Experimental and Theoretical Studies

on the Fluid Properties Required

for Simulation of Thermal Processes
S.T. Lee, Gulf Research and Development CO.
R.H. Jacoby, * SPE, Gulf Research and Development Co.
W. H. Chen, Gulf Research and Development Co.
W.E. Culham, SPE, Gulf Research and Development Co.

Abstract
Experimental phase equilibrium data are presented peri mental data and theoretical correlations on phase
for three reservoir oils at conditions approximating equilibria approximating these systems are virtually
those encountered in in-situ thermal recovery nonexistent. Early work by White and Brown 2 dealt
processes. The fluid systems involved consist of three with high boiling-point hydrocarbon phase
major groups of components: flue gas, water, and equilibria. However, the highest pressure studied was
crude oil. Data were measured at temperatures from 6894.8 kPa (1,000 psia) and the lightest component
204.4 to 371.1 °C (400 to 700°F) and pressures from was pentane. Poettmann and Mayland, 3 on the basis
6996.0 to 20785.6 kPa (1,000 to 3,000 psia). Ex- of an empirical correlation, 4 constructed charts of
perimental phase equilibrium data were used to equilibrium coefficients, or K values, as functions of
develop a correlation of binary interaction coef- pressure and temperature for various boiling-point
ficients of crude-oil fractions required for the Peng- fractions. But the maximum pressure studied was
Robinson equation of state. Phase equilibrium data 6894.8 kPa (1,000 psia). Later, Hoffmann et al. 5
predicted using the Peng-Robinson equation of state, studied phase behavior of a gas-condensate system
using our interaction coefficients, are compared with with the highest pressure reaching 20684.3 kPa
experimental data. Generally, the Peng-Robinson (3,000 psia) but the highest temperature investigated
equation of state predictions were in close agreement was only 94.2°C (201°F). In 1963, Grayson and
with the experimental data. Effect of feed gas/oil Streed 6 reported experimental vapor /liquid
ratio and water/oil ratio on the equilibrium coef- equilibrium data for high-temperature and high-
ficients was examined through the Peng-Robinson pressure hydrocarbon systems. They also extended
equation of state. A study on the feasibility of the Chao-Seader correlation to cover the higher
representing the crude oil by only two fractions was temperature ranges. However, the. major light
made also. This study includes a procedure for lump- component in Grayson and Streed's system was
ing the crude-oil fractions and examples showing the hydrogen. Recently, because of the increasing ac-
importance of mixing rules in determining the tivity in carbon dioxide flooding processes, the phase
pseudocritical properties of lumped fractions. equilibria of systems involving carbon dioxide and
crude oil has received attention. Simon et al. 7 studied
Introduction phase behavior and other properties of carbon-
The steady growth of commercial thermal recovery dioxide/reservoir-oil systems. Shelton and Yar-
processes 1 has created a need for basic data on phase borough 8 examined phase behavior in porous media
equilibria that involve water and hydrocarbons during carbon dioxide or rich-gas flooding. No
ranging from methane to high boiling-point frac- extensive data on equilibrium coefficients were
tions. The in-situ thermal recovery processes often reported in those papers, and the temperature ranges
are operated at pressures above 6800 kPa (1,000 psia) (out of physical reality) were below 93.5°C (200°F).
and temperatures above 200°C (400°F). Ex- None of these papers surveyed included water as a
component.
• Now with Pennsylvania State U. The purpose of this paper is threefold: (1) to
0197· 7520/81/0010·8393$00.25
provide reliable experimental phase equilibrium data
Copyright 1981 Society of Petroleum Engineers of AIME for systems encountered in in-situ thermal processes,
OCTOBER 1981 535
 ~r
r-
I

PACKED BED

,
O':)lTACTOR

,
I 11
,, r1
t.j LOW "'ESSURE
SEPARATOR
I

! '--+-f--,---'---<8-------, ~
<0 L-_OVEN ____
1
~J~
A
CRUDE
LOW ,,"ESSU.E
SEPARATOR

Fig. 1 - Flow chart of SPE 8393 content. Fig. 2 - Schematic of flow apparatus.

(2) to use experimental data to correlate an equation each high-pressure stream from the separator was
of state, and (3) to use the correlation obtained to reconstructed by recombining the three streams (gas,
predict phase equilibrium properties on other oil, and water) obtained from each by means of the
systems. Our work relies on the Peng-Robinson atmospheric pressure separators. Thus, the bottom
equation of state. 9 A feasibility study of representing stream analysis and flow rate were the sum of the
the crude oil by only two fractions is included also. high-pressure oil (L 1) and water (L 2 ) phases.
The structure of this paper is illustrated by a flow Control of the operation was carried out as
diagram (Fig. 1). follows. The crude oil and distilled water under slight
gas pressure were injected by positive displacement
Experimental pumps whose rates were set. The flue-gas feed was
Apparatus and Procedure regulated by an Annin valve which in turn was ac-
A flow apparatus (shown schematically in Fig. 2) was tivated by an orifice meter-differential pressure cell
used to obtain the vapor/liquid equilibrium data. In through a controlled air supply to the Annin. The
brief, a crude oil, distilled water and flue gas under flow rate thus could be set by adjusting the desired
pressure were injected into a common line feeding pressure drop at the orifice controller.
into the top of a packed-bed contactor, where the A liquid level in the high-pressure (HP) separator
fluids were mixed and heated to run temperature. was maintained by a capacitance probe in a Jerguson
From the bottom of the contactor, the mixture gauge outside of the oven but hydraulically parallel
entered the middle of a vertical pipe separator, where with the HP separator. The capacitance probe was
a controlled level of the gas/liquid interface was linked through a controlled air supply to another
maintained. The contact or and separator were in the Annin valve that regulated the flow of the high-
thermostatically controlled oven. The gas phase pressure liquid mixture from the HP separator to the
flowed from the top of the separator through fine atmospheric separator.
mesh screens to prevent carryover of entrained The high-pressure separator-gas flow rate was
liquid, and the combined oil and water phases flowed controlled simply by manually adjusting a throttling
from the bottom of the separator. These phases were valve in its line; this was the primary effluent control
considered equilibrium phases at the temperature and for given feed rate settings and through it the system
pressure conditions prevailing in the separator. No pressure was maintained at the desired level.
attempt was made to separate the two liquid phases There were, of course, many more experimental
at the test conditions. details and difficulties which cannot be elaborated
Each of the high-pressure streams from the here. Only a few of the most relevant ones can be
separator then was flowed through control valves to mentioned.
a low-pressure separator at atmospheric pressure to Runs were made by operating the unit for a
break down the stream to gas and liquid phases minimum of 1 hour to observe its performance.
which could be analyzed and measured more readily. When it appeared to be in a steady-state condition,
The low-pressure gas from each stream was measured two consecutive test periods of approximately 75
with a small wet-test meter and the liquid portion was minutes each were conducted (duplicate ob-
collected in receivers. Since each liquid portion servations) before shifting to another pressure.
consisted of both oil and water (often emulsified), it Operation in anyone day was limited to one tem-
was centrifuged to ensure good separation before the perature. The range of pressures covered was 6996 to
separate oil and water volumes were observed and 20785 kPa (1,014.7 to 3,014.7 psia) at temperatures
recorded. Samples of the atmospheric gas and oil -of 204.4 to 371.1 °C (400 to 700°F). Gas samples
streams recovered were analyzed, and the analysis of taken during the test period were analyzed routinely
536 SOCIETY OF PETROLEUM ENGINEERS JOURNAL
 TABLE 1 - PROPERTIES OF CRUDE OILS USED IN EXPERIMENTAL WORK'

Crude Oils rAPI)

Disti Ilation A (26.2) B (34.8) C (19.4)
Ranges Yield Specific Mole Yield Specific Mole Yield Specific Mole
Fractions (0C) (wt%) Gravity Weight (wt%) Gravity Weight (wt%) Gravity Weight
1 IBP to 148.9 7.11 0.7150 97.5 14.27 0.7416 100.0 2.01 0.7567 105.0
2 148.9 to 204.4 9.00 0.7760 146.1 11.73 0.7901 130.0 4.38 0.8095 134.0
3 204.4 to 260.0 7.02 0.8054 159.0 12.82 0.8208 166.0 7.08 0.8504 173.0
4 260.0 to 315.6 8.68 0.8418 198.0 11.41 0.8413 209.0 8.31 0.8762 204.0
5 315.6 to 371.1 8.49 0.8783 245.0 10.89 0.8607 249.0 9.94 0.9007 239.0
6 371.1 + 59.70 0.9888 675.0 38.88 0.9334 508.0 68.38 0.9806 609.0
·Distillation ranges are true boiling-point values and molecular
weights were determined using ASTM D·2503 procedures.
Note: The boiling point range for each fraction, except Fraction 1, is
55.56'C (100'F).

by mass spectrometer. Oil phase samples were the weight fractions of the flue-gas components as
submitted for specific gravity and simulated well as those of the crude-oil fractions should be
distillation by the chromatographic method ASTM consistent both at feed and equilibrium. Consistency
D2887. The latter re~uires some explanation. calculations of this type have been performed for all
It has been shown 0 that the simulated-distillation experimental runs. Inconsistent experimental data
test is not representative to the extent that the were found mostly for experimental runs at
heaviest fraction of the oil is not propagated through equilibrium temperatures equal to or higher than
the column. In this work, actual true boiling-point 31S.6°C (600°F). Violations of the consistency
fractionations were compared with simulated- principle are attributable to inaccuracies in analytical
distillation results (on a weight fraction basis) for measurements and to unstable liquid-flow conditions
each crude oil used to establish the amount of the and possible thermal cracking at higher equilibrium
heaviest portion of the oil that was being lost in the temperatures. The results reported in this paper are
column. Appropriate corrections then were made to mostly in the temperature ranges of 204 to 260°C
the simulated-distillation analysis of each bottoms (400 to SOO°F) and reflect an experimental error of
sample when determining equilibrium vapor and less than 1 wt% for each component.
liquid compositions. The results of simulated
distillation after the correction procedure reflect an Consistency Test on the Equilibrium Coefficients. In
average absolute deviation of 0.34070 for each the literature, two procedures usually are described
fraction of the three crude oils compared with those to locate inconsistencies in K values and to allow for
by true boiling-point fractionation. The equivalent extrapolations of K values of other fractions. The
deviation without the correction procedure is 6.98%, first procedure was discussed by Hoffmann et al. 5 in
with the largest deviations contributed by the their work on gas-condensate systems. By the
heaviest fractions of the three crude oils. Clausius-Clapeyron equation 11 and the assumption
of an ideal solution, one can relate log KiP linearly to
Experimental Fluid Materials a boiling-point function called F( T bi ).
The experimental fluid system consists of three major
groups of materials: flue gas, water, and crude oil.
Pci )
log ( 101.3
(1Tbi - T1)
The flue gas used in our experimental work was F( T bi ) == 1 1 ' ....... (1)
obtained from Matheson Gas Products and was a
high-pressure mixture of approximately 84% N 2 , (T - T )
bi ci
11 % CO 2 , and S% CH 4 by moles. Three crude oils
were used in our work. The properties of these crude where Pci is the critical pressure in kPa (substitute
oils are given in Table 1. 14.7 for 101.3 for P ci in units of psi) of the ith
component, and Tbi and Tci are, respectively,
Consistency of Experimental Data boiling-point and critical temperatures (in degrees
Data consistency of multicomponent mixture systems Kelvin) for component i. In our work, a linear
can be tested through the Gibbs-Duhem equation. 11 regression subroutine was developed to fit log KiP as
However, the Gibbs-Duhem equation depends on a linear function of F( T bi ) for each set of ex-
experimental values of activity coefficients. Errors in perimental data. Obviously inconsistent data were
evaluating these data would render the testing excluded manually so as not to influence the
procedure invalid. In this regard, we resort to two regression procedure. Fig. 3 is a typical plot resulting
methods of checking the consistencies of ex- from this procedure. Except for N 2 , CO 2 , and the
perimental data. last fraction, our results indicate that for the crude-
oil systems studied, log (KiP) and F( T bi) can be
Consistency Test of the Component Weight Frac- fitted quite well by a straight line.
tions. Assuming no thermal cracking or chemical The second procedure is to express log (K i ) or log
reactions occurred during the experimental process, (KiP) as a polynomial function of the normal
OCTOBER 1981 537
 :\
10000
or J~ 10000
~~
~o

~
1000
/
0/1 ~
1000
\~ \~~
0\ '\
"-
><

/;/ "-
><
~o
\\
100

d: 1 PRESSURE, k Po

°
6
6996.0
10443.4
100

EXPERMENTAL PREDICTED
~~"..~~
o~
10
•
10
°6 . 699604
10443.42
6,\

0
• 13800.80
..
N
""
z "'~
N
z
I
u
. 8
~
i=

~
~
~
~
Q
t
~
i=
u
~

~
i=
\I
~

-2 -I o I 2 3 4
F(Tbi) 500 600 700 800
Tb' OK

Fig. 3 - Consistency of experimental equilibrium K values Fig. 4 - Consistency of experimental equilibrium K values
vs. boiling-point fractions, Crude B at 204.4°C vs. boiling points, Crude B at 204.4°C (400°F).
(400°F).

boiling-point temperature of the component, T bi . paraffinic, naphthenic, and aromatic (PNA) analysis
The idea stems from the physical reasoning that for ofthat fraction. We adapted the Bergman procedure
an equilibrium mixture, the K values should be a to calculate Pc' T c' and w for the crude-oil fractions.
continuous function of certain characteristic Details of the Bergman procedure have been
properties of the components, such as the boiling- described elsewhere in the literature. 13 ,17 In our
point temperatures. In Fig. 4, we plot log (KiP) with work, the Bergman procedure was modified slightly
respect to T bi . Except for the CO 2 data, most of the as outlined in the following.
data can be fitted by a smooth curve for the three sets Instead of using Bergman's linear relationship of
of data shown. the fractional aromaticity, fa' vs. the Watson
characterization factor, K uop - i.e.,
Theoretical
Characterization of Crude-Oil Fractions fa = O.7(l2.1-Kuop ) . . . . . . . . . . . . . . . . . . . . (2)
The Peng-Robinson equation of state requires that
for each component in a mixture the critical - we used a correlation developed by Jacobyl8 who
properties Tc. Pc. acentric factor w, and binary correlated fa as functions of specific gravity at 20°C
interaction coefficients kij's, be defined. The binary (68°F) and the molecular weight. The reasons for
interaction coefficients tor crude-oil fractions are such modification are (l?:
Eq. 2 was obtained from
discussed in the next section. rather scattered data, 3 and (2) the Watson
The pseudocritical properties, as well as the characterization factor is very insensitive to even a
acentric factor, of a crude-oil fraction can be wide range of boiling-point and specific-gravity
characterized by certain physicochemical properties values.
associated with the particular fraction, such as
specific gravity at lS.S6°C (60°F), weight-averaged Regression Calculations To Obtain
boiling-point temperature, and fractional aroma- Interaction Coefficients
ticity, etc. Several correlations for calculating critical Binary interaction coefficients between methane and
proper!ies of hy~othetical fractions are availa~le in C 6 + fractions were published by Katz et al. 19 in
the hterature. 2-16,30,31 Hopke and Lm,12 their study of phase behavior of condensate/ crude-
Bergman,13 Erbar,14 and most recently Robinson oil systems using the Gas Producers Assn. (GPA)
and Peng l6 have proposed methods to determine the Peng-Robinson program. However, the binary inter-
critical properties of the C 7 + fractions based on a action coefficients between crude-oil fractions are
538 SOCIETY OF PETROLEUM ENGINEERS JOURNAL
 0.1 r - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - ,

0.1
9
8
7
klj
6 001

2
0.001

o VALUES OBTAINED BY KATZ ET AL'·

a VALUES OBTAINED BY US
DATA USED fOR REGRESSION
0.01 C3
YARBOROUGH 21 MIXTURES 3, 4, 5,6
9
24 CASES
8 C.
65.56 ~ T,·C ~ 121.11
7
6L-~ __-L__ ~ __ L-~ __- L__~__L _ - L_ __L~ 3185. 4 ~ P, kPo ~ 22097.7
100 150 200 250 300 350 400 450 500 550 600 650
o 0001 '----:'40':-~-60~--'--:'OO:--~-:1-:c00:--L-...lI20-:--~-I40L--L~I60~
MOLECULAR WEIGHT MOLECULAR WEIGHT Of N-PARAffIN, GM/GM-MOLE

Fig. 5 - Binary interaction coefficients for oil fraction Fig. 6 - Binary interaction coefficients between CH 4 and
pairs as a function of molecular weight of frac· n·paraffins as a function of molecular weights of
tions. n·paraffins.

not available in published literature to our during the regression procedure. This is especially
knowledge. In this paper, we have used a general- true when the vapor-phase mole fractions of the
purpose nonlinear optimization package,20 together heavy fractions are much smaller than those of the
with part of our experimental data to determine the flue-gas components. Fig. 5 shows our kij's. In the
kij's for crude-oil fractions. figure, the binary interaction coefficients between the
This was achieved by first defining the objective ith fraction and other fractions, namely, kij's for all
function: j.c i, are plotted vs. the molecular weights of other
fractions. The linear connections in Fig. 5 serve only
to show the trend of kij's vs. the molecular weights.
We also used the multicomponent hydrocarbon
no nc mixture data published by Yarborough 21 to
determine the kij's between methane and other
+Wy
j=1
E
E i=1 wyi \Y Ei,j - Yi,j
PR \
hydrocarbons. The results are shown in Fig. 6, from
which it is seen that the kij's between methane and
no nc other n-paraffins can be fitted quite well linearly by a
+Wk
j=1
E
E i=1 Wki\K7,j-K[l\· .... (3) log kij' vs. M· plot. By comparison, Peng and
Robinson 9 used kij = 0 for all methane/n-paraffin
pairs. Improvement of the equilibrium K values
We minimized Y with respect to the interaction
coefficients k ij of the crude-oil fractions. To simplify resulting from our kij for methane is shown in Table
2 for a methane-rich mixture studied by Yarborough
the regression problem further, the following relation
at T = 121.11°C (250°F) and p = 7122.28 kPa
was used to correlate kij as functions of the
(1,033 psia).
molecular weights of the pair involved.
1M· M· Discussion of Results
kij = Xi..,) 100:0 x ~ , ............... (4)
In this section we show the effects of several im-
where Xi is the independent variable to be deter- portant variables namely, equilibrium tem-
mined by the regression subroutine and Mi and M j peratures and pressures, crude-oil specific gravities,
are molecular weights of fractions i and j. In our and feed molar gas/oil ratio (Fg) and water/oil ratio
regression work, it was found that the weighting (F w) on the equilibrium-phase mole fractions.
factors are essential in establishing a local minimum Whenever experimental data are available, we
OCTOBER 1981 539
 TABLE 2 - EFFECT OF k jj OF METHANE ON PREDICTED K VALUES
Average
Absolute
CH 4 C2 H6 C3 HS nC5 nC6 Toluene nC 10 Deviation
-- --
Experimental K values 21 4.1096 1.6742 0.8906 0.2606 0.0866 0.0670 0.0146
Kvalues' 3.8990 1.6622 0.9011 0.2696 0.0906 0.07671 0.01981 0.0371
Kvalues" 4.1211 1.6600 0.8963 0.2670 0.0891 0.07610 0.01850 0.0076
'Predicted by the original GPA Peng·Robinson program. 9
"'Predicted with our kij's for methane.

plotted the experimental K values together with those pressure increases. However, for light and medium
predicted by the Peng-Robinson equation of state. crude-oil fractions, namely fractions 1 to 4, absolute
This is to demonstrate that the given equation of minimum values can be located in the isothermal K-
state, with appropriate characterization of the value curves. Also, for the heavy fractions, the K
parameters associated with the crude-oil fraction, values are rather insensitive to pressure changes at
can predict basic phase equilibrium data satisfac- lower pressures and then increase slowly as pressure
torily at conditions encountered in thermal recovery increases. These phenomena are indicative of the
processes. fluid system being operated near' the retrograde
region.
Effect of Pressure on Equilibrium Coefficients
Figs. 7 and 8 compare predicted and experimental K Effect of Temperature on Equilibrium Coefficients
values and illustrate the effect of pressure on Comparison between predicted and experimental
equilibrium K values at 204.44 and 260.0°C (400 and data and the effect of temperature on the K values of
500°F) for Crudes A and C, respectively. In general, the components are shown in Fig. 9 for Crude B at
predicted and experimental data agree closely. Also, 10 443.42 kPa (1,514.7 psia). Again, for this set of
as can be seen from these figures, the K values for the conditions there is good agreement between predicted
flue-gas components decrease monotonically as and experimental data. Fig. 9 also shows that as

1000 1500 2000 2500 3000 p(psia) 1000 1500 2000 2500 3000 p(psia)
10.-----,-----,-----,-----,-----,------. 10.0,-----...
,-----.-----,-----,-----,----'----'--.
:::: AI

~~
G:--~ ___~~~
D~~ _ _ •
o ~ _ _6~
~-Q
o

1.0

PREDICTED. , . , •
EXPERIMENTAL 0 , 6, , 0

NOTE: K VALUES OF
FRACTION 6 OUT OF
SCALE, NOT SHOWN

'" 0.1 '" 0.1

•_____ _____
o FRACTION 3
~ ~----_.-----i
0

0.01 0.01 o
PREDICTED: ., ., •
FRACTION 4 O_i~
EXPERIMENTAL: 0, 6, 0 ·----"i~---·

A
.-.--~~
FRACTION 6

0.001 ........._ __..l._ _ _ _ _ _ _- ' - -_ _ _ _ _ _----'-_ _ _ _ _ __ _ _ ' 0.001 L....--'----_ _ _ _ _ _---'-_ _ _ _ _ _ _ _-'--_ _ _ _ _ _-'--_ _ _ _ _ _----'
L
.
.
.
.
-
-
'
-
-
~
5000 10000 15000 20000 5000 10000 15000 20000
p(kPo) p(kPo)

Fig. 7 - Effect of pressure on equilibrium K values for Fig. 8 - Effect of pressure on equilibrium K values for
Crude A at 260°C (500° F). Crude C at 204.4°C (400°F).

540 SOCIETY OF PETROLEUM ENGINEERS JOURNAL

 TABLE 3 - EXPERIMENTAL EQUILIBRIUM K VALUES FOR DATA PRESENTED IN FIGS. 7 THROUGH 9

Equilibrium K Values

Component Crude A (Fig. 7) Crude B (Fig. 8) Crude C (Fig. 9)

N2 4.391 4.183 3.595 2.715 7.430 5.421 4.501 3.893 3.289 4.701 3.862 3.250 3.603
methane 3.312 3.180 2.670 1.980 5.440 3.310 2.901 2.423 2.010 2.996 2.810 2.453 2.433
CO 2 2.653 2.708 2.380 1.890 2.897 2.352 2.210 1.840 1.605 2.361 2.420 2.157 2.434
Fraction 1 0.423 0.299 0.214 0.261 0.116 0.0919 0.0938 0.0981 0.0980 0.135 0.277 0.492 0.673
Fraction 2 0.158 0.114 0.104 0.115 0.0601 0.0398 0.0379 0.0427 0.0411 0.0472 0.129 0.215 0.320
Fraction 3 0.0782 0.0578 0.0501 0.0748 0.0274 0.0221 0.0205 0.0232 0.0230 0.0228 0.0737 0.131 0.196
Fraction 4 0.0459 0.0369 0.0340 0.0433 0.0104 0.0066 0.0076 0.0076 0.0088 0.0093 0.0281 0.0646 0.145
Fraction 5 0.0184 0.0218 0.0175 0.0251 0.0021 0.0022 0.0026 0.0020 0.0031 0.0037 0.0129 0.0320 0.0718
Fraction 6 0.0010 0.0009 0.0013. 0.0020 0.00007 0.00007 0.00009 0.00015 0.00017 0.0010 0.0024 0.0061 0.0186

Fuel Characteristics
Fg 2.21 1.74 4.32 3.88 2.47 3.09 3.24 3.42 4.10 1.23 2.65 2.70 1.72
T, OF 500 500 500 500 400 400 400 400 400 400 500 600 700
P, psia 1,014.7 1,514.7 2,014.7 3,014.7 1,014.7 1,514.7 2,014.7 2,514.7 3,014.7 1,514.7 1,514.7 1,514.7 1,514.7

temperature increases, the K values for the flue-gas Effect of Oil Specific Gravity
components decrease, while those of the crude-oil on Equilibrium Coefficients
fraction increase. However, as has been observed Fig. 10 is a plot of experimental K values vs. normal
before, 21 ,22 K values for the flue-gas components are boiling point for each component or fraction at
expected to show a maximum at temperatures below 260·C (500°F) and l3 890.80 kPa (2,014.7 psia) for
our range. This behavior reflects the tendency of the the three crude oils investigated. According to the
phases to approach each other in composition as the testing of experimental data consistency described
critical region is approached. earlier in this paper, for the data reported in Fig. 10
Experimental K values reported in Figs. 7 through the average absolute experimental error is 0.68 wtOJo
9 are summarized in Table 3. for the flue-gas component and 0.57 wt% for the

" 0.1

~ 0.1

CRUDE OIL
A 0

0.01 B +
C 6

0.01

,.t" /
/
v
4.~"?'
0/ r------------.,
PREDICTED: ., . , •
EXPERIMENTAL: 0, 6, 0
O.OO.IL---~~-~ _ _~_ _~_ _~____~ Fig. 10 - Effect of oil specific gravity on experimental
200 250 300 350 400 equilibrium K values at 260°C, 13890.80 kPa
(500°F; 2,014.7 psia) for Crudes A, 8, and C.
T(·C)
Fig. 9 - Effect of temperature on equilibrium K values for
Crude 8 at 10443.42 kPa (1,514,7 psia).

OCTOBER 1981 541

 TABLE 4 - AMOUNTS OF OIL VAPORIZED AND GAS K values was discussed by several authors. 24-27
DISSOLVED FOR CRUDE OILS A, B, AND C Standing and Katz,24 in their work on vapor/liquid
AT 260°C AND 13 890.8 kPa (500°F and 2,014.7 psia)
equilibria of natural gas and crude-oil systems,
Crude Oil (Feed F g) compared K values for different F 's and stated that
A (4.34) B (2.37) C (3.02)
"at pressure above 6894.8 kPa (1,600 psia) the effect
Dissolved gas in
of the composition of the system on the equilibrium
oil·rich phase, mol fraction 0.1503 0.1611 0.1463 constant-pressure relationships becomes very im-
Vaporized oil in portant." Jacoby and Yarborough,27 in their paper
gas phase, mol fraction 0.0429 0.0656 0.0238 on equilibrium properties of reservoir fluid systems,
presented results of K values for a hydrocarbon
system with varying feed gas/oil ratios. On the basis
crude-oil fractions. These error data suggest that of their results with pressure from 7170.5 to 14044.6
experimental errors are unlikely to influence the kPa (1,040 to 2,037 psia) and F from 2,052 to
trend of the curves in the figure. The results indicate 24,015, Jacoby and Yarborough 2 contended that -r
that the lightest crude-oil, Crude B, has the smallest "the total mixture composition (or mixture GOR
K values for the flue-gas components and the largest [FgD does not affect the K values noticeably until the
K values for the oil fractions. In other words, the fraction of oil components in the mixture reaches
amount of gas dissolved and the amount of oil very small values."
vaporized are larger for lighter crude-oil systems In this part, we first compare experimental and
(also see Table 4). This observation confirms the predicted equilibrium-phase mole fractions-i.e., L J
common belief that the lighter crude oils can be for crude-oil-rich phase, L2 for water-rich phase,
swelled . and stripped more easily in enhanced and V for vapor phase, with variable feed Fg and F w'
recovery processes. Then we present results showing the effect of Fg and
F w on the equilibrium K values for a vapor/crude-
Effect of Feed Gas/Oil Molar Ratio (Fg ) oil/water system based entirely on the predictions of
and Water/Oil Molar Ratio (F w) the GPA-Peng-Robinson program. The experimental
on Equilibrium-Phase Mole Fractions L J , L 2 , and V values are determined according to the
and Equilibrium KValues following three linearly independent mass-balance
The influence of feed Fg and F w on the equilibrium equations.

1500 2000 2500 3000 p(psia)

10r---~-----.----'----'-------------' 400 500 600
100
+-+ FEED Fg
9 .-e PREDICTED V/L 1 90
0-0 EXPERIMENTAL V/L 1

8 .--. PREDICTED V/L 1

WITH Fg = 3.52
80
15
7 70 14
e
13
6 \ 60\
- 12
-1
..... \ + + FEED Fw
> II
\ > \
ti 5 \ - 50 \- o [}. EXPERIMENTAL L 2 .V
10
II.. •\. N
-1
e • PREDICTED L 2 • V
9
4 40
\ • PREDICTED DEW POINT
FOR L2 8
~
II..

\
\ 7
3 30 \ 6

,/
/'" 5
2 20 --+-/ 4
3
10 2

OL-__L-____ ~ ______ ~ __ ~~~--~

0 0
10000 15000 20000 25000 250 300 350 400
p{kPo) T (OC)
Fig. 11 - Effect of feed FJl on equilibrium-phase molar Fig. 12 - Effect of feed Fw on equilibrium-phase mole
ratio V/L 1 , Crude c; at 260°C (500°F). fractions Vand L2 , Crude Bat p 10443.42 kPa =
(1,514.7 psia).

542 SOCIETY OF PETROLEUM ENGINEERS JOURNAL

 V+LI +L2 = 1.0, .................... (5) and L2 values are close. The predicted water-rich
phase dewpoint temperature is 280.6°C (537.8°F) at
YH 20 v+xk 2oLI +x1-IzaL 2 = ZH 20 ' ...... (6) the given pressure. It can be seen from the figure that
the phase quantities change quite drastically within a
Yi V+Xil LI +x?L2 = zi' ir!'H 20, ....... (7) small temperature range. For the case discussed in
this figure, about 90070 of phase change occurs for
both water and vapor phases between 204.4 and
where zi' Yi' xl, x? are, respectively, experimental
282.2°C (400 and 540°F).
mole fractions of the ith component at feed, vapor
phase, LI phase, and L2 phase. Experimentally, we In Figs. 13 through 16, we examine the effects of
varying feed Fg and F w on various phase equilibrium
assumed x1-I 2o = 1.0.
properties based entirely on predictions from the
Fig. 11 shows the effect of feed F w on the' GPA-Peng-Robinson program. The ranges of Fg and
equilibrium VI L 1 molar ratios at different pressures F w investigated are, respectively, 1 to 6 and 0 to 30.
for Crude C. These feed variations were introduced Fig. 13 illustrates the effect of feed F on the
by the experimental necessity of having enough of the equilibrium K values. As can be seen from tbe figure,
phases for reasonable measurements. It can be seen the K values are rather insensitive to the Fg's except
from Fig. 11 that except at the 10 443.5-kPa for the heavy fractions. For the last fraction, in-
(1,514.7-psia) point, the theoretical predictions creasing the Fg from 1 to 6.5 reduces the K value
match the experimental measurements satisfactorily. from 0.005 to 0.00276, or by about 45%. Table 5
One also can see from the figure that increasing the gives relative changes of K values subject to
feed Fg - i.e., atp= 17338.2 kPa (2,514.7 psia) - variations of F g . These re.lative changes were o~­
results in an increased VI L 1 ratio despite a tained from data reported m Refs. 24 and 27. It IS
simultaneously increasing pressure. Had the feed Fg illustrated through results given in the table that the
been held constant, as it was at 10 443.4 kPa (1,514.7 heavy fractions in an equilibrium mixture are indeed
psia), the equilibrium VI L 1 ratio would be as shown more sensitive to variations of feed F g • Figs. 14
by the dashed curve in the figure. through 16 illustrate the effect of feed F w on various
In Fig. 12, V, L 2 , and Fw are plotted vs. tem- equilibrium properties. In Fig. 14, mole fractions of
perature for Crude B at 10 443.4 kPa (1,514.7 psia). the crude-oil rich phase (L I) and the water-rich phase
For this set of data, the predicted and measured V (L 2 ) are plotted vs. F w. We note here that the L2
4.0 ,-----------------,100
N2
CH4
90
CO 2

80
1.0

70
26

24 ----+ 60
FRACTION I

50
N
...J
FRACTION 2
16 40
~ 0.1
FRACTION 3 8 14

~ 12

FRACTION 10
4
8 20
~ 6
FRACTION 5 4
---_-l 10

0.0 I
l-
t- Fw

~ FRACTION 6
Fig. 14 - Effect of feed Fw on equilibrium·phase mole
fractions L1 and L2, Crude C at 260·C, 17338.18
kPa (500·F; 2,514.7 psia).

0.00 I
2 3 4 5 6
Fg

Fig. 13 - Effect of feed Fg on equilibrium Kvalues, Crude B

at 260·C, 17338.18 kPa (500· F; 2,514.7 psia).

OCTOBER 1981 543

 TABLE 5 - EFFECT OF FEED Fg ON K VALUES reduce the number of fractions representing the
crude oil to minimize computing time. Through the
Relative Change of Component
KValues(%)
Peng-Robinson equation of state, we now examine
the feasibility of representing the crude oil by only
Case Fg/FgJ Methane Heaviest Component-
two lumped fractions. The feasibility is established if
a 4.92 4.3 9.1
b 1.96 11.0 44.0
a two-fraction scheme gives a set of results fairly
c 2.0 2.1 5.9 close to that predicted by the original six-fraction
d 6.0 15.0 45.0 scheme. This is considered in two parts in the
Notes:
following discussion.
a = Jacoby and Yarborough 27 ; p = 14044.6 kPa, T = 93.3°C (2,037
psia,199.9°F). A Procedure for Lumping Crude-Oil Fractions
b = Standing and Katz 24 ; p = 17236.9 kPa, T = 4S.9°C (2,500 psia,
120°F). We have devised a simple procedure that can serve as
c, d = our study, p = 17 33S.2 kPa, T = 260.0°C (2,514.7 psi a, 500°F).
°n·decane in Case a; designated as C7 + in Case S, with
a guideline for lumping the oil fractions. The idea for
l' = 0.S50, M = 199.0; designated as Fraction 6 in Cases c and d, with the procedure lies in the physical reasoning that
= l' =
0.9334, M 50S.0.
crude-oil fractions having relatively close
phase appears only after the F w exceeds a certain physicochemical properties can be represented fairly
value - in this case, 3.2. In Fig. 15, the effect of Fw accurately by a single fraction. The closeness of these
on the amount of oil stripped into the gas phase as properties can be reflected by the slopes of curves
well as that of the gas dissolved into the L 1 phase is when these properties are plotted against some
examined. Data in Fig. 15 show that before the characteristic independent variable. Since almost all
appearance of the L2 phase, increasing F w values is the physical properties of oil fractions are measured
unfavorable to both stripping and swelling of the L 1 and defined according to their boiling-point ranges,
phase. The effect of feed F w on the K values is shown it seems appropriate to consider the weight-averaged
in Fig. 16. These data illustrate that the effect of F w boiling-point (tb ) of each fraction as an in-
on the K values is rather insignificant when the L2 dependent variable.
phase is present. This statement also is illustrated With this argument, the procedure consists of the
numerically by data in Table 6. following five steps.
1. Plot the available physicochemical properties,
Feasibility of Representing such as specific gravity, molecular weight, fraction
the Crude Oils by Two Fractions aromaticity, etc., of each original fraction vs. its tb'
In the application of a thermal simulator or a 2. Compute numerically the slope for each
compositional simulator, one always wishes to fraction at each t b (i.e., m ij)' where i = 1 ... n!, j = 1
100

00210
ca=FW AT WHICH L, PHASE APPEARS I 028
N,
CH.

- CO,

026 10

> "'"' \ -"

~
'"
<f)

024 FRACTION I
~ 00200 '"...J:::>
...
~

o
...o - ...0 FRACTION 2
'" 01
z z
o Q
i= I- FRACTION 3
u 022 U
« «
...
ex:
...
Q:

~ 00195
o
'"...J FRACTION 4
~
"
0.20
FRACTION 5
00 I
- ~
00190

- 018
I
c 5 10 15 20 25 30 35 40
Fw

000 I
o 04 O.B 1.2 16 20 2.4 2.B
Fig. 15 - Effect of feed Fw on amount of oil vaporized Fw
(stripping) and amount of gas dissolved in oil-rich
phase (swelling), Crude C at 260°C, 17338.18 kPa Fig. 16 - Effect of feed Fw on equilibrium K values, Crude
(500°F; 2,514.7 psia). C at 260°C, 17 338.18 kPa (500°F; 2,514.7 psia).

544 SOCIETY o,F PETROLEUM ENGINEERS JOURNAL

 TABLES-EFFECT OFFEED Fw ON K VALUES

Without Presence of L2 Phase With Presence of L2 Phase

Relative
Fw 0 3.0 Change (%) Fw 3.8 38 Change (%)
KCH4 2.8056. 2.1745 -22.5 KCH4 2.6157 2.6198 0.16
3 0.4514 0.9524 111 3 0.8975 0.8866 -1.21
KFraction 6 X 10 KFraction 6 X 10

... np ' and nf = number of original oil fractions and Approach 2.

np = number of available properties.
3. Define and compute the normalized, absolute given
inij - i.e., properties
of
in·· = Im··1
II
original
IJ maxi=l...n)mijl fractions
4. Compute the weighted sum of slopes,

lf 1
np Pei
Ali = E
j=1
Wjinij'
correlations J
:~i
where Wj is a weighting factor associated with thejth
property - normally, Wj = lInp. _
5. Judge the numerical values of Mi for each
fraction and group those fractions which have similar
Ali values.
In this way, Ali represents the averaged change of mixing rule TeL =E Ii Tei
physicochemical properties of the crude-oil fractions
along the boiling-point axis. Applications of this
procedure are shown later in this section.
Rules for Determining Pseudocritical Properties
and Acentric Factors of the Lumped Fractions Subscripts i and L refer, respectively, to the original
Two approaches to determine the pseudocritical and lumped fractions, M and l' refer to molecular
properties and acentric factors, PeL' TeL> wL> for the weight and specific gravity at 15.56°C (60°F), andh
lumped fractions are considered. These two ap- can represent a weight fraction, a mole fraction, or
proaches are described schematically as follows. some property-weighted fraction of the ith com-
ponent.
In our study, we found that Approach 2 gives more
accurate results than Approach 1. Thus, the results
Approach 1. discussed later in this paper are based on Approach
2. The mixing rule needed in both approaches plays
an important role in the accuracy of the results. The
given influence of the mixing rule enters through the
properties parameter Ii' whose definition was given in the
of
original preceding paragraph. We have examined and
'Yi compared results with Ii selected as mole fraction
fractions
and as a weight fraction, and found that the mole-
fraction mixing rule gives better results. Our findings
further indicate that while the predicted equilibrium
properties - e.g., molecular weight and density of
oil phases - are not sensitive to different mixing
rules, the vaporization rates (i.e., the equilibrium K
values) depend strongly on the mixing rule.
Regardless of the mixing rules discussed above, the K
value of the lumped lighter fraction is much smaller
than that obtained from the six-fraction scheme for
similar boiling-point ranges. To overcome this

f 1 deficiency caused by using a two-fraction scheme, we

l J. conducted a sensitivity study on the parametric effect

PeL
correlations :~ of TeL> PeL> and wL on the predicted results. We
found that TeL is the most influential. Thus, to
improve vaporization prediction in the two-fraction
OCTOBER 1981 545
TABLE 7 - Mi VALUES FOR CRUDE OILS A, B, AND C

Oil
Fraction Crude A Crude B Crude C
1 0.4099 0.4440 0.6925
2 0.3629 0.3572 0.6195 1500 2000 2500 3000
3 0.3925 0.2737 0.5048 070
4 0.4224 0.2552 0.4374
5 0.5191 0.4311 0.5292
6 0.9703 1.0000 0.8919 0.65

0.60

TABLE 8 - DEVIATION OF EQUILIBRIUM FLUID

PROPERTIES WITH DIFFERENT LUMPING SCHEMES
.
-
055
AND MIXING RULES, CRUDE OIL AT 400°F;
2,014.7 PSIA (204 °C, 13 890.8 kPa)
NO OF MIXING
050 CURVE FRACTION RULE
Lumping , 6
Scheme Weight- Mole- 2 2 ""'WEIGHTEO
Component Fraction Fraction pc-Weighted 0.05'-----------1
3 2 MOLE FRACTION
i-j, k+ Mixing Rule Mixing Rule Mixing Rule 4 2 WEIGHT
FRACTION
1-2 3+ 8.20 1.75 2.21
1-4' 4 + 4.59 2.18 1.45
1-5: 6+ 4.14 2.36 2.03
003
-" _ _ _ 3

002
--- ----
TABLE 9 - DEVIATION OF EQUILIBRIUM K VALUES WITH _ _ _ _4 _ _ _ _ _ _ _ _
DIFFERENT LUMPING SCHEMES AND MIXING RULES,
CRUDE OIL A AT 400°F, 2014.7 PSIA 001

(204°C, 13 890.8 kPa)

Lumping Weight· Mole- ,= 15000

Scheme Fraction Fraction pc-Weighted P (kPaJ
i-j, k+ Mixing Rule Mixing Rule Mixing Rule
1-2 3+ 112.13 95.95 72.11 Fig. 17 - Effect of different mixing rules on liquid and
1-4' 4 + 115.61 132.58 115.53 vapor mole fractions of the lumped fraction - i.e.,
1-5: 6 + 67.70 35.39 7.04 X 1.5 and Y1-5 - Crude B at 204.4 °C (400° F).

system, we devised a critical-pressure-weighted fraction lumping scheme, the optimal lumping is to

second-order mixing rule for TcL only as follows. consider Fractions 1 to 5 as a lumped fraction and
Defining Fraction 6 as another for the three crude oil samples.
To validate this statement, we have conducted
, for T ci ' . . . . . . . . . . . . . . . (8) simulations with various lumping schemes. We then
compared results of these various schemes with
respect to those predicted by the original six-fraction
then scheme. The comparisons were made in two major
areas:
Tci = L L fJ) (TciTc) Yz(1-kij) , ....... (9) 1. Equilibrium fluid properties: molar volume,
i ) molecular weight, density, volume percent, mole
percent of both oil-rich and gas phases are compared;
where fi and k ij are, respectively, the mole fraction
and binary interaction coefficient for the original oil
these properties are designated and pl
where L pl,
and Trefer to lumped and true - i.e., that predicted
fractions. A concise treatment of mixing rules is by the original six-fraction scheme.
given in Ref. 28. The introduction of the critical- 2. Equilibrium K values: the K values by the
pressure-weighted fi and the binary interaction lumped scheme are designated K7_) and Kl+,
coefficient, kij' reduces the critical temperature, where the subscript i - j denotes lumping from ith to
T cL ' slightly for the lumped fraction. This con- jth fraction and k + ; from kth fraction on to the last
sequently leads to a more reasonable K value for the fraction, the K values of the six-fraction scheme are
lumped fraction. designated K7_) and KI+, where
Example Illustrations
) 6
In this part, we present some examples to sub-
stantiate the two preceding discussions. Based on the L Ym L Ym
lumping procedure that we developed, the Mi values
are listed in Table 7 for each oil fraction for Crudes
{-j t
KT . _ _ m:.:.:..,..=.:...i__
K Tk+ -
-
__
m-;o=_k_ _
6

A, B, and C. From Table 7, it is clear that for a two- m=i

L
m=k

546 SOCIETY OF PETROLEUM ENGINEERS JOURNAL

 T("F)
300 400 500 600

100,000
10,000

-
10,000
1000

p(psia}
1500 2000 2500 3000

o PREDICTED BY
6 100 6-FRACTION SCHEME

+ PREDICTED BY
2-FRACTION SCHEME

/
NO Of MIXING
CURVE FRACTION RULE

f<-WEIGHTED

MOLE FRACT ION

WEIGHT
FRACTION

100 I
[AJ-

Ii 10

j! v

O~-I~OOOO~--~'5=OOO~--~20000~
10
l
200 250 300
p IkPa)
T (OC)

Fig. 18 - Effect of different mixing rules on the relative Fig. 19 - Prediction of phase envelopes of Crude B with
deviations of fluid properties predicted by two- six· and two-fraction schemes.
fraction scheme, Crude Oil Bat 204.4 'C (400'F).

Tables 8 and 9 give relative percentage deviations composltlOns used for this study were the ex-
for various two-fraction lumping schemes and for perimental feed compositions for the 315.6°C,
three mixing rules. The results confirm the lumping 13 890.8-kPa (600°F; 2,OI4.7-psia) case, with a feed
procedure we established from the Mi-value study. Fg = 1.61 and an Fw = 14.36. In Fig. 19, the phase
Overall, the results indicate that lumping of Frac- envelopes are shown. It is demonstrated through this
tions 1 to 5, plus the pc-weighted fraction scheme, example that the two-fraction scheme predicts phase
gives the most accurate results as compared with the envelopes very close to those obtained from the six-
six-fraction scheme. fraction scheme for a vapor/crude-oillwater system.
In Fig. 17, the liquid- and vapor-phase mole We now apply this lumping procedure and the
fractions of the lumped fraction, X~_5 and Y~-5' are mixing rules to a gas/crude-oil system studied by
plotted vs. pressure for Crude B at 204.4°C (400°F). Hariu and Sage. 29 In total, 15 fractions were used by
For the same data as in Fig. 17, the relative those authors to represent the crude oil in addition to
deviations of the fluid properties are shown in Fig. four light hydrocarbon components in the mixture.
18. Both figures, together with Tables 7 and 8, The physical properties associated with each fraction
confirm that through a lumping procedure and a - i.e., average boiling point, molecular weight, and
special mixing rule the crude oil can be represented specific gravity, together with the mixture feed
fairly accurately by only two fractions. composition - also were given by Hariu and Sage.
Fusse1l 23 recently attempted to predict phase According to these physical properties, we calculated
equilibria of three coexisting hydrocarbon phases the Mi values (see Table 10). Judging from the data
using a lumped C 7 + fraction. Here our lumping in Table 10, it is clear that no distinct changes of Mi
procedure, as well as the pc-weighted mixing rule is are observed along the boiling-point axis. This is
applied to predict the phase behavior of a vapor caused by the narrower boiling-point ranges used to
(V)/crude-oil (L 1 )/water (L 2 ) system. The feed define a fraction. In Table 11, we show the effect of
OCTOBER 1981 547
 different fraction-lumping schemes and two mixing
rules on the prediction of bubble-point temperature
at 10 342.1 kPa (1,500 psia) for the gas/crude-oil
TABLE 10-M; VALUES FOR CRUDE OIL IN REF. 29 system studied by Hariu and Sage. It is demonstrated
through these results that the pc-weighted-mixing
Average rule converges for all lumping schemes presented and
Crude Oil Boiling Point gives a slightly better result when both cases con-
Fraction ("C) M; verge. The results also show that the bubble-point
1 60.0 0.6943 temperature prediction is not sensitive to the lumping
2 90.0 0.5894
3 114.4 0.5543 schemes.
4 133.3 0.5170
5 154.4 0.4289 Conclusions
6 175.6 0.3553 Experimental phase equilibrium data have been
7 199.4 0.2717
8 223.9 0.4805
obtained for systems involving flue gas, water, and
9 247.8 0.5608 crude oil over a range of pressures from 6996.0 to
10 272.8 0.3173 20785.6 kPa (1,014.7 to 3,014.7 psia) and tem-
11 311.1 0.4175 peratures from 204.4 to 371.1 °C (400 to 700°F). A
12 367.7 0.5633 nonlinear regression approach for determining the
13 425.0 0.5787
14 494.4 0.6395 equation-of-state parameters has been utilized in
15 582.2 0.7089 conjunction with the Peng-Robinson equation of
state. Comparisons of predicted and experimental
phase data were made. Our study shows the
following.
1. The Peng-Robinson equation of state
TABLE 11 - PREDICTION OF BUBBLE·POINT satisfactorily predicts phase properties for
TEMPERATURE WITH VARIOUS LUMPING SCHEMES
AND TWO MIXING RULES FOR THE GAS/CRUDE OIL vapor/crude-oillwater equilibria at in-situ thermal
SYSTEM IN REF. 29, p= 10 342.1 kPa (1,500 psia) recovery conditions.
2. According to the predictions of the Peng-
Predicted Bubble Point Robinson equation of state, varying the feed Fg
Lumping Mixing Temperature (0C)
Scheme Rule
appreciably affects the equilibrium coefficients of
(True Value = 271.29)
heavy crude-oil fractions. This observation is also
1 A
1 B 273.68
valid for the effect of feed F w but only for systems in
2 A 274.02 which a liquid water phase has not been formed.
2 B 273.15 3. A simple procedure for lumping the crude-oil
3 A fractions is presented. Importance of an appropriate
3 B 274.73
4 A
mixing rule for determining the pseudocritical
275.35
4 B 272.4 properties of the lumped fractions is demonstrated
Notes: Lumping Scheme 1 combines Fractions 1 through examples.
through 4, 5 through 11, 12 through 15; Scheme
2 combines Fractions 1 through 5, 6 through 10,
4. Under the temperature and pressure ranges of
11 through 15; Scheme 3 combines Fractions 1, our study, it is feasible to represent the crude oil by
2 through 12, 13 through 15; Scheme 4 combines
Fractions 1 through 7 and 8 through 15. only two fractions using the Peng-Robinson equation
Mixing Rule A uses mole·fraction mixing rule for of state.
the critical temperatures of the lumped frac·
tions, TcL , while Mixing Rule B uses a Pc'
weighted mixing rule for TcL '
Nomenclature
·No numerical convergence. A,B,C = crude oil specifications
fa = weight fraction aromaticity of the
oil fraction, Eq. 2
fi = representing weight fraction, mole
fraction, or pc-weighted fraction
F(Tb ) I
= boiling-point function, Eq. 1
Fg = feed gas/oil molar ratio
Fw = feed water/oil molar ratio
kij = binary interaction coefficient
K,K i = equilibrium coefficient or K value
Kuop Watson characterization factor
L 1 ,L 2 oil- and water-rich phases,
respectively, also phase mole
fractions
m··IJ slope of the curve of property vs.
boiling point
Mi referring to averaged change of
properties vs. boiling points
548 SOCIETY OF PETROLEUM ENGINEERS JOURNAL
 M = molecular weight and I.C. Keyes, who performed the experimental
total number of components, oil work. Also the assistance of J.T. Swansiger in
fractions, and available developing and assembling the experimental ap-
physicochemical properties, paratus and performing the initial shakedown runs is
respectively greatly appreciated. We also thank the management
p pressure, kPa (psia) of Gulf Research and Development Co. for per-
mission to publish this paper.
referring to ith physicochemical
property
T = temperature, K
References
I. Prats, M.: "A Current Appraisal of Thermal Recovery," J.
Tb = boiling-point temperature, K Pet. Tech. (Aug. 1978) 1129-1136.
Tb = weight-averaged boiling point 2. White, R.R. and Brown, G.G.: "Phase Equilibrium at High
V = vapor phase; phase mole fraction Temperatures," Ind. and Eng. Chern. (Oct. 1942) 34, 1162-
1174.
weighting factor, Eq. 3 3. Poettmann, F.H. and Mayland, B.J.: "Equilibrium Constants
for High Boiling Hydrocarbon Fractions of Varying
weighting factor used in lumping Characterization Factors," Petroleum Refiner (July 1949) 28,
101-112.
procedure 4. Kirkbride, e.G.: "Method for Predicting Phase Equilibrium
mole fraction of Component i in Constants," Petroleum Refiner (Dec. 1945) 24, 97-104.
oil-rich phase 5. Hoffman, A.E., Crump, J.S., and Hocott, e.R.:
"Equilibrium Constants for A Gas-Condensate System,"
component mole fraction in oil- Trans., AIME (1953) 198,1-10.
rich phase 6. Grayson, H.G. and Streed, e.w.: "Vapor-Liquid Equilibria
for High Temperature, High Pressure Hydrogen-Hydrocarbon
independent variable for Systems," Proc., Sixth World Pet. Cong., Frankfurt/Main
nonlinear regression, Eq. 4 (1963) 20-PD7, 233-245.
mole fraction of component i in 7. Simon, R., Rosman, A., and Zana, E.: "Phase Behavior of
vapor phase COz-Reservoir Oil Systems," Soc. Pet. Eng. J. (Feb. 1978) 20-
26.
Y m = component mole fraction in vapor 8. Shelton, J.L. and Yarborough, L.: "Multiple Phase Behavior
phase in Porous Media During CO2 or Rich-Gas Flooding," J. Pet.
Y = objective function, Eq. 4 Tech. (Sept. 1977) 1171-1178.
9. Peng, D.Y. and Robinson, D.B.: "A New Two-Constant
Zi = feed mole fraction of ith com- Equation of State," Ind. Eng. Chern. Fund. (Jan. 1976) IS,
ponent 59-64; also available as GPA Peng-Robinson Program, Gas
Processors Assn., Tulsa.
'Y = specific gravity at lS.S6°C 10. Philyaw, L.E., Krc, A.E., and O'Neal, M.J.: "Gas
w acentric factor Chromatographic Analysis of Samples Containing Both
Volatile and Non-Volatile Organic Components," Anal.
Superscripts Chern. (May 1971)43, 787-789.
E = referring to experimental data, 11. Denbigh, K.G.: Principles of Chemical Equilibrium, Cam-
Eq. 3 bridge U. Press, New York City (1957) 230-235 and 214-216.
12. Hopke, S.W. and Lin, C.J.: "Application of the BWRS
L referring to lumped value Equation to Absorber Systems," Proc., 53rd GP A Con-
PR referring to value predicted by vention, Denver (1974) 63-71.
13. Bergman, D.F.: "Predicting the Phase Behavior of Natural
Peng-Robinson equation of Gas in Pipelines," PhD dissertation, U. of Michigan, Ann
state, Eq. 3 Arbor (1976).
T = referring to value based on six- 14. Erbar, J.H.: "Prediction of Absorber Oil K-Values and
Enthalpies," Research Report 13, Gas Processors Assoc.,
fraction scheme
Tulsa, OK (1977); available through GPA.
1,2 = referring, respectively, to L 1 and 15. Cavett, R.H.: "Physical Data for Distillation Calculations -
L2 phases, Eqs. S through 7 Vapor-Liquid Equilibria," Proc., 27th API Mid-Year
Meeting, San Francisco (1962).
Subscripts 16. Robinson, D.B. and Peng, D.Y.: "The Characterization of the
Heptanes and Heavier Fractions," Research Report 28, Gas
B = boiling point Processors Assoc., Tulsa (1978).
c = critical point, also pseudocritical 17. Bishnoi, P.R., H.eidmann, R.A., and Shah, M.K.:
"Calculation of Thermodynamic Properties of Bitumen
property Systems," Oil Sands (1977) 248-253.
i,j, k, n component index or property 18. Jacoby, R.H.: "NGPA Phase Equilibrium Research Project,"
index Proc., API Refining Div. (1964) 288.
19. Katz, D.L. and Firoozabadi, A.: "Predicting Phase Behavior
L lumped of Condensate/Crude-Oil Systems Using Methane Interaction
i - j = lumping oil fractions from ith to Coefficients," J. Pet. Tech. (Nov. 1978) 1649-1655.
jth fraction 20. "COMPUTE III - A General-Purpose Optimizer for
Continuous, Nonlinear Models," Gulf Management Science
k+ lumping oil fraction for kth Group, Gulf Oil Corp., Pittsburgh (Nov. 1977).
fraction to the last fraction 21. Yarborough, L.: "Vapor-Liquid Equilibrium Data for Multi-
Component Mixtures Containing Hydrocarbon and
Acknowledgments Nonhydrocarbon Components," J. Chern. and Eng. Data
(Feb. 1972) 17,129-133.
We gratefully acknowledge the assistance of R.M. 22. Jacoby, R.H. and Rzasa, M.J.: "Equilibrium Vaporization
Henry, who helped analyze the experimental data, Ratios for N2, CH4 , CO2 , C2~ and H 2S in Absorber Oil -

OCTOBER 1981 549

 Natural Gas and Crude Oil/Natural Gas Systems," Trans., APPENDIX
AIME (1952) 195, 99-110.
23. Fussell, L.T.: "A Technique for Calculating Phase Equilibria Mathematical Notation
of Three Coexisting Hydrocarbon Phases," paper SPE 6722 Relative Deviation
presented at SPE 52nd Annual Technical Conference and
Exhibition, Denver, Oct. 9-12,1977.
24. Standing, M.B. and Katz, D.L.: "Vapor Liquid Equilibria of
Natural Gas-Crude Oil Systems," Trans., AI ME (1944) 155,
232-245.
25. Eilerts, C.K. and Smith, R. V.: USBM R.1. 3642 (April 1942).
26. Yarborough, L. and Vogel, J.: "A New System for Obtaining
Vapor and Liquid Sample Analyses to Facilitate the Study of
Multicomponent Mixtures at Elevated Pressures," paper
presented at AIChE Meeting, Salt Lake City, May 1967. I: i ----- (100) for Table 7 .
27. Jacoby, R.H. and Yarborough, L.: "PVT Measurements on
Petroleum Reservoir Fluids and Their Uses," Ind. and Eng.
pt
Chern. (Oct. 1967) 59, 48-62.
28. Reid, R.C., Prausnitz, J.M., and Sherwood, T.K.: The KL .-KT .

29.
Properties of Gas and Liquids, (third edition), McGraw-Hill
Book Co., New York City (1977) Chap. 4.
Hariu, O.H. and Sage, R.C.: "Crude Split Figured by
(
I
I-)

KT .
I-}
I-}
I) (100)
Computer," Hydrocarbon Processing (April 1969) 143-148. for Table 8.
30. Kesler, M.G., Lee, B.I., and Sandler, S.I.: "A Third
Parameter for Use in Generalized Thermodynamic
Correlations," Ind. Eng. Chern. Fund. (Jan. 1979) 18, 49. SI Metric Conversion Factors
31. Riazi, M.R. and Danbert, T.E.: "Simplify Property o API 141.5/(131.5 + API) 0 g/ cm 3
Prediction," Hydrocarbon Processing (March 1980) 115. OF CF - 32)/1.8 °c
Original manuscript received in Society of Petroleum Engineers office Aug. 8, psia,psi x 6.894757 E+OO kPa
1979. Paper accepted for publication June 18, 1980. Revised manuscript
received Aug. 24, 1981. Paper (SPE 8393) first presented at the SPE 54th Annual
Technical Conference and Exhibition, held in Las Vegas, Sept. 23·26,1979. SPEJ

550 SOCIETY OF PETROLEUM ENGINEERS JOURNAL