Sie sind auf Seite 1von 6

A New Mixing Rule for Petroleum

Mixture Compositional Processes


D.X. YOUNG
United Oil & Gas Consulting Ltd.

such problems have not been solved with satisfaction, reservoir


Abstract simulators that use EOS to model reservoir fluid compositional
A new mixing rule of a realistic vector form is proposed. It is behaviour are most likely to inherit the problems. When this hap-
used for vapour-liquid equilibrium calculations of petroleum pens in simulation, a lot of times, it is not clearly understood. The
mixtures with the commonly used Peng-Robinson (PR) and reason that this is important is because reservoir fluid vapour-liq-
Soave-Redlich-Kwong (SRK) Equations Of State (EOS). The uid equilibrium states reflect its energy and petroleum recovery
results are compared with those from the conventional calcula- potential.
tions with the same EOS and the original matrix form of mixing Another problem with compositional reservoir simulators that
rule, and the experiment data obtained for the chosen petroleum use an EOS model is that the computing time and storage require-
mixtures with and without added CO2. ment increase sharply with the number of components used for
The new vector form of mixing rule is a successful “trans- each grid and iteration step, and soon become a restriction in a
form” of the old matrix form of mixing rule based on the idea of simulation(4). Even with the new generations of computers, this is
component deviation from their pure-state, vapour-liquid equi- still the case.
librium behaviour when they are present in a petroleum mixture. Many techniques have been developed to tackle convergence,
Such transformation will allow us to eliminate binary interaction accuracy, computation time, and storage problems. Among these
coefficients with the more practical component deviation coeffi- are numerous lumping schemes(5, 6, 7) and highly sophisticated
cients for petroleum mixture phase equilibrium problems mathematical optimization methods(8). In this paper, a new mixing
encountered in compositional simulation and relevant chemical rule that can contribute a lot towards solving such problems is
engineering calculations. used with the PR and SRK EOS. The rationale is to quantify the
The utilization of the new mixing rule has lead to results with phase behaviour deviations from their pure state behaviour for
improved computation efficiency, convergence behaviour and each component present in a petroleum mixture when it is inter-
accuracy for the calculations performed. However, convergence acting with the rest of the constituents.
problems cannot be solved completely with the proposed mixing The above conceived scheme is realized for phase equilibrium
rule as the problem does not stem from there only. Further computation of two reservoir fluids from different fields, for
investigation results in a new critical criterion that has not been which lab measurements on composition and saturation pressure
satisfied by many EOS, including the PR and SRK EOS, on both are conducted for the fluid samples alone and when various
analytical and numerical basis for all pure components. Thus, amounts of CO2 are added.
convergence problem has never been really solved no matter
how sophisticated mathematical methods have been employed to
get the “converged” results. The Mixing Rule
The new critical criterion is established with a newly devel-
oped three term cubic EOS, which can be easily reduced to The new mixing rule is first introduced to improve computa-
encompass many two term cubic EOS of similar forms as the PR tional efficiency of phase equilibrium calculations that use an
or SRK EOS. A discussion on the new EOS is given. With the EOS by decomposing the binary interaction coefficients into com-
new EOS, it becomes obvious that the critical convergence prob- ponent deviation factors in a way that is suitable for petroleum
lem is not purely a mathematical one, the EOS models them- mixtures. The reasons for doing this are to overcome the draw-
selves may have some inherent restrictions for the phase equilib- backs of the binary interaction assumption for complex mixture
rium computation schemes to work for complex mixtures when systems.
the critical points are approached. When we assume that binary interactions are valid for petrole-
um mixtures, we are inviting trouble for modelling such systems
although this might be the right thing to do at some point for rela-
tively simple mixture systems to establish certain methodology.
Introduction First, the basis is on very limited binary mixture experimental
data, which is extended to complex mixtures as an approximation.
The PR and SRK EOS have been in use for phase equilibrium However, in reality, we all know that naturally occurring reservoir
calculations for over twenty years since they were published(1, 2). fluids are composed of thousands of hydrocarbon constituents.
However some problems still exist for petroleum mixtures con- Second, it is not always easy to control the binary interaction
taining more heavy fractions and near critical point conditions or coefficients, and a lot of times, such data may not be available at
across phase boundaries(3). In some cases, it is even hard to get all. Third, the limited binary interaction information may not be
converged results without considering their accuracy. Because comprehensive enough to incorporate all important factors that

February 2000, Volume 39, No. 2 35


TABLE 1: Reservoir fluids composition. Where Ki is the deviation coefficient for component i. If we
want Equation (7) to perform the similar function as Equation (5)
Res. Fluid Sample-1 Res. Fluid Sample-2 in EOS vapour-liquid equilibrium calculations, we can expand
Component (Mol%) (Mol%) Equation (7) and make the two equations equivalent by the
following:
H2S 5.86 0.00
CO2 0.30 2.97
N2 0.81 0.40 ....................................................................................(8)
C1 34.92 8.61
However, we should notice that the decomposed component
C2 11.46 7.39
deviation coefficient Ki in Equation (7) bears different meaning
C3 5.58 7.64 now and Ki = 1 represents no deviation for the component.
I- C4 1.28 0.95 As expected, the new mixing rule does not present any problem
N-C4 2.62 6.27 with respect to the fundamental thermodynamics. The justifica-
I- C5 1.32 1.59 tions for such modification may be partially attributed to the regu-
N- C5 1.44 3.84 lar changes of the physical and chemical properties of the con-
C6 2.63 4.06 stituents in a petroleum mixture.
C7 or C7+ 3.04 56.28 As the coefficient Ki in Equation (7) presents a vector form
C8 4.11
instead of the matrix form in Equation (5), obviously, the advan-
tages of using it are to increase computation efficiency and reduce
C9 4.09
the required storage. For a mixture system with n species, we only
C10 3.43 need n component deviation factors Ki compared to the previous
C11 2.60 n-by-n binary interaction coefficient matrix kij. When we imagine
C12 2.02 that this is for the computation of each grid and iteration of the
C13 1.99 simulation, its effect will be substantial.
C14 1.56 The proposed mixing rule is used to calculate the bubble point
C15 1.33 pressure and dew point temperature of two reservoir fluids with
C16 0.99 different CO2 concentrations. The two reservoir fluids are special-
ly chosen for the research project. Their compositions are given in
C17 0.89
Table 1. Sample-1 is a light sour crude, and 20%, 40%, and 60%
C18 0.81
CO2 in mole concentrations are added to the sample. Sample-2 is
C19 0.74 a relatively heavy crude, and 30%, 45%, and 60% CO2 are added
C20+ 4.18 to it.
The above-mentioned phase equilibrium calculations are per-
may contribute a lot to the vapour-liquid equilibrium state of a formed for the eight cases using the two crudes and six derived
petroleum mixture in the real engineering world because such data samples with the PR or SRK EOS to verify the validity of the new
are obtained for relatively ideal binary systems. Four, it causes mixing rule. The only difference in the calculations is the mixing
low computation efficiency for compositional reservoir and rules used. In one case, it is the original matrix form with binary
process simulations. interaction coefficients, and in the other, the new vector form with
The mixing rule proposed is for the numerator, aα, of the sec- component deviation factors. The results are presented in Table 2,
ond term in an EOS of a similar form as the PR or SRK EOS. It Table 3, and Table 4 along with the available experimental data
can be used to replace the corresponding parameter in such an for the specified calculations.
EOS for vapour-liquid equilibrium calculations. The PR and SRK As we can see from Table 2 or 3, the calculated bubble point
EOS have similar forms except the denominator of the second pressure results with the new mixing rule are in good agreement
term. Both EOS are commonly used for compositional process with the experimental data and the results from the conventional
calculations. Here we use the PR EOS given by Equations (1) – calculations. Table 4 compares the calculated dew point tempera-
(6) in standard form as the example to introduce the new mixing tures from the two schemes as there are no experimental data
rule as follows: available. Good consistency has been achieved as well.
One observation that is made when performing the above com-
putations is that the convergence behaviour of the compositional
phase equilibrium calculations is improved as an unexpected
................................................................(1) result of using the new mixing rule. This is indicated by the bub-
ble point pressure calculation for the case of sample-2 with 60%
CO2. For this case, the conventional calculation with the PR EOS
..........................................................................(2)
and the old mixing rule fails to converge, but the new scheme
using the same EOS and the new mixing rule works out quite well
.............................................................................(3)
as a result. With several other cases of high CO2 concentrations,
convergence problem is also observed for the conventional
scheme, but converged results are obtained after many trials. This
is not the case with the new scheme. However, a convergence
....................................(4)
problem does exist for the new scheme as we perform the calcula-
tion for the last case in Table 2 where the converged result is
achieved after some trials. The problem may have been attributed
.......................................(5) to the near critical point situation due to the very high CO2 con-
centration in the reservoir fluid.
...............................................................................................(6) In the calculations performed in this study, values for the com-
ponent deviation coefficients Ki are easy to control compared
For a mixture system with n components, the newly proposed with binary interaction coefficients. In general, values used for
mixing rule given by Equation (7) below can be used to replace light components have relatively large deviations from one.
Equation (5): Typical values used are between 0.92 – 0.99 for hydrocarbon
components with light components, such as methane and ethane,
having values close to the lower side. For a specific component,
.............................(7) only small variations in the Ki values are observed in different

36 Journal of Canadian Petroleum Technology


TABLE 2: Bubble point pressure calculation results.

Exp. Conventional New


Results Calc. Results Method
Cases (kPa) (kPa) (kPa)
Sample-1 18,609 18,533 18,588
Smp-1+20%CO2 21,043 20,960 21,091
Smp-1+40%CO2 26,745* 26,614 26,889
Smp-1+60%CO2 35,984* 38,176 32,950
Sample-2 3,902 3,709 3,827
Smp-2+30%CO2 6,881 6,840 6,991
Smp-2+45%CO2 8,267 8,225 8,225
Smp-2+60%CO2 9,198 not converged 9,266

Average error % 1.37% 0.87%


Note:
(1) * are Core Lab calculated bubble point results.
(2) The conventional calculation results are from the standard PR EOS and mixing rules.
(3) The new method results are from the PR EOS and the new mixing rule and the mixing rule given by Equation (6).
(4) The results are at temperature 220˚ F.

cases. The difference is normally within 0.02. For example, the Ki may change a bit if different grouping techniques and different
values used for propane is between 0.97 and 0.99. This is actually computer programs are employed when more extensive studies
similar to the cases when using binary interaction coefficients. For are carried out in the future.
CO2, H2S, and N2, the deviation coefficients used a range from
0.82 – 0.91 with variations of about 0.02 for each individual com-
ponent in different cases. It should be pointed out that the compo- The Three Term Cubic EOS
nent deviation coefficients provided in this paper are the findings
of the pioneer simulation work. It is possible that these values The criticality problem may have existed as long as the history

TABLE 3: Predicted bubble point pressure.

Bubble point Predicted Bubble Point Pressure (kPa)


Sample-1 Sample-1+20% CO2
Temperature Old New Old New
Deg. C Method Method Method Method

15.6 12,562 12,569 12,211 12,307


37.8 14,472 14,493 15,051 15,162
71.1 16,816 16,851 18,478 18,602
104.4 18,533 18,588 20,960 21,091
137.8 19,685 19,747 22,615 22,739
171.1 20,312 20,367 23,518 23,642
193.3 20,464 20,512 23,746 23,863
226.7 20,298 20,333 23,559 23,649
260.0 19,691 19,726 22,780 22,822
293.3 18,643 18,664 21,415 21,401
315.6 17,699 17,706 20,119 20,084
329.4 19,009 18,947

TABLE 4: Predicted dew point temperature.

Bubble point Predicted Bubble Point Temperature (Deg. C)


Sample-1 Sample-1+20% CO2
Pressure Old New Old New
kPa Method Method Method Method

3,447 445 444 439.4 438.3


6,205 442 441 438.9 437.8
8,963 426 425 427.2 426.7
12,411 393 393 403.9 403.3
14,479 370 369 386.1 385.6
15,858 348 349 371.1 371.1
17,237 313 314 356.7 356.1
18,616 339.4 338.9
Note:
(1) Old method in Table-3 and Table-4 is the standard PR EOS and mixing rules. New method is the PR EOS and the new mixing rule.

February 2000, Volume 39, No. 2 37


of EOS for several hundreds of years, regardless of what kinds of
temperature or volume related corrections have been tried to
improve it. This is the fact caused largely by the imperfection of
all EOS and the mixing rules for extended mixture applications.
Numerous lumping schemes, property correlations, and math
techniques also come into the play. The same is true for the
vapour-liquid equilibrium calculations in this study.
To gain deeper insight into the critical convergence problem, .........................(13)
further efforts are devoted to it, and some important findings
come out with the help of a newly developed three term cubic To further reduce Equation (13), we choose b = 0, which is a
EOS. It reveals how the volume related correction for pure com- condition required by the new EOS at the critical point for all pure
ponents and mixtures could give rise to convergence problem components, to get:
when the critical point is approached.
As we know, uncertainties can be introduced into phase equi-
librium calculations that use an EOS, when the application of an
EOS is extended from pure components to mixtures to get the
approximate mixture parameters by mixing rules. However, to
make an EOS work near the critical point, such approach will fail
in most cases due to the extreme sensitivity of an EOS near the
critical point. For such mixture applications, the problem arises
mainly from the second mixing rule that appears as parameter b in ........................(14)
the denominator of an EOS of similar forms as the PR or SRK
EOS. The problem will become very clear as we present the new Where,
generalized three term EOS obtained on analytical basis:
.........................................................................................(15)

....................................................................................(16)

............................(9) ...............................................................................................(17)

In Equation (9), a, b, c, m, and n are treated as constants so far; We can notice that “c” has been eliminated from the EOS when
m and n are introduced to generalize the EOS and to adjust the b is set to zero. Here we need to emphasize that b = 0 at the criti-
functional behaviour of the EOS. The use of m and n gives more cal point is a very important finding that must be satisfied by the
flexibility for α, which is a parameter similar to the α in the PR or new EOS on both analytical and numerical basis at the critical
SRK EOS, to be correlated more reasonably; X and Y are inter- point for not just one or two components, but for all pure compo-
mediate constants given by the restrictive relationships below: nents, which is a tough condition that has not been met strictly by
any EOS on the same basis.
...................................................................................(10) If we want Equation (13) to become a two term cubic EOS, we
can make the constant c equal to zero to arrive at:

....................................................(11)
...........................(18)

....................................................................................(12) Where,

As we can see, although a, b, and c are treated as constants, ..................................................................................(19)


they are not really independent of each other. This is a very
important characteristic of the new EOS that is not reflected in the
other EOS. Another important fact is that (Zc– b) in the first term
of the new EOS is used to replace the constant coefficient of the ...........................................................(20)
parameter b of the corresponding term in the PR or SRK EOS to
make the new EOS more generalized.
Equation (9) is in its complete generalized form, which is
seemingly complicated. But the interesting fact about the new ...........................................................................................(21)
EOS is that it can be reduced consistently. When we set c = 0, it
reduces to a two term EOS, and when (Zc – b) is set to a constant Again, we find that “a” has been eliminated in Equation (18)
close to what is used in the PR or SRK EOS (m = 0), we get new when c is set to zero. As we have mentioned earlier that the newly
EOS that are very close to the two EOS in form and performance. proposed EOS can generate EOS that are very close to the PR or
We will deal with these aspects in more detail as we start to SRK EOS. Here we will set an example when we choose the typi-
reduce the new EOS. cal values Zc = 0.3 and b = 0.22 to make (Zc – b) fall between
First, we will notice that m and n will disappear from Equation 0.0778 and 0.0866, which are the constant coefficients that are
(9) at the critical point. We can set m and n at different values to used in the PR and SRK EOS. The result of doing this is the fol-
try to better correlate the parameter α, which appears in the sec- lowing EOS in the same form as the two mentioned:
ond term of the EOS as the most appropriate place for such corre-
lation. To keep it simple, we only need to set m = 0 and n = 0 to
get the “normal” form of equation with the same a, b, c, X, and Y
below: .........................(22)

38 Journal of Canadian Petroleum Technology


When we compare Equations (1) through (4) with Equation compositional process behaviour. The possibility to extend it to
(22), we see the minor difference for such an instance. This is not mixtures other than petroleum is not investigated.
just a coincidence if we analyse the values of typical critical com- As we know, in the literature, experiment and research for mix-
pressibility factors Zc and the general applicable range of an EOS, tures are conducted extensively on ideal binary pairs and some
but rather we are dealing with a more generalized EOS. The b ternary systems that yield very limited information. To verify the
value used in Equation (22) and the required value at the critical new mixing rule, we may use modern lab measurement technolo-
point may help us to understand the convergence problem and the gies for petroleum mixtures along with simulation tools. For
mixing rules better. instance, experiments can be run by mixing the light components
Now we will look into the EOS described by Equation (18) in of a petroleum mixture in vapour state and the heavy components
more detail. If we apply Equation (18) for some commonly found in liquid state without the light components in it.
components in a petroleum mixture at the critical points and allow The component deviation coefficients used in this paper are
b to take different values, we will be able to find out that “b” has generated by simulation match for the experimental data from the
to be zero in order for the EOS to work for all the pure compo- specified reservoir fluids. The method is very effective and easy
nents at their critical points. When this criterion is enforced to to control as we perform the calculations. With today’s simulation
Equation (18), a two term EOS that will work at the critical points capabilities, component deviation factors will be easier to handle
for pure components is created: as more research data flow out.
In general, the new mixing rule can yield results that indicate
equivalent or improved accuracy and convergence behaviour with
higher computational efficiency. The efficiency improvements are
............................(23) mainly observed for the cases that are difficult to converge. For
these cases, the number of iterations to achieve a converged result
Where, has about 70% reduction from around 30 iterations with the old
program to approximate nine with the modified program, when
the same real and pseudo-components are used. Thus, lumping
........................................................................................(24)
factor is excluded out of study.
The other mixing rule given by Equation (6) is not changed in
.....................................................................................(25) the calculations performed in this paper. However, its effect in
causing convergence problems is discussed with the help of the
newly developed three term generalized cubic EOS. The signifi-
....................................................................................................(26) cant finding is the new critical criteria required at the critical point
on both analytical and numerical basis. It is true that this is the
When we compare Equations (18) and (23) with the PR or SRK observation with the new EOS, but the clue it provides on conver-
EOS, the difference is mainly caused by the requirement of setting gence problem has significant meaning for all the EOS that are
b = 0 at the critical point to satisfy the critical condition. This can encompassed by the new generalized EOS.
be verified on a rigorous numerical basis using analytically It is author’s opinion that the new EOS can be shaped with
obtained EOS applied for different components at the critical great flexibility for general or specialized applications, either for
points. We will be able to find out how sensitive the EOS will be vapour or liquid phase in liquid, two-phase or superheated
when the critical point is approached, and no longer rely on the regions, and for common system conditions or localized applica-
analytical EOS only. tions. For petroleum mixtures, critical compressibility correlation
If an EOS has not worked for the individual components that can be developed. These are not the topics of this paper.
are present in a mixture, it is not reasonable to expect it to
improve when it is extended to mixtures. Obviously, when the
mixing rule given by Equation (6) is used with the PR or SRK
EOS, it is less likely that they will produce terms close to those Conclusions
given in Equation (23) at the critical point. Thus, poor conver- 1. A new vector form of mixing rule based on the idea of com-
gence performance and accuracy are expected near the critical ponent deviation from their pure state vapour-liquid equilib-
points for these EOS no matter how other masking factors may rium behaviour when they are present in a petroleum mix-
also contribute. ture, has been proposed and utilized successfully with the
The above-stated points are virtually true for all EOS of similar commonly employed PR and SRK EOS for reservoir fluid
forms. We can certainly use different mixing rules and lumping compositional process phase equilibrium calculations. It
schemes; again the point is whether the right EOS or mixture replaces the corresponding matrix form of mixing rule with-
parameters can be generated with these methods for the system out presenting any theoretical problem, and thus eliminates
conditions. One implication that is still there is that single fixed the use of binary interaction coefficients with component
correction for the b constant in the new EOS or the b parameter in deviation factors. Its use for saturation pressure and temper-
the PR or SRK EOS under all system conditions may not always ature calculations of the selected petroleum mixtures has
work out as indicated by the difference of Equation (23) and the resulted in improved computation efficiency, convergence
two EOS mentioned. behaviour, and accuracy. The proposed mixing rule provides
a more realistic treatment of the constituents of a petroleum
mixture for relevant reservoir and chemical engineering
Discussion problems. It reduces the required computation storage, and
its potential in increasing compositional simulation efficien-
In the paper, we introduced the new mixing rule given by cy will be expected to be enormous if applied successfully in
Equation (7) on the basis of component deviation for petroleum such a simulator.
mixtures, and utilized it with the PR and SRK EOS for the initial 2. The newly developed three term generalized cubic Equation
investigation for reservoir or chemical engineering processes of State along with the newly established critical criterion on
involving equilibrium or compositional behaviour. Its effective- both analytical and numerical basis, provides unique insight
ness is supported by the comparable calculation results and avail- into the critical convergence problem and possible solutions
able experimental data. to this tough existing problem.
The significance of the new mixing rule may be clearly seen by
professionals who develop simulation models and simulator soft-
ware, and by engineers who have dealt with the binary interaction
NOMENCLATURE
coefficients with difficulty. In the real engineering world, it may EOS = Equation Of State
allow us to incorporate more practical factors that contribute to PR EOS = Peng-Robinson EOS

February 2000, Volume 39, No. 2 39


a, b, c = EOS parameters or new EOS constants
i, j = subscript index for component i or j Author’s Biography
kij = binary interaction coefficient for components i
and j Derek Young is a reservoir engineer with
KI = component deviation coefficient for component i United Oil and Gas Consulting Limited. He
Kij = 1 – kij worked as a senior reservoir engineer for
m, n = new EOS constants several oil companies and was involved in
P = pressure reservoir simulation and chemical process
Pc = critical pressure engineering simulation research before
R = gas constant joining UOG. He holds a B.Sc. degree in
T = temperature petroleum engineering, and studied for his
Tc = critical temperature M.Sc. and Ph.D at the University of
Tr = reduced temperature Wyoming in chemical and petroleum
V = molar volume engineering.
xI = mole fraction of component i
X = symbol
Y = symbol
Zc = critical compressibility factor
α = EOS parameter
ω = acentric factor

REFERENCES
1. SOAVE, G., Equilibrium Constants from a Modified Redlich-
Kwong Equation of State; Chem. Eng. Sci, vol. 27, p. 1197, 1972.
2. PENG, D.Y. and ROBINSON, D.B., A New Two Constant Equation
of State; I. & E. C. Fundamentals, Vol.15, p. 59, 1976.
3. PEDERSEN, K.S. et al., Thermodynamics of Petroleum Mixtures
Containing Heavy Hydrocarbons. 1. Phase Envelope Calculations by
use of the Soave-Redlich-Kwong Equation of State; Ind. Eng. Chem.
Process Des. Dev, Vol. 23, pp. 163-170, 1984.
4. SCHLIJPER, A.G., Simulation of Compositional Processes: The Use
of Pseudo-components in Equation of State Calculations; SPE Res.
Eng., September 1986.
5. NEWLEY, T.M.J. and MERRILL, R.C., Pseudo-component
Selection for Compositional Simulation; SPERE, p. 490, November
1991.
6. DANESH, A., XU, D., and TODD, A.C., A Grouping Method to
Optimize Oil Description for Compositional Simulation of Gas-
injection Processes; SPE Res. Eng., August 1992.
7. LI, Y.K., NGHIEM, L.X., and SIU, A., Phase Behaviour
Computations for Reservoir Fluids: Effect of Pseudo-components on
Phase Diagrams and Simulation Results; Journal of Canadian
Petroleum Technology, pp. 29-36, 1985.
8. LITVAK, M.L., New Procedure for the Phase-equilibrium
Computation in the Compositional Reservoir Simulator; 12th SPE
Symposium on Reservoir Simulation, New Orleans, LA, USA,
February 28 – March 3, 1993.
9. HAYNES, H.W. and MATTHEWS, M.A., Continuous Vapour
Liquid Equilibrium Computations Based on True Boiling Point
Distillations; I. & E. C. Res., Vol. 30, p. 1911, 1991.
10. WILLMAN, B. and TEJA, A.S., Prediction of Dew Points of Semi-
continuous Natural Gases and Petroleum Mixtures. 1.
Characterization by Use of an Effective Carbon Number and Ideal
Solution Predictions; I. & E. C. Res., Vol. 26, p. 948, 1987.
11. COTTERMAN, R.L. and PRAUSNITZ, J.M., Flash Calculations for
Continuous or Semi-continuous Mixtures Using an Equation of
State; Ind. Eng. Chem. Process Des. Dev., Vol. 24, p. 434, 1985.

Provenance—Original Petroleum Society manuscript, A New


Mixing Rule for Petroleum Mixture Compositional Processes
(97-116), first presented at the 48th Annual Technical Meeting,
June 8 – 11, 1997, in Calgary, Alberta. Abstract submitted for
review November 25, 1996; editorial comments sent to the
author(s) November 19, 1998; revised manuscript received
January 11, 1999; paper approved for pre-press February 16,
1999; final approval January 27, 2000.

40 Journal of Canadian Petroleum Technology

Das könnte Ihnen auch gefallen