SPE 81432

The Coefficient of Isothermal Compressibility of Black Oils

Muhammad Ali Al-Marhoun, SPE, King Fahd University of Petroleum and Minerals Dhahran 31261, Saudi Arabia

Copyright 2003, Society of Petroleum Engineers Inc.

c o = −(1/V )( ∂V / ∂P) T (1)
This paper was prepared for presentation at the SPE 13th Middle east Oil Show & Conference or
to be held in Bahrain 5-8 April 2003.

This paper was selected for presentation by an SPE Program Committee following review of c o = −(1/B o )(∂B o / ∂P) T (2)
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 When reservoir fluid studies are available, oil compressibility
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 can be calculated from the slope of an isotherm of the
SPE meetings are subject to publication review by Editorial Committees of the Society of
Petroleum Engineers. Electronic reproduction, distribution, or storage of any part of this paper
logarithm of the oil formation volume factor versus pressure
for commercial purposes without the written consent of the Society of Petroleum Engineers is curve as:
prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300
words; illustrations may not be copied. The abstract must contain conspicuous c = −∂ (lnB ) / ∂P (3)
o o
acknowledgment of where and by whom the paper was presented. Write Librarian, SPE, P.O.
Box 833836, Richardson, TX 75083-3836, U.S.A., fax 01-972-952-9435. This procedure will give instantaneous compressibility at a
particular pressure, but in the oil industry for material balance
Abstract calculations, usually the oil compressibility is calculated as an
This paper presents a new correlation for the coefficient of average value between two pressures of interest:
isothermal compressibility of black oils at pressures above the c = −(1/B )(B − B ) /(P − P ) (4)
o o o2 o1 2 1
bubble point. The correlation is expressed as an empirical Where
function of oil relative density at bubble point pressure, (5)
Bo = (Bo1 + Bo 2 )/2
reservoir temperature, bubble point pressure and the reservoir
pressure. A total of 3412 data points from 186 laboratory PVT
analyses from Middle East fields were used to develop the oil When the oil compressibility is used in the calculation of
compressibility correlation. The data encompassed a wide undersaturated oil density and oil formation volume factor
range of gas-oil ratios, oil and gas relative densities, reservoir above the bubble point, the following equations are applied:
pressure, and reservoir temperature.
Multiple linear and nonlinear regressions were used to
B = B e co (Pb −P)
o ob
(6)

develop this model. This model is chosen from large number and
of models tested. The evaluation of the correlation includes co (P − Pb )
ρo = ρob e (7)
comparative studies of the new and existing mathematical
where
models of oil compressibility correlation. P
The model is validated by using three different data sets 1
from other geographical region of the world not used in the co =
p − pb ∫c
Pb
o (p)dp (8)
development of the model.
The newly developed mathematical model and correlation The average compressibility is one value to be used for
outperforms the existing mathematical models and correlations convenience to approximate the compressibility between
for oil compressibility based on low value of average absolute bubble point pressure and the pressure of interest. To use the
percent relative error and standard deviation. correlations to estimate the average oil compressibility, the
pressure in the correlation is replaced by arithmetic average
Background pressure (P+Pb)/2.
The isothermal oil compressibility is an important physical In general, to avoid the calculation involved in Eq. 8 the
property in the design of high-pressure surface equipment and average oil compressibility between two pressures can be
reservoir calculations. Higher accuracy of oil compressibility calculated from the point function or instantaneous oil
estimates will improve the accuracy of the design of high- compressibility as follows:
pressure surface equipment and material balance calculations. c = c {(P + P )/2} (9)
o o 1 2
The isothermal oil compressibility is defined as the unit or
change of volume with pressure. It is a point function
as shown: co = {c o (P1 ) + co (P2 )}/2 (10)
1
Standing pointed out that minor differences exist in the
compressibility between liquids of the same density but
2 SPE 81432

different composition, but are not regular enough to take into used pseudoreduced pressure and temperature to determine
account. Calhoun2 stated that for a homologous series of undersaturated oil compressibility. Vasquez and Beggs4 in
compounds, compressibility varies more or less regularly with 1980 presented an empirical correlation to estimate
the variation in molecular size. This gives rise to the general compressibility using the available reservoir parameters. A
observation that for oils of similar base, the compressibility total of 4486 data points were used in the development of the
decreases as the density of oil increases. Values of oil compressibility correlation using the model:
compressibility rarely exceed 35 x 10−6 psi−1.
c o = ( a1 + a 2 + R s + a 3T + a 4 γ g + a 5γ api ) / P (19)
Compressibility Pressure Relationship
Under the same conditions of reservoir temperature and fluid
compositions, the isothermal compressibility of an Petrosky and Farshad5 in 1993 developed another correlation
undersaturated crude oil is a function of pressure only. The for oil compressibility using the same parameters used by
function is derived from the equation relating formation Vasquez and Beggs4 but in a different arrangement using the
volume factor and pressure. The following four equations are following mathematical model:
presented here:
1.Quadratic form: c o = a 1R sa 2 γ ag3 γ aapi
4
T a5 P a6 (20)
V = a1 + a 2 P + a 3P 2 (11)
Data Acquisition
When the oil volume is described by a quadratic equation, the The analyses of 186 bottomhole samples from the Middle East
isothermal oil compressibility will be oil fields were made available for this study. The
− (a 2 + 2a 3 P) experimentally obtained data were 3412 data points in 494
Co = 2
(12) data sets for oil FVF above bubble point pressure. The
a1 + a 2 P + a 3P isothermal oil compressibility was calculated from these data
2.Hyperbolic form: points. A description of the data utilized in the development of
the correlation is shown in Table 1.
a1 + a 2 P
V= (13)
1 + a 3P Table 1: Summary of Middle East PVT Data
When the oil volume is described by a hyperbolic equation, Property Xmin Xmax X s
the isothermal oil compressibility will be
Oil FVF At Pb 1.02 1.89 1.23 0.160
− a2 a3
Co = + (14) Bubblepoint Pressure 106.00 3331.00 1209.46 828.38
a1 + a 2 P 1 + a 3P Temperature, ºF 71.00 240.00 139.96 44.63
Gas-Oil Ratio 24.00 1453.00 400.27 316.97
3.Exponential form: Gas Relative Density 0.75 1.59 1.00 0.15

V = a 1e a 2P + a 3 (15)
Oil Relative Density 0.80 0.95 0.87 0.03
Oil Gravity 17.50 44.60 31.34 5.29
When the oil volume is described by an exponential equation, 6
Compressibilityx10 3.45 31.11 8.66 3.40
the isothermal oil compressibility will be
Pressure above Pb 175.00 5015.00 2541.20 1065.40
− a 1a 2 e a 2P
Co = (16)
FVF above Pb 1.002 1.88 1.22 0.16

a 1e a 2P + a 3
For each set of data, the data of FVF above
4.Logarithmic form:
bubblepoint pressure are fitted using the quadratic model:
lnV = a 1e a 2P + a 3 (17)
When the oil volume is described by a logarithmic equation, B o = a1 + a 2 + P + a3 P 2 (21)
the isothermal oil compressibility will be
C o = −a 1a 2 e a 2P (18) Then c o is calculated as
1 dB o − (a 2 + 2a 3 P) (22)
Literature Review co = − =
Calhoun2 in 1947 conducted the earliest research when he B o dp a1 + a 2 P + a 3 P 2
presented a graphical correlation for determining the The oil volume and the calculated Co are checked for the
isothermal compressibility of an undersaturated crude oil. physical trend as follows:
This correlation relates a value of average compressibility to • Oil volume is decreasing with increasing pressure;
the oil relative density at bubble point pressure. There is a
• Co is decreasing with increasing pressure;
single value of oil compressibility for all pressures above the
• The slope of Co, (d Co /d p) is negative; and
bubble point. Trube3 in 1957, for his graphical correlation,
SPE 81432 3

• The slope is decreasing with increasing pressure in Using least square linear regression, the following
absolute value. equation is found to be the best form that minimizes the
If these conditions are not met then data set is put aside and deviation from experimentally determined data:
considered not valid because it does not follow the 3
ln co = a1 + a2/ γ ob + a3 (P − Pb)/γ ob + a4/(T+460) (27)
physical trend.
The same number of data points used in the curve fit were where
generated using the equation above. a1 = − 14.1042
a2 = 2.7314
To validate the correlations obtained, four data groups were a3 = − 56.0605 x 10-6
used. First one is 78 PVT files analysis from Canada, the a4 = − 580.8778
second is 22 PVT files from Pakistan6, the third is 25 PVT
files from Yemen7, and 18 PVT files from Vasquez thesis8. Statistical Evaluation of Models and Correlations
The data groups are summarized in Tables 2. These data are The accuracy of models and correlations relative to the
checked for the physical trend test. The data failing this test experimental values is determined by various statistical means.
were excluded from the validation test. The criteria used in this study were average percent relative
error, average absolute percent relative error, maximum
absolute percent relative error, standard deviation, and the
Table 2: Summary of PVT Validation Data
correlation coefficient.
Data
Number of Number of Number of Table 3 presents the comparison of errors relative to
Data Points Data Sets PVT Files the experimentally determined undersaturated isothermal oil
Middle East 3412 494 186
compressibility of 3412 data points estimated from the models
Canada 495 78 78 and correlations. The model for oil compressibility of this
Pakistan 246 22 22 study achieved the lowest error and standard deviation, with
Yemen 182 25 25 the highest correlation coefficient accuracy of 0.9829.
Vasquez 115 18 18 Petrosky and Farshad5 model stood second in accuracy, with a
correlation coefficient of 0.9222. Vasquez and Beggs4 model
showed poor accuracy, with the highest error and the lowest
Correlation Development correlation coefficient of 0.3056.
Calhoun2 assumed that the oil compressibility is a function of
oil relative density at bubble point
c = f(γ ) (23) Table 3: Statistical Accuracy of Oil
o ob
This will yield a single value for compressibility Compressibility for Middle East Data
above the bubble point regardless of pressure.
Vasquez and Beggs4 assumed the following general relation of Correlations Er Ea Emax s r
undersaturated isothermal oil compressibility correlation as a Vasquez & Beggs 11.56 43.42 429.76 57.96
function of the reservoir fluid properties: Petrosky & Farshad 6.13 25.45 138.08 31.74

c o = f ( Rs , γ g , γ api , T , P ) (24) Models

Vasquez & Beggs 17.84 34.53 244.77 40.33 0.3056
It was found that different flash separation laboratory
Petrosky & Farshad -0.73 9.47 73.18 12.12 0.9222
tests yield different combination of data set for Rs, γg, γapi.
Therefore, different values of co will result for a given crude This Study -0.25 5.46 26.87 7.06 0.9829
at the same reservoir condition. Because of this shortcoming,
new parameters were chosen such that regardless of different The most important indicator of the accuracy of an
separator tests results, it leads to the same value of Co for a empirical model is the average absolute percent relative error.
given crude. It is used here as a comparative criterion for testing the
The undersaturated isothermal oil compressibility accuracy of Vasquez and Beggs4 model, Petrosky and
correlation was developed based on the following assumed Farshad5 model, and the model of this study. The three models
general relationship: were applied to the five different data sets shown in Table 2.
c o = f (γ ob , T , P, Pb ) (25) Table 4 presents the comparison of the average
absolute percent relative errors relative to the experimentally
where
determined undersaturated isothermal oil compressibility of
γ ob = (γ o + 2.18 x10 −4 R s γ g ) / Bob (26) the five sets of data estimated from the models and
correlations. The model for oil compressibility of this study
achieved the lowest error. Petrosky and Farshad5 model stood
The independent variables γo, Rs, γg, Bob were second in accuracy. Vasquez and Beggs4 model showed
obtained from laboratory separator tests conducted on oil poor accuracy.
samples at bubblepoint pressure.
4 SPE 81432

Figures 6-10 illustrate scatter diagrams of the

Table 4: Average Absolute Percent Relative Error predicted versus experimental Middle East Co values. These
Of Oil Compressibility cross plots indicates the degree of agreement between the
experimental and the predicted values. If the agreement is
Vasquez & Petrosky &
Beggs Farshad
This Study perfect, then, all points should lie on the 45º degrees line on
the plot. Compared to other cross plots, Fig. 10 shows the
Correlations
most tight cluster of points around the 45º degrees line
Middle East Data 52.37 24.97 5.56
indicating the excellent agreement between the experimental
Canadian Data 24.26 18.04 10.09 and the calculated data values. The most scattered points were
Pakistani Data 30.81 37.45 16.30 found in Fig. 6 and 8, representing Vasquez and Beggs4
Yemeni Data 48.97 47.74 17.24 Correlation and model respectively. The calculation of new
Vasquez Data 17.03 12.75 23.43
constants for Petrosky and Farshad5 model improves the
performance and reduces errors as shown in Fig. 7 and 9.
Models
However, the improvement of Vasquez and Beggs4 model
Middle East Data 36.63 9.34 5.56 with new constants is insignificant. Again, this indicates the
Canadian Data 22.24 8.89 4.77 superior performance of the new model compared to other
Pakistani Data 21.25 8.22 5.31 empirical correlations and models.
Yemeni Data 38.77 16.53 10.97
Conclusions
Vasquez Data 12.85 7.60 7.13 1. A new correlation for predicting the coefficient of
undersaturated isothermal oil compressibility has
The correlation developed by Vasquez and Beggs4 been developed.
correlation generally underestimates isothermal oil 2. Equation 27 estimates the coefficient of undersaturated
compressibility by as much as 50% at high pressure. Accuracy isothermal oil compressibility from bubble point oil
is improved at pressures near bubble point. Vasquez relative density, bubble point pressure, reservoir pressure,
correlation failed in predicting oil compressibility in 492, 9, 42 and reservoir temperature.
cases in the Middle East, Canadian, and Yemeni data 3. The comparative study of the correlation in the literature
respectively where oil compressibility is negative. Vasquez shows that the deviations as average absolute percent
and Beggs correlation will not guarantee applicability because relative error were lower for this study than for
of its nature and its structure. Other correlations and models estimations based on correlations in the literature.
do not fail. 4. The comparative study of the mathematical models of the
correlation in the literature shows that the deviation from
Graphical Evaluation of Models and Correlations experimentally determined data, indicated as average
The new correlation is a function of oil properties that do not percent relative error, average absolute percent relative
depend on separator test data that depend on the number of error, and standard deviation, were lower for this study
separator stages and separator pressure and temperature. The than for estimations based on other mathematical models
current empirical correlations will give different results in the literature. Also, the correlation coefficient of the
depending on the conditions of the separator test. Therefore, model of this study is closer to 1 than those of
the new correlation is not sensitive to the condition of the other models.
separator test. The form of the model shed light on the
relationship among variables for better understanding of the Nomenclature
physical behavior. Bo = oil FVF above
3
bubblepoint
3
pressure,
For the graphical presentation of errors, different RB/STB (m /m )
ranges of API gravity, reservoir temperature, oil relative Bo = average FVF
density at bubblepoint pressure, pressure difference (P − Pb), Bob = oil FVF at bubblepoint pressure,
3 3
and measured isothermal oil compressibility were selected to RB/STB (m /m )
test the accuracy of the considered models and correlations Co = oil compressibility above bubble point,
−1
with Middle East PVT data. psi−1 (kPa )
The statistical accuracy grouped by oil API gravity is
shown in Fig. 1. The new model gives the best accuracy for co = average oil compressibility
oil API gravity for the whole range of data. Petrosky and Ei = percent relative error
Farshad model stood second in accuracy. Vasquez and Beggs = 100( xm − xc ) / xm
correlation, Vasquez and Beggs model, and Petrosky and
Farshad correlation show poor accuracy respectively. Er = average percent relative error
Figures 2-5 show the accuracy of models and Ea = average absolute relative error
correlations grouped by reservoir temperature, oil relative Emax = maximum percent relative error
density at bubblepoint pressure, pressure difference, and s = percent standard deviation
n
measured isothermal oil compressibility, respectively. The
same trend discussed in Fig. 1 is observed in these figures.
= ∑ (E i − E r ) 2 /(n − 1)
SPE 81432 5

P = pressure, psia (kPa) SI metric conversion factors

Pb = bubble point pressure, psia (kPa)
°API 141.5 /(131.5 + °API) = g/cm3
R = correlation coefficient 3 3 bbl x 0.1589873 = m3
Rs = solution gas oil ratio, SCF/STB (m /m )
ft3 x 0.02831685 = m3
T = temperature, oF (K)
cp x 1 = mPa.s
v&v = volume & average volume(m3) °F (°F +40) / 1.8 - 40 = °C
∆P = pressure difference = P - Pb, psia (kPa) °C (°C +40) x 1.8 - 40 = °F
γ API = (141.5/γo)−131.5 psi x 6.894757 = kPa
3
= stock-tank oil gravity, ºAPI (g/cm ) °R / 1.8 =K
x&x = property & average property scf/bbl x 0.1801175 = std m3/m3
γg = average gas relative density (air = 1) Appendix - Undersaturated Isothermal Oil
γo = stock-tank oil relative density (water = 1) Compressibility Models and Correlations

γ ob ,gb = bubblepoint oil relative density(water= 1) Vasquez and Beggs4

n = number of data points
co = ( a1 + a2 Rs + a3 T + a4 γ g + a5γ api ) / P (A1)
Acknowledgement where
The author is grateful to the Department of Petroleum
Engineering at the King Fahd University of Petroleum and Original Model constants
Minerals, Dhahran, Saudi Arabia, for the facilities Coefficient correlation for Middle East
utilized to perform the present work and for coefficients data
their support. a1 -14.33x10-3 2.3386 x10-3
a2 50x10-6 36.9769 x10-6
References a3 0.172x10-3 55.3945 x10-6
1. Standing, M. B., Volumetric and Phase Behavior of Oil
Field Hydrocarbon Systems; 9th Printing, SPE, Dallas, a4 -11.80x10-3 - 8.1716 x10-3
Texas, 1981. a5 0.1261x10-3 58.2514 x10-6
2. Calhoun, J.C., Jr.: Fundamentals of Reservoir Engineering,
U. of Oklahoma Press, Norman, Oklahoma, 35, ( 1947). Petrosky and Farshad5
3. Trube, A.S.: “Compressibility of Undersaturated
Hydrocarbon Reservoir Fluids,” Trans., AIME (1957) 210,
pp 341-44. c o = a 1R sa 2 γ ag3 γ aapi4 T a 5 P a 6 (A2)
4. Vasquez, M.E. and Beggs, H.D.: ‘Correlations for Fluid where
Physical Property Prediction,: JPT (June 1980) 968-970,.
5. Petrosky, G.E. Jr. and Farshad, F.F.: “Pressure-Volume-
Original Model constants
Temperature Correlations for Gulf of Mexico,” paper SPE
26644, presented at the 1993 SPE Annual Technical Coefficient correlation for Middle East
Conference and Exhibition, Houston, Oct. 3-6. coefficients data
6. Mahmood, M. M. and Al-Marhoun, M. A.: “Evaluation of a1 0.1705x10-6 0.097856 x10-6
empirically derived PVT properties for Pakistani crude a2 0.69357 0.1998314
oils,” J. Pet. Sci. & Engg. 16 (1996) 275. a3 0.1885 -0.1936419
7. Dallag, M.M.: Reservoir Fluid Properties of Yemeni Crude a4 0.3272 0.5778032
Oils, MS Thesis, King Fahd University of Petroleum and a5 0.6729 0.5293904
Minerals. (1997). a6 -0.5906 -0.1633186
8. Vasquez, M.E. : Correlations for fluid physical property
prediction. MS Thesis, U. of Tulsa, Oklahoma. (1976).
This Study
ln co = a1 + a 2 / γ ob + a3 ( P − Pb ) / γ 3 ob + a 4 /(T + 460) (A3)
where

Model constants for

Coefficient Middle East data
a1 -14.1042
a2 2.7314
a3 -56.0605 x10-6
a4 -580.8778
6 SPE 81432

80 80

70 Vas-corr 70 Vas-corr

Average absolute relative error

Pet-corr
Average absolute relative error

Pet-corr 60
60 Vas-model
Vas-model
50 Pet-model 50 Pet-model
This study This study
40 40

30 30

20 20

10
10
0
0
T< 95 95<T<115 115<T<135 135<T<155 155<T<175 T>175
API<20 20<API<25 25<API<30 30<API<35 35<API<40 API>40
(494) (751) (471) (330) (615) (751)
(60) (312) (1066) (1260) (535) (179)
Ranges of oil temperature
Ranges of oil API gravity

Fig. 1: Errors of Co grouped by oil API gravity. Fig. 2: Errors of Co grouped by oil temperature.

100 100
90 90
Vas-corr
Average absolute relative error

Vas-corr

Average absolute relative error

80 Pet-corr 80
Pet-corr
70 Vas-model 70 Vas-model
Pet-model
60 60 Pet-model
This study
50 This study
50
40 40

30 30
20
20
10
10
0
0 DP<300 300<DP<700 700<DP<1100 1100 <DP<1500 1500 <DP<1900 DP>1900
gb<.73 .73<gb<.76 .76<gb<.79 .79<gb<.82 .82<gb<.85 gb>.85 (546) (549) (534 ) (513) (416) (854)

(751) (558) (492) (340) (637) (634) Ranges of pressure difference(P-Pb)

Ranges of bubblepoint oil relative density

Fig. 3: Errors of Co grouped by bubblepoint oil relative

Fig. 4: Errors of Co grouped by pressure difference.
density.
100
30
90 Vas-corr
Average absolute relative error

80 Pet-corr
Predicted Co in 10E 6/psi

70 Vas-model
Pet-model 20
-

60
This study
50
40
30 10
20
10
0
0
Co<5 5<Co<6.5 6.5<Co<8 8<Co<9.5 9.5<Co<11 Co>11
0 10 20 30
(194) (834) (765) (522) (416) (681)

Ranges of Co in10E-6 1/psi Measured Co in 10E-6/psi

Fig. 5: Errors of Co grouped by its experimental values. Fig. 6: Cross Plot of Vasquez & Beggs Correlation.
SPE 81432 7

30 30
Predicted Co in 10E 6/psi

Predicted Co in 10E 6/psi

20 20
-

-
10 10

0 0
0 10 20 30 0 10 20 30
Measured Co in 10E-6/psi Measured Co in 10E-6/psi

Fig. 7: Cross Plot of Petrosky & Farshad Correlation. Fig. 8: Cross Plot of Vasquez & Beggs Model.

30 30
Predicted Co in 10E 6/psi

Predicted Co in 10E 6/psi

20 20
-

-
10 10

0 0
0 10 20 30 0 10 20 30
Measured Co in 10E-6/psi Measured Co in 10E-6/psi

Fig. 9: Cross Plot of Petrosky & Farshad Model. Fig. 10: Cross Plot of this study Model.