SPE 6719

SPE
Society of Petroleum Engineers of AIME

CORRELATIONS FOR FLUID

PHYSICAL PROPERTY PREDICTION
by Mi 1ton Vazquez, Member SPE-AH~E,
INTEVEP, and H.D. Beggs, Member
SPE-AIME, U. of Tulsa

Copyright 1977, American Institute of Mining, Metallurgical, and Petroleum Engineers, Inc.
This paper was presented at the 52nd Annual Fall Technical Conference and Exhibition of the Society of Petroleum Engineers of AIME, held in Denver, Colorado, Oct. 9-12, 1977. The material is subject to
correction by the author. Permission to copy is restricted to an abstract of not more than 300 words. Write: 6200 N. Central Expy, Dallas, Texas 75206.

ABSTRACT resistance of the fluid to flow at a given pressure

and temperature.
Empirical correlations for approximating phase
behavior and flow properties for hydrocarbon gas-oil Increasing emphasis is being placed on the
systems are used in many procedures for design of necessity of having accurate estimates of physical
petroleum production and reservoir engineering properties in advance of a PVT laboratory analysis of
operations. In many cases, the accuracy of a complex reservoir fluids, and therefore, these parameters
design is seriously reduced by relying on outdated, must be calculated by using empirical correlations
inaccurate fluid property correlations for key that relate readily measurable parameters in the
parameters. field such as pressure, temperature, gas specific
gravi ty, and oil API gravity, to the unknown
The scope of the proposed research consisted of physical properties. The purpose of this study is
developing new and improved correlations for the to ~tilize a large base of measured data in the
fluid properties: solution gas-oil ratio, fluid development of improved correlations to replace those
formation volume factor (below and above bubble commonly in use today developed many years ago from
point), and viscosity of oil (above bubble point), al limited data.
functions of parameters measured in the field such as
crude oil API gravity, gas specific gravity, One of the first studies aimed at prediction of
reservoir temperature, and pressure. formation volume factor was performed by Katz 12 in
1942. He prepared a correlation from data on
The correlations were developed using data from Mid-Continent crudes. The procedure for calculating
laboratory fluid property analyses. Multiple formation volume factors using Katz's empirical
regression analysis techniques were used to process curves is rigorous due to the limitations imposed on
and correlate the data. Statistical results of the the data used in drawing the curves and the lack of
newly developed correlations are presented. gas gravity as a parameter.
INTROIUCTION In 1947, Standing 16 presented the results of a
laboratory study of the PVT behavior of 22 different
The evaluation of physical properties of crude oil-natural gas mixtures from California oil
hydrocarbons is one of the main concerns in the fields. He developed correlations for bubble-point
handling of the different stages of oilfield pressures and oil formation volume factors as
operations such as in the determination of pressure empirical f~ctions of solution gas-oil ratio, gas
gradients occurring in multiphase fluid flow gravity, oil gravity, pressure, and temperature
situations, the design of surface operation derived from 105 experimental data points. The
facilities, and in general, in the design of every correlation for dissolved gas has been found to be
piece of equipment involved in the production and inaccurate for high-gravity crude oils, but it is
transportation of crude oils. Some of these still one of the most widely used methods. His
properties are: solution gas-oil ratio, oil forma- correlation for oil formation volume factor, although
tion volume factor, and oil viscosity. The solution developed from limited data, is the method most
gas-oil ratio is the amount gas in stru1dard cubic widely used in the petroleum industry today.
feet dissolved at a certain pressure and temperature
per barrel of residual oil. The oil formation volume Borden and Rzasa4 in 1950 presented correlations
factor is the ratio of the amount of reservoir oil at of bottom-hole sample data for bubble-point pressure
a given pressure and temperature that would yield one and oil formation volume factor predictions. Use of
barrel of oil at standard conditions (14.7 psia and the correlations is somewhat restricted by the
60 OF). Viscosity, in general, is a measure of requirement of using the gas gravity calculation
. References and illustrations at end of paper.
2 CORRELATIONS FOR FLUID PHYSICAL PROPERTY PREDICTIONS
instead of using separator gas gravity. This
requires a chemical analysis of the gas-oil system,
which is impractical.
Lasater 13 in 1958 presented a correlation of
the bubble-point pressure for black oil systems
developed using standard physical-chemical equations
of solutions. The correlation is based on 158
experimentally measured bubble-point pressures of
137 independent systems. Lasater's correlation has
been found to be more accurate than Standing's
correlation for high-gravity crude oil systems, and
Chierici et al.7 suggest using Standing's correlation
for crudes of API gravity less than 15 and the
Lasater correlation for lighter crudes.
10
In 1972, Cronquist developed empirical rela-
tions for gas evolution and oil formation volume
factor 11sing differential liberation data from PVT
analyses of a large number of Gulf Coast crude oils.
Application of the method is restricted because it
requires knowledge of dissolved gas or formation
volume factor at one pressure in order to predict
the values at other pressures.
Extensive research has been done in the deter-
mination of crude oil viscosities below the bubble-
point. In one of the early works found in the
literature, Bea1 2 in 1945 presented correlations for
dead oil viscosity as a function of API gravity and
temperature; viscosity of saturated oil as a function
of dissolved gas and dead oil viscosity; and air and
water viscosities as functions of pressure and
temperature. He also developed a correlation for
viscosity of undersaturated crude oils.
The Beal correlation for gas-free or dead oil
viscosity is used widely throughout the oil industry
and is considered to be fairly accurate. One of the
limitations found in his presentation is the lack
of analytical expressions to evaluate viscosities of
crude oil systems below and above the bubble point.
6 extrapolating the correlation to be used to crude
In 1959, Chew and Connally extended Beal's
correlation to cover a wider range of dissolved gas-
oil ratios and dead oil viscosities. The correlation
was presented in the form of an equation and also in
a graphical form.
In 195~, Trube
18
presented a correlation aimed
at determining the isothermal coefficient of expan-
sion of an undersaturated hydrocarbon that may be
contained in a particular reservoir. This quantity
commonly is used in the determination of the' formation
volume factor of undersaturated fluids. The use of
these charts is limited to high-gravity mixtures due
to the difficulty found in obtaining values of
critical pressure and temperature for fractions of
C'7 compounds.
A method for predicting reservoir fluid
viscosity as a function of pressure, temperature, and
composition was presented by Lohrenz et al. 5 in 1964.
This correlation is applicable when compositional
material balance is available. Therefore, it would
have limited applications in the field.
Little and Kennedy14 in 1966 presented a corre-
lation pertaining to the prediction of viscosity of
complex hydrocarbon systems from pressure, tempera-
ture, and composition. Although the correlation was
found reliable for both liquid and gases, it requires
SPE 6719
the compositional analysis of the mixtures.
Beggs and Robinson3 in 1975 presented a corre-
lation to estimate the viscosity of crude oil systems
based on PVT analysis of reservoir fluid samples.
They developed a correlation for the dead oil
viscosity as a function of temperature and oil
specific gravity. An expression was given to calcu-
late live oil viscosity. One advantage found when
using this correlation is that it neglects the
dependence of oil viscosity on composition.
Houpeurt and Thellie~ll in 1976 developed a
procedure for predicting the viscosities of hydro-
carbon liquid phases from their composition. The
procedure is similar to that used by Standing and
Katz 17 for determining the density of reservoir oils.
The technique was checked experimentally using two
mixtures, and accurate results were obtained,
although a limited number of heavy components were
investigated.
The literature review revealed that the most
widely used correlation for dead oil viscosity was
developed in 1946 by Beal, and for live oil
viscosity in 1959 by Chew and Connally. Correlations
developed in 1947 by Standing and in 1958 by Lasater
are the most commonly used for predicting dissolved
gas and oil formation volume factors. The main
difficulty found in Standing's and Lasater's
correlations is that neither of them specifies the
nature of the gas specific gravity involved, whether
it is from the stock tank or from the first-stage
separator. Also, if it is from the separator, they
did not mention at what pressure.
Other limiting factors commonly found through-
out the research are the lack of mathematical expres-
sions for ease of handling of computations and the
use of data from specific geographical regions in
the development of correlations. Hence, a bias
factor is induced in the results obtained when
oils of different characteristics. The correlations
involving material-balance computation for crude oils
are regarded as impractical for field calculation
because of the requirement for costly compositional
analysis.
DATA ACQUISITION AND PREPARATION
Sampling Technigues
1
There are essentially three sampling techniques
for obtaining reservoir fluid samples for analysis
of pressure, volume, and temperature (BVT) relations.
These are commonly known as (1) bottom-hole sampling,
(2) recombination sampling, and (3) split-stream
sampling. A brief description of the techniques is
as follows.
1. Bottom-hole sampling. Steps toward well
preparation must be followed prior to the perform-
ance of sampling routines. The procedure for
preparing the well indicates that it should have been
producing with a stabilized gas-oil ratio. It also
suggests that the well be as new as possible so as to
minimize free gas saturation. A tubing pressure sur-
vey is conducted to locate the gas-oil and water-oil
interfaces, and a special sampling bomb is run on a
wire line. This bomb is activated at the surface so
as to retrieve a bottom-hole fluid sample under
SPE 6719 MILTON VASQUEZ AND H. DALE BEGGS
pressure. The sample collected should be representa-
tive of the reservoir fluid.
2. For the recombination-sampling-technique,
the reservoir fluids are collected at the surface.
Samples of separator oil and separator gas are
collected and these samples are recombined in the
laboratory in the proper proportions, as determined
by production characteristics measured at the surface
during sampling operations. The recombination method
of sampling is as good as the bottom-hole-sampling
technique for reservoirs where the flowing pressure
exceeds the bubble-point pressure of the reservoir
fluid.
3. The split-stream-sampling method is
primarily used in sampling of gas condensate wells.
A small-diameter tube is inserted in the middle of
the flow stream. Part of the flow is diverted
through this tube into either an auxiliary separator
or sampling bottles. In most cases, this sample
is obtained by inserting the tube in tubing to a
depth of 8 or 10 ft below the surface wellhead
connections or in the flow stream just upstream of thE
separator. The split-stream method of sampling loses
its accuracy with high-liquid-content fluids. It is
difficult to insure the proper entry of gas and
liquids into the sampling tube for high-flowing
liquid-gas ratios.
Once the samples have been collected, they are
shipped to a laboratory for complete analysis. The
type of laboratory analysis is dependent upon the
type of reservoir and the information desired.
A typical PVT laboratory analysis of reservoir
fluid samples is presented in Table A-1 of Appendix A.
The volumes given in Column 2 are the results of the
gas flash liberation process. In the flash libera-
tion process, all of the gas evolved during a
reduction in pressure remains in contact and in
equilibrium with the liquid phase from which it is
liberated. Below the bubble-point pressure of 3,891
psig, the volumes include the volume of liberated
gas and are therefore two-phase volume factors.
Since the stock-tank oil remaining at atmospheric
pressure depends upon the pressure, temperature, and
separation stages by which the gas is liberated at
lower pressures, the volumes are reported relative
to the volume at bubble-point pressure, Vsat '
Column 3 records the results of a viscosity test run
at reservoir pressure and temperature of the gas
flash liberation process. In Columns 4, 5, and 6 are
the results of a differential process. In the
differential process the gas evolved during a pres-
sure reduction is removed from contact with the
liquid phase as rapidly as it is liberated. Columns
4 and 5 indicate the gas-oil ratio liberated and in
solution per barrel of residual oil at every
pressure reduction below the bubble-point conditions.
Also reported is the differential formation volume
factor. In order to relate the reservoir volumes to
the stock-tank oil volumes, additional tests are
performed on other samples using small-scale
separators that are operated in the range of pres-
sures and temperatures used in the field separation
of the gas and oil.
Table A-2 of Appendix A shows the results of
five tests at separator pressures of 60, 300, 750,
and 1,000 psig and a separator temperature of 90 OF.
At 60 psig and 90 OF, the tests indicate that 509 scf
3
are liberated at the separator and 12 scf in the
stock tank, or a total of 521 scf/STB by two-stage
separation, if the stock tank is called a stage.
Then the initial solution gas-oil ratio, Rst, is 521
scf/STB. The tests also show that under these separ-
ation conditions 1.252 bbl of fluid at the bubble
point yields 1.00 STB of oil. The stock-tank gravity
of residual oil is reported as well as the specific
gravity of the flashed gas.
Preparation of Fluid Analysis PVT Data for
Use in the Development of Correlations
Flash separation data are used in conjunction
with the values of the flash and differential process
to calculate the fluid properties needed for reser-
voir calculations as well as to provide a data bank
for further use in the development of correlations.
This manipulation involves the computation of
combination formation volume factors and gas-oil
ratios. These parameters are calculated assuming
that differential liberation occurs in the reservoir
and that flash liberation occurs between the reser-
voir and the stock tank. In order to calculate the
combination fluid-analysis properties from standard
analysis data, certain assumptions are required. 1
General Assumptions
1. Standard cubic feet of gas in solution per
barrel of bubble-point oil is defined by the flash
liberation test to separator pressure and tempera-
ture.
2. The standard cubic feet of gas liberated
per barrel of bubble-point liquid is defined by a
differential liberation process at reservoir con-
ditions.
3. The standard cubic feet of gas remaining in
solution at reservoir conditions that will be
liberated upon producing that liquid to the separator
by a flash liberation process is the difference
between the original gas in solution and the differ-
entially liberated gas corrected for the reservoir
shrinkage of the fluid.
R R (1)
s sp
where Rs gas in solution per stock-tank barrel
of oil, scf/STB .
total gas liberated at the separator
per stock-tank barrel of oil by
flashing bubble-point oil, scf/STB
the standard cubic feet of gas liber-
ated by a differential process per
barrel of residual oil at standard
conditions, scf/STB
bubble-point oil required to yield 1
bbl of stock-tank oil when flashed
through the separator to stock-tank
conditions, bbl/STB
bubble-point oil required to yield 1
bbl of stock-tank oil when differen-
tially liberated to stock-tank con-
ditions, bbl/STB
Fig. 1 is a plot of pressure in psig vs solution gas-
oil ratio of four fluid samples after applying
Assumption 3.
4 CORRELATI:ONS FOR FLUl:D PHYS,l.GAL PREDICTIONS
4. The formation volume factor of the bubble-
point liquid is determined by the flash liberation
process to separator conditions and then to the
stock-tank.
5. The relationship between the flash and
differential liberation processes is assumed to be
constant at any reservoir pressure. The combination
flash-differential formation volume factor can be
expressed as
B
B
ofb
(2)
o Bodb Bod ' included in Table 2. The laboratory data have been
where Bod = oil at reservoir pressure p required to
yield 1 bbl of stock-tank oil when
differentially liberated to stock-tank
conditions, bbl/STB
Bo oil at reservoir pressure p required to
yield 1 bbl of stock-tank oil when
flashed through the separator, bbl/STB;
the term often referred to as simply
the flash formation volume factor
6. The oil viscosity above bubble-point pres-
sure, Pb' is defined by the flash liberation test.
A summary of the ranges of data obtained from
these data preparation procedures is shown in Table
1.
DEVELOPMENT OF CORRELATIONS
In the development of the correlations, multiple
regression analysis was applied using reservoir pres-
sure and temperature, gas specific gravity at 100-
psig separator pressure and API gravity of stock-tank
oil as independent variables for each of the proper-
ties. The first step toward the definition of the
mathematical model to use in the regression analysis
was a study of the nature of the relationship between
the dependent variables and the independent variable
as the coefficient of correlation in the covariance
matrix.
Liberated Gas Specific Gravity Correlation
Introdllction
This study arose from a need for a standardized
surface separator condition for further data manipula
tion in the development of correlations to evaluate
crude oil physical properties. This correlation is
an attempt to estimate the effect of varying field
separator pressure in terms of a change in the
liberated gas gravity from any given separator
pressure to a separator operated at 100 psig.
To develop correlations for predicting the
physical properties of crude oils requires knowledge
of liberated gas specific gravity that would result
from a single-stage separation at 100-psig separator
pressure. Ordinarily, field separator pressures may
range from 45 to 500 psig at various temperatures.
The effect of varying separator pressures
already has been studied by several investigators in
order to provide information on variations in physica
properties of produced reservoir oils. 8 To ac-
complish this, flash-gas liberation experiments
SPE 6719
conducted in the laboratory on portions of the sub-
surface oil samples were used. The study showed that
variations in separator operating conditions will
affect production gas-oil ratios, stock-tank oil
gravities, gravity of produced gas, and relative oil
volume.factors.
Procedure
The current study has been carried out on
separator tests (flash data) of 27 reservoir fluid
samples. A description of the data involved is
reported for each fluid sampl~ at a constant tempera-
ture between stages of separation.
Fig. 2 shows a typical plot of separation pres-
sure vs oil shrinkage. The figure demonstrates that
each crude oil has an optimum operating separator
pressure at which oil shrinkage is minimum (close to
unity) and oil recovered per barrel of reservoir oil
in place is maximum.
On this basis, optimum separator pressures have
been found by means of flash data separator tests.
The variations of optimum separator pressures found
led to the selection of 100 psig as the median and
practical value for separator pressure operation
within the ranges of the data involved and the field
operations.
In the development of the correlation to predict
the change of the specific gas gravity, it was found
that for a given sample the change in gas gravity
determined as the ratio of the gas gravity at 100
psig to the gas gravity measured at any other given
separator pressure will result in a fairly straight
line when plotted against the logarithm of the ratio
of the absolute separator pressures involved.
Mathematically,
'Y (114.7))
( gyg(P) = c 1 + c 2 log (11f.7) , .. (3)
where y g (114.7) = gas specific gravity that would
result from a separator pressure
of 100 psig (114.7 psia) and
separator temperature T, OF
gas specific gravity measured at
the given separator pressure, p
Regression Analysis
A linear regression analysis was performed on
the data choosing Eq. 3 as the mathematical model.
The regression analysis revealed that the intercept
of the equation equals 1, which is in agreement with
what is expected where p is equal to 114.7 psia.
The product of the stock-tank oil gravity (API
degrees) times the separator temperature in degrees
Fahrenheit influences the slope of the line due to
the fact that the change in gas gravity is propor-
tional to the degree of volatility of the crude oil
and the separatton temperature. The regression
equation is
Yg (114.7)) =
( Yg(p) l.
+ 0.5912 (API) (T) log (llt 7) 10. -4 (4)
SPE 6719 MILTON VASQUEZ AND H. DALE BEGGS 5

Yg (114.7) =
(
Y (114.7))
gyg(P) (yg(p)) (5)
log C:) = 10g(c 1 ) + c
2
10g(p) + c
3
~~I). : (8)
API stock-tank API oil-gravity
T separator temperature, of Eq. 8 now can be used in the linear regression
analysis in order to evaluate the C constants.
Table 3 summarizes the statistical results of
the correlation. Fig. 3 is a plot of the observed Regression Equations
vs the calculated gas specific gravity using Eqs. 4
and 5. The form of Eq. 8 was found to be the one that
fitted the data most accurately on the basis of the
Solution Gas-Oil Ratio Correlation least standard error of the estimated value and
average error when compared with the correlations of
Introduction Standing and Lasater. The following are the final
correlations for each set of data.
The basic form in which the variables were
related to the independent variable solution gas-oil Correlation for API~30:
ratio (Rs) was found by observation of the degree
of linearity evaluated as the coefficient of correla- ( Yg ) ( P )1.0937 API )
tion in the covariance matrix. This previous R 10. 1 1.172 ( T+460. w
s 27.64
analysis suggested that (9)

Rs = (p, lIT, Yg , API) (6) Correlation for API>30:

The complexity of the determination of the

function relating the independent variables in order R
(
Yg
) ( )1.187
P 10.393 (T-t!!~. )
to predict the dependent variable Rs necessitated the s 56.06 10. '. (10)
~tudy of the data i~ terms ~f the degree of volatil~
lty. It was found ln the Ilterature~ that experience Table 4 gives a summary of the statistical
indicates that low-volatility conditions existed 1iJhen analyses of the developed correlations compared with
the stock-tank gravity was below 35 oAPI, the solu- the standing and Lasater work. Because of the
tion gas-oil ratio was less than 500 scf/STB, and the uncertainty found by extrapolating values on the
reservoir temperature was below 150 OF. It was curves provided by Lasater, fewer points were
pointed out that these were approximate limits. possible to calculate when testing his correlation
against the present study. Fig. 4 is a plot of the
Based on the previous discussion, the data were observed values of Rs vs the calculated values when
divided on the basis of API gravity to provide two using Eqs. 9 and 10, respectively.
sets of data: one set compiling all the crude oil
samples for which API gravity was less than or equal Formation Volume Factor Correlation
to 30 and another set for the crude oils above 30 (Below Bubble Point)
API. This division provided the following average
values. ' Introduction
API ~30 API,>30
The data preparation of formation volume factors
R 245. 567. was accomplished by means of Assumptions 4 and 5
s given in an earlier section. The basic dependent-
B 1.153 1.340
independent variable relationships were drawn through
0 the study of the correlation matrix evaluated in the
.81 .81
two sets of data. It was found that
Yg
API
for API-::;30,
26. 40.
Bo = (p, Rs' T, Yg , API) , (11)
T 162. 180.
Another indication of the degree of volatility and for API>30,
of the oil is the proximity of the formation volume
factor to 1. Under these conditions, it is likely B
o
f(p, Rs' T, l/y g , API) (12)
that only slight differences will exist between the
flash and differential liberation data. Procedure

Procedure It also was found that the formation volume

factor is more highly related to the solution gas-
The mathematical model to use as a means of oil ratio than to the pressure. This led to the
defining the regression equation was definition of Bo in terms of four independent vari-
ables:
Bo = f(Rs' T, Yg , API) (13)
(7)
After testing several mathematical models, the
equation selected was:
Taking the common logarithm of the expression, the
equation is linearized in the following form:
6 CORRELATIONS FOR LFUID PHvSICAL PKUI-'I:K I Y PREDI CTIONS SPF f\71q

-c =1 (dV) T
V .dp . (18)

Eq. 19 is a numerical approximation of Eq. 18

for a finite pressure change:
(14)
c =
VI - V2
--:-=----=--:-
Va (P2 - PI)
, . . . . .'. . . . . . (19)
Regression Equations
in which P2>Pl and V1>V2 and V = (Vi + V2)/2. The
Eq. 14 was the model that most closely repro- instantaneous compressibility ~ay be estimated also
duced the data on the basis of having the highest from the slope of a curve on which pressure is
coefficient of multiple regression, least standard plotted vs natural log V. Fig. 6 is a plot of pres-
error of estimated value, and least average error. sure in psig vs in Bo' The constant relationship
It also meets the requirements for the formation observed between p and In Bo is just an apparent
volume factor of 1. at 60 OF and atmospheric pres- relation, and it has been estimated by
sure (Rs = 0.). The equations found for the sets of
data are: In GOb )o
for APk;30, C = (P _ p) ( 20 )
-4 b
B 1. + 4.677 (R ) 10. Fig. 7 is a plot of pressure (psig) vs the
o s observed compressibility factors of some of the
4 samples plotted on Fig. 6.
+ .1751 (T-60.) (A~~) 10.-
Procedure
A study of Fig. 7 indicates clearly the effect
- 1. 8106 (Rs) (T-6V.) (A~~) 10. -8 , (15) of solution gas-oil ratio and API gravity of crude
oil on the observed compressibility factor. The
correlation matrix evaluated on the data showed that
and for API>30, c = f(l!p, K) (21)
-4
B
o
= 1. + 4.67 (R ) 10.
s
K= feR ,
s
T, API~ l!y g ) (22)
-4
+ .11 (T-60.) (A~:) 10. K is a constant related to the initial solution gas-
oil ratio, reservoir temperature, API stock-tank oil,
and 100-psig separator gas specific gravity. The
-8
(A~:)
best mathematical model found that combined these
+ .1337 (R ) (T-60.)
s
10. , (16) four parameters was a linear equation of the form

K = a1 + a
2
(RJ + a3 (T) + a
4
(API) + as (Y ) ,
g
Fig. 5 is a plot of the observed BQ values vs
the calculated Bo's using Eqs. 14 and 1). (23)
Table 5 gives a summary of the statistical and the best estimating equation for the variable C
analyses of the developed correlations, as well as is
Standing's work, tested on the basis of the same
data.
c = (~) (24)
Formation Volume Factor
(Above Bubble Point) Regression Equation
Introduction After performing the analysis of regression,
the equation arrived at was
The determination of the formation volume factor
of crude oils at pressures above the bubble-point '""-1433. + 5.(R) + 17.2 (T

J
s
conditions (undersaturated region of the phase
diagram) implies the estimation of the liquid com- - 1180. (Yg ) + 12.61 (API)
pressibility or isothermal coefficient of expansion c = 5
(25)
c to be used in the following equation: (p) 10.

c(Pb-p) Fig. 8 is a plot of observed formation volume

Bo = Bob e ( 17) factor vs calculated formation volume factor when
using the correlating equations, and Table 6 gives
The isothermal coefficient of expansion or com- a summary of the statistical analyses of the correla-
pressibility ofa substance is defined as 18 tion.
SPE 6719 MILTON VASOUEZ AllD H DALE
Introduction
Viscosity of a fluid is a measure of its
resistance to internal deformation or shear. In the
study of the movement of fluids in pipelines, con-
solidated rocks, and other porous media, viscosity
plays an important role in influencing the ability
of the fluid to flow. Darcy's law states that the
velocity of a homogeneous fluid in a porous medium
is proportional to the pressure gradient and
inversely proportional to the fluid viscosity,9 or
v =- ~1-1 (~)
ds'
(26
The fluid viscosity" !-L, is expressed in centipoises,
and the pressure gradient, dp/ds, is in atmospheres
per centimeter, taken in the same direction as v.
The proportionality constant k is the permeability
of the rock expressed in darcy units. The negative
sign indicates that if the flow is taken as positive
in the positive s-direction, then the pressure
decreases in that direction so that the slope dp/ds
is negative. Viscosity is also involved in several
dimensionless parameters to calculate flow regimes,
friction factors, and pressure gradients in multi-
phase fluid flow problems.
For gas-saturated crude oils, it has been found
in previous studies 2 that:
1. Most crude oils follow a general pattern of
decline in oil viscosity as a function of dissolved
gas relative to the viscosity of a gas-free crude
oil at reser~oir temperature; that is, viscosity of
gas-free crude oils will be decreased by dissolving
gas until saturation is reached.
2. The greater reduction in viscosity occurs at
low values of gas saturation in low-gravity crude
oils, and this reduction decreases as the gravity
of the crude and pressure increase.
3. The amount of gas dissolved in crude oil for
a particular oil gravity has a more important bearing
on gas-saturated crude oil viscosity than has pres-
sure.
Procedure
The study performed on viscosity of under-
saturated crude oils (above bubble point) indicates
that:
1. For a given saturated crude oil at bubble-
point conditions, viscosity increases proportionally
to pressure.
2. The rate of increase of viscosity is a
function of pressure.
~ = (p) . (27) one available in equation form, as well as the corre-
b.p
3. Since no gas is going into solution and the
composition of crude oil is constant, pressure
becomes of primary importance when estimating crude
oil viscosity in the undersaturated region.
The foregoing discussion has been verified by
the analysis of the correlation matrix evaluated on
the data. The following relationships were found to
be appropriate for the regression analysis.
RJ:"~~S 7
1-10 \.L
ob
(L)m ............
Pb .
(28)
m f(p)
, ..............
where!-Lo viscosity (cp) of undersaturated oil
!-LOb viscosity of bubble-point oil
p reservoir pressure above bubble point
Pb = bubble-point pressure of the crude oil
The best form of Eq. 29 found was
log(m)
1 2
=
c + c (p) + c 3 log(p). (30)
Regression Equation
The basis for selecting Eq. 30 as the regression
equation was the high degree of correlation shown by
p and log (p) in defining log (m). Again, a high
coefficient of multiple regression was observed as
well as a low standard error of the estimated value.
The regression equation has the form
-3
m 2.6 (p)1.187 (10.)(~.039 (p) 10. - 5.)
(31)
Fig. 9 is a plot of observed oil viscosity vs
calculated viscosity using Eqs. 31 and 28. Table 7
gives a summary of the statistical analyses of the
correlation.
CONCLUSIONS
1. In a particular reservoir, the solution
gas-oil ratio and other fluid properties were found
to be dependent upon the manner by which the gas is
liberated from the oil. Since the data used to
develop the correlations were the result of a flash-
differential liberation combination of PVT laboratory
data, it is concluded that the correlations will
pertain to the estimation of values for similar
processes when used in the field.
2. The behavior of crude oils is a function of
their composition. The degree of volatility of crude
oils governs the differences found between the flash
and differential processes. Since the API gravity
of oil is a function of oil composition, the separa-
tion of the data into two ranges (API530 and API>30)
took into account some of the effects due to oil
volatility.
3. The newly developed correlations are
definitely an improvement over the existing corre-
lations. The correlation to estimate the liquid
compressibility factor for the evaluation of forma-
tion volume factor of undersaturated oil is the only
lation to estimate viscosity of undersaturated oil.
R.ECOMMENDATIONS
Since the correlations have been drawn from
combination-type (flash-differential) PVT data on
reservoir fluid samples, it is advisable to test the
correlations using reservoir production data
collected from different types of reservoirs and
define the reservoir conditions that will be best
8 CORRELATIONS FOR FLUID PHYSICAL PROPERTY PREDICTIONS SPE 6719

reproduced by the correlations. Bottom Hole Sample Data," Trans., AIME (1950)
189, 345-348.
NOMEN"CLA'IURE 5. Bray, B. G., et al.: "Calculating Viscosities
of Reservoir Fluids From Their Compositions,"
API stock-tank oil gravity, API J. Pet. Tech. (Oct. 1964) 1171-1176.
B formation volume factor, bbl,STB 6. Chew, J. and Connally, C. H., Jr.: "A Viscosity
c compressibility factor, psig- 1 Correlation for Gas Saturated Crude Oils,"
f function Trans., AIME (1959) 216,.
p = pressure, psig Chierici, G. L., et al.: "Two-Phase Vertical
Rs = solution gas-oil ratio, scf/STB Flow in Oil Wells - Prediction of Pressure
T = temperature, of Drop," Trans., AIME (1959) 23.
/: :" = difference 8. Cook, A. B., et al.: "Change in Gas-Oil Ratios
~ viscosity, cp With Variations in Separator Pressures and
r = specific gravity Temperatures," Pet.~. (March 1954) No.3,
B-77-82.
Subscripts Craft, B. C. and Hawkins, M. F.: Petroleum
Reservoir Engineering, Prentice-Hall, Inc.,
b = bubble point Englewood Cliffs, N. J. (1959).
g = gas 10. Cronquist, C.: "Dimensionless PVT Behayior of
i = initial Gulf Coast Crude Oils," J. Pet. Tech.
o = oil (May 1973) 1-8.
od = differential liberated oil 11. Houpeurt, A. H. and Thelliez, M. B.: "Predict-
odb = bubble-point differential liberated oil ing the Viscosity of Hydrocarbon Liquid Phase
ofb = bubble-point flash liberated oil From Their Composition," J. Pet. Tech. (Feb.
sp = separator conditions 1976) 223-231.
st = standard conditions 12. Katz, D. L.: "Prediction of the Shrinkage of
Crude Oils," Drill. and Prod. Prac., API
ACKNOWLEWMENTS (1942) 137.
13. Lasater, J. A.: "Bubble Point Pressure Corre-
The author wishes to express his appreciation lation," Trans., AIME (1958) 213, 379-81.
to Core Laboratories, Inc., Dallas, Tex_, for furnish- 14. Little, J. E. and Kennedy, H. T.: "A Corre:-
ing the PVT laboratory data used in this study, to lation of the Viscosity of Hydrocarbon Systems
the University of Tulsa Fluid Flow Projects in with Pressure, Temperature, and Composition,"
providing computer time, and the Instituto Soc. Pet. Eng. J. (June 1968) 157-162.
Tecnol6gico Venezolano del Petr6leo (INTEVEP) for 15. Parsons, R.: Statistical Analysis: A
making the publication of this paper possible. DeCision-Making Approach, Harper & Row,
Publishers, Inc., New York (1974).
REFERENCES 16. Standing, M. B.: "A Pressure-Volume-
Temperature Correlation for Mixtures of
1. Amyx, J. W. et al.: Petroleum Reservoir Engi- California Oil and Gases," Drill. and Prod.
neering: Physical Properties, McGraw-Hill Book Prac., API (1947) 275-86.
Co., Inc., New York (1960). 17. Standing, M. B. and Katz, D. L.: "Density
2. Beal, C.: "The Viscosity of Air, Natural Gas, of Crude Oils Saturated With Natural Gases,"
Crude Oil and Its Associated Gases at Oil Field Trans., AIME (1942) ]1&, 159-165.
Temperature and Pressures," Trans., AIME (1946) 18. Trube, A. S.: "Compressibility of Under-
165, 94-112. saturated Hydrocarbon Reservoir Fluids,"
3. Beggs, H. D. and Robinson, J. R.: "Estimating Trans., AIME (1957) 215, 341-44.
the Viscosity of Crude Oil Systems," J. Pet. 19. W. J. Dixon, ed., BMD: Biomedical Computer
Tech. Forum (Sept. 1975) 1140. Programs, Los Angeles, Calif: School of
Borden, G. and Rzasa, M. J.: "Correlation of Medicine, U. of California (Jan. 1973).
 TABLE 1 - GENERAL INPUT DATA DESCRIPTION

CORRELATION TYPE DATA PRESSURE Rs Bo GAS-SPECIFIC OIL-GRAVITY VISCOSITY

POINTS psig SCF/STB bbl/STB GRAVITY API cp

SOLUTION GAS-OIL RATIO API~30 1141 0.-4557. O. -831. 1. 042-1. 545 511-1. 35.1 15.3-30
AND FORMATION VOLUHE
FACTOR (below bp) API>30 4863 0.-6040. 0.-2199. 1.028-2.226 .530-1. 259 30,6-59.5
FORl-'tA TION VOLUME 4486 126.-9500. 9.3-2199. 1.006-2.226 .511-1. 351 15.3-59.5
FACTOR (above bp)
VISCOSITY OF UNDER- 3593 126. -9500. 9.3-2199. .511-1.351 15.3-59.5 .117-148.
SATURATED OIL

TABLE 2 - SEPARATOR DATA DESCRIPTION

Separator Stock-Tank Specific Separator

Pressure Oil-Gravity Gravity of Temperature
psig API Produced Gas deg. F.

45. -550, 17. -45. .58-1.2 76. -150.

TABLE 3 - STATISTICAL ANALYSES OF CORRELATION

Data Points 124

Standard Deviation of Observed Value .092

Standard Deviation of Predicted Value .086

Standard Error of Estimate .032

Average Percent Error -.545

TABLE 4 - SUMMARY OF STATISTICAL ANALYSES OF Rs CORRELATION

AVERAGE STANDARD DEVIATION STANDARD

CORRELATIONS
DATA
POINTS
PERCENT
ERROR
OBSERVED
VALUE, SDO
I PREDICTED
VALUE, SDP
ERROR
ESTIMATE, SEE
bubble-eoint data
PRESENT STUDY 455 -2.5 359. 311. 138.
STANDING 455 11.1 359. 350. 147.
LASATER 437 14.9 357. 330. 182.

all data Eoints

PRESENT STUDY 5008 '-.7 279. 265. 96,
STANDING 5008 12.1 279. 296. 100.
LASATER 4190 15.8 282. 274. 122.
 TABLE ,5 - SUMMARY OF STATISTICAL ANALYSES OF Bo CORRELATION (BELOW'
BUBBLE-POINT)

CORRELATION
DATA
POINTS
AVERAGE
PERCENT
ERROR
OBSERVED l
STANDARD DEVIATION
PREDICTED
VALUE, SDO VALUE, SDP
STANDARD
ERROR
ESTIMATE, SEE
bubb1e-Eoint data
PRESENT STUDY 455 .4 .189 .185 .038
STANDING 455 -1.8 .189 .203 .042

all data Eoints

PRESENT STUDY 5008 4.7 .182 .187 .118
STANDING 5008 5.3 .182 .196 .120

TABLE 6- SUMMARY OF STATISTICAL ANALYSES OF Bo CORRELATION (ABOVE

BUBBLE-POINT)

CORRELATION
DATA
POINTS
AVERAGE
PERCENT
ERROR
OBSERVED l
STANDARD DEVIATION
PREDICTED
VALUE, SDO VALUE, SDP
STANDARD
ERROR
ESTIMATE, SEE
all data Eoints
PRESENT .STUDY 4036 -.284 .195 .191 .035

TABLE 7 - SUMMARY OF STATISTICAL ANALYSES OF ~o CORRELATION (ABOVE

BUBBLE-POINT)

AVERAGE STANDARD DEVIATION STANDARD

CORRELATION
DATA
POINTS
PERCENT
ERROR
OBSERVED I
PREDICTED
VALUE, SDO VALUE, SDP
ERROR
ESTIMATE, SEE
all data points
PRESENT STUDY 3143 -7.541 7.94 8.70 1.434
 1000
+ 1S3,F. .5 ~ 26 AFI
x 176 F. .6 ~ 30 AFI
x 195 F . 6 ~ 37 AFI
y 140 F . . 5 r,. 44 AFI
BOO

,...... SOD
1!:l
Cfl
---..
4-<
()
til
'-'
til
~ 400

200

o
o 1000 2000 3000 4000 5000
P (psig)
FIG, 1 - PRESSURE VS SOLUTION GAS-OIL RATIO FOR FOUR
RESERVOIR FLUID SAMPLES (AFTER AMYX J ET AL),

.85

+ 200 SCF V5TB. 42 AFI

.84 / --- --
V
/
<])
00
oj
.:.<
~
'M
~
.84
.c.::

/
Cfl
..-l
'.-1
0

.83
/

.83
.00
/ lS.67 3:3.33 50.00

Separator Pressure, psig

66.67 83.33 100.00

FIG, 2- SEPARATOR PRESSURE VS OIL SHRINKAGE,

 v
1.20 1600'~ __________~__________~________~r-______~~

v
-f.
1.10

1200.~ __________~__________~________~~~______~

v
1.00

-f.
.; """ -f.
.-i -f.
os
<J .90 'I-

~
bIl
>- -f.
-f.

.BO
+

.70 V
~
+

.SO ,00
.sa .70 .so .90 1.00 1.10 1.20 .00 400.00 . 600.00 1200.00 1600.00
Ygobs.
Rs (scf/STB)calc.
FIG. 3 - OBSERVED VS CALCULATED GAS SPECIFIC GRAVITY,
FIG, 4- OBSERVED VS CALCULATED SOLUTION GAS-OIL RATIO,

.e7
+ 340 S.

,
26 API
x 55<: S. 29 API
<l> 55<: S. 32 API
~ 600 S. 32 Ai='I
2.20 4' 226 S. 44 API
+ BELOW BU BLE FOINT
-f.
.24
"'-l '-
~~
1.90
-f. '"'
Q
.-1
.el
.p- o
~
;l ~
til .0
.-.,<J .g
~ .0

~ 1.60 Q)
:>
.-i 0 .1S
.0
e f;l
0
P'l III
.g
......
0
P'l
1.:30 '6'
.-I .15 ""-.
~
r--..
1.00
.~
1.00 1,:30 1.60 1.90 2.20
Bo (bbl/STB)obs.
~
5- ~
FIG, OBSERVED VS CALCULATED FORMATION VOLUME FACTOR,
.0S
o 1000 2000 3000
r-------.
4000 5000 6000

p (psig)

FIG. 6 - PRESSURE VS OBSERVED FORMATION VOLUME FACTOR

(ABOVE BUBBLE-POINT),
 1.20

1.07

.93

.....
I

.~
~

.80
'"I

~
M

Ul
.0
0
U

.67

.53

.40
2000 3000 4000 5000 6000

p (psig)

FIG. 7- PRESSURE VS OBSERVED COMPRESSIBILITY FACTOR C.

1.50 1.0a
+ ABOVE: BUSBlE: PO I NT + ABOVE: BUBBLE: PO NT
+

1.40 .80

U
..... 1.30 .GO
U
~ .....
t11

~
U

~
.0 ~
6 1.20 0 .40
:'I.
0
J'Q

1.10 .20

1.00 .00
1.00 1.10 1.20 1.30 1.40 1.50 .00 .20 .40 .60 :80 1.00
~o (cp)
Bo (bbl/STB) obs. obs.

FIG. 8 - OBSERVED VS CALCULATED FORMATION VOLUME FACTOR FIG. 9 - OBSERVED VS CALCULATED VISCOSITY OF UNDERSATURATED OIL FOR:
(ABOVE BUBBLE-POINT). .1 00 2. 1.
 TABLE A-I - RESERVOIR FLUID SAMPLE TABULATED FORM

Re se rvoir Fluid SAMPLE TABULAR DATA

PRESSURE. VOLUME
OIFF"ERENTIAL L.IBERATION @ 187 'F.
VISt:OSITY
RE.l ATION OF OIL.
PRESSURE @ 187 'F .. 'F ..
GAS/OIL RATIO GAS/OIL RATIO RELATIVE OIL.
@ 187 LICERATED IN SOLUTION VOLUME.

I
PSI GAUGE
REL.ATIVE VOL.UME OF'
OIL. AND GAS. V/VSAT.
CENTIPOISES
PER DARREL. OF
RESIDUAL. OIL.
PER BARREL. OF'
RESIDUAL. OIL.
I V/VR

6020 1. 53
6000 0.9836 1. 230
5610 1. 49
5500 0.9872 1. 235
5200 1. 45
5000 0.9910 1. 240
4810 1. 41
4500 0.9949 1. 245
4430 1. 37
4400 0.9957 1. 246
4300 0.9965 1. 247
4200 0.9974 1. 248
4100 0.9982 1.33 1. 249
4000 0.9990 1.250
3891 L 0000 1. 32 0 518 1. 251
3881 1. 0005
3859 1. 0015
3817 1. 0034
3749 1. 0068
3700 1. 32
3660 1.0118
3650 30 488 1. 241
3500 1. 35
3490 1. 0217
3360 69 449 1. 226
3213 1. 0416
3200 1. 42
3057 110 408 1.211
2855 1.0748
2800 1. 54
2661 161 357 1. 193
2473 1'. 1250
2400 L 68
2262 213 305 1. 174
2104 1.1978
2000 1. 85
1858 267 251 1. 154
1710 1.3197
1500 2.11
1437 322 196 1. 134
1305 1.5328
1045 372 146 1. 116
1000 2.46
953 1. 8810
715 2.3224
415 103 1. 100
704
550 2.8762
500 2.96
456 62 1. 086
395
385 3.8768
200 483 35 1.075
4.34 518 0 1. 055
0
@ 60 0 F. = 1. 000

0 0
Gravity of re!:llduat oil = 26.9 API @ 60 F.

v = Volume at given pressure

VSAT. = Volume at sa.turation pressure and the specified temperature.
VR = Residual oil volume at 14.7 PSI absolute and 60 0 F.
 TABLE A-2 - SEPARATOR TESTS OF RESERVOIR FLUID SAMPLE

SHRINKAGE FORMATION SPECIFIC

SEPARATOR SEPARATOR SEPARATOR STOCK TANK STOCK TANK VOLUME
FACTOR. GRAVITY OF
PRESSURE. GRAVITY. FACTOR.
TEMPERATURE, GAS/OIl. RATIO O .... S/OIL.. RA riO FLASHED GAS
Vrt/VS,,"T. VSAT./Vrt
PSI GAUGE
F.
API frY 60 F.
See Food'lote (1) ~~e F""I :-;"I~ \ 1) See Foot :-fote (2) See Foot N-otf!' (3)

60 90 509 12 26.9 0.7987 1. 252 0.6~8

300 90 451 58 27. 1 0.8052 1. 242 0.590

750 90 373 134 27.2 0.8050 1. 242 0.588

1000 90 340 168 27. 1 0.8049 1. 242 O. 534

(1) S~p.!\ralor ::md ~tot!k

tank Gas/Oil Ratio in cubic feet of ga~ @ 60 F. and 14.7 PSI absolute pel' barrel of
stock tank oil @ 60" !1'.
(2) Shrinicag-c Fador: VRlVuT. is barrels of stock tank oil @ 60 0 F. per barrel of saturated oil @ 3 Sq J
PSI ga!'ge and-13.1_'_0 F.
(3) Formntion 'tv' olume Fndor = 'JuT./VA ig barrels of saturated oil @ .3 8g 1 PSI gauge and 1R7 0 F. per
barrel or stock tank oil @ 60 F.