabilities we must estimate these potentials. The only way
186
of doing this is to take some sort of average of the fine grid block phase potentials.
Unfortunately, there is no unique way of doing this. The choice of method leads
to the different types of pseudo-function methods commonly used. In general we
can write,
o =
wo k o k
wo k
w =
ww k w k
ww k
and
where wok and wwk are, yet to be defined, phase potential weighing functions. The
following averaging methods are examples of the more commonly used averaging
methods.
Eclipse Method
wok = wwk = kk zk
All the methods yield similar pseudo-relative permeability curvesin most practical
cases.
187
different relative permeability curve which would have been measured in the lab
using a representative core plug. In the present example we have simplified a
little by specifying that all the layers have the same rock or lab measured relative
permeability curves. The only reason for doing this is to highlight the differences
between rock curves and the pseudo curves.
After calculating the pseudo relative permeability curves we would like see how
well they work. Since we have averaged down a column of blocks the psuedos allow us to represent the 10x1x10 100 grid-block fine grid model by a
10x1x1 10 grid-block coarse grid model. This requires us to convert the file
DOWNXsection.dat into a 10x1x1 model file that I have called DOWNpseudo.dat.
The rock curves in the pseudo file are replaced by the pseudo relative permeabilities in PSEUDOkr.prn.
Figure 4.5 shows dynamic pseudo relative permeabilities calculated using the Kyte
and Berry method for data generated using the cross-sectional model shown in
Figures 4.6 and 4.7. The input data file for the cross-sectional model is called
Pseudos.dat in folder Dynpseudo. You should run this file for the three cases
shown in Figure 4.5 - uniform permeability, upward coarsening sequence and
downward coarsening sequence. The actual pseudo relative permeabilities are
calculated using the Mathcad spreadsheet DYPSUDO.mcd in folder Psuedo. The
spreadsheet reads the necessary simulator output arrays contained in the files
Uniform.prn, Up.prn and Down.prn. Study these files carefully to see how they
have been modifies so that Mathcad can understand them - Mathcad reads a
matrix of numerical data and ignores all non-numerical inputs.
As you may have anticipated from the theory, the uniform sand Psuedo relative
permeabilities are the same as the rock curves. However, the Psuedo relative permeabilities for the upward and downward coarsening sequences are quite different
from the rock curves - particularly the Psuedo relative permeability for water. In
both cases, the Psuedo relative permeability for water is considerably higher that
the corresponding rock values making it much easier for water to move in the
heterogeneous formations. This is a very important point and makes it very difficult - if not impossible - to history match simulation models which use only rock
relative permeability curves.
The Psuedo relative permeability curves shown in Figure 4.1 were determined at
a flooding rate of 25,000 STB/day. Calculate the Psuedo relative permeability
curves when the flooding rate is reduced to 2,500 STB/day.
189
Figure 4.8: Permeability distribution for cross-sectional fine-grid model - downward coarsening sequences.
Figure 4.9: Water saturation for the cross-sectional fine-grid model - downward
coarsening sequences.
191
192
Solution hint
Re-run the cross-sectional model input data file, Pseudos.dat, for a production
rate of 250 RB/day.
Make a print of file Up.prn. Make a copy of the file and name it Up250.prn.
Using the last time-step output arrays in the cross-sectional output results file,
Pseudos.out, re-construct the Up250.prn so that it contains the data for the low
rate flood.
Run DYPSUDO.mcd for the new file and see the resultant Psuedo relative permeabilities.
Repeat the procedure for the other cases.
What do you conclude?
193
4.2.2
194
e
kro
PN
e
hi ki krw
i
h
k
i=1 i i
PN
e
hi ki kroi
i=1 hi ki
i=1
PN
i=1
PN
The following simple equations are used to calculate the average water saturation and average relative permeabilities, pseudo relative permeabilities, at the
production face.
195
Swp n =
Pn
i=1
hi i (1 Sor i ) + N
i=n+1 hi iSwc i
PN
i=1 hi i
P
The above equation simply states that when layer n has watered-out, the average
water saturation in the reservoir is the sum of the water volumes in the wateredP
out layers ( ni=1 hi i (1 Sor i )) and water volumes in the layers yet to be wateredP
out ( N
water saturation is the total water volume
i=n+1 hi i Swci ). The average
PN
divided by the total pore volume ( i=1 hi i ).
krwp n =
krop n =
Pn
e
hi ki krw
i=1 hi ki
i=1
PN
PN
e
i=n+1 hi ki kro
PN
i=1 hi ki
196
197
15
11
2
9
22
1
5
10
9
20
5
kn (mD)
Swc
Sor
krwe
kroe
350
250
500
450
150
1000
300
600
250
150
650
0.21
0.2
0.23
0.23
0.18
0.24
0.21
0.23
0.2
0.19
0.24
0.25
0.28
0.24
0.24
0.27
0.24
0.27
0.24
0.27
0.28
0.25
0.27
0.27
0.27
0.27
0.27
0.27
0.27
0.27
0.27
0.27
0.27
0.3
0.3
0.3
0.3
0.3
0.3
0.3
0.3
0.3
0.3
0.3
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
A symmetry element of the line drive waterflood is shown in the above figure.
The flood is conducted at a pressure above the bubble point pressure. The oil and
water formation factors are Bo = 1.475 RB/STB and Bw = 1.03 RB/STB. The
oil and water specificgravities are o = 0.78 and w = 1.1. The water viscosity is
w = 0.5 cp.
198
Figure 4.16: Oil-water pseudo relative permeability curves for Vertical Equilibrium (VE) waterflood.
Solution guide
1. Pseudo relative permeabilities and fractional flow curves
Most of the data for the problem is in the file Pseudo1.prn.
The layer flooding sequence is layer-1 to layer-11 (bottom-up). The averaging
equations discused in the previous sections are written in Mathcad spreadsheet
Pseudo1.mcd. This spreadsheet calculates and plots the pseudo relative perme199
Figure 4.17: Pseudo fractional flow curves for oil viscosities of 50, 5 and 0.5 cp.
200
Figure 4.18: Computed production profiles for oil viscosities of 50, 5 and 0.5 cp.
following figure. The figure shows oil rate as a function of time for the first
10 years of production. The 0.5 cp viscosity case with mobility ratio less than
one (M = 0.3) is still producing clean oil after 10 years. This is indicative of
a very efficient piston-like displacement. For the intermediate viscosity case of
5 cp the mobility ratio is moderately unfavourable (M = 3) and water breaks
after approximately1.5 years of injection with the oil rate declining as the water
cut increases For the highest viscosity, 50 cp, oil the mobility ratio is highly
unfavourable (M = 30). Water breakthrough occurs almost immediately and the
oil rate declines rapidly with most of the oil being recovered at high water cuts.
The following figure gives an impression of the mature of the displacement front
in the reservoir for the three cases. For M = 30 there is severe under-running
of the oil by water as gravity causes most of the injected water to slump to the
base of the sand. This results in early water breakthrough. For M = 3 viscous
forces become more important and counter to some extent the adverse effects of
gravity segregation. This behaviour is common for displacements in uniform or
uniform acting sands.
201
Figure 4.19: Sketch of water displacement fronts for oil viscosities of 50, 5 and
0.5 cp.
Figure 4.20: Permeability distribution: (a) actual log-core data showing a downward coarsening 94 foot section (b) 10 layer model for the downward coarsening
sequence (c) inverted 10 layer model to investigate an upward coarsening sequence.
203
accross the section which suggests that the displacement will be dominated by
gravity and occur under vertical equilibrium conditions.
A relative permeability was measured for only one of the core plugs (a common
occurance in actual practice) and we will assume that this data is characteristic
e
e
for the section. The end-point values are krw
= 0.33 and kro
= 1.0 and an average
residual oil saturation Sor = 0.33.
We are required to determine the following:
1. Pseudo relative permeabilities for the downward coarsening section and for
the upward coarsening sequence shown in (c).
2. Fractional flow curves and production profiles for oil viscosities of 1.24 cp and
20 cp. The corresponding mobility ratios for the displacements are 1 and 16.
3. Compare the above results for the case of a uniform sand having the same
average properties as the upward and downward coarsening sequences.
Formation properties and thickness averaged parameters for the 10
layer downward coarsening model
7
14
14
11
6
8
3
13
9
8
33
320
32
650
718
1244
74
1560
2000
2840
Swc
Sor
0.213
0.22
0.215
0.227
0.228
0.235
0.22
0.253
0.25
0.259
0.21
0.196
0.205
0.195
0.187
0.18
0.192
0.175
0.165
0.168
0.33
0.33
0.33
0.33
0.33
0.33
0.33
0.33
0.33
0.33
Figure 4.21: Pseudo relative permeability curves for 10 layer downward coarsening
sequence and 10 layer upward coarsening sequence models.
Also shown in the figure is a sketch of typical laboratory measured relative permeability curves. The comparison shows that:
1. The pseudo relative permeability curves for the upward and downward
coarsening sequences are very different and both are different from typical rock measured curves.
2. The mobility of water is much greater for the downwards coarsening sequence where gravity causes water to segregate to the higher permeability
layers at the base of the interval causing early breakthrough of water. The
effects of both gravity and heterogeneity are detrimental to displacement
efficiency.
3. The mobility of water is much lower for the upwards coarsening sequence
where gravity causes water to segregate to the lower permeability layers at
the base of the interval delaying water breakthrough. Here gravity mitigates
the detrimental effect of heterogeneity on displacement efficiency.
The following figures show the fractional flow curves and the corresponding oil
production profiles for the downward and upward coarsening sequences. The
fractional flow curves for both favourable and unfavourable mobility ratios for
205
Figure 4.22: Comparison between pseudo relative permeability curves for 10 layer
downward coarsening sequence and 10 layer upward coarsening sequence models
and typical rock relative permeability curves.
206
Figure 4.23: Comparison between the fractional flow curves for 10 layer downward
coarsening sequence and 10 layer upward coarsening sequence models .
207
Figure 4.24: Comparison between oil production profiles for 10 layer downward
coarsening sequence and 10 layer upward coarsening sequence models.
4.2.3
L
qn
=
t
An n (1 Sorn Swcn )
qn =
e
kn krw
An p
w
L
Since w and p/L are the same for all the layers we can write,
vn
e
kn krw
n (1 Sor n Swc n )
The flooding order is determined by this velocity. The layer with the highest
velocity floods out first and that with the lowest velocity last.
209
Figure 4.26: Vertical equilibrium flooding sequence with layers in order of decreasing permeability.
kn
n
or simply,
vn kn
The above figure shows the flooding sequence for a simple layered system where
all the layers have the same properties other than permeability. (a) shows the
positions of the displacement front in individual layers before any of the layers
has watered-out. The higher the layer permeability the closer the displacement
front to the production face. (b)-(f) show the positions of the displacement front
at the time each layer waters-out. The layers water out in order of decreasing
permeability.
Having established the flooding sequence we know the water saturations at the
production face at the times each layer waters out. We can therefore use the
previously derived averaging equations to calculate pseudo relative permeabilities. The only difference the previously studied vertical equilibrium case and the
present viscous doninated case as far as calculating pseudo relative permeabilities
is concerned is the flooding order.
Since there is no cross-flow between layers the layers can be ordered in any convient manner. The attached figure shows the form of ordering which is most
convenient for computation of average values.
210
Figure 4.27: Permeability distribution for a deltaic sand section in the only cored
well in the field.
Problem 4.4 - Water-oil displacement in a stratified reservoir under no
cross-flow conditions [Pseudos.mcd]
The previously discussed vertical equilibrium conditions commonly prevail in marine or beach type depositional enviornments in which sands are relatively clean
and largely free from areally extensive barriers to vertical flow (cross flow). In
non-marine deltaic depositional enviornments reservoirs commonly have numerous channel sands separated both areally and vertically from each other by shale
and silt intervals.
The following figure shows the permeability distribution across a deltaic sand
section intersected by a well. The section is divided into a set of 8 noncommunicating layers based on permeability.
211
Layer
hn (ft)
kn (mD)
Swc
Sor
1
2
3
4
5
6
7
8
10
8
14
4
2
8
10
13
174
103
487
73
141
904
1223
70
0.22
0.18
0.21
0.22
0.18
0.21
0.20
0.20
0.37
0.41
0.35
0.42
0.39
0.34
0.32
0.42
0.28
0.28
0.28
0.28
0.28
0.28
0.28
0.28
e
e
The end-point relative permeabilities are kro
= 1 and krw
= 0.39. The fluid
viscosities are o = 1.0cp and w = 0.3cp.
212
Figure 4.28: Pseudo relative permeabilities for 8 layer deltaic sand section.
213
Figure 4.29: Comparison between actual field and calculated water cut development.
Figure 4.30: Comparison between calculated oil rate and actual field rate (inset).
214
Chapter 5
WATER AND GAS CONING
Oil and gas production from many fields is affected by water coning, gas coning
or both simultaneously. For this reason wells in reservoirs where there is bottom
water are completed in the top part of the oil zone, provided there is no gas cap.
Wells in reservoirs with a gas cap are completed at the bottom of the oil zone,
provided there is no bottom water. In reservoirs subject to both water and gas
coning the well is completed in the central part of the oil interval.
The only way of accurately predicting the coning performance of vertical, inclined
and horizontal wells is to use reservoir simulation. However, a number of useful
correlations are available to make order of magnitude estimates of the critical
coning rate - the maximum rate at which a well may be produced without coning
water or gas. At the critical coning rate, the water or gas cone is static with the
top of the cone at the base of the perforated interval for water coning and at the
top of the perforated interval for gas coning.
On first inspection the published coning correlations they may appear complex
and unwieldy. However they can be readily understood by a simplified analysis in
which the viscous forces at the wellbore which cause the cone to form are balanced
by the gravitational force which acts against the deformation of the contact.
5.1
CONING
WELLS
BEHAVIOR
AT
VERTICAL
For simplicity we consider the case of water coning at a vertical well. The attached
figure shows a schematic of the critical cone - the well is producing at the critical
rate where the viscous force is just sufficient to keep the water cone static at the
base of the perforations. Assume that capillary forces are negligible (po = pw ).
215
Figure 5.2: Pressure distribution for a water cone below a vertical well
216
217
possible - we assume that the viscous pressure drop is given by the steady-state
radial flow equation,
2kh hp
qo =
o Bo ln
re
rw
where the effect of the water cone on the pressure drop is ignored.
Substituting the previously determined relationship, p = ghc , we obtain,
qo =
2kh hghc
o Bo ln
re
rw
qo =
o Bo ln
re
rw
(h[h hp ])
This equation allows us to calculate the critical coning rate for a vertical well. All
the published water/gas coning correlations have the above form. As an example
consider the coning correlation proposed by Meyer, Gardner, and Pirson (1977)
which may be written as,
qo =
2kh g
Bo o ln
re
rw
(h2
h2p )
The similarities between the above two equations are clearly obvious. In fact, all
the correlations can be expressed in the following form,
qo =
2kh g
Bo o ln
re
rw
(h, hp , kv /kh )
where the additional viscous pressure loss due to the presence of the cone is
incorporated into the function (h, hp , kv /kh ). The dependence of on vertical
permeability results from the combined effects of partial penetration, and the
presence of the water cone.
218
5.1.1
qo =
kh g
Bo o ln
re
rw
(h2
h2p )
Chaperon (1986)
qo =
kh g
Bo o ln
where
re
rw
(h2 qc )
h
qc = 0.7311 + 1.9434
re
!s
Schols (1972)
qo =
kh g
Bo o ln
219
re
rw
kh
kv
where
2
= (h
h2p )
"
0.432 +
ln(re /rw )
h
re
!0.14
qo =
where
"
hp
= h2 1
h
88.9kh g
Bo o ln
#2 1.325
re
rw
!0.238
h
re
ln
re
rw
1.99
Chierici (1995)
qo =
where
=
2kh g
Bo o ln
re
rw
A
B + C ln rDe
and
rDe
re
=
h
kv
kh
hp
h
0.00119
A = 0.993 +
0.769 bD
1
B=
0.302 0.053bD 0.366b2D
bD =
i
220
kv /kh = 1
209 STB/day
168 STB/day
kv /kh = 0.1
278 STB/day
223 STB/day
kv /kh = 0.01
413 STB/day
332 STB/day]
Solution hints
Use [Coning.mcd] - some of the correlations have weird units.
221
5.2
222
Figure 5.4: Pressure distribution for a water cone below a horizontal well
Chaperon (1986)
kh g
qo =
Bo o
L
(h hv )2F
ye
where
F = 3.962 + 0.0616h 0.000542h
ye
h =
h hp
kv
kh
kh g
qo =
Bo o
L
(h hv )2 F
ye
where
1
F = 1
6
223
h hv
2ye
!2
Joshi (1991)
This correlation is a little different the others in that it provides the ratio of
critical rates for horizontal and vertical wells.
ln(rev /rw ) (h2 h2p )
qoh
=
qov
ln(reh /rwh ) (h2 h2v )
where
reh L/2a
q
rwh =
1 + 1 (L/2a)2 (h/2rw )h/L
L
a = 0.5 +
2
2reh
0.25 +
L
4
0.5
Joshi suggests using the Meyer, Gardner and Pirson (1977) correlation to calculate the critical rate for a vertical well
qov =
kh g
Bo o ln
re
rw
(h2
h2p )
As for the vertical well correlations, the correlations for horizontal wells display
quite large differences in the predicted critical rate. Despite this they do show a
significant increase in the critical rate for horizontal wells over vertical wells.
224
kv /kh = 0.1
260 STB/day
1571 STB/day
kv /kh = 0.01
509 STB/day
1515 STB/day
Solution hints
Use [Coning.mcd] - some of the correlations have weird units.
The results of this calculation clearly show that horizontal wells perform markedly
better than vertical wells in terms of coning performance. The relative superiority
of the horizontal well decreases with decreasing kh /kv ratio, going from about 10:1
for kh /kv = 1 to about 3:1 for kh /kv = 0.01.
225
Chapter 6
NATURAL WATER INFLUX
Material balance calculations clearly show the important effect of water influx
on reservoir performance. Although material balance calculations allow us to
calculate past water influx volumes, they cannot be used to predict future influx
volumes which are necessary to predict future reservoir performance.
In contrast with the reservoir where there is usually good well control, the description of the aquifer is much more uncertain because of the large extent of
the aquifer and the lack of adequate well control. Porosity, permeability, thickness, geometry and areal continuity are all subject to great uncertainty. A good
knowledge of basin geology is essential to developing a satisfactory reservoir description. Dry holes and exploration seismic data provide additional data which
may be valuable in characterizing the aquifer.
van Everdingen and Hurst proposed a simple analytical model for predicting water
influx from an aquifer. The reservoir-aquifer system is modeled as two concentric
cylinders or sections of cylinders - the same model as previously used to model
radial flow into a well. The reservoir takes the place of the wellbore and the outer
radius becomes the outer boundary of the reservoir. The flow equations for the
aquifer are identical to those for the radial well models.
6.1
6.1.1
The inner radius represents the boundary between the reservoir and the aquifer
and is equivalent to the reservoir radius, ro . The outer radius, re , is the effective
226
Figure 6.1: Radial flow model for water influx from an aquifer
radius of the aquifer or the radial extent of the aquifer.
Water pressures in the aquifer zone are governed by the radial diffusivity equation
which may be written in dimensionless form as,
1
pD
rD
rD rD
rD
pD
tD
2kh
(pi p)
q
kt
cro2
r
rD =
ro
tD =
The radial diffusivity equation may be solved for a constant pressure or a constant rate at the inner (reservoir) boundary and a no-flow condition at the outer
boundary ie., a closed or finite extent aquifer. For the constant terminal pressure
case the solution is presented as a dimensionless water influx rate, qD (rD , tD ).
For the constant terminal rate case the solution is presented as a dimensionless
pressure at the boundary, pD (rD , tD ). The two solutions are related by Darcys
law which may be written in dimensionless form as,
pD
qD =
rD rD =1
227
The van Everdingen and Hurst aquifer model is based on the constant terminal
pressure solution, qD . The solution gives the water rate into the reservoir, qD (tD ),
for a constant pressure drop, pD or p = pi p, at the reservoir-aquifer boundary.
qD =
q
2khp
where q is the water influx rate and p is the pressure drop at the aquifer-reservoir
boundary.
The dimensionless cumulative water influx, WD , is obtained by integration,
WD =
tD
qD dtD =
2khp
k
cro2
!Z
qdt =
2khp
k
W
cro2
enchroachment angle
360o
and is used to model cases where the aquifer shape is better represented by a
segment of the radial system.
In field units the above equations are given as,
tD =
kt
0.00634cro2
U = 1.119fhcro2
In the above equations time is in days and the aquifer constant has units of
RB/psi.
228
Figure 6.2: Dimensionless water influx for constant terminal pressure solution
(Hurst and Everdingen)
6.1.2
X
r2 1
en tD J12 (n rD )
WD = D
2
2
2
2
2
n=1 n [J0 (n ) J1 (n rD )]
where J0 and J1 are Bessel functions of the first kind of order 0 and 1, respectively.
n are the roots of
[J1(n rD )Y0 (n ) Y1 (n rD )J0 (n )] = 0
where Y0 and Y1 are Bessel functions of the second kind of orders 0 and 1.
In practice, it is only necessary to carry the first two terms in the summation
to achieve a result which is within a fraction of a percent of the actual value.
Klins, Bouchard and Cable (SPERE February 1988) have fitted polynomials to
approximate the van Everdingen-Hurst solutions.
You can look at the Mathcad spreadsheet [PtD.mcd] to see the form of the
[J1(n rD )Y0 (n ) Y1 (n rD )J0(n )] function and the accuracy of the Klins et
al. correlation for the first two roots 1 and 2 . You can also look at the spreadsheet to see how we can use the Klins et al. polynomials to estimate WD for
229
rD > 2 and any value of tD . You can compare the spreadsheet values with tables
and graphs published in various text books and handbooks. This Mathcad function makes it very easy to perform accurate water influx calculations without the
tedium of looking-up graphs or tables.
=
=
=
=
=
50 ft
0.25
50 md
0.4 cp
9106 psi1
Answer:
Years We MMRB
0.5
0.1874
1
0.0.2458
1.5
0.2671
2
0.2750
3
0.2789
Solution hints
If you are still not comfortable with Mathcad you have a major problem here.
You can do it by hand as is done in Exercise 9.1 (p313) of Dake, Fundamentals
of Reservoir Engineering - check out the [WEvHE1.mcd] solution with that given
in Dake.
230
6.2
FIELD WATER
TIONS
INFLUX
CALCULA-
In the previous section we used the constant terminal pressure solution to the
radial diffusivity equation to calculate cumulative water influx for a step change
in pressure at the reservoir-aquifer boundary. In field applications we assume
that this pressure change is equal to the change in average reservoir pressure.
6.2.1
Principle of Superposition
pi p1
pi + p1
=
2
2
pi + p1 p1 + p2
pi p2
=
2
2
2
p1 + p2 p2 + p3
p1 p3
p2 = p2 p3 =
=
2
2
2
p1 = p1 p2 =
and in general,
pi = pi pi+1 =
pi1 pi+1
pi1 + pi pi + pi+1
=
2
2
2
The cumulative water influx to time tn , is the sum of the individual water influxes
due to each of the pressure changes. This sum is written as,
We (t) = U {p0WD (tDn ) + p1 WD (tDn tD1 ) + p2 WD (tDn tD2 ) + ......
...... + pi WD (tDn tDi ) + ...... + pn1 WD (tDn tDn1 )
We may write this in compact form as,
We (t) = U
n1
X
pi WD (tDn tDi )
232
6.2.2
The following example will take you, step-by-step, through a typical water influx calculation requiring the use of superposition. Consider an aquifer-reservoir
system with the following properties,
h
c
ro
f
=
=
=
=
=
=
=
50 ft
0.21
500 md
0.3 cp
6.6106 psi1
2000 ft
1
233
2500
2480
2470
2464
2460
The average pressure over the first year is the sum of the pressures at the beginning of the year (initial reservoir pressure) and at the end of the year divided by
two,
2500 + 2480
= 2490
2
The average pressure over the second year is
2480 + 2470
= 2475
2
These pressures and those for years 3 and 4 are shown in the table below.
The step changes in reservoir pressure at the oil-water contact for the first year
of production is the initial reservoir pressure minus the average reservoir for the
first year,
2500 2490 = 10
The step change in reservoir pressure for the second year of production is the
average reservoir pressure for the first year minus the average reservoir pressure
for the second year,
2490 2475 = 15
These pressure step changes and those for years 3 and 4 are also shown in the
table below.
The principle of superposition states that the cumulative water influx is the sum
of the influxes due to each pressure step change as if this pressure step change was
acting alone (4 individual P s). At the end of the 4th. year the first pressure
step change (10 psi) has been active for 4 years, the second (15 psi) for 3 years,
the third (8 psi) for 2 years and the fourth (5 psi) for 1 year. These periods are
also shown in the table below.
The cumulative water influx for each pressure step change can be calculated using
the spreadsheet used for the previous exercise. These values are entered in the
table and summed to give the cumulative water influx at the end of the 4th. year.
234
2500
2480
2470
2464
2460
Pave
(psia)
2500
2490
2475
2467
2462
10
15
8
5
4
3
2
1
2.1978
2.5646
0.9623
0.3320
6.0567
The above procedure is easily computerized to make the calculation very convenient. This is done in the spreadsheet used to solve the next exercise.
You can follow this calculation using [WEvHE1.mcd]. If you do this an aquifer
size reD = 1000 is effectively an infinite aquifer. The principle of superposition
reduces the problem of determining the cumulative water influx after 4 years to
summing the cumulative water influxes for a 10 psi pressure drop after 4 years,
for a 15 psi pressure drop after 3 years, for a 8 psi pressure drop after 2 years,
and for a 5 psi pressure drop after 1 year. Each of these terms can be calculated
using [WEvHE1.mcd] in exactally the same way you did for Problem 6.1. It is
important to understand this.
235
c
ro
=
=
=
=
=
=
=
19.2 ft
0.209
275 md
0.55 cp
6106 psi1
5807 ft
180o
The reservoir has been producing for a period of 549 days during which time
the average reservoir pressure has declined from an initial pressure of 3793 psia
to 3643 psia. The following table shows the pressure decline as a function of
production time.
Time
(days)
P
(psia)
0
91.5
183.0
274.5
366.0
457.5
549.0
3793
3788
3774
3748
3709
3680
3643
Use the van Everdingen-Hurst Method to calculate the cumulative water influx
from the aquifer. Since the size of the aquifer is unknown perform the calculation
for the following reservoir sizes; rD = , 10, 8, 5, 2.
Note: reD = 1000 is equivalent to an infinite reservoir.
236
reD = 1000
reD = 10
91.5
183
274.5
366
457.5
549
0.0064
0.0349
0.1057
0.2322
0.4145
0.6505
0.0064
0.0349
0.1057
0.2317
0.4129
0.6460
reD = 8 reD = 5
reD = 2
0.0064
0.0349
0.1053
0.2302
0.4079
0.6336
0.0017
0.0082
0.0218
0.0418
0.0651
0.0896 ]
0.0063
0.0333
0.0971
0.2039
0.3447
0.5093
Solution hints
[WEvHESP.mcd] is only a minor modification of the previous spreadsheet to allow
the superposition summation to be carried out automatically. This removes all
of the tedium in the calculation. The reservoir time-pressure data is read from
file [SP6-2.prn].
The results show two very important characteristics of aquifers. Firstly, the
aquifer response is time-dependent with only minor influx volumes early in the
life of the reservoir and gradually building up to much greater volumes at later
times. Secondly, the larger the aquifer the greater the water influx.
237
6.3
The objective of history matching an aquifer is to determine aquifer size (rD ) and
the value of the aquifer constant (U). This is done by fitting the aquifer model
to water influx data calculated from the material balance equation using field
production data. The aquifer model can then be used to predict future water
influx rates and this together with the material balance equation or reservoir
simulation can be used to predict future reservoir performance.
6.3.1
History matching the aquifer does not result in a unique solution and a number
of different combinations of aquifer size and aquifer constant may give an equally
acceptable fit to material balance influx data. It is therefore necessary to limit
the range of adjustable parameters to ensure that they are consistent with the
best estimates (guesses) of geological parameters such as likely aquifer extent,
average thickness, permeability and porosity. The following comments provide a
general guide to a systematic history matching procedure.
Cumulative water influx is given by,
We = UpWD (tD , reD )
For a given aquifer size, reD , cumulative water influx is directly proportional to
the aquifer constant, U.
U = 2fhcro2
The shape of the water influx response (the aquifer time-constant) is determined
by dimensionless time, tD .
kt
tD =
cro2
In order to vary these parameters independently, change h to vary U and change k
to vary tD . The changes are made after performing the first-pass calculation using
the best-estimate parameters based on geological, seismic and dry-hole data.
238
c
ro
cf
cw
Swc
=
=
=
=
=
=
=
=
=
=
100 ft
0.25
200 md
0.55 cp
7106 psi1
9200 ft
140o
4106 psi1
3106 psi1
0.05
The average reservoir pressure and cumulative production data over the first 10
years of operation is given in the table below and in the production history file
[H6-2.prn]. PVT data for the reservoir oil is given in the table below. This
data and the production history are used to calculate cumulative water influx by
material balance. You should examine the spreadsheet in folder MTB6-3 to satisfy
yourself that you understand the material balance calculation for this reservoir.
Use the aquifer history match to determine the size of the aquifer, reD .
239
PRODUCTION HISTORY
Time
P
(year) (psia)
0
1
2
3
4
5
6
7
8
9
10
2740
2500
2290
2109
1949
1818
1702
1608
1535
1480
1440
Np
Gp
(MMSTB) (BSCF)
0
7.88
18.42
29.15
40.69
50.14
58.42
65.39
70.74
74.54
77.43
Wp
(MMSTB)
0.00
5.99
15.56
26.82
39.67
51.39
62.22
71.60
79.23
85.35
89.82
0
0
0
0
0
0
0
0
0
0
0
PVT DATA
P
(psia)
1440
1480
1535
1608
1702
1818
1949
2109
2290
2500
2740
Bo
Rs
(RB/STB) (SCF/STB)
1.273
1.276
1.280
1.287
1.294
1.303
1.316
1.329
1.349
1.374
1.404
364
371
383
398
418
448
471
507
545
592
650
240
Bg
(RB/SCF)
Bw
(RB/STB)
0.00182
0.00176
0.00170
0.00160
0.00150
0.00139
0.00128
0.00117
0.00107
0.00098
0.00093
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
Solution hints
1. The first step in history matching an aquifer is to calculate historical water
influx volumes from reservoir pressure-production data. This calculation was
studied in the Reservoir Engineering A course. For the present example the
pressure-production data is contained in file [H6-3.prn]. The PVT data is in
file [PVT6-3.prn]. The actual material balance calculation is performed by the
spreadsheet [WEvMBE.mcd]. All these files are located in folder MTB6-2.
2. The material balance water influx volumes are entered in file [HM6-2.prn]
located in the main problem folder. This file is read by the aquifer history match
spreadsheet [WEvEHHM.mcd].
3. The history match spreadsheet [WEvEHHM.mcd] input data consist of the best
estimates of aquifer properties. The spreadsheet allows you to manually tune the
values of k and h. These are the parameters which are most difficult to estimate
and which are therefore most uncertain. These parameters should be changed
with great care. The match to material balance influx volumes is not unique
and it is possible to achieve a good match with k and h values which are far
from reality. This would give an incorrect aquifer size. Other than this, history
matching an aquifer with [WEvEHHM.mcd] is simply identifying the curve which
best matches the material balance influx data.
4. Play around with different values of k and h to get a feeling for the uniqueness
of the match - see if you can match with k = 100md and h = 200ft, same value
of kh. Compare the matches for k = 200md and h = 97ft with k = 35md and
h = 300ft - can you tell the difference?
241
ro
cf
cw
Swc
=
=
=
=
=
=
=
=
=
30 ft
0.19
70 md
0.4 cp
6000 ft
140o
5106 psi1
3106 psi1
0.248
242
PRODUCTION HISTORY
Time
(day)
P
(psia)
0
50
100
150
200
250
300
350
400
450
500
550
600
650
700
4217
3915
3725
3570
3450
3390
3365
3600
3740
3850
3900
3920
3720
3650
3640
Np
Gp
Wp
(MMSTB) (BSCF) (MMSTB)
0.00000
0.34270
0.63405
1.06898
1.36788
1.64397
1.91465
1.91465
1.91465
1.91465
1.91465
1.91465
2.11402
2.24921
2.33133
0.00000
0.17821
0.32970
0.55587
0.71130
0.85487
0.99562
0.99562
0.99562
0.99562
0.99562
0.99562
1.09929
1.16959
1.21229
0.00000
0.00000
0.00121
0.02636
0.07041
0.12218
0.17304
0.17304
0.17304
0.17304
0.17304
0.17304
0.20368
0.25094
0.31690
PVT DATA
P
(psia)
3365
3465
3565
3665
3765
3865
3965
4065
4165
4265
Bo
Rs
(RB/STB) (SCF/STB)
1.2511
1.2497
1.2483
1.2469
1.2455
1.2441
1.2427
1.2413
1.2399
1.2385
510
510
510
510
510
510
510
510
510
510
Bg
(RB/SCF)
Bw
(RB/STB)
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
[Ans: This is a difficult one to fit but the following parameters fit reasonably well
and are consistent with the estimated values of h and k - k = 70md, h =24ft,
and reD = 8]
243
6.4
6.4.1
Carter and Tracey approximated the water influx rate using a series of constant
rate intervals - in the same way that we have previously approximated a continuous pressure decline by a series of constant pressure steps. The cumulative water
244
j1
X
qn (tn+1 tn )
n=0
This may be written as the sum of two sums through the ith interval and between
the ith and jth intervals:
We (tj ) =
i1
X
qn (tn+1 tn ) +
n=0
j1
X
qn (tn+1 tn )
n=i
or,
We (tj ) = We (ti ) +
j1
X
qn (tn+1 tn )
n=i
The solution for a single rate change - the constant terminal rate solution - may
be written as,
We (tD,j ) = U
tD,j
0
p()
d
[WD (tD )] d
d
The above equation is known as the convolution integral for the constant terminal
rate solution.
Carter and Tracy solved the above equation for the cumulative water influx in
terms of the cumulative pressure drop, pn , using Laplace transforms. The mathematical details of the solution are beyond the scope of this course. Interested
students may refer to the original paper - R.D Carter and G.W. Tracy An Improved Method for Calculating Water Influx, Petroleum Transactions AIME, 219,
416, 1960. The solution is written as,
0
U
p
We
P (tD,n )
0
P (tD,n )
tD
=
=
=
=
=
=
aquifer constant
total pressure drop (pi p) in psi
cumulative water influx in RB
dimensionless constant terminal rate solution
dimensionless time derivative of the constant terminal rate solution
dimensionless time
245
The major advantage of the above equation over that of the van Everdingen Hurst solution is that it does not require the use of superposition. To make
machine calculations easier, Fanchi(1985) fitted the following regression equation
to tabular dimensionless pressure functions presented by van Everdingen and
Hurst,
P (tD ) = a0 + a1tD + a2 ln tD + a3 (ln tD )2
The coefficients are,
reD = re /ro
1.5
2.0
3.0
4.0
5.0
6.0
8.0
10.0
a0
0.10371
0.30210
0.51243
0.63656
0.65106
0.63367
0.40132
0.14386
0.82092
a1
1.66657
0.68178
0.29317
0.16101
0.10414
0.06940
0.04104
0.02649
-0.000368
a2
-0.04579
-0.01599
0.01534
0.15812
0.30953
0.41750
0.69592
0.89646
0.28908
a3
-0.01023
-0.01356
-0.06732
-0.09104
-0.11258
-0.11137
-0.14350
-0.15502
0.02882
246
6.4.2
Fetkovich (1971) used the well productivity equation to considerably simplify the
water influx calculation from the aquifer. Fetkovich assumed that the aquifer is
finite and that semi-steady state conditions exist in the aquifer - this can only be
true for small aquifers. The water inflow equation may be written as,
qw =
dWe
= J (
pa p)
dt
where,
qw
J
p
pa
=
=
=
=
The productivity index is given by the pseudo-steady state radial flow equation
as,
2khf
J= r
ln roe 34
Cumulative water influx for a specified step change in reservoir pressure (pi p)
at the reservoir-aquifer (inner) boundary may be determined by integrating the
equation for qw . This requires that we know the average aquifer pressure, pa , as
a function of time. The required relationship is obtained from a material balance
for the aquifer,
We = cWi (pi pa )
where pi is the initial pressure and c is the average compressibility of the aquifer.
Wi is the initial water volume in the aquifer and is therefore related to the sizes
of the aquifer and reservoir,
Wi = (re2 ro2 )hf
We define Wei as the maximum possible water influx (
pa = 0),
Wei = cWi pi
Using the above equations, the aquifer material balance may be rearranged to
give,
We
pa = pi 1
Wei
247
d
pa
J pi
=
dt
pa p
Wei
This equation may be integrated from t = 0 (We = 0, pa = pi ) to give an expression for the average pressure in the aquifer as a function of time,
ln(
pa p) =
J pi
t+C
Wei
Jpi
Wei
Jpi
Wei
1 dWe
J dt
dWe
= J (pi p)e
dt
Integrating this equation and evaluating the constant of integration from the
condition We = 0 at t = 0, gives
h
Wei
We =
(pi p) 1 e
pi
Jpi
Wei
i!
t
This is the cumulative water influx for a step change in pressure at the reservoiraquifer boundary as a function of time. The equation may be used together
with the principle of superposition to calculate cumulative influx for an arbitrary
pressure profile at the reservoir boundary.
Fetkovitch has shown that a difference form of the equation may be used for influx
calculations which eliminates the need for superposition.
248
Wei
=
(pi p1 ) 1 e
pi
Jpi
Wei
t1
where p1 is the average reservoir boundary pressure over the time interval t1.
For the second time step (t2 = t2 t1) we can write,
We 2
Wei
=
(
pa,1 p2 ) 1 e
pi
Jpi
Wei
t2
where pa1 is the average aquifer pressure at the end of the first time interval t1.
This pressure may be calculated from the material balance equation as,
pa,1 = pi
We1
1
Wei
Wei
(
pa,n1 pn ) 1 e
=
pi
pa,n = pi 1
Pn
1 Wej
Wei
Jpi
Wei
tn
pn1 + pn
2
Fetkovitch has demonstrated that the above equations give results which closely
match those obtained with the Hurst-Everdingen procedure when the assumption of semi-steady state is valid. Although this is a valid assumption for small
aquifers, it is not valid for infinite or infinite-acting aquifers (large aquifers) where
flow conditions are transient.
249
Chapter 7
DECLINE CURVE ANALYSIS
Decline curve analysis is probably the most widely used procedure for forecasting
future production rates for a well, group of wells or a full field on the basis of
past performance i.e., it is a procedure for extrapolating past performance. The
method can only be applied for unprorated fields where production rates are only
limited by the ability of the reservoir to produce and not by limitations imposed
by restricting well rates or surface production limitations.
For decline curve analysis to be valid wells must be producing with constant
choke size or constant wellhead pressure, pumping wells must be pumped-off or
producing at constant fluid level, skin-factors must remain constant, the number
of wells does not change, the drainage area does not change and the reservoir drive
mechanism remains the same. The requirement that the drainage area remains
unchanged implies that production is occuring at pseudo-steady state conditions
i.e., the effect of boundaries on production is fully developed. These conditions
are normally satisfied only late in the production life of a field when wells are
producing at capacity and operating constraints and production facilities remain
unchanged.
Decline curve analysis is commonly used to assess the effectiveness of improved oil
and gas recovery projects. These include infill drilling, fluid injection, acidizing
and fracturing. Decline curve analysis can be used to estimate future well or
field performance without the changes. Decline curve analysis can also be used
to analyse early (months to a year) stimulated well or field production data.
Comparing the two decline curves provides technical and economic measures of
the success of the stimulation project.
Although the decline functions commonly used have some basis in the physical
laws governing flow through porous media, decline curve analysis is essentially a
simple curve-fitting procedure. The major advantages of decline curve analysis
are that they are easy to use, they require no explicit knowledge of reservoir pa250
7.1
DECLINE RATE
Figure 7.9 shows that when the production rate is plotted as a function of cumulative production, Np , the result is often a straight line. The equation for the
straight line can be written as,
q = DNp + C
where D and C are constants. Differentiating the above equation with respect to
251
Figure 7.2: Typical plot of oil production rate as a function of cumulative oil
production.
time gives,
dq
dNp
= D
dt
dt
and since dNp /dt = q, we can write that,
1 dq
= D
q dt
The expression (1/q)dq/dt is known as the loss ratio and D is referred to as the
nominal decline rate. It represents the fractional change in production rate per
unit change in time.
For cases where a plot of production rate as a function of cumulative production
does not produce a straight line, the loss ratio is found to be a function of the
production rate i.e.,
D qb
252
q
q0
!b
0<b<1
or,
D = D0
q
q0
!b
D0 is the initial decline rate and q0 is the corresponding initial production rate.
The values of D0 and q0 can be those corresponding to the actual start of the
decline period or they can be the values for any point on the decline curve.
Different values of b produce different decline curves. The following terms are
used to describe the different decline curves,
Exponential decline; b = 0
Hyperbolic decline; 0 < b < 1
Harmonic decline; b = 1
Typical Decline Curves
Figure 7.3 shows the general form of the decline curves for three values of the
decline exponent b. Exponential decline decline is the most severe and Harmonic
decline the least severe. Hyperbolic decline produces a family of curves for different values of b lying between the two extreems of exponential and harmonic
decline.
In fitting decline curves to production data we consider q0, D0 and b to be variables
which must be determined in the curve fitting process.
7.1.1
Consider the case of a single well in an ideal closed volumetric oil reservoir at
pseudo-steady state conditions when the flowing bottom hole pressure is held
253
q=
(
p pmin
2kh
w )
B ln(re /rw ) 0.75 + s
2kh
1
B ln(re /rw ) 0.75 + s
d
p
dq
= C1
dt
dt
or,
d
p
1 dq
=
dt
C1 dt
Since the reservoir is closed or volumetric we can write a material balance equation
as,
Np =
co N
(p0 p)
B
where co is the oil compressibility, N is the initial oil in place and p0 is the initial
reservoir pressure.
Differentiating this eauation with respect to time gives,
dNp
co N d
p
=
=q
dt
B dt
Combining this with the previous result for d
p/dt gives,
q=
co N dq
BC1 dt
This equation may be rearranged to give the nominal decline rate for the reservoir
as,
1 dq
co N
=
= constant = D0
q dt
BC1
This corresponds to the previously discussed exponential decline rate commonly
used for production decline analysis of solution drive reservoirs.
7.1.2
Exponential decline; b = 0
1 dq
q dt
D0 =
q0
dq
=
q
D0 dt
which gives,
q
ln
q0
= D0 t
or,
q = q0 eD0 t
which gives the curve the name exponential decline.
rate-time analysis
Taking natural logarithms of the exponential decline rate equation gives,
ln q = ln q0 D0 t
or,
log q = log q0
D0
t
2.303
A plot of log q vrs t produces a straight line with slope M = D0 /2.303 and
intercept B = log q0. These are used to determine the valuse of D0 and q0,
D0 = 2.303M
and
q0 = 10B
256
t2
t1
qdt
where t1 and t2 are the times associated with rates q1 and q2.
From the definition of decline rate,
q=
1 dq
D0 dt
qdt =
and
Np =
Np =
q2
q1
1
dq
D0
1
(q1 q2 )
D0
257
1
(q0 q)
D0
which gives the cumulative production from the starting rate of the analysis to
when the rate declines to a rate q. The rate q is usually taken as the minimum
economic production rate which makes Np the remaining reserves from the start
of the analysis period.
rate-cumulative production analysis
Cumulative production , as previously, is
Np =
qdt =
q0 exp (D0 t)
1
(qo q)
D0
A plot of q vrs Np produces a straight line with slope M = 1/D0 and intercept
B = q0/D0 . These are used to determine the valuse of D0 and q0 ,
D0 =
1
M
and
q0 = BD0
7.1.3
Harmonic decline; b = 1
q
q0
Figure 7.5: Graphical analysis of exponential decline of rate-cumulative production data (Do = 0.005, q0 = 5000).
1 dq
q
= D0
q dt
q0
Separating variables,
dq
D0
=
dt
2
q
q0
Integrating,
Z
q0
dq
D0
=
2
q
q0
dt
which gives,
q0
q 1 q01
1
1
= D0 t
or,
q=
q0
(1 + D0 t)
1
1
D0
=
+
t
q
q0
q0
A plot of 1/q vrs t produces a straight line with slope M = D0 /q0 and intercept
B = 1/q0 . These are used to determine the valuse of D0 and q0,
1
B
q0 =
and
D0 = q0 M
Cumulative production
We can evaluate the cumulative production between two rates, say q1 and q2 .
Np =
t2
t1
260
qdt
q0 dq
D0 q
which gives,
q0
Np =
D0
dq
q
q2
q1
and finally,
q1
q0
Np =
ln
D0
q2
D0
Np
2.303q0
A plot of log q vrs Np produces a straight line with slope M = D0 /2.303q0 and
intercept B = log q0. These are used to determine the valuse of D0 and q0,
q0 = 10B
and
D0 = 2.303q0 M
7.1.4
D = D0
261
q
q0
!b
= D0
q dt
q0
!b
Separating variables,
dq
D0
= b dt
b+1
q
q0
Integrating,
Z
q0
dq
D0
= b
b+1
q
q0
dt
which gives,
"
1 1
b qb
q0
q
#q
q0
!b
D0
q0b
dt
= 1 + bD0 t
262
or,
q=
q0
(1 + bD0 t)1/b
1
1
= b (1 + D0 t)
b
q
q0
A plot of 1/q b vrs t produces a straight line with slope M = D0 /q0b and intercept
B = 1/q0b . These are used to determine the valuse of D0 and q0,
1
B
q0b =
and
D0 = q0b M
Since b is not known beforehand the analysis must be carried out for a range of
b values and value selected is that which produces the best straight line for the
data plot.
Cumulative production
We can evaluate the cumulative production between two rates, say q1 and q2 .
Np =
t2
t1
qdt
q0b dq
D0 q b
263
which gives,
Np =
q0b
D0
q2
q1
dq
qb
and finally,
Np =
q0b
(q 1b q21b )
D0 (1 b) 1
Np =
q0b
(q01b q 1b )
D0 (1 b)
D0 (1 b)
Np
q0b
264
q0b M
1b
Again, since b is not known beforehand the analysis must be carried out for a
range of b values and value selected is that which produces the best straight line
for the data plot.
265
7.2
Fetkovich (1980) introduced decline type-curve matching as a useful form of decline curve analysis. Defining dimensionless production rate (qD ), dimensionless
time (tD ) and dimensionless cumulative production (NpD ) as,
qD =
q
q0
tD = D0 t
NpD =
NpD0
q0
the equations for production rate decline and cumulative production derived in
the previous section can be expressed in dimensionless form.
266
qD =
NpD =
1
(1 + btD )1/b
1 1b
qD 1
qD 2 1b
1b
Harmonic decline
qD =
1
1 + tD
267
NpD
qD 1
= ln
qD 2
Figure 7.10 shows dimensionless decline curve plots on a log-log graph for hyperbolic decline for b=0 to 1 in increments of 0.1. Note that in the limit of the
decline exponent b approaching 0 or 1, the hyperbolic curve approaches the limits
of exponential or harmonic decline, respectively.
Taking logs of both sides of the equations defining dimensionless rate and dimensionless time gives,
log qD = log q log q0
and
log tD = log t + log D0
Since log q0 and log D0 are constants it is obvious that the shapes of a dimensional
decline curve (log q vrs. log t) and the dimensionless decline curve (log qD vrs.
log tD ) when plotted on the same scale are identical but shifted on the rate and
time axes by log q0 and log D0 , respectively. Matching the dimensional data i.e.,
the actual production data, to the dimensionless decline curve allows the values of
b and D0 to be determined. This procedure is called decline type-curve matching
and is illustrated schematically in Figure 7.12. The procedure is illustrated in
the following exercise.
Solution hint
Use [DECLINE.mcd] or any other decline rate analysis which you may have access
to. If you use [DECLINE.mcd] read the explanatory notes on determining b and
D0 carefully.
7.2.1
Using reservoir simulation and actual field production data, Fetkovich (1980)
showed that wells in simple single-layer reservoirs displayed values for b between
0 and 0.5 depending on the reservoir drive mechanism. For layered reservoirs
with no cross-flow between layers the value for b can be between 0.5 and 1. For
these reservoirs Fetkovich showed that the closer the value of b is to 1, the greater
the volume of reserves remaining in the lower permeability layers and the greater
268
System description
0.3
269
3060
2580
2100
2090
1780
1860
1470
1510
1250
1330
1220
1090
1150
982
940
883
850
713
700
743
270
[Answer:
(i) D0 =0.12 month1 , b = 0.65, q0=2900STB/day.
(ii) economic limit rate 100 STB/day: Np =0.41 MMSTB, 6.9 years.
(iii) economic limit rate 10 STB/day: Np =0.999 MMSTB, 39.9 years.]
271
0
3
6
9
42
42
42
42
1321
1245
1176
1113
1993 Jan
April
July
Oct
12
15
18
21
42
42
42
42
1050
987
937
887
1994 Jan
April
July
Oct
24
27
30
33
42
42
42
42
836
786
742
704
1995 Jan
April
July
Oct
36
39
42
45
42
42
42
42
667
627
591
560
1996 Jan
April
July
Oct
Dec
48
51
54
57
60
50
50
50
50
50
943
855
780
711
648
272
[Answer:
Incremental oil recovery is 0.086 MMSTB. Remaining field life reduced from 7.6
years to 6 years ]
Solution hint
You will need to analyse two separate decline curves. One for the data prior to
the field stimulation project and one for the data after.
273
274