Download as pdf or txt
Download as pdf or txt
You are on page 1of 15

SPE 81021

Three-Phase Relative Permeability of Petroleum Reservoirs


Dragan Pejic and Brij B. Maini/University of Calgary

Copyright 2003, Society of Petroleum Engineers Inc. always behaves as a non-wetting phase and oil and water
This paper was prepared for presentation at the SPE Latin American and Caribbean Petroleum
could be wetting or intermediate wetting. It was also noted
Engineering Conference held in Port-of-Spain, Trinidad, West Indies, 27–30 April 2003. (Leverett 1941, Corey 1956, Saraf 1966) that the relative
This paper was selected for presentation by an SPE Program Committee following review of permeability of the wetting and non-wetting phase is primarily
information contained in an abstract submitted by the author(s). Contents of the paper, as
presented, have not been reviewed by the Society of Petroleum Engineers and are subject to
a function of its own saturation and relative permeability of
correction by the author(s). The material, as presented, does not necessarily reflect any the intermediate-wetting phase is strongly affected by the
position of the Society of Petroleum Engineers, its officers, or members. Papers presented at
SPE meetings are subject to publication review by Editorial Committees of the Society of saturation history and the saturations of other two phases.
Petroleum Engineers. Electronic reproduction, distribution, or storage of any part of this paper
for commercial purposes without the written consent of the Society of Petroleum Engineers is
prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300
Conducting three-phase flow experiments is a tedious task
words; illustrations may not be copied. The abstract must contain conspicuous and only a limited number of such experiments dealing with
acknowledgment of where and by whom the paper was presented. Write Librarian, SPE, P.O.
Box 833836, Richardson, TX 75083-3836 U.S.A., fax 01-972-952-9435. three-phase relative permeability measurements have been
reported. Most of the reservoir engineering calculations for
processes involving three-phase flow are based on predictive
models that attempt to estimate the three-phase relative
Abstract
permeability from more easily obtained two-phase oil-water,
This paper presents a critical review of recent oil-gas and/or water-gas relative permeability measurements.
experimental and theoretical developments in the area of
Review of recent experimental data
three-phase relative permeability along with a comparative
evaluation of several models for estimating the oil relative This review focusses on the data reported in the ninties
permeability under three-phase flow conditions. and later. Oak (1990) presented several sets of steady-state
experimental data for both two-phase and three-phase relative
Reported experimental data on three-phase relative
permeabilities of water-wet Berea sandstone. Three different
permeability were screened for reliability and completeness
cores having 200 md, 800 md and 1000 md permeability were
and three representative data sets were selected for use in
tested. X-ray absorption method was used for measurement of
evaluation of the prediction techniques. Eight different
liquid saturations; gas saturation was calculated as
prediction models were selected for evaluation. Each model
complement. Eight cases of saturation history were
was tested with the three selected data sets for its ability to
considered. The agreement between the experimental data and
match the experimentally determined oil relative permeability
the Stone's models I and II was not good. Oak suggested that
under three-phase flow conditions.
the difficulty was in using the two-phase saturation history to
It was found that none of the models predicted all of the model the much more complex three-phase one. He also
tested data sets with high accuracy. However, there are pointed out that displacement of trapped oil by gas can not be
models that give significantly more reliable predictions of neglected in a water-wet medium. The critical oil saturation
three-phase oil relative permeability compared to the needed to initiate oil flow was not always the same for the
other models. two-phase and three-phase data. He observed that the three-
phase relative permeability to water was dependant only on its
Introduction own saturation and the relative permeability to gas was a
When studying a production from an oil reservoir, an function of gas saturation and the saturation history. Oil
investigator can sometimes successfully limit the scope of his relative permeability was dependent on saturation of all phases
research on one or two flowing phases. However, there are as well as on the saturation history.
many situations where the flow of all three phases - oil, gas Maini et al. (1990) reported measurements of 3-phase
and water must be considered. The water drive, steam drive, relative permeability at elevated temperature. They used
underground combustion, solution gas drive and gas cap drive steady-state and unsteady-state measurement techniques with
often involve simultaneous flow of all three phases. Ottawa sand. It was observed that water and gas relative
Application of enhanced oil recovery techniques that utilize permeability were function of its own saturation, and oil
injection of gas or light hydrocarbon are frequently relative permeability was function of saturations of all phases.
accompanied with three-phase flow. Analyzing relative Hysteresis was obvious in the case of gas, not significant for
permeabilities and hysteresis effects in a flowing oil-water-gas the oil and absent in the case of water. There were significant
system, numerous investigators have pointed out that gas differences in values of oil relative permeability between the
2 SPE 81021

steady-state and unsteady-state techniques. The characteristics After a series of steady-state tests, it was noticed that the core
of the sand-pack used in this work are presented in the Table had become less oil-wet, which was attributed to stripping of
1. The oil isoperms were concave toward the oil apex when crude oil polar compounds during the flow of the refined oil.
plotted on a ternary diagram. Figure 1. shows the oil isoperms Initial water saturation did not affect endpoint water relative
for the DDI history (water decreasing, oil decreasing, gas permeability. The authors were unable to draw conclusions on
increasing) using steady-state technique. hysteresis in case of the three-phase relative permeabilities.
Oak (1991) conducted an investigation on intermediate- Baker (1995) presented a large collection of three-phase
wet Berea sandstone. He used the steady-state measurement relative permeability data for oil-wet, water-wet and
technique for the oil relative permeabilities keeping ratio of intermediate-wet sandstones; the latter two also included the
water and oil flow rate constant during the experiments. The work of Oak (1990 and 1991). The oil wet cores were
core properties are presented in the Table 2. Water, dodecane naturally oil-wet, and had absolute permeabilities between 126
(oil phase) and nitrogen were used as fluids and the X-ray and 208 md. X-ray absorption method was used to measure
absorption method was used to measure liquid saturations. The saturations of the brine and dodecane; gas saturation was
wettability of the core was altered by treating it with bis- calculated as complement. Baker's study mainly confirmed
dimethylamino-dimethylsilane. Both DDI and IID type tests observations by previous investigators regarding dependence
were conducted. The oil relative permeabilities are shown in of oil relative permeability on saturations of the other phases.
the Figure 2. Three-phase relative permeability of the wetting phase was
seen to be a function of its own saturation and its hysteresis
Contrary to the earlier results on a water-wet Berea
was negligible. The relative permeability of the non-wetting
sandstone (Oak, 1990) he found that gas injection did not
phase was dependant on both the saturation of that phase and
reduce the residual oil saturation after water flood. The Stone
the saturation history. The relative permeability of the
Model I was found to overpredict the actual oil relative
intermediate-wetting phase -- oil in water-wet case and water
permeability and the Stone Model II had both over- and
in oil-wet case -- was dependent on more than one saturation
underpredicted regions with significant deviations.
and the hysteresis was more pronounced. The shape of oil
Skauge and Larsen (1994) conducted unsteady state isoperms in water-wet tests was concave and the oil isoperm
WAG (water alternating gas) experiments on sandstones and curvature decreased with increasing oil-wetness. He also
compared them to gas injection and waterflooding observed that the Stone Model I was more suitable than
experiments. They used cores of three different wettability: saturation weighted interpolation for the water-wet cores and
water-wet, oil-wet and intermediate-wet. To change vice verse for the intermediate- and oil-wet cores.
wettability of a core, they treated it with the solution of
Kalaydjian et al. (1997) studied effect of oil spreading on
dimethyldichlorsilane in hexane. For the water-wet cores,
water in presence of gas. The oil-spreading coefficient (S) is
relative permeability to water was only function of its own
function of water/gas, water/oil and oil/gas
saturation with practically insignificant hysteresis. The oil
interfacial tensions:
isoperms were concave when plotted on a ternary diagram and
no conclusion could be made about the hysteresis in oil S = γ WG − ( γ WO + γ OG ) (1)
relative permeability, because all of the displacements were
toward reducing oil saturation. Gas relative permeability were If this coefficient has a positive value then oil spreads on
strongly influenced by the saturation path and showed water, which contributes to continuity of the oil phase and
tendency toward depending on saturation of other phases. In could lead to lower residual oil saturations. Two different oils
oil-wet cores, relative permeability to water displayed were used to introduce both positive (Isopar M) and negative
hysteresis and the oil relative permeability was primarily a (Soltrol 170) spreading coefficients. It was seen that for oil
function of its own saturation in the first few displacements. injection, the oil relative permeability was lower in case of
However, due to the problems with maintaining the altered positive spreading coefficient compared to the tests where it
wettability, it was difficult to make more certain conclusions. was negative. In the case of gas injection, positive spreading
Dependence of trapped gas saturation on core wettability was coefficient lead to higher oil recoveries. Although consistent
not observed, however, an increase in the initial gas saturation with unsteady-state trends, the steady-state technique
led to increase in trapped gas saturation. Compared to the two- produced higher oil relative permeability values. Kalaydjian et
phase experiments, the three-phase flow experiments resulted al. tested their data with a theoretical model based on fractal
in lower residual oil saturation. Simulation of oil production in theory of pore structure, details of which are given by Moulu
the WAG displacement based on Stone's models was in et al. (1997). For spreading conditions, they reported their data
reasonable agreement with the experimental results. to be in good agreement with the model.
Eleri et al. (1995) used X-ray computed tomography, a
non-destructive imaging technique for determination of fluid
distribution in Berea and Clashach sandstone cores. The Review of three-phase oil relative
objective was to study the hysteresis effects and the influence permeability correlations
of initial water saturation on the relative permeability. They Corey (1956) presented first empirical correlation for
observed hysteresis for the water phase relative permeability prediction of oil relative permeability in a three-phase system:
in both steady-state and unsteady-state measurements,
however, it was more pronounced in the unsteady-state tests.
SPE 81021 3

(S L − S w ) 3 Later, Stone (1973) developed another model (known as


k ro = (S w + S L − 2S Lr ) (2) Stone II Model) for prediction of three-phase oil relative
(1 − S Lr ) 4 permeability that made use of all four two-phase relative
permeability relationships form the oil/gas and oil/water
where Sw is water saturation, SL represents sum of oil and system. He defined σw and σg as probabilities of contributions
water saturation and SLr is residual liquid saturation. to flow in the two-phase systems:
Stone (1970) introduced a model based on the channel
σw = krow + krwo (9)
flow theory which treats the porous medium as an assemblage
of flow channels and assumes that in any flow channel there is σg = krog + krgo (10)
mostly only one mobile fluid. Wettability determines the
occupancy of different sized channels. The intermediate where krwo and krgo are relative permeability to water and
wetting phase separates the wetting phase in smallest channels relative permeability to gas, respectively, when oil is the
from the non-wetting phase in the largest. As a consequence, second phase. Under three-phase flow conditions:
having the same saturation history and the saturation of the kro + krgo + krw = σw⋅σg (11)
wetting phase (water), it will occupy the same flow channels
in a three-phase system as in a two-phase system. It also and solving for kro, one gets:
means that water relative permeability will be a function of its kro = ( krow + krwo )⋅( krog + krgo ) - ( krwo + krgo ) (12)
own saturation and it will be the same in a two-phase system
as in a three-phase system. Similarly, for the same saturation Hirasaki (Dietrich and Bondor, 1976) defined the
history, the relative permeability of the non-wetting phase reduction in oil relative permeability due to the presence of a
(gas) will not change in a three-phase system compared to the third phase as:
two-phase one. The relative permeability of the intermediate
wetting phase would depend on which intermediate channels it
∆ = (k row max − k row ) + (k row max − k rog ) − (S w + S o ) ⋅
occupied, which in turn would depend on the saturations of (k row max − k row )(k row max − k rog )
water and gas. He developed the following relationship for oil ⋅
relative permeability: k row max

k ro = S onor ⋅ β w ⋅ β g (3) (13)


where krowmax is two-phase oil relative permeability measured
S o − S om at connate water saturation. The first and second term account
S onor = (4)
for the blockage of oil due to presence of water and gas, and
1 − S wc − S om
the third term is an adjustment for the interaction between
where Som is three-phase residual oil saturation and Swc the water and gas blockage mechanism. The resulting expression
connate water saturation. Function βw depends only on water for the oil relative permeability becomes:
saturation and βg is function only of gas saturation:

k row (k row max − k row )(k row max − k rog


βw = (5) k ro = k row + k rog − k row max + (S w + So )
k row max
1 − S wnor
(14)
k rog
βg = (6) Aziz and Settari (1979) modified both Stone's models to
1 − S gnor overcome their limitation in that the original models would
reduce to the two-phase data only if end-point relative
S w − S wc permeabilities were equal to one. The modified Stone I
S wnor = (7) model is:
1 − S wc − S om
Sg
S gnor = (8) kro = krowmax⋅Sonor⋅βw⋅βg (15)
1 − S wc − S om
k row
In the above expressions, krow and krog represent the βw = (16)
relative permeability to oil in oil-water system and the relative
k row max (1 − S w nor )
permeability to oil in oil-gas system, respectively. Stone
pointed out that Som can be a function of saturation but could k rog
βg = (17)
be taken to be constant as an approximation. Som can also be k row max (1 − S gnor )
used as an adjustable parameter to minimize the deviations
between experimental and calculated values of the oil
relative permeability.
4 SPE 81021

Stone's second model was modified as: where Sgro is residual gas saturation in the oil/gas two-phase
system. The assumption in deriving this relationship is that the
 k  k rog   (18)
k ro = k row max   row + k rwo  + k rgo  − (k rwo + k rgo ) end points of the two-phase data are the same as in the three-
 k row 
 max  k row max   phase system. The weighting parameter (Sw-Swc) could also be
so chosen to match permeabilities at some other
Fayers and Matthews (1984) proposed a correlation for weighting point.
finding Som, the free parameter in the Stone I model:
Pope (Delshad and Pope, 1989) proposed a three-phase
Som = α⋅Sorw + (1 - α)⋅Sorg (19) model in which the two-phase relative permeability do not
Sg appear explicitly:
α = 1− (20)
1 − S wc − S org k ro =
1
2
[
k rowmax (Sonor )eog ⋅ (1 − Swnor )eow − eog + (Sonor )eow ⋅ (1 − Sgnor )eog − eow ]
It has been observed that the above correlation leads to (28)
significant over-prediction of residual oil saturations in The exponents eow and eog are found by fitting
presence of trapped gas. For that particular situation, Fayers two-phase data:
and Matthews proposed following relationship:
Som = Sorw - 0.5⋅Sg (21)
krow = krowmax⋅(Sownor)eow (29)
In their analysis of the three-phase relative permeability eog
models, they concluded that the Stone I model is superior to krog = krogmax⋅(Sognor) (30)
the Stone II.
Sow − Sorw
Aleman and Slattery (1988) developed the following Sownor = (31)
1 − Swc − Sorw
model for the evaluation of oil relative permeability:
Sog − Sorg
k (1 − k rgos )(k rows − k rwos ) − k rwos (1 − k rows )(k rgos − k rgos ) Sognor = (32)
k ro = k row max ⋅ Sonor ⋅ rogs
(1 − k rgos )(k rows − k rwos ) − (1 − k rows )(k rgos − k rogs )
1 − Swrog − Sorg − Sgro
(22) where Swrog is the residual water saturation at which the oil/gas
two-phase experiment has been conducted. The three-phase oil
residual saturation Som (included in Sonor) could be estimated
where the normalization of two-phase functions has been done using Fayers and Matthews relationship or it could be used as
according to free, adjustable parameter. If history data matching is primary
goal then exponents in the equation as well as the constant
k row / k row max (1/2) can be also replaced by free parameters.
k rows = (23)
S o nor Kokal and Maini (1989) addressed two limitations of
Stone's first model as modified by Aziz and Settari: 1) the
k rog / k rog max measurements of two-phase oil/gas relative permeability are
k rogs = (24) not always done at connate water saturation, and 2) often the
S o nor
measured values of krowmax and krogmax are not equal. They
modified Stone's first model according to:
k rwo / k rwo max
k rwos = (25)
k rog
S w nor k row
k ro = Sonor ⋅ ⋅
k row max (1 − Swnor ) k rog max (1 − Sgnor )
k rgo / k rgo max
k rgos = (26)
S g nor
k rog max Sgnor + k row max Swnor
Testing the model on data of Saraf and Fatt (1967), Corey ⋅ (33)
1 − Sonor
et al. (1970) and Dalton et al. (Stone, 1970), the authors
observed a marginal difference compared to Stone model I as Hustad and Hansen (1995) presented a model that
modified by Aziz and Settari (1979). includes all six residual values from three two-phase
Baker (1988) used saturation-weighted interpolation experiments. They used the following interpolation scheme for
between oil-water and oil-gas data to obtain three-phase oil oil isoperm:
relative permeability: Sg
Sw
k ro (Somnx x ) = ⋅ k row (Somnx ) + ⋅ k rog (Somnx )
(Sw − Swc )k row + (Sg − Sgro )k rog S w + Sg S w + Sg
k ro = (27)
(Sw − Swc ) + (Sg − Sgro ) (34)
SPE 81021 5

Oil saturation is normalized between the values Somn and Somx: − DL


R 
N k ∝  k  (45)
 Ro 
So − Somn
Somnx = (35) Having capillary pressure equal to
Somx − Somn

S w ⋅ Sorw + Sg ⋅ Sorg + Sorg ⋅ Sorw (So − 1) pc = (46)
Somn = (36) Rk
Sg (1 − Sorw ) + S w (1 − Sorg )
and saturation of the wetting phase (water) given as
Sw ⋅ Sgro + Sg ⋅ Swro + Sgro ⋅ S wro (So − 1) 2− D L
Somx = (37) R 
Sg ⋅ Swro + S w ⋅ Sgro S w =  k  (47)
 Ro 
Goodyear and Townsley (Balbinski et al., 1997) proposed
following relationship: the fractal dimension is found from the slope of the capillary
pressure - water saturation curve in log-log coordinate system
k ro (So ) = k rog (Sogn ) f ( α ) ⋅ k row (Sown )1−f ( α ) (38)
1
p c ∝ (Sw )DL −2 (48)
defining values of α as

Sg Finally, applying Poiseuille's law to each capillary, Moulu


α= (39) et al. came to the following three-phase oil relative
Sg + Sw − Swc permeability relationship:

k ro = k row  (So + Sw − Sorg )2−DL − (Sw + Sorw )2−DL 


and arguments of the two-phase oil relative permeability  4− D L 4−D L

functions as functions of the normalized saturations  
Sogn = Sorg + Sonor (1 − SLrg ) (40) (49)

Sown = Sorw + Sonor (1 − S wc − Sorw ) (41) The authors extended the model to oil-wet and
intermediate-wet conditions (Moulu et al., 1999) making use
The function f(α) could be chosen arbitrarily but must of the wettability index WI (equals one for water-wet and -1
satisfy the conditions that f(0)=0 and f(1)=1. The form f(α)=α for oil-wet medium):
also satisfies these conditions.
 4− D L 4− D L

Moulu et al. (1997) extended the work from Vizika k ro = (1 − m) So 2−DL − Soi 2−DL  +
(1993) proposing a theoretical model that treats rock porous
 
structure as a set of fractal pores. The model is applicable for
+ m ⋅ k row  (So + Sw − Sorg )2−DL − (S w + Sorw )2−DL 
4− D L 4− DL
strongly water wet medium and for oil spreading coefficient
greater than zero. The porous medium is considered as bundle  
of parallel capillary tubes with a fractal cross-section. An
iterative procedure is required for constructing such cross (50)
section - a half perimeter of a circle is divided into η parts and where m is defined as
then each part is replaced by half a perimeter of another
smaller circle. Starting with some tube radius Ro, at some step WI + 1 (51)
m=
k there are Nk new groves with radius Rk and cross section 2
area Ak: Blunt (1999) introduced a three-phase oil relative
permeability model based on saturation-weighted interpolation
π k
Rk = ( ) ⋅ Ro (42) that uses "flowing saturations" as arguments in two-phase
2η permeability functions. The model claims to account for oil
trapping and film drainage effect. The film drainage model
k
2 π 
1 2 treats oil saturation as composed from bulk oil (Sob) and oil in
A k = πR o   (43) the films (Sol):
2  4η 
So = Sob + Sol (52)
Nk = ηk (44) The oil films can exist only between the gas phase in the
At the same time, number of grooves is related to the so called pore center and the water phase wetting the solid surface, and
fractal dimension DL in following way: its saturation is given by
6 SPE 81021

 Sg ⋅ Sodrain  saturation is given by:


Sol = min , So  (53)
 S  Sofb = min ( So-Sol, max ( Shf-Sgf-Sol, 0 ) ) (63)
 gdrain 
The subscript "drain" indicates situation when the oil exists
only in layers; Sodrain would be equal to the minimum oil Finally, the three-phase oil relative permeability is
obtained by:
saturation reached in the experiment. The oil relative
permeability in the film flow is proportional to the relative (S w − S wc ) ⋅ (a o ⋅ k row (Sofb ) + b o ⋅ k rgw (Sofb ))
permeability to gas: k ro (So ) = α ⋅ +
(S w − S wc ) + (Sg − Sgr )
k rg ⋅ k rogdrain 2
k ol = 2
⋅ Sol (54)
k rgodrain ⋅ Sodrain (Sg − Sgr ) ⋅ (a o ⋅ k ob (Sofb ) + b o ⋅ k rgo (Sofb ))
+α⋅ +
The bulk oil relative permeability is given by following (S w − S wc ) + (Sg − Sgr )
difference:

So
kob(Sob) = krog(So) - kol(Sol) (55) + α ⋅ k ol (Sol ) + β ⋅ (a o ⋅ k row (S hf ) + b o ⋅ k rgw (Shf ))
So + Sg
Trapping is considered by introducing flowing and
trapped fractions of the saturation (64)

S = Sf + St (56) The α, β, ao and bo are functions of interfacial tension and


fluid density that take into account compositional consistency
and for the water-wet rock, flowing gas saturation is equal i.e. variation in fluid properties due to change in pressure and
to: composition. If this effect is not significant, the functions are
assigned values: 1, 0, 1, 0, respectively. Blunt tested his model
1  4 
Sgf = (Sg − Sgr ) + (Sg − Sgr ) 2 + (Sg − Sgr )  on data from Oak(1990) and observed improvement compared
2  Cg 

to the model from Baker (1995) that uses saturation-based
interpolation only.
(57) Data and Models Selected for Testing
Residual gas saturation is found from In order to be able to test the correlations of interest in
this work, one needs not only the three-phase data itself but
Sg max also at least four relative permeability curves from the two-
Sgr = (58)
1 + C g ⋅ Sg max phase oil/water and oil/gas displacements: krow, krwo, krog
and krgo. Only a limited number of published experimental
studies provide this information. In this work, the
experimental results reported by Maini et al. (1990), Oak
and the trapping constant is equal to (1991), and Donaldson and Dean (1966) were used for testing
1 various correlations. The first two have been mentioned
Cg = − Sg max (59) previously; Donaldson and Dean conducted gas-displacement
Sgrw tests of oil and water on Berea sandstone using distilled water,
a paraffinic oil and air as the fluids. The core had 41.6 %
Similarly, the flowing hydrocarbon saturation is porosity, 200 md absolute permeability and connate water
saturation of 31 %. After each test, the core was cleaned with
1  4 
 isopropyl alcohol, dried to its original weight, and saturated
Shf = (S − S ) + (S − S ) 2
+ (S − S )
2  
h hr h hr h hr with water for the next run. The terminal saturation that
Ch  authors referred to, represents saturation at the end of the core
(60) not the average one. The relative permeabilities were
evaluated by extension of Welge's (1952) two-phase
with residual hydrocarbon saturation unsteady-state method to three-phase flow.
Sh max In selecting the models for testing there were certain
Shr = (61) conditions to be fulfilled. One of them is, of course, to be have
1 + C h ⋅ Sh max
all data required by a particular model from the published
and the hydrocarbon trapping constant experimental work on three-phase oil relative permeability.
For the models tested by others on different data sets, it was
Ch = min ( Co, Cg ) (62) necessary that the model gave reasonable predictions
The Shmax, maximum hydrocarbon saturation, is a free compared to other models for that particular data set.
parameter to be optimized. Now, the bulk oil flowing
SPE 81021 7

The original versions of Stone I and Stone II Models, Pope (1989) Model
although they represent fundamental work and had inspired
The Figures 15., 16. and 17. show predictions of Pope's
many other authors, were not tested because it has been shown
model. The residual oil saturation (Som) for this model was a
in the past that their modified versions posses better
free parameter and found by minimizing squared difference
prediction capabilities.
between experimental and calculated values of oil relative
Results permeability. This model matched Donaldson's data good even
at very low oil saturations. However, the model had much
Hirasaki (1976) Model
difficulties in predicting other two data sets.
The oil relative permeability prediction of this model is
Kokal and Maini (1989) Model
illustrated in Figures 3, 4 and 5. The prediction is not very
good for any of the three data sets. The largest errors are in the The oil relative permeabilities calculated using Kokal and
range of low oil saturations (which is actually the biggest Maini Model are shown in Figures 18., 19. and 20. The match
challenge for every model). Data from Donaldson&Dean and between calculated and experimental values for the Maini et
Oak are both over- and underpredicted depending on the al. data set is good including the most challenging part - low
saturation range; Maini et al. data are consistently oil saturation range. In the case of Donaldson's data, the model
overpredicted. No free parameters are contained in this model predictions are better at higher oil saturations than at lower
to improve the fit with experimental data. ones. This model, being modification of Stone I Model that is
developed for water-wet medium, did not do well in the case
Aziz and Settari (1979) Modification of Stone’s Model I
of Oak's data most probably due to the intermediate wet
The predictions from this model are illustrated in Figures rock property.
6, 7 and 8. The model did well for the Maini et al. experiment
Goodyear and Townsley (1999) Model
and especially for the most difficult part - low oil saturations.
The results for Donaldson and Dean experiment are very The prediction capability of the model of Goodyear and
reasonable except the overprediction at the water saturations Townsley is illustrated in Figures 21., 22. and 23. When
close to connate. Recalling the fact that this model was neither water nor gas is close to its residual value, the oil
developed for water-wet medium, it is not surprising that the relative permeabilities calculated by this model match data of
prediction of Oak's data (intermediate-wet core) is poor. Maini et al. relatively good. In the case of intermediate core
(Oak), the predictions are better close to the residual values of
Aziz and Settari (1979) Modification of Stone’s Model II
water or gas than otherwise. Matching of data for the
The effectiveness of this model is shown in Figures 9, 10 Donaldson's experiment is considered as absence of prediction
and 11 and it appears to be less satisfactory than the previous capability completely; the model was not able to predict the
model in matching these data sets. The model over-predicts two highest isoperms and the remaining isoperms (with one
oil relative permeability compared to the experimental results exception) were not reasonable.
of Maini et al. This over-prediction is more pronounced when
Blunt (1999) Model
water and gas saturations are comparable and is worse at low
oil saturations. With the other two data sets, both over- and The prediction ability of the Blunt's model for the three
under-prediction of the oil relative permeability is seen. The data sets is illustrated in Figures 24., 25. and 26. The
model predictions become less reliable in the more important maximum hydrocarbon saturation in three-phase flow (Shf)
zone of low oil saturations. was treated as free parameter and optimized in order to
minimize sum of squared differences between the
Baker (1988) Model
experimental and calculated values of oil relative
The predictions of this model for the experimental data permeability. That worked well for the Oak's data, however, in
sets tested are given in Figures 12, 13 and 14. The calculated the Donaldson's case, the lower oil relative permeability, the
values of oil relative permeability appear to be very more difficulties by predicting it with this model. Moreover,
reasonable. The model is remarkable for its simplicity and the model was not capable in predicting the lowest three oil
although the deviation between calculated and experimental isoperms in Maini et al. case.
values are not negligible and vary depending on a saturation
region, overall, the model did better than many other much
more complicated models. In the original version (equation 6. Models Comparison
27.) krow and krog are regarded as functions of single
In an attempt to quantify the effectiveness of each model
saturations Sw and Sg, respectively. In this study, normalized
for a particular experimental data sets, the sum of squared
saturations have been used and residual oil saturation Sor had
deviations between the experimental and calculated values of
entered the equation as a part of the argument of the two-phase
three-phase relative permeabilities has been used
functions. Treating Som as free parameter and finding its
optimum value which would minimize the squared differences
between experimental and calculated values of three-phase
n
relative permeabilities, increased the accuracy of the model to ΣDEV = ∑ (kro exp,i − kro calc,i ) 2 (65)
some extent. i=1
8 SPE 81021

where the constant n represents the total number of Sw – water saturation


experimental data points for a given experimental data set. So Swc – connate water saturation
defined, ΣDEV should be used to compare the models only WI – wettability index
within the same data set. Values of ΣDEV are given in the
Table 3., and the graphical illustration of it is in the Figure 27. Greek Letters
∆ - difference
It is obvious that none of the models was the best choice
α - defined by eqn. 20.
for all three data sets. One can only look for models that more
γ - interfacial tension
frequently performed better than others. Choosing only three
models per data set that performed better than others, one can βw - defined by equation 5.
see that following models would be chosen twice: Baker's (for βg - defined by equation 6.
Donaldson and Oak data), Kokal and Maini ( for Donaldson σ - proability funcion defined by eqn. 9. or 10.
and Maini et al. data) and Goodyear & Townsley model (for ΣDEV – sum of deviations defined by eqn. 65.
Maini et al. and Oak data). The disadvantage of the last
mentioned model is its inability to predict all the isoperms of Subscripts
the remaining third data set. On the other side of the coin, the g - gas
models that appeared to have least success in predicting oil L - liquid
relative permeability of the selected data sets were the Lr - residual liquid
Hirasaki model and the Aziz & Setttari II (modification of max – maximum
Stone II model). It is possible that these models will work n – normazlization according to eqns. 40. and 41.
quite well with other data sets. It is also apparent that models nor – normalized saturation
that work well with water-wet systems perform poorly with o - oil
intermediate wettability. OG – oil/gas system
s – normalized relative permeability according to eqns. 23.-26.
Conclusions t - trapped
• The accuracy of current three-phase oil relative w - water
permeability models is strongly influenced by the WG – water/gas system
wettability of the medium being considered. WO – water/oil system

• No single three-phase model was capable of satisfactorily References


matching the three selected data sets. Aleman, M.A., Slattery, J.C., 1988. Estimation of three-phase relative
permeabilities. Transport in Porous Media, 3: 111.
• There is a need for a more robust three-phase relative Aziz, K., Settari, A., 1979. Petroleum Reservoir Simulation. Applied
permeability model that would be applicable for various Science Publishers, London.
wettabilities. Baker, L.E., 1988. Three-phase relative permeability correlations.
SPE/DOE Enhanced Oil Recovery Symposium, Tulsa, Apr. 17-20
Nomenclature
Baker, L.E., 1999. Three-Phase Relative Permeability of Water-Wet,
C – Land trapping constant
Intermediate-Wet, and Oil-Wet Sandstone. New Developments in
DL – fractal dimension Improved OIl Recovery. Geol. Soc. London Special Publication
eow – fitting parameter in oil/water system No. 84, 51
eog – fitting parameter in oil/gas system
Balbinski, E.F., Fishlock, T.P., Goodyear, S.G., Jones, P.I.R., 1999.
kro – three-phase relative permeability to oil Key characteristics of three-phase oil relative permeability
kro,exp – experimental value of oil relative permeability formulations for improved oil recovery predictions. Petroleum
kro,calc – oil relative permeability as predicted by a model Geoscience, 5, 4: 339.
krow – two-phase relative permeabilty to oil in Blunt, M.J., 1999. An Empirical Model for Three-Phase Relative
oil/water system Permeability. SPE Annual Technical Conference and Exhibition,
krog – two-phase relative pernmeability to oil in Houston, Oct. 3-6.
oil/gas system Corey, A.T., 1954. The interrelation between gas and oil relative
krwo – two-phase relative permeabilty to water in permeabilities. Producers Monthly, Nov., 38.
oil/water system Corey, A.T., Rathjens, C.H, Henderson, J.H., Wyllie, M.R.J., 1956.
krgo - two-phase relative pernmeability to gas n oil/gas system Three-phase relative permeability. Petroleum Transactions,
m – defined by eqn. 51 207: 349.
p - pressure Delshad, M., Pope, G.A., 1989. Comparison of the three-phase oil
S - saturation, spreading coefficient relative permeability models. Transport in Porous Media, 4: 59.
Sh – hydrocarbon saturation Dicarlo, D. A., Sahni, A., Blunt, M.J., 2000. The Effect of
Shr – residual hydrocarbon saturation Wettability on Three-Phase Relative Permeability. Transport in
Sob – bulk oil saturation Porous Media, 39, 3: 347.
Sol – oil saturation in layers Dietrich, J.K., Bondor, P.L., 1976. Three-phase oil relative
Som – three-phase residual oil saturation permeability models. 51st Annual Fall Tecnical Conference and
Sorw – two-phase residual oil saturation in oil/water system Exhibition of the Society of Petroleum Engineers of AIME, New
Sorg – two-phase residual oil saturation in oil/gas system Orleans, Oct. 3-6.
SPE 81021 9

Donaldson, E.C., 1966. Two- and three-phase relative permeability Maini, B.B., Kokal, S.L., Nicola, F., 1988. Measurements of Three-
studies. U.S. Dept. of the Interior, Bureau of Mines, Report 6826. Phase Relative Permeability at Elevated Temperatures and
Donaldson, E.C., Kayser, M.B., 1981. Three-phase fluid flow in Pressures. PRI Research Report 1988-15, Calgary, Canada.
porous media. U.S. Department of Energy, DOE/BETC/IC-80/4. Maini, B.B., Kokal, S., Jha, K., 1989.Measurements and correlations
Eleri, O.O., Graue, A., Skauge, A., 1995. Steady-State and Unsteady- of three-phase relative permeability at elevated temperatures and
State Two-Phase Relative Permeability Hysteresis and pressures. 63rd Annual Technical Conference and Exhibition of
Measurements of Three-Phase Relative Permeabilities Using the Society of Petroleum Engineers,
Imaging Techniques. SPE Annual Technical Conference & San Antonio, Oct. 8-11.
Exhibition, Dallas, Oct. 22-25. Maini, B.B., Nicola, F., Goldman, J., Sarma, H.K., 1990.
Eleri, O.O., Graue, A., Skauge, A., Larsen, J.A., 1995. Calculation of Measurements and Estimation of Three-Phase Relative
Three-Phase Relative Permeabilities from Displacement Permeability. Petroleum Recovery Institute, Calgary,
Experiments with Measurements of In-Situ Saturation. Soc. Core Canada, October.
Anal. Int. Symp., San Francisco. Moulu, J.C., Vizika, O., Kalaydjian, F., Duquerroix, J.P., 1998. A
Fayers, F.J., Matthews, J.D., 1984. Evaluation of normalized Stone's New Model to Calculate Three-Phase Relative Permeabilities:
methods for estimating three-phase relative permeabilities. Society Application and Validation for Sandstone. Revue de l'Institut
of Petroleum Engineers Journal, April. Francais du Petrole, 53, 4: 395.
Fayers, F.J., 1987. Extension of Stone's method I and conditions for Moulu, J-C., Vizika, O., Kalaydjian, F., Duquerroix, J-P., 1997. A
real characteristics in three-phase flow. 62nd Annual Technical New Model for Three-Phase Relative Permeabilities Based on a
Conference and Exhibition of the Society of Petroleum Engineers, Fractal Representation of the Porous Medium. SPE Annual
Dallas, Sept. 27-30. Technical Conference and Exhibition, San Antonio, Oct. 5-8.
Hui, M.H., Blunt.M.J., 2000. Effects of Wettability on Three-Phase Moulu, J-C., Vizika, O., Egermann, P., Kalaydjian, F., 1999. A New
Flow in Porous Media. J. Phys. Chem. B, 104, 116: 3833. Three-Phase Relative Permeability Model For Various Wettability
Hustad, O.S., Hansen A.G., 1995. A consistent correlation for three Conditions. SPE Annual Technical Conference and Exhibition,
phase relative permeabilities and phase pressures based on three Houston, Oct. 3-6.
sets of two phase data. 8th EAPG Impr. Oil Recovery European Naar, J., Wygal, R.J., Henderson, J.H., 1962. Imbibition Relative
Symp., Vienna, 1: 289. Permeability in Unconsolidated Porous Media. Society of
Jerauld, G.R., 1997. General Three-Phase Relative Permeability Petroleum Engineers Journal, March.
Model for Prudhoe Bay. SPE Reservoir Engineering - Society of Oak, M.J., Baker, L.E., Thomas, D.C., 1988. Three-phase relative
Petroleum Engineers, 12, 4: 255. permeability of Berea sandstone. SPE/DOE Enhanced Oil
Kalaydjian, F.J-M, Moulu, J-C., Vizika, O., Munkerud, P.K., 1997. Recovery Symposium, Tulsa, Apr. 17-20.
Three-phase flow in water-wet porous media: Gas/oil relative Oak, M.J., 1990. Three-Phase Relative Permeability of Water-Wet
permeabilities for various spreading conditions. Journal of Berea. SPE/DOE Seventh Symposium on Enhanced Oil Recovery,
Petroleum Science and Engineering, 17, 3: 275. Tulsa, Apr. 22-25.
Kokal, S., Maini, B., 1989. An improved model for estimating three- Oak, M.J., 1991. Three-phase relative permeability of intermediate-
phase oil-water-gas relative permeabilities from two-phase oil- wet Berea sandstone. 66th Annual Technical Conference and
water and oil-gas phase. 40th Annu. Cim. Petrol. Soc. Tech. Mtg., Exhibition of the Society of Petroleum Engineers, Dallas, Oct. 6-9.
Banff, 2: 56-1. Parker, J.C., Lenhard, R.J., Kuppusamy, T., 1987. A Parametric
Land, C.S., 1968. Calculation of imbibition relative permeability for Model for Constitutive Properties Governing Multiphase Fluid
two- and three-phase flow from rock properties. Society of Conducting in Porous Media. Water Resources Research, 23: 618.
Petroleum Engineers Journal, June, 149. Robinson, R.L., Slattery, J.C., 1994. Estimation of Three-Phase
Land, C.S., 1971. Comparison of calculated with experimental Relative Permeabilities. Transport in Porous Media, 16, 3: 263.
imbibition relative permeability. Society of Petroleum Engineers Sahni, A., Guzman, R., Blunt, M., 1996. Theoretical Analysis of
Journal, Sept., 419. Three Phase Flow Experiments in Porous Media. SPE Annual
Larsen, J.A., Skauge, A., 1995. Comparing Hysteresis Models for Technical Conference and Exhibition held in Denver, Oct. 6-9.
Relative Permeability in WAG (Water Alternating Gas) Studies. Saraf, D.N., Batycky, J.P, Jackson, C.H., Fisher, D.B., 1982. An
Soc. Core Anal. Int. Symp., San Francisco. experimental investigation of three-phase flow of water-oil-gas
Leverett, M.C., Lewis, W.B., 1940. Steady flow of gas-oil-water mixtures through water wet sandstones. California Regional
mixtures through unconsolidated sands. Petroleum Technology, Meeting of the Society of Petroleum Engineers, San Francisco,
May, 107. March 24-26.
Table 1. Core Data in Maini et al. Experiment
sand 140 - 200 mesh
core length 46.7 cm
area 25 cm2
pore volume 409 ml
absolute permeability 3.6 Darcy
pore pressure 3.45 MPa
10 SPE 81021

Table 2. Core Data in Oak Experiment


Core length 7.6 cm
Diameter 5 cm
Absolute permeability 1010 md
Wettability altering agent organosilane

Table 3. Values of ΣDEV of the oil relative permeability models

Experimental Data Set

Donaldson&Dean Maini et al. Oak


1966 1990 1991
Hirasaki
0.09344 0.45020 0.09574
1976
Aziz&Settari-I
0.05305 0.04196 0.08359
1979
Aziz&Settari-II
0.09168 0.46770 0.09816
1979
M o d e l

Baker
0.03863 0.40546 0.06198
1988
Pope
0.03536 0.67184 0.06503
1989
Kokal&Maini
0.03820 0.02773 0.09491
1989
Goodyear and
1.05187 0.39000 0.06179
Townsley, 1999
Blunt
0.07519 0.43240 0.02080
1999
SPE 81021 11

Gas Gas

0
0 100
100 kro,exp=0.001
kro,exp=0.01
kro,exp=0.1
kro,exp=0.01
kro,exp=0.2 kro,exp=0.05
kro,exp=0.4
20 kro,exp=0.6 20
80 80

40 40
60 60

60 60
40 40

80 80
20 20

100 100
0 0
Water 0 20 40 60 80 100 Oil Water 0 20 40 60 80 100 Oil

Figure 1. Oil isoperms of the experiment of Maini et al. Figure 2. Oil isoperms of the experiment of Oak

Gas Gas

0 0
100 100
kro,exp=0.01 kro,exp=0.01
kro,exp=0.08 kro,exp=0.1
kro,exp=0.2 kro,exp=0.2
kro,exp=0.4 kro,exp=0.4
kro,exp=0.65 20 kro,exp=0.6
20 80 kro,calc
80 kro,calc

40 40
60 60

60 60
40 40

80 80
20 20

100 100
0 0
Water 0 20 40 60 80 100 Oil Water 0 20 40 60 80 100 Oil

Figure 3. Hirasaki's model prediction of Donaldson and Dean Figure 4. Hirasaki's model prediction of Maini et al. experiment
experiment

Gas Gas

0
0 100
100 kro,exp=0.01
kro,exp=0.001 kro,exp=0.08
kro,exp=0.01 kro,exp=0.2
kro,exp=0.05 kro,exp=0.4
kro,calc kro,exp=0.65
20
20 80 kro,calc
80

40
40 60
60

60
40
60
40

80
20
80
20

100
0
100 Water 0 20 40 60 80 100 Oil
0
Water 0 20 40 60 80 100 Oil

Figure 5. Hirasaki's model prediction of Oak experiment Figure 6. Aziz and Settari I Model prediction of Donaldson and
Dean experiment
12 SPE 81021

Gas
Gas
0
100 0
kro,exp=0.01 100
kro,exp=0.1 kro,exp=0.001
kro,exp=0.2 kro,exp=0.01
kro,exp=0.4 kro,exp=0.05
kro,exp=0.6 kro,calc
20
80 kro,calc 20
80

40
60 40
60

60 60
40 40

80 80
20 20

100 100
0 0
Water 0 20 40 60 80 100 Oil Water 0 20 40 60 80 100 Oil

Figure 7. Aziz and Settari I Model prediction of Maini et al. Figure 8. Aziz and Settari I Model prediction of Oak exper
experiment
Gas Gas

0 0
100 100
kro,exp=0.01 kro,exp=0.01
kro,exp=0.08 kro,exp=0.1
kro,exp=0.2 kro,exp=0.2
kro,exp=0.4 kro,exp=0.4
20 kro,exp=0.65 kro,exp=0.6
80 20
kro,calc 80 kro,calc

40 40
60 60

60 60
40 40

80 80
20 20

100 100
0 0
Water 0 20 40 60 80 100 Oil Water 0 20 40 60 80 100 Oil

Figure 9. Aziz and Settari II Model prediction of Donaldson & Figure 10. Aziz and Settari II Model prediction of Maini et al.
Dean expe experiment
Gas
Gas

0
100 0
kro,exp=0.001 100
kro,exp=0.01
kro,exp=0.01
kro,exp=0.08
kro,exp=0.05 kro,exp=0.2
kro,calc kro,exp=0.4
20 20 kro,exp=0.65
80 80 kro,calc

40 40
60 60

60 60
40 40

80 80
20 20

100 100
0 0
Water 0 20 40 60 80 100 Oil Water 0 20 40 60 80 100 Oil

Figure 11. Aziz and Settari II Model prediction of Oak Figure 12. Baker's Model prediction of Donaldson & Dean
experiment experiment
SPE 81021 13

Gas
Gas
0
100
0 kro,exp=0.001
100
kro,exp=0.01 kro,exp=0.01
kro,exp=0.1 kro,exp=0.05
kro,exp=0.2 kro,calc
kro,exp=0.4 20
20 kro,exp=0.6 80
80 kro,calc

40
40 60
60

60 60
40 40

80 80
20 20

100
0 100
0 20 40 60 80 100
0
Water Oil 0 20 40 60 80 100
Water Oil

Figure 13. Baker's Model prediction of Maini et al. experiment Figure 14. Baker's Model prediction of Oak experiment
Gas Gas

0 0
100 100
kro,exp=0.01 kro,exp=0.01
kro,exp=0.08 kro,exp=0.1
kro,exp=0.2 kro,exp=0.2
kro,exp=0.4 kro,exp=0.4
20 kro,exp=0.65 20 kro,exp=0.6
80 kro,calc 80 kro,calc

40 40
60 60

60 60
40 40

80 80
20 20

100 100
0 0
Water 0 20 40 60 80 100 Oil Water 0 20 40 60 80 100 Oil

Figure 15. Pope's Model prediction of Donaldson & Dean Figure 16. Pope's Model prediction of Maini et al. experiment
experiment
Gas Gas

0 0
100 100
kro,exp=0.001 kro,exp=0.01
kro,exp=0.01 kro,exp=0.08
kro,exp=0.05 kro,exp=0.2
kro,calc kro,exp=0.4
20 20 kro,exp=0.65
80 80 kro,calc

40 40
60 60

60 60
40 40

80 80
20 20

100 100
0 0
Water 0 20 40 60 80 100 Oil Water 0 20 40 60 80 100 Oil

Figure 17. Pope's Model prediction of Oak experiment Figure 18. Kokal & Maini Model prediction of Donaldson & Dean
experiment
14 SPE 81021

Gas
Gas

0
0 100
100 kro,exp=0.001
kro,exp=0.01 kro,exp=0.01
kro,exp=0.1 kro,exp=0.05
kro,exp=0.2 kro,calc
kro,exp=0.4
kro,exp=0.6 20
20 80
80 kro,calc

40
40 60
60

60
60 40
40

80
80 20
20

100 100
0 0
0 20 40 60 80 100 Water 0 20 40 60 80 100 Oil
Water Oil

Figure 19. Kokal & Maini Model prediction of Maini et al. Figure 20. Kokal & Maini Model prediction of Oak experiment
experiment

Gas Gas

0 0
100 100
kro,exp=0.01 kro,exp=0.01
kro,exp=0.08 kro,exp=0.1
kro,exp=0.2 kro,exp=0.2
kro,exp=0.4 kro,exp=0.4
kro,exp=0.65 20 kro,exp=0.6
20 80
80 kro,calc kro,calc

40 40
60 60

60 60
40 40

80 80
20 20

100 100
0 0
Water 0 20 40 60 80 100 Oil Water 0 20 40 60 80 100 Oil

Fig. 21. Goodyear & Townsley Model prediction of Donaldson & Fig. 22. Goodyear & Townsley Model prediction of Maini et al.
Dean experiment experiment
Gas Gas

0 0
100 100
kro,exp=0.001 kro,exp=0.01
kro,exp=0.01 kro,exp=0.08
kro,exp=0.05 kro,exp=0.2
kro,calc kro,exp=0.4
20 20 kro,exp=0.65
80 80 kro,calc

40 40
60 60

60 60
40 40

80 80
20 20

100 100
0 0
Water 0 20 40 60 80 100 Oil 0 20 40 60 80 100
Water Oil

Fig. 23. Goodyear & Townsley Model prediction of Oak Figure. 24. Blunt's Model prediction of Donaldson and Dean
experiment experiment
SPE 81021 15

Gas

0
100
kro,exp=0.01
kro,exp=0.1
kro,exp=0.2
kro,exp=0.4
20 kro,exp=0.6
80 kro,calc

40
60

60
40
0.4

80
0.6 20

100
0
Water 0 20 40 60 80 100 Oil

Figure 25. Blunt's Model prediction of Maini et al. experiment


Gas

0
100
kro,exp=0.001
kro,exp=0.01
kro,exp=0.05
kro,calc
20
80

40
60

60
40

80
20

100
0
Water 0 20 40 60 80 100 Oil

Figure 26. Blunt's Model prediction of Oak experiment

Hirasaki
1 Aziz&Settari I
Aziz&Settari II
Baker
0.8 Pope
Kokal&Maini
Goodyear&Townsley
0.6
DEV

Blunt

0.4

0.2

0
Donaldson&Dean data Maini et al. data Oak data

Figure 27. Models Comparison

You might also like