Journal o f Petroleum Science and Engineering, 5 (1990) 197-200 197

Elsevier Science Publishers B.V., Amsterdam - - Printed in The Netherlands

An improved temperature-viscosity correlation for crude oil

systems

Emmanuel O. Egbogah and Jack T. Ng*

American Hydrocarbon Technologies, Inc., Box 422, Station T, Calgary, Alta. T2H 2G9 (Canada)
(Received April 1, 1988; revised and accepted September 10, 1988 )

ABSTRACT

Egbogah, E.O. and Ng, J.T., 1990. An improved temperature-viscosity correlation for crude oil systems. J. Pet. Sci. Eng.,
4: 197-200.

In the application of Beggs and Robinson dead-oil viscosity-temperature empirical correlation to several hundred dead-
oil systems, we have found a consistently significant deviation between measured and calculated viscosity values. This
paper presents a modified Beggs and Robinson empirical viscosity correlation. The proposed new correlation has been
verified using a data base of about four hundred oil analysis reports (oil systems), and it shows a significantly better
correlation, with a standard deviation and average error of at least 12% and 55%, respectively, better than the existing
correlation of Beggs and Robinson. We have also expanded the applicability of this correlation further by the inclusion of
pour point temperature in the empirical equations. The new correlation also appears to apply to a wide range ofoil specific
gravities and pour point temperatures.

Introduction nany (1959) for live-oil systems. While Beal

correlated dead-oil viscosity as a function of
One of the most important physical proper- API gravity and temperature, Chew and Con-
ties of crude oils is viscosity, which is a critical nally presented a correlation for the effect of
parameter required in various aspects of pe- dissolved gas on oil viscosity. Other authors
troleum engineering analysis. A laboratory (Lohrenz et al., 1964; Little and Kennedy,
PVT analysis of a crude oil provides viscosity 1968; Houpeurt and Thelliez, 1974 ) have con-
information at reservoir temperature and sidered the effect of the chemical composition
pressure. However, difficulties arise when vis- of crude oil on its viscosity. More recently,
cosities are required at other temperatures. Beggs and Robinson, (1975) presented an
Therefore, if viscosity at a desired tempera- empirical viscosity correlation equation for
ture can be predicted by a less expensive labo- dead or gas-free crude oil as a function of API
ratory measurement technique, it will be ex- gravity and temperature. The correlation for
tremely useful in the reservoir performance dead-oil viscosity which was developed by
prediction, particularly for thermal recovery plotting t°log (T) versus l°logl°log (POD+ 1 )
processes. on Cartesian coordinates, enjoys extremely
Until recently the most widely used meth- wide application in the petroleum industry. We
ods for predicting oil viscosity were those of have examined the correlation equation pre-
Beal (1946) for dead-oil and Chew and Con- sented by Beggs and Robinson and found a sig-
nificant deviation between measured and cal-
*Now with Northern Telecom, Ottawa. culated viscosity values. This paper presents a

0920-4105/90/$03.50 © 1990 Elsevier Science Publishers B.V.

198 E.O. EGBOGAH AND J.T.NG

modified Beggs and Robinson empirical vis- Tp -- pour point temperature ( ° C)

cosity correlation which incorporates the pour y = specific gravity
point characterization of crude oils. T = temperature (°C)
Equation 2 is the result of previous analysis in
Pour point temperature that the selection of proper parameters and
combination of cross products was performed
In this study, a new parameter, pour point by plotting viscosity versus the above-men-
temperature, is introduced into the viscosity tioned parameters. For example, a plot of
correlation. Pour point temperature, (Tp), by l°logl°log (/too+ 1) versus l°log(T-Tp)
definition, is the lowest temperature at which showed a series of straight lines. A plot of these
the oil is observed to flow when cooled and ex- slopes against pour point temperatures showed
amined under conditions prescribed in ASTM a linear relationship.
D97. Pour point temperature of crude oil ap-
pears to be related to the paraffin content; the
Data used for regression analysis
pour point temperature increases with crude
oil paraffin content. An attempt has been made
to incorporate pour point temperature into the All viscosity data for the generation of
viscosity correlation together with specific regression equations are from the Reservoir
gravity and temperature. In doing so, we be- Fluids Analysis Laboratory of AGAT Engi-
lieve that important chemical compositional neering, Ltd. The total number of oil systems
aspects of crude oil has been reflected in the used is 394. API gravity range is from 5 to 58
viscosity correlation. including pour point temperature range of
- 50 ° to - 15 °C and temperature range of 15 °
Multiple regression models to 80 ° C.

In order to find a functional relationship be-

Results and discussion
tween viscosity of crude oil and various other
parameters, such as temperature, specific
gravity and pour point temperature, multiple- The empirical equation thus developed is
regression analysis is required. The regression given by:
model of Beggs and Robinson is:
l°logl°log (/tOO + 1 ) = -- 1.7095--0.0087917Tp +
101ogl01og(/tOo + 1 ) =a~ +a2~o +a3 l°log Tf 2.75237+ ( -- 1.2943+0.0033214Tp +
(1) 0.9581957)1°1og (T-Tp) (3)
where: The modified Beggs and Robinson equation is
Yo = oil gravity (°API) given by:
= Temperature ( ° F )
al,a2,a3 = constants
1°log l°log(flo D + 1 ) =
The equation of the regression model used in 1.8653--0.0250867o --0.56441 l°log (Tf)
this study is:
(4)
l°log lO1og(/tOD + 1 ) = a l +a2Tp+a3Y+
Table 1 shows statistical comparison of cor-
(a4 +as Tp +a67)l°log ( T - To) (2) relations. The error reported here is the rela-
where: tive error between the measured and calcu-
/tOD = viscosity of dead oil (cP) lated viscosity values. Standard deviation is
a~ to a6 = constants that of relative error from average percentage
AN IMPROVED TEMPERATURE-VISCOSITY CORRELATION FOR CRUDE OIL SYSTEMS 199

TABLE 1

Statistical comparison of correlations

API Average error Standard deviation

Range (%) of percent errors

Dead-oil data used to develop this correlation (394 cases) 5-58

Beggs and Robinson 61.00 67.40
This study:
Without pour point Temp. -5.13 55.51
With pour point Temp. -4.30 52.79

Dead-oil data collected from literature ( 12 cases) 5-58

Beggs and Robinson - 11.20 66.10
This study:
Without pour point temp. 5.85 54.63
With pour point temp.

error. Due to the wide range of viscosity values The equation with pour point temperature was
used, no attempt was made to analyze the ab- not included in the comparison of data from
solute error. Of the three regression equations the literature because data from other sources
under consideration, the one containing pour did not contain information on pour point
point temperature shows the least-average er- temperatures.
ror and standard deviation, followed by the An attempt was made to determine whether
modified Beggs and Robinson correlation. or not the equation with pour point tempera-
Clearly, from Table i, the original Beggs and ture is superior to the modified Beggs and Ro-
Robinson correlation is the worst of the three binson correlation. Some viscosity data with
cases compared. This result reaffirms the orig- pour-point temperature which were not used
inal data comparisons of the authors. For ex- for the generation of the modified equation
ample, when Beggs and Robinson applied their were utilized in this investigation. From the
equation to data collected from the literature data of Table 2, it is easily seen that the equa-
( 93 cases ), there appeared to be a tremendous tion with pour point temperature appears to be
average error (114.27%) and extremely large slightly better than the modified Beggs and
standard deviation (530%). The authors did Robinson for purposes of viscosity prediction.
not explain the reason for the large errors but Both the modified Beggs and Robinson cor-
simply warned that extrapolation outside the relation and the new pour point temperature-
range of the data used to develop the correla- viscosity correlation do not provide accurate
tions should be done with caution. viscosity prediction. Accurate viscosity esti-
For purposes of a further test of the superi- mation of crude oil is inherently difficult by
ority of our proposed correlation we have in- nature. The most important reason for the in-
cluded limited literature data in Table 1. Again, accuracy is that the viscosity of crude oil sys-
needless to say, the modified Beggs and Robin- tems, unlike gas systems, has no physical range.
son correlation of this study has considerably Furthermore, at a high viscosity range, oil sys-
lower errors. These results may suggest that the tems often become non-Newtonian and vis-
viscosity data used by Beggs and Robinson in cosity measurements by the application of a
their initial correlation analysis were system- power law and other models become extremely
atically different from other data from other difficult. This can also lead to non-unique vis-
sources, and may have been of poor quality. cosity values.
2OO E.O. EGBOGAH AND J.T.NG

TABLE 2

Pour-point temperature statistical comparison of correlations

EPI Averageerror Standard deviation

range (%) of percent errors
Viscosity data with pour-point temperature not used for
generation of correlation equation (42 cases) 5-58
Beggs and Robinson -74.7 49.5
This study:
Without pour-point temp. -6.6 43.9
With pour-point temp. 2.4 45.8

Conclusions References

Beal, C., 1946. Viscosity of air, water, natural gas, crude

( 1 ) T h e Beggs a nd R o b i n s o n e q u a t i o n o f oil and its associatedgases at oil-fieldtemperatures and
dead-oil viscosity was m o d i f i e d using new vis- pressures. Trans. AIME, 165:94-115.
cosity data. T h e m o d i f i e d e q u a t i o n was veri- Beggs, H.D. and Robinson, J.R., 1975. Estimating the
fied by use o f nearly four h u n d r e d oil systems viscosity of crude oil systems. J. Pet. Technol. (Sep.):
1140.
and the results o b t a i n e d show a substantial
Chew, J. and Connally, C.A., 1959. Viscosity correlation
i m p r o v e m e n t o v er the original Beggs and Ro- for gas-saturated crude oils. Trans. AIME, 216: 23-25.
binson correlation. Houpeurt, A.H. and Thelliez, M.B., 1974. Predicting the
( 2 ) An additional pa r am e t e r , p o u r point viscosity of hydrocarbon liquid phases from their
composition. S.P.E. 5057 presented at SPE-AIME 49th
t e m p e r a t u r e was i n c o r p o r a t e d into the modi-
Annu. Fall Meeting, Houston, Tex.
fied viscosity correlation with a slight Lohrenz, J., Bray, B.C. and Clark, C.R., 1964. Calculat-
improvement. ing viscosities of reservoir fluids from their composi-
( 3 ) T h e new correlation is t hought to be ti- tions. J. Pet. Technol. (Oct.): 1171-1176; Trans.
mely when considering the cur r ent develop- AIME: 231.
Little, J.E. and Kennedy, H.T., 1968. Correlation of the
m e n t s in a wide range o f heavy oil reservoirs viscosity of hydrocarbon systems with pressure, tem-
a n d high p o u r p o int crudes where viscosity be- perature and composition. Soc. Pet. Eng. J. (Jun.):
h a v i o u r is ex tr eme l y i m p o r t a n t . 157-162; Trans. AIME: 243.