Society of Petroleum Engineers

SPE 28182

Quantitative Analysis of Radial Effects on Centrifuge Capillary

Pressure Curves
P.L. Forbes,* lnst. Franc;ais du Petrole, and Z.A. Chen and D.W. Ruth,* U. of Manitoba
* SPE Members

Copyright 1994, Society of Petroleum Engineers, Inc.

This paper was prepared for presentation at the SPE 69th Annual Technical Conference and Exhibition held in New Orleans, LA, U.S.A., 25-28 September 1994.

This paper was selected for presentation by an SPE Program Committee following review of information contained in an abstract submitted by the author(s). Contents of the paper,
as presented, have not been reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material, as presented, does not necessarily reflect
any 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 Petroleum Engineers. Permission to copy is restricted to an abstract of not more than 300 words. Illustrations may not be copied. The abstract should contain conspicuous acknowledgment
of where and by whom the paper is presented. Write Librarian, SPE, P.O. Box 833836, Richardson, TX 75083-3836, U.S.A. Telex, 163245 SPEUT.

Abstract and a number of problems associated with fundamental

Much progress has been made during the past decade on the
centrifuge capillary pressure technique. However, many
fundamental problems concerning flow mechanisms in a
centrifuge experiment need to be further understood. These
problems include the definition of boundary conditions on
both endfaces of a core plug, the gravity degradation effect
at low speeds, the characterization of core sample
heterogeneity, and the effect of the radial centrifugal field
distribution inside a core sample. In this paper we present
a quantitative analysis of the radial effect of the centrifugal
field in core plugs for capillary pressure curves. Using the
N factor defined by Christiansen UJ, we re-derive the
equation for converting mean saturations into local
saturations. The results show that when the radial effect is
taken into account, the capillary pressure curve is shifted
to higher values (not lower as Christiansen indicated) with
the magnitude of the changes of saturations being as high
as 3/4 N. Differences higher than 3 saturation units can be
obtained for a standard centrifuge equipment and core
sample geometry.
Introduction
The centrifuge technique has been widely used in the oil
and gas industry for determining petrophysical properties
of reservoir rock core samples, and especially for
measuring experimentally capillary pressure behaviour.
Other applications include measuring wettability, electrical
properties and relative permeability, and performing
recovery studies.
During the past decades, advances have been made in many
aspects of the centrifuge technique, particularly in data
interpretation. However, people have realized in recent
years that the centrifuge technique has not been perfected,
assumptions made in the traditional Hassler-Brunner
analysis remain to be clarified. Recently, a survey on
centrifuge capillary pressure measurements organized by
the Society of Core Analysts (SCA) has shown that
various problems exist with the experimental design,
running, recording, data processing and reporting of
centrifuge experiments. A review addressing pitfalls and
problems in light of the SCA survey can be found
elsewhere [2].
The most pressing problems with the centrifuge technique
are focused on the validity of the fundamental assumptions
associated with the Hassler-Brunner equation. These
include (1) the assumption of zero-capillary pressure at the
outlet endface of the core plug. A great deal of effort has
been made to investigate this assumption but so far no
definitive conclusions have been drawn [3-51; (2) the
omission of core sample heterogeneity effects on the flow
mechanism; (3) the assumption of no vertical gravitational
acceleration effect on the horizontal centrifugal field [6];
and finally (4) the assumption of a one-dimensional
distribution of centrifugal field inside core samples, which
has turned out to be inaccurate. Christiansen [ 1] singled
out the significance of this phenomenon, and established a
two-dimensional model to describe the effect. However, he
did not provide any solutions to quantify the effect on
capillary pressure-saturation relationships. In addition, one
of his conclusions, according to our derivations, turns out
to be wrong. Ayappa et al [7] also proposed a capillary
pressure model for arbitrary shaped samples, but they did
not use their models to develop practical applications.
Others researchers have also pointed out the significance of
the radial effect on capillary pressure information [8-101.
The present paper gives a detailed analysis to quantitatively
characterize the effect of the radial centrifugal field on
capillary pressure curves determination. The factor N

References and illustrations at end of paper

9
2 QUANTITATIVE ANALYSIS OF RADIAL EFFECfS ON
CENTRIFUGE CAPILLARY PRESSURE CURVES
defined by Christiansen [11 is used in a strict derivation of
the saturation equation which is used to convert
experimental datasets from mean saturation values into
local values. Furthermore, modified algorithms based on
any one-dimensional solution are proposed to account for
the radial centrifugal field effect in deriving a capillary
pressure curve.
Hassler-Brunner Theory
Modem centrifuge techniques are primarily based on the
pioneer work by Hassler and Brunner in 1945 [121,
although centrifuges were used much earlier than Hassler-
Brunner [131. The centrifuge method consists in measuring
mean saturation values, S, versus capillary pressure, PCi•
at the inlet end face of a sample at equilibrium during
rotation at various angular velocities, ro. Several
assumptions have to be made in order to obtain an accurate
approximation of explicit solutions (an exact closed-form
solution is unavailable) for a capillary pressure curve, the
most important of which is that a one-dimensional
centrifugal force field exists inside the core plug as shown
in Figure 1. Then the classic Hassler-Brunner saturation
equation, also known as the fundamental equation, is
written as
1 Pc=!~p
S(Pci)
1 +~ J·_5~dx, ........................... (1)
-v 1-Bx
0 endface zero-capillary pressure condition should hold, that
where PCi is the capillary pressure at the inlet endface of
the core plug,
Pci =!~p ro2 (r 0 2- ri 2) ...................................... (2)
and P c the capillary pressure at location r from the
rotating axis inside the core plug.
Pc =!~p ro2(ro2_ r2) ...................................... (3)
Here r E [ri, r 0 ], B = 1- c-!'i)2, x = PPc , ri and r0 are the
'ro Ci
inlet and outlet rotation radii, ~p is the difference in the
fluid pair densities.
Many approximate solutions to obtain S(P c) from Eq. 1
have been proposed [7, 11, 141. The most simplified and
widely-used is the solution suggested by Hassler and
Brunner [121 which is expressed as
dS(Pci)
S(PCi) = SHB(PCi) = "S(PCi) + p Ci dP Ci ••••.•.•• (4)
This solution assumes B = 0. This means in fact that the
centrifugal aspect of the force field is neglected and that the
field is considered to have a constant (uniform) acceleration
inside the whole core plug, as is shown in Figure 1. Other
solutions in the literature considered the B ::~: 0 case,
assuming a one-dimensional distribution of the centrifugal
force field (see Figure 2). One of these solutions, which is
simple and has also proved to be very accurate, was
proposed by Forbes [11]:
10
SPE 28182
- P Ci dS(Pci)] B
S(Pci) = [ S(Pci) + 1+ a dPc; (1- 2)
1
+ ~ (I+ J3)JxJ3 SHB(xPc;) dx ......... (5)
where the factors a and J3 are defined as
' 1--vl-B 2
a= 1+2-vl-B and J3 = ~ ........... ;..................... (6)
Radial Centrifugal Field
As many workers recognized [1,2, 8-10, 151, the
centrifugal field in a centrifuge experiment is not one-
dimensional but has a radial component. Equations 1-3 in
the Hassler-Brunner analysis hold for a short, slender core
plug for which R/L = 0 and ri =r 0 . Here R is the radius
of the sample and L is its length. However for larger
values of R/L, the effect of the core plug width will
become significant. This concern can be visualized by
considering the top-view configuration of the core sample
given in Figure 3. When radial effects are considered, the
capillary pressure is better expressed as
ro2(q2- r2) ........................................ (7)
where f4 is defined as the maximum radius of
rotation from the axis and it is postulated that at f4 the
is, P c = 0. In addition, it is obvious that the maximum
capillary pressure takes place at the minimum rotation ·
radius rt = ri where
Pc =! ~p ro 2(r4 2 - r1 2) ...................................... (8)
1
When taking into account the radial effect, a dimensionless
geometric factor, N, was defined by Christiansen et al
[1,151 as
P CJ radial r42- rt 2
N= - 1 - 1................. (9)
P c1 non-radial r3 2 - rt 2
or in short
R2
N=r32-r12 .................................................... (10)
N is, according to Christiansen [11, a measure of departure
from the capillary pressure distribution, of Eqs. 2 and 3,
caused by the non-linear nature of the actual centrifugal
field, Eqs. 7 and 8.
In order to minimize the radial effect in a centrifuge
experiment, it has long been recommended that short,
narrow core samples be used in the experiment. However,
for the oil and gas industry, the most popular centrifuges
operate with conventional 1" x 1" (2.54 em x 2.54 em)
sample . Under most circumstances the configurations and
sizes used for the rotor buckets and core sample ensure that
radial effects are important in the centrifugal field.
To assess the significance of radial effects, new models
must be constructed. For the saturation material balance
SPE 28182 P. L. FORBES, Z. A. CHEN and D. W. RUTH 3

equation, the following equation can be written (neglecting capillary pressure curve. For a given set of data, the P c1
gravity effects): from Eq. 8 is always greater than from Eq. 2, which
r4 82(r) indicates a move of P c1' at a fixed S, to a higher value.

S(Pc1 )=~ JS(r) J-vR 2- r2Sin(8)2 rd8dr (11-a)

Therefore, as a new insight into data interpretation (Figure
4 shows such a procedure), the capillary pressure curve
1tLR r 1 81 (r) should also be shifted upward to both higher pressure and
where saturation values (Figure 5). This comprehensive change is
81(r) = 0, r1:$;r~3 illustrated by Figure 6, in which three typical solutions are
used for a comparative analysis:
e1(r) = cos- 1( r3/r) r3:$;r:$;r4 ......................... (11-b)
and
• The Hassler and Brunner solution (Eq. 4 and Figure 1),
82(r) = cos- 1( r 1/r ), r1:$;r~2
which assumes B == 0 and N = 0 for a constant acceleration
e2(r) = sin-1( R/r) r2:$;r:$;q ......................... (11-c) field in the core sample;

Both Christiansen et al [1,15] and Ayappa et al [7] • Forbes' improved solution (Eq. 5 and Figure 2), for a
established the above equation. As N == 0, the above one-dimensional centrifugal acceleration distribution in the
double integral will be reduced to the case corresponding to sample (B :;tO, in which the factor N remains zero); and
Eq. I [1] or

-o 1 JI'3 • A new solution of Eq. 11, which will be detailed in the

S (Pc 1)=L S(Pc(r))dr ..................................... (12) present paper, that takes into account the radial distribution
f} of the centrifugal field, for which case neither B nor N is
Here the 0 is added at the upper-right of the saturation
II II
zero (Figure 3).
S(Pc 1). indicating N = 0, and this symbol will be used in
the following discussion. Data Interpretation Techniques

The Effect of the Radial Field on Raw We have discussed the effect of the radial centrifugal
Experimental Drainage Data acceleration on the field distribution in a centrifuge
experiment. This effect requires that capillary pressure
curves for a given set of experimental data must be
The previous analysis gives a mathematical description of interpreted quantitatively on the basis of new equations. In
the effect of the radial centrifugal field on the centrifuge Eq. 7 Pc is a function of r while in Eq. 8 P c1 is not. It
measurements. The significance of such an effect can be
characterized, as Christiansen pointed out Ul, by the fact follows that
that for a single drainage capillary pressure curve, the _(q2- r2)
Pc=Pc1 ( 2 2) ......................................... (13)
synthesized mean saturations obtained with Eq. 11 are q-q
always lower than those obtained from Eq. 1. In other
words, Eq. 1 gives a larger change between the local If we choose B, N, and r3 as variables, all other parameters
saturation curve and the mean saturation curve. can be expressed as functions of these variables.
Specifically
In practice, interpreting centrifuge capillary pressure curves
q=r3ns
from raw experimental data is an inverse process, that is to
say, we measure a fixed set of centrifuge production r2=r3...J,....1--B-+-NB-
perfonnance data, that is mean saturation data, "S(P c1). and r3=r3
we need to convert this set of data into local saturation q=r3..J 1+NB
curve, S(P c1). Neglecting gravitational effects, Eq. 11
R=r3-fNB
gives the correct prediction of mean saturation for a given
capillary pressure curve. It follows that, because Eq. 1 L=r3(1--{i:B)
gives a larger change between the local saturation curve 1+NB- (r/r3)2
and the mean saturation curve, inverse techniques based on Pc=Pc1 B(I+N) ...................................... (14)
Eq. 1, that is techniques which neglect radial effects, will
always underpredict the local saturation. As discussed by A new pair of integrating variables, x and a., are introduced
Forbes [9] and Forbes et al. [16] this conclusion is opposite to convert the integration domain of the core plug to a
to that one of Christiansen [1], which stated that domain with normalized boundaries. As shown in Figure
"centrifuge data that are reduced without accounting for 7, these variables are defined by
radial effects will give capillary pressure relationships that
are shifted to higher-than-correct saturations ...... ". 1- (r/r3)2Cos(8)2
X= B ............................... (15)

(r/r3)~n(e)2 ' O.s.a..s. ~ ............... (16)

Note that the above discussion is based on saturation
conversion only. The radial effect also embodies a change and Sin(a.)2 =
of reference capillary pressure P c1 • For the double
integral, Eq. 8 instead of Eq. 2 has to be used to plot a It follows that

11
4 QUANTITATIVE ANALYSIS OF RADIAL EFFECfS ON SPE 28182
CENTRIFUGE CAPILLARY PRESSURE CURVES

1
Pc =Pel x+N1:o;(a)2 ..................................... (17) =<> -- 1+V 1-B fS(xPcl) dx
S (Pel)
2 ...J 1-Bx
~ =" 1-Bx+BN Sin(a)2 ................................... (18)
0
S(bPc 1) +a [ S(abPc1 )- S(bPc1) ] .... (24)
Sin(S) __ fNB .
Cos(S) - 'J 1-Bx Sm(a) ................................... (19)
a = 3/4
1+{1-B
2
N
( 1+N )
and Where
{
...!L. ................... (25)
b =1+ 0.23 <1+N )
2 r dr d8
r
32
'I,.---m3
-B Cos(a) _
1-Bx dx da .................. (20)
Equation 24 therefore allows the data set to be corrected in
Substituting these definitions into Eq.ll results in a straightforward manner. Note that N = 0 leads to the
0
S(Pc1)= usual approximation ofS = S (that is a= 0 and b = 1).
7t
To test Eq. 24, different capillary pressure curves S(P c1 )
1 2 have been chosen to cover the range of typical curves
Js(P c x+NCos(a) ) ~ic
2 which are encountered using various centrifuge geometries,
1+-fliJ ~
(a )2d a B =0.1 and0.9,andN =0.1. Then theexact~(Pc 1 )and
2 1 1+N _
'I 1-Bx 1t
OS
0 0 S(P c1 ) have been calculated using Eqs. 12 and 23
................. (21)
respectively. Finally, ~ (P c1 ) has been evaluated from
S(Pc1 ) according to Eq. 24. Figure 8 and 9 show different
0
S (P c1 ) curves (solid lines) calculated from Eq. 12 and
their evaluations (the filled circles) according to Eq. 24 for
N = 0.1 (an expected maximum value). Excellent
agreement is obtained in all cases, which shows that Eq.
24 (filled circles) is a very accurate evaluation of~ (P c1 ) •

Quantification of the Radial Effect

As demonstrated in Figures 8 and 9, Equation 24,
accounting for the radial effect, consists in replacing
0
S (Pc 1 ) =S(Pc 1) ............................ : ................... (1)
To obtain S from these equations, we can either by
i). develop some methods to directly invert 0

these equations; or
ii). develop a method for "correcting" the
mean saturation data to obtain an ~ (P c1) that has the
property assumed in a non-radial interpretation, that is
1
-o
S (P c1 ) 1+-fli f S(Pc) .
_~ dx, (correspondmg to Eq. 1
2 'I 1-Bx
0 factor of 1 + 0.23N and that ~ is shifted to a higher
and 12). This equation may then be inverted using one of
the many accurate semi-analytical solutions already
0
available, using S instead of S as non-radial mean
saturation.
We have chosen the second way in order to take advantage
of existing techniques and to provide a simple
methodology to correct for radial effects. The usual
approaches, neglecting the radial effects, consist in
0
assuming N = 0, i.e., S (P c1 ) = S(P c1). In the Appendix
0 of saturation by as much as 3 to 5 saturation units for the
there is an explanation of how S (P c1 ) can be evaluated
for various values of N (N is less than 0.1 for practical
applications). From the Appendix, we obtain
S (P c1 fb) =S(P c1) +a [ S(aPc1)- S(P c1) ] ............. (26)
If we make an assessment, approximating "a" by its
maximun value, 3/4 N, and '"b" by 1 + 0.23N, and
inserting the maximum value of
[ S(aP CJ)- S(P c1) ], ........................................... (27)
that is 1, we will find that P c1 will shift downward by a
value by an amount as high as 3/4 N. In addition, we have
to consider the upward shift of P c1 by a factor of (1 + N)
due to the change in the zero-capillary pressure condition
(Eq.9). Therefore, the final shift in P c1 will be an upward
shift by a factor of about (1 + 0.77N). Finally the
difference in saturation values can be higher than 10
saturation units (Figure 5) depending on N, on the sample
geometry and on the shape of the capillary pressure curves.
This leads us to the fact that the traditional assumption of
a non-radial centrifugal field can cause an under-evaluation
conventional Beckman geometry (1" x 1" (2.54 em x 2.54
em) sample, r3 = 8.6 em). Radial effects are therefore
significant and should be accounted for in data
interpretation methods.

12
SPE 28182 P. L. FORBES, Z. A. CHEN and D. W. RUTH 5

G(k p) = A1 F1(k 1 p) + A2 F2(k2 p) will be transformed

Simple Procedures To Account For
Radial Effects into G(k p) =AJ F(kJ p) + A2 F(k2 p).

This means that the transformation, F(p) --->F(p),

PRE-PROCESS CORRECTION
commutes with any linear -transformation or a
The pre-process correction method is based on Eq. 24. multiplication of the variable p by a constant. According
Starting with the measurement (S,co), we first calculate
to the scheme Eq.28, the evaluation of{~. Pc 1 jb} is
(S, P c1 ) according to Eq.8 instead of Eq.2. This consists
in multiplying the traditional P c1 values by (1 +N). Then obtained by such a linear correction, r, with S= F1 = F2 ,
{SO, P cJfb} is evaluated based on Eq. 26. 1+N
1 A1 = 1- a, A2 =a, kJ = 1, k2 =a, and k = b
•
Fmally
=-OS
= 1+ V1-B JS(Pclx) . . "II .
_,-:-:_- dx IS mverted 10 owmg
{sP c r
2 -v 1-Bx non-radial ~
1
0
one of the usual, accurate methods available in the
literature. S(Pc1non-radial)+a[S(aPc1non-radial)-S(Pc1non-radial)]
{ Pc1 non-radial
This procedure can be summarized as follows: (1+N) b
• Non-Radial Procedures: ........... (31)
(S, co}--> (S,Pc1 non-radial}
~ {Current Analysis } ~ (S, Pc1 )
Therefore the evaluation of {S, P c) curves will be obtained
Techniques [11,12,14] as well when applying the current analysis techniques
before the correction, r, is implemented:
• Radial Procedures: _ { Current Analysis }
_ _ _ Pc1 radial {S, co}~ Techniques [11,12,14]
{S, co} --> {S,Pc1 radial}--> {S 0 , b }
r
~ {Current Analysis } ~ (S, Pc1 ) ~ {Snon-radial , Pc non-radial }~ {S, P c} ·········· (32)
Techniques [11 ,12, 141
·················(28) that is
where
Pc1 radial= (1+N)Pc1 non-radial
~c 1 non-radial
r
{
SO (P c1 radial fb) = · { ~

S(P c1 radial)+ a [ S(aP c1 radial)- S(P c1 radial)]

................. (29) S(Pc1non-radial)+a[S(aPc1non-radial)-S(Pc1non-radial)]
This implies that the radial effect correction can be done { Pc1 non-radial
very simply as a pre-treatment of datasets before any of the (1+N) . b
techniques already in use are implemented. ........... (33)

Equation 33 represents a post-process which allows the

POST-PROCESS CORRECTION
Because centrifuge capillary pressure curves available in
the literature have been determined without accounting for
radial effects, it is of interest to propose a post-process to
directly correct these curves. Referring to the previous
scheme Eq.28, it can be observed that what are called the
"current analysis techniques" consist in the transformation
of a function F(p) into F(p) according to
direct correction of conventional capillary pressure curves
(Snon-radial, Pc non-radial) obtained without accounting
for radial effects.
Imbibition Capillary Pressure Curves
The previous analysis focused on drainage capillary
pressure determination. By analogy, a similar approach and
similar calculations can be used to account for the radial
effect in imbibition experiments.

-F(p) = 1+m
2
f1
.EQ&.
...J -Bx dx .............................. (30)
1
The procedure for the pre-process correction is outlined as

{S,Pc3 }~{S(Pc3 )+a[S(aPc3 )-S(Pc3 )]. P~3 }

0
while the post-process correction will be
Obviously for any constant parameters A 1, A2, k 1, k2,
Pc3 }
and k, it follows that {S, Pc3 }~ { S(Pc3 ) +a [ S(aPc3 )- S(Pc3 )] .-b-

13
6 QUANTITATIVE ANALYSIS OF RADIAL EFFECfS ON
CENTRIFUGE CAPILLARY PRESSURE CURVES
with ................. (34)
Note that P c1 is replaced by P c3 , the capillary pressure at
the outlet face and that P c3 non-radial= P c3 radial= P c3
because the zero capillary pressure is imposed at the center
line of the inner face for both the non-radial and radial
considerations.
In practical applications (r3 ~ 4 rl, i.e. b <<1), the same
kind of conclusions can be drawn for imbibition capillary
pressure curves as for drainage curves.
Examples
Two examples are used in this section to show the
significance of the radial effect on data interpretation.
In l;he first example, Bentsen and Anli's data [17] (Sample
B) Is used to show the proposed technique. The geometric
parameters include: 2R = 2.44 em, q = 12.45 em, r3 =
16.31 em. Figure 10 shows a comparison among various
methods. Little difference is found in this figure. It can be
concluded from this example that for core samples of large
radii and small diameters, the shift of capillary pressure
curves is not pronounced. Under such circumstances
previously-proposed, semi-analytical, one-dimensionai
solutions, such as Forbes' solution (Eqs. 5 and 6), are
good approximations, and the regular algorithms hold for
experimental data reduction.
The dataset for the second example is taken from a recent
paper by Ajufo et al [18]. Data from their Fig.3 were re-
analyzed with three different interpretation techniques:
Hassler-Brunner's solution (Eq. 4), Forbes' improved
solution (Eq. 5), and the new solution developed in the
present paper accounting for the radial effect (pre- and post-
process corrections have been applied). The centrifuge
geometrical parameters include q = 10.42 em, L= 4.39 em
and 2R = 3. 79 em. The final results are illustrated in
Figure 11. It can be seen from this example that the curves
obtained from both Hassler-Brunner's solution and Forbes's
improved solution do have discrepancies with the curve
from a porous plate experiment, the most widely
recognized standard technique for capillary pressure
measurements, while the curves accounting for the radial
effect are in excellent accord with the porous plate curve.
This example supports the observations that (1) for rock
samples of small outside endface radius (say~ 10 em) and
large diameter (2R ~ 2.54 em or 1"), the errors due to
radial effects are significant; and (2) the newly-developed
method does account for the radial effect and provides a
14
SPE 28182
much improved agreement with the porous plate capillary
pressure curve.
It is interesting to note that Figure 11 in Christiansen and
Cerise's original paper [15] clearly shows that for the
inverse process of data reduction, the double integral
successfully recovers the true capillary pressure, while
both the single integral method and the truncated Hassler-
Brunner solution give local saturations that are lower than
the true capillary pressure curve. This figure supports the
derivation and analysis of the present paper.
Conclusions
A new solution has been presented for the centrifuge
saturation equation. This solution has allowed a
quantitative characterization of the effect of the radial
centrifugal acceleration on capillary pressure curves. The
analysis shows that when the radial effect is taken into
account, the resulting drainage capillary pressure curve will
be shifted toward higher saturation and pressure values than
the traditional Hassler-Brunner curve, not lower as
predicted previously. Current analysis techniques that do
not account for radial effect can lead to an under-estimation
of up to 3 to 5 saturation units for conventional centrifuge
geometries. A new inversion technique, which accurately
accounts for the radial effect, has also been proposed. The
new technique entails correcting for the radial effect on S
datasets and processing data with any current, proven
method. Therefore the technique can be implemented
simply as a pre-process or a post-process to any method
already ~n use. It has been shown that the new technique
accountmg for the radial effect significantly changes
interpretations of data from the current (Beckman)
centrifuges, and then significantly improves the agreement
between centrifuge and porous plate capillary pressure
curves. Previous disagreement, and agreement as well,
presented in the literature between these two methods,
should be re-analyzed accounting for radial effects.
Acknowledgement
Sincere thanks are extended to J.D. Warne, Core
Laboratories, for providing the dataset of the second
example in the present paper. One of authors (Z. Andy
Chen) is supported by a grant from Imperial Oil.
Nomenclature
constant parameter
constant parameter
parameter defined in Eq. 25
parameter defined in Eq. 25
f'
geometric factor, B = 1- ~~)2
constant parameter
constant parameter
constant parameter
core plug length, em
SPE 28182 P. L. FORBES, Z. A. CHEN and D. W. RUTH
geometric factor defined in Eq. 9
capillary pressure
capillary pressure at rl
capillary pressure at r3
capillary pressure at ri
the radius of core cylinder plug, em
rotation radius, em
the radius to the inlet endface, em
the radius to the outlet endface, em
the radius to the inlet endface, em
the radius to the edge of the inlet endface, em
the radius to the outlet endface, em
the radius to the edge of the outlet endface, em
wetting phase saturation
mean wetting phase saturation
mean wetting phase saturation when N =0
variable, x = Pc I Pc1
y variable, y = cos2(a)
a variable defined in Eq. 16
~ parameter defined in Eq. 6
0 parameter defined in Eq. 6
E free parameter defined in Eq. A-1
~p density difference of fluid pairs
e angles defined in Eq. 11
ro angular rotation speed
References
1. C_hristiansen, R.L., 1992, Geometric Concerns for
Accurate Measurement of Capillary Pressure
Relationships with Centrifuge Methods, S P E
Formation Evaluation, 7, 311-314.
2. Ruth, D.W. and Z.A. Chen, On the Measurement and
Interpretation of Centrifuge Capillary Pressure Curves:
The SCA Survey Data, accepted by The Log Analysts
for publication.
3. Wunderlich, R.W., 1985, Imaging of Wetting and Non-
Wetting Phase Distribution: Application to Centrifuge
Capillary Pressure Measurements, SPE paper 14422
presented at The 60th SPE Annual Technical
Conference and Exhibition, Las Vegas, Sept. 22-25.
4. Baardsen, H., V. Nilsen, J. Leknes and A. Hove, 1991,
Quantifying Saturation Distributions and Capillary
Pressures Using Centrifuge and Computer
Tomography, RESERVOIR CHRACTERIZATION II:
(L.W. Lake, et al., ed.), Academic Press, Inc., San
Diego, California.
5. O'Meara, D.J., Jr., G.J. Hirasaki and J.A. Rohan, 1992,
Centrifuge measurements of capillary pressure: Part !-
Outflow boundary condition: SPE Reservoir
Engineering, 7, 133-142.
6. Chen, Z.A. and D.W. Ruth, Gravity Degradation Effect
on Low-Speed Centrifuge Capillary Pressure Data,
accepted by AIChE Journal for publication.
15
7
7. Ayappa, K:G., H.T. Davis, E.A. Davis, and J. Gordon,
1989, Capillary Pressure: Centrifuge Method Revisited,
A/ChE Journal, 35, 365-372.
8. Blackwell, R.J., 1991, Capillary Pressure
Measurements of Heterogeneous Cores Using a
Centrifuge, SCA paper 9105 presented (orally) at The
Society of Core Analysts Annual Technical
Conference, San Antonio, Texas, August 21-22.
9. Forbes, P.L., 1993, Taking Into Account Radial
Effects, presentation at The Centrifuge Capillary
Pressure Workshop, The Society of Core Analysts
Annual Technical Conference, Houston, Texas, August
9-11.
10. Hirasaki, G.J. and J.A. Rohan, 1993,
Recommendations for Determining Production Data,
presentation at The Centrifuge Capillary Pressure
Workshop, The Society of Core Analysts Annual
Technical Conference, Houston, Texas, August 9-11.
11. Forbes, P.L., 1991, Simple and Accurate Methods for
Converting Centrifuge Data into Drainage and
Imbibition Capillary Pressure Curves, SCA paper 9107
presented at The Society of Core Analysts Annual
Technical Conference, San Antonio, Texas, August 21-
22.
12. Hassler, G.L. and E. Brunner, 1945, Measurement of
Capillary Pressures in Small Samples, Petroleum
Transaction, AIME, 160, 114-123.
13. Boucher, E.A., 1987, Centrifugation of Porous Solids
and Compacts Containing Imbibed or Capillary-
Condensed Fluid, Journal of Colloid and Interface
Science, 120, 289-291.
14. C~en, Z.A. and D.W. Ruth, 1993, Centrifuge
Capillary Pressure Reduction with Modified Forbes
Method, Journal of Petroleum Science and Engineering,
9, 303-312.
15. Christiansen, R.L. and K.S. Cerise, 1987, Capillary
Pressure: Reduction of Centrifuge Data, paper 47d
presented at the 1987 AIChE Annual Meeting, New
York City, Nov.15-20.
16. Forbes, P.L., D.W. Ruth and Z. Chen, 1994,
Discussion of "Geometric Concerns for Accurate
Measurement of Capillary Pressure Relationships with
Centrifuge Methods", SPE Formation Evaluation, 9, 1,
79-80.
17. Bentsen, R.G. and J. Anli, 1977, Using Parameter .
Estimation Techniques to Convert Centrifuge Data Into
a Capillary-Pressure Curve, Soc. Pet. Eng. Journal, 17,
57-64.
18. Ajufo, A.O., D.H. Daneshjou and J.D. Warne, 1993,
Capillary Pressure Characteristics at Overburden
Pressure Using the Centrifuge Method, SPE paper
26148 presented at The SPE Gas Technology
Symposium, Calgary, Alberta, Canada, June 28-30.
8 QUANTITATIVE ANALYSIS OF RADIAL EFFECfS ON SPE 28182
CENTRIFUGE CAPILLARY PRESSURE CURVES

Appendix 1

We start from Eq. 22:

S(Pc1)=
1 +~J .~ [S<Pc,x)i ts(o.)2da
1t "'J 1-Bx 1t 0
1 2 0
1+ m 0J2
____!L
...J 1-Bxo
Js(P CJ x+N Cos(a)\ icos(a)2 da
1+N 1t
1t
2
+ S'(Pc 1x)Pc 11 NN i Jcos(a.)4da
................. (22) + 1t0
Replacing Pc 1 by Pc 1 (1+e 1~N) (where e is a free
1t
2
parameter for the moment) results in - S'(Pc 1x)Pc 11NN (1-e) x i Jcos(a.)2da.] ........ (A-5)

-8 N
(Pci1+£1+N) )=
1+m
2
1
J dx
...J 1-Bx
Using
1t
+ 1t0

1t
0
4 2 4 2
1t - fcos(a.) 2da. = 1, - fcos(a.)4da = 3/4,
2 1t0 1t0
Js(p (1+e1+N
N )
CJ 1+N
2
x + N CosCa) ) Cos("')2
u.
d"'
1t u.
i 1
1+-{li JS(xPcl> d -soP
0 2 ...J 1-Bx x - ( c1)
......................... (A-1) 0
Using
2 and
P ·( 1+e_l!_) x + N CosCa) P
CJ 1+N 1+N CJX

+Pc 11~N (Cos(a) 2- (1- e) x) +( 1~N)2Pc1 e (Cos(a.)2- x)

......................... (A-2)
( we use S 0
'(P c) to represent
dS0 (P c) ) then we get
and expanding the saturation term of the integrand through · dPc
the regular Taylor series to the first order in
1 ~N , we
- N - 0
S(Pc1(1+£1+N))= S (PcJ)- (1-e)Pc11+N S '(Pcl)
N -0
obtain
1
S(P (1 +e.J:L) x ~ N 1+N
Cos(a)2 )= S(P x)
+ 3/4l~Nl+~ J~~Bx S'(Pc x)Pc
CJ 1+N Cj 1 1 .............. (A---{j)

+ S'(Pc 1x) Pc 11 ~N (Cos(a.) 2 - (1- e) x) 0

Now consider the integral term on the right-hand side of
.......................... (A-3) Eq. A-6
(note that we use S'(Pc) to denote d%c)) 1
1+-{1i
2 -~
J____!L
S'(P CJ x)P CJ
where the last term on the right-hand side of Eq.A-2 is "'J 1-Bx
eliminated because it is second order. Then we will find 0
that
- N
S(P c1(1+e 1+N))=
1
1+mJ ____!L 4
2 ...J 1-Bx 1t
0
1t
2
JCos(a)2[S(Pc x)+S'(Pc x)Pc11:N(Cos(a)2-(1-e)x)]da.
1 1
1
0
.........................(A--4) =I+~ Jdx S'(Pc1x)Pc,·Ji .. Bx ..................... (A· 7)
0

16
SPE 28182 P. L. FORBES, Z. A. CHEN and D. W. RUTH 9

Integrating by parts
1
1 +~J"' dx. S'(Pe 1x)Pe 1 - BPe1 S '(Pc 1)
0 For drainage capillary pressure curves 0 ~ B ~ 1 and if we
1-Bx plot eo as a function of B, we will find that there exists a
0
relationship
= [ 1+-{1-B S(P
2 elX
)V!-BxX ]x=1
x=O 0.21 S. Eo s. 0.25 .......................................... (A-17)

2
1
+ 1+"' 1-B B/2 -~
-v 1-Bx
f
dx. S(Pe x) .......................(A-8)
1
As an approximation, we can take eo = 0.23 for any B
value. Therefore
-
S(Pe/1+ 0.23 1+/'i))=
N
0

- 1 +~ [S(Pe1 )~- S(O)] + B/2 S 0

(Pe 1 ) ..... (A-9) S0 (Pe 1 )+3/4 1 ~N 1 +~ [S (Pe 0
1)- S(O)] ..... (A-18)

Defming the parameters

Using B= (1 +~ ) (1- Vl-B )and re-arranging 1+{1-B N
1
+~"' 1-B [S(Pe )-S (Pe
{ a= 3/4 2N ( l+N ) ....................... (A-19)
0
= 1 1 )] b =1+ 0.23 ( 1+N )
+ 1+ ~ [S (P e 0
1
) - S(O)] ............................{A-10) we can obtain an approximation of the solution to Eq.10
to quantitatively characterize the radial centrifugal
acceleration field (note that S(O) = S 0 (0) = S (0) ).
that is
1
1 +-{1-BJ~S'(P )P
2 "'1-Bx elX el
0 extracting S0 {P e1) gives
_ S (0) - S (bPe 1 )
=BPe1 S0 '(Pe 1 ) +
1 +~ V1-B [S(Pe )-S0 (Pe )] 0
S (P el) = S(b pel)+ a 1 + a ...... (A-21)
1 1

+ 1+ ~ [S (Pe1)- S(O)]0
.......... (A-11)

Recall that a simple approximate solution of Eq. 12 has

been proposed by Forbes [111 as
Now consider
- Pel -
S(Pe 1) = S 0 (Pe 1 ) + + S 0 '(Pe 1) ............... (A-12) f(a) = (1 +a) (S (aPe 1) - S (P e1 )) .................... (A-23)
1 8
where 8 is defined in Eq. 6,
Then we have then for small values of a, f(a) = f(O)
~0 Pel - 0
As far as "a" is of the same order of magnitude as N <<1,
S<Pe 1)- S (Pe 1 ) = + S '(Pe 1) •••••••••••••••• {A-13) Q!is is eg_uivalent to _ _
1 8
S (0)- S (Pe 1) = (1 +a) (S (aPe 1)- S (Pe 1))
Using Eqs.A-11 and A-13 to re-arrange Eq.A-6 we obtain or
- N S (0) - S (P e1 ) _ _
S(Pe1(1+e 1+N))=
1
+ a = S (aPe 1 )- S (Pe 1 ) ••••••••••• (A-24)

S 0
(Pe 1 )+ 3/4 1 ~N 1 +~ [S (Pe 0
1)- S(O)] Putting this expression into Eq. A-22 results in
0
S (Pe 1) = S(bPe1) +a [S(abPe1)- S(bPe1)] •••• (A-25)
N [ 1A(B 1+"' 1-B "'1-B) 1 ] p -5 o, p )
+ 1+N 3,.... + 2 1+ 8 - + e e1 < e1
........................ (A-14) The above equation contains both the single and double
integral terms corresponding to the solutions of two cases,
Hassler-Brunner's solution and the solution accounting for
As we have let e be a free parameter, we can choose the radial effect. This equation replaces the fundamental
equation for predicting centrifuge capillary pressure curves
e =eo= 1 - 3/4 (B t 1+-{1-B ..J 1-B) = 7 - 3B ...JI-B - which take into account the radial effect. The evaluation
2 1+ 8 8(2+fu)
A-25 is preferred to A-21 because it accounts for S(P e1)
........................ (A-15)
which makes = S 0 (Pe 1 ) = S (Pe 1 ),
when Pe 1 = oo, which can be
- N deduced from exact equations 11 and 12.
S(Pe/1 +Eo 1+N))=

17
10 QUANTITATIVE ANALYSIS OF RADIAL EFFECT'S ON SPE 28182
CENTRIFUGE CAPILLARY PRESSURE CURVES

Iso-Pc
Contour

Axis of Rotation

ri

L 'InletFace H-B Analysis

TOP- VIEW B=O & N=O
Figure 1. Assumption of A Linear Field Distribution

Axis of Rotation

ri

L 'InletFace Improved Analysis

TOP- VIEW B>O & N=O
Figure 2. Assumption of One Dimensional Distribution For Centrifugal Field

Iso-Pc

Axis of Rotation

L Radial Analysis
TOP- VIEW B>O & N>O
Figure 3. Exact Radial Distribution of Centrifugal Field

18
SPE 28182 P. L. FORBES, Z. A. CHEN and D. W. RUTH 11

{ro}
-
{S}

{~ ~}non-radial {S} non-!C' ~ial

' '/

Pet 1 + 0.77 N

s
Figure 5. Capillary Pressure Curve Shift

--- Hassler & Brunner

------ Forbesirnproved
--- Radial Effect
--- Mean Saturation

1
Wetting Phase Saturation (frac.)

0 0.2 0.4 0.6 0.8 1

Figure 6. Typical Solutions For A 1"xl" Sample and A Centrifuge Geometry with roulet =8.6 em

19
12 QUANTITATIVE ANALYSIS OF RADIAL EFFECfS ON SPE 28182
CENTRIFUGE CAPILLARY PRESSURE CURVES

r/8
(a)

x/a
(b)
r

Figure 7. Transformation of Integration Domains

20 - B=O.l, 0.9
N=O.l

Figure 8. Evaluation of Equation 24

20 ·-
·n• • ~ \ r~·<\, 8=0.1, 0.9
•• • \\ • • N=O.l
~
~15
\\1\\ K\\\\ '<~:'.
n \} \ ·,~· '~,
"''\
·-

~ \
'-'

~ 10 ·- \\'!\\ ~·~ ........

\\ ..
ell

J: .,.~~~.
.,(~)~
·~~
., .~..==, J
~ 5
=
·~
u
.. ·-
~·..
•.:>~·~
Wetting Phase Saturation'~A
·~
,.~.
~~.
-·~·-·,1
(frac:r==a.•.-.-·-~
---..
~

:..•
!

0 : : : I ;

0 0.2 0.4 0.6 0.8 1

Figure 9. Evaluation of Equation 24

20
I ' '

SPE 28182 P. L. FORBES, Z. A. CHEN and D. W. RUTH 13

7
..Radial
·r-·

.:
.:
• Non-radial
• Bentsen & Anli
6
~
.:
~
L!
·\·· • •
Regenerated

l!
e:
L!·
l!
[~
~
'\ ..
·~
2 ·- ~ ,...~..
•.

~-~·..
L! [~[·.::~ ,.
........ .
·~
•~ •
~ L!l•. [i"~ L!. .!J·~·
•'~·
1 •-
Wetting Phase Saturation (frac.)
0 : : : : I
I
I
I
I

0.25 0.35 0.45 0.55 0.65 0.75 0.85 0.95

Figure 10. Bentsen and Anli's Data: Sample B (ref. 17)

45
• Hassler & Brunner
• Forbes improved
40 c Radial Effect, pre
"' and post process
35 <> Porous Plate
Ajufo et al' Data
(Ref.18)

•I ~

10

5
·~~- (frac.)
""'-+~"" -A
•ot~"::fr. o •r£
0 +---------------------+---------------------~--------------------~------=~~
0.15 0.25 0.35 0.45 0.55
Figure 11. Ajufo et al's Data: Sample 13A (ref. 18)

21