SPE 69420

Oil Mixture Viscosity Behavior: Use in Pipeline Design

G. L. Degiorgis, SPE/Repsol-YPF, S. Maturano, Cuyo University, M. Garay, SPE/ Cuyo University, G. R. Galliano,
SPE/Repsol-YPF, A. Fornes/ Cuyo University

Copyright 2001, Society of Petroleum Engineers Inc.

This paper was prepared for presentation at the SPE Latin American and Caribbean Petroleum
Engineering Conference held in Buenos Aires, Argentina, 2528 March 2001.
This paper was selected for presentation by an SPE Program Committee following review of
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
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
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
for commercial purposes without the written consent of the Society of Petroleum Engineers is
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
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.

Abstract
A properly understanding of how oil mixtures behaves as they
flows in a pipeline is a key factor for piping design.
Facilities engineers from oil companies currently use
commercial pipeline simulators in order to perform pressure
and temperature gradients predictions.
Nearly all the simulators apply the general mixing rule
when they need to calculate the viscosity of a certain oil
mixture.
As viscosity depends both on molecular characteristics and
molecular interaction, and oil is a complex mixture of
components, it is not surprising that this rule gives poor
results. Besides, non-Newtonian behavior makes things more
complex.
This paper discusses experimental results of mixing at lab
several different oils and measuring their viscosity. Each oil
had its own composition and rheological behavior. Some
mixtures were prepared in order to reproduce real situations.
Once the mixtures viscosities were obtained they were
introduced as input data in a commercial pipeline network and
nodal analysis software.
Then, pressure and temperature gradients were calculated
a) Using the experimental mixture viscosity
b) Using the general mixing rule viscosity estimated
by the software
Results showed, in some cases, dramatic differences
between the two alternatives. Higher deviations were obtained
where the mixtures were composed of crudes of different
properties.
In order to improve the accuracy other mixing rules were
tried and some of them fit better.

A proper understanding of crude oil properties seems to be

essential for successfully applying complex simulators in
pipeline design
Introduction
Viscosity of liquid mixtures have been studied for a long
time(1) but, unfortunately, this knowledge has not always been
taken into account in the development of commercial pipeline
simulation software.
Viscosity has a molecular origin, and it is highly
dependent on the molecular interaction. That is the case of
complex multiphase and multicomponents fluids such as oil.
If an engineer needs to design an oil pipeline, there are
several commercial software available to predict pressure and
temperature drop. It is only necessary to introduce a viscosity
curve as an input to the soft. But the problem is presented
when several pipes, which transport different oils, form the
system to design.
Most of this technical software makes use of the general
mixing rule as the main alternative to calculate the mixture
viscosity.
This paper presents a study of viscosity mixtures of several
crude oils. In each case individual crude oil viscosities were
measured at a particular shear rate and shear stress, and
different temperatures. The viscosity mixture measurements
were made at similar conditions, and the same values were
calculated using a commercial pipeline network and nodal
analysis software.
The results show that there is no satisfactory equation,
which can be applied to all the practical cases. The simulator
generally overestimates the pressure drop through the pipeline.
Therefore this could carry out economical problems.
Theory
The general mixing rule is currently used to obtain an average
value of many physical properties, including viscosity. The
rule states that the property of a mixture is obtained by the
weight average of the properties of each component. However,
this equation is only valid when the fluids that are mixed do
not interact with each other. That is not the case of oil
mixtures. Besides it is not valid when dealing with emulsions
or water-oil flow. These particular situations (also found in
pipeline design) are not in the scope of this work.

G. L. DEGIORGIS, S. MATURANO, M. GARAY, G. R. GALLIANO, A. FORNES

Some researchers proposed different approaches to

improve the accuracy of the general mixing rules.
Lederer1,2 published an equation for mixture viscosity that
was a modified version of Arrhenius expression. Lederer
equation is described by Eq. 1 trough 3
ln = x A ln A + x B ln B ... (1)
xA =

VA
(2)
VA + VB

xB = 1 - x A

.. (3)

is an empirical constant obtained when Eq. 1 is fitted to

viscosity data of a specific mixture
Shu1 developed in 1984 a correlation to obtain (Eq. 4)
from the specific gravity and viscosity of each component

17.04 SG 0.5237 SG A
=

ln A
B

3.2745

SG B

1.6316

...... (4)

He claimed his correlation was valid for mixtures of heavy

oils, bitumen and petroleum fractions. Shu based his method
in published and in-house data. The correlation was tested
with heavy oil solvent blending data.
Arrhenius3 (Eq. 5), Bingham3 (Eq. 6) and Kendal and
Monroe4 (Eq. 7) have also proposed procedures to obtain
blend viscosities of different oil streams
log = VA log A + VB log B .. (5)
1 VA VB
.... (6)
=
+
A B
3

n
1/ 3
.. (7)
=
xi i
i =1

The API Procedure 11.A3.15 also recommends equation 7

that seems to be created to estimate the viscosity of mixtures
at low pressures when their individual components are known.
Besides, due to the API report, the procedure only can be used
when the components are similar in molecular weight and
general characteristics. These constraints dramatically limit
the method scope. API claims that procedure 11.A3.1 has an
average error of 5.7 % based on 1,300 data points.
The API Procedure 11.A4.56 offers an accurate method of
estimating the viscosity of a blend of two hydrocarbon liquids
of undefined composition. The weight fraction of each
component must be known. Procedure is described by Eq. 8.
The average error was 3.3% for approximately 300 blends.
The API states that the error increases to approximately 11.1%
if one component is a very light oil and the other is a very
heavy oil.
ln [ln ( + 0.7 )] = m (ln T ) + b ............... (8)

SPE 69420

Experimental Tests
In order to assess the accuracy of some of the published
mixing rules, an oil sample survey was designed and
performed. Four different groups of crude oils from Repsol
YPFs Mendoza Basin were selected. Groups were called
group A, B, C, and D. The basic rheological study of each
individual group of crudes was already known from previous
studies7.
Table I contains the main average properties of each group
of crude oils. Viscosity was measured with a Haake RV20
viscometer with a RC20 rheocontroler, M rotors (selected for
each case) and M5 or M10 headers, connected to a computer.
It was chosen due to its accuracy and scope. This viscometer
can measure viscosities ranging from 1 cp to 1,000,000 cP and
shear rates from 0.001 1/s to 1,100 1/s.
Temperature was controlled and monitored with a Haake
F9 cryo-thermostate. The equipment has a special device that
keeps the temperature constant within a 0.1C accuracy.
The viscosity of each individual crude was measured at
different shear rates and temperatures. In the cases were the oil
presented a non-newtonian behavior, a rheological model was
determined, and then the shear rate was calculated depending
on the flow, diameter of the actual or proposed pipeline, and
the rheological model.
Figs. 1 trough 4 shows the viscosity curves of one crude of
each group. The shear rate corresponds to the actual operating
conditions.
A complete rheological study was performed with several
mixtures, which were prepared at the laboratory. All the
mixtures were prepared in a volumetric basis
Discussions and Experimental Results
All of the presented mixing rules were tried. Both equations
recommended by the API did not give accurate prediction so
the results are not shown. The rest of mixing rules methods
where applied to each mixtures, and the measured result was
compared with the predicted one. The Lederer equation was
used with the correlation developed by Shu.
Table 2 and table 3 show a comparison between measured
and calculated viscosity of crude oils from group A. These
crude oils belong to North Mendoza basin. They are paraffinic
and very viscous at low temperatures. However this behavior
changes as the temperature increases between 40C and 50C.
Two mixtures were prepared. One of them is a mixture of 65%
of crude oil 1 and 35% of crude oil 2. The other is a mixture of
45% of crude oil 1, 25% of crude oil 2 and 30% of crude oil 3.
As can be seen the measured viscosity seems to be closer to
the viscosity of the heaviest oil, which was involved in the
mixture. The Lederer equation gave the best results for all the
cases. At high temperatures Bingham and Lederer equations
predict mixing viscosities closer to the measured ones, but due
to the low viscosity ratio the general mixing rule also predict a
good result. Viscosity ratio was defined as:

SPE 69420

Viscosity Ratio =

OIL MIXTURE VISCOSITY BEHAVIOR: USE IN PIPELINE DESIGN

A
B

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

Table 4 shows a comparison between measured and

calculated viscosity of crude oils from group B. It can be seen
that the Bingham correlation gives the best approach to the
measured data. In spite of the fact that the asphaltenes content
is not very high, deposition problems have been reported in
some wells.
Table 5 presents the results from group D. The mixture
was prepared with two crude oils that have different
rheological properties. One of them is much more viscous than
the other. It appears to be that Arrhenius gives the best results
at all temperatures. In addition, it can be seen that when the
viscosity ratio is very high the Bingham equation predicts a
viscosity value that is closer to the lower one that is involved
in the mixture.
Table 6 and table 7 show the mixture of a heavy and a light
crude oil. In both cases the general mixing rule predicts high
viscosity values that do not fit with the measured ones. In the
other hand the Lederer equation gives the best approach in
these cases, although no correlation accurately match the
measured viscosity in the second one. Besides Bingham
equation again failed predicting viscosity values when the
viscosity ratio is too high.
Effect on pressure drop
In order to calculate the pressure drop of an horizontal or
nearly horizontal pipe the key factor is the calculation of the
pressure drop due to friction. It depends on the Reynolds
number, which is defined for a Newtonian fluid by the
following expression:

vd
.... (10)

If the flow regime is laminar (Re <2100), the influence of

the viscosity on the calculation will be significant as the
friction factor is directly proportional to the viscosity as can be
seen by the Moody(8) equation:
Re =

64
(11)
Re
For turbulent flow regimes, the influence of viscosity is
reduced as pipe roughness appears now as an important
parameter in the friction factor calculation as shown in the
Chen(9) equation
f=

1.1098 7.149 0.8981
5.0452
= 2 log

log
+

Re
f
3.7065
2.8257 Re

Viscosity also affects the temperature gradient calculation

that will affect the pressure drop calculation.
As can be seen, a miscalculation of the mixture viscosity
can have a high effect on pressure drop calculations and can
lead to a lot of technical and economical problems, such as
wrong selections of pumps, pipe diameters, heaters, etc.

Simulation Discussion and results

To evaluate the effects mentioned before, some of the
measured and calculated viscosities were introduced in a
popular commercial pipeline network and nodal analysis
software and the pressure gradients were calculated and
compared. The software used was a compositional simulator
that only allowed two pairs of viscositytemperature data as
input. The available alternatives for viscosity blending were
volumetric (general mixing rule) and index. The last
option could not be used because no literature about it was
found.
Table 8 shows the error in pressure drop calculations using
the viscosity blending obtained by the general mixing rule
instead of the measured ones. Furthermore, these values are
compared with those calculated using the correlation that best
fitted the actual ones. Deviation percent was defined as:
Error =

P (measured viscosity) - P(mixing rule)

x 100
P (measured viscosity)

The following cases were analyzed:

Case 1: an actual 21km long, 12 diameter pipeline was
simulated. The characteristics of the crude oil mixture, which
is flowing, are shown in table 2. The actual temperature and
pressure gradients were reproduced and then recalculated with
the mixture viscosity predicted with the general mixing rule
and the Lederer equation.
As can be seen, the pressure drop calculated with the
measured mixture viscosity is very similar with the one
calculated with the general mixing rule, and with the Lederer
equation
Case 2: it was assumed that an oil mixture with the viscosity
data of table 5 was flowing in the same pipeline of case 1. It
can be observed that there is an important deviation between
the value predicted using the general mixing rule and the
measured one. In the other hand the simulator gives an
acceptable error using the Arrhenius equation.
The high deviation predicted using the general mixing rule
is due to the characteristics of each individual component that
were mixed.
Case 3: in this case, it was assumed that the oil mixture of
table 4 was flowing in the same pipeline of case 1.
Similar results as in case 2 were found using Bingham
instead of the Arrhenius correlation.
Case 4: to transport the heavy oil of group C to a gathering
network a pipeline design was needed, one of the alternatives
to reduce the pressure drop was the addition of a solvent. This
case is a simulation of 40km long, 10 diameter pipeline with
the viscosities from table 7.
Predicted pressure drop was four times different depending
on the value which was used. If the general mixing rule had
been used, the pipeline would have been certainly oversized. It
would have represented an investment increase of
approximately U$S 1,000,000.

G. L. DEGIORGIS, S. MATURANO, M. GARAY, G. R. GALLIANO, A. FORNES

Conclusions
1. In Mendoza Province, most of the mixing viscosity
methods, which are contained in commercial pipeline
simulators, can only be used with confidence in light and not
viscous oil mixtures. In these cases error in pressure drop
should be less than 10 %.
2. The error in pressure drop prediction can reach 300% or
more when one or more of the crude oils in the mixture are
viscous and/or non-Newtonian.
3. Based on these results, it could not be found one
correlation that can be applied to all of the cases. Lederer and
Bingham equations were the most accurate, but they were not
precise for the mixture of group C and the solvent.
Unfortunately, the studied commercial simulators do not
include these equations.
4. Further studies must be performed to check if similar
mixing behavior can be find on gel strength measurements.
Gel strengths are a very important parameter in the design of
the maximum allowable pressure of the pipe.
5. Dynamics test in a flow loop will be performed to assess
the effect of continuous flow.
Nomenclature
d

= Pipe Diameter, cm

= Friction factor

m, b = Parameters for predicting the viscosity of the

blend determined from experimental viscosity temperature data of blend components
SG = Specific Gravity
T

= Absolute Temperature, R

= Mole fraction

= Volume fraction

= Fluid velocity, cm/s

= Density, gr/cm 3
SG = SG A - SG
P = Pressure drop, kPa

= Dynamic Viscosity cP
= Kinematic Viscosity cSt
= Pipe relative roughness
Subscripts
A = The more viscous component
B = The less viscous component
References
1.
2.

Shu, S.H.:A Viscosity Correlation for Mixtures of Heavy Oil,

Bitumen, and Petroleum Fractions SPE paper 11280.
Lederer, E.L.: Proc., World Pet. Cong., London (1933) 2, 52628.

3.

4.

5.
6.
7.

8.
9.

SPE 69420

Adamson, Arthur W.: The Difussional Behavior and

Viscosity of Liquid Mixtures, Soc. Pet. Eng. (1963), paper
1348, volume 219, p. 158.
Kendall-Monroe, Viscosity of Liquids II. The Viscosity
Composition Curve for Ideal Liquid Mixtures, J.ACS (1917)
39, 1787 1807.
STD. API Petroleum TDB Chapter 11 Engl 1992, p. 30-32.
STD. API Petroleum TDB Chapter 11 Engl 1992, p. 45-49.
Galliano, G.; Degiorgis, G.; Lpez, M.; Forns, A.: Modeling
an optimizing oil and gas production and gathering systems:
Cerro Fortunoso Field Case, Soc. Pet. Eng., paper SPE 39877
presented at the 1998 SPE International Petroleum Conference
and Exhibition, Villahermosa, Mexico, 3-5 March, 1998.
Moody, L.F.: Friction Factors for Pipe Flow Trans. ASME,
V.66, p.671, 1944
Economides, M.; Hill, A.D.; Ehlig-Economides, C.:
Petroleum Production Systems, Chapter 7, page 142,
Prentice Hall Petroleum Engineering Series, (1994).

SI Metric Conversin Factors

cP x 1.0
E-03 = Pa.s
R (R-459.67-32)/1.8 = C

SPE 69420

OIL MIXTURE VISCOSITY BEHAVIOR: USE IN PIPELINE DESIGN

25000

Viscosity (cP)

20000
15000
10000
5000
0
20

30

40
Tem perature (C)

50

60

Figure 1 Viscosity curve of crude 1 Group A

35000

Viscosity (cP)

30000
25000
20000
15000
10000
5000
0
10

20

30

40

50

60

Temperature (C)

Figure 2 Viscosity curve of crude 6 Group B

80000

Viscosity (cP)

70000
60000
50000
40000
30000
20000
10000
0
30

35

40

45

50

55

Temperature (C)

Figure 3 Viscosity curve of crude 10 Group C

60

G. L. DEGIORGIS, S. MATURANO, M. GARAY, G. R. GALLIANO, A. FORNES

SPE 69420

1200

Viscosity (cP)

1000
800
600
400
200
0
20

30

40

50

60

Temperature (C)
Figure 4 Viscosity curve of crude 8 Group D
Group of
crude oils
A

B
C

Region

API
Gravity

Kuop

Sulfur
Content
(Wt %)

North Mendoza

29.3

12.7

0.14

Cerro Fortunoso
(South Mendoza)

14.4

11.7

2.58

2.6

Llancanello
(South Mendoza)

12.4

10.8

2.71

9.1

No of crudes
Crude oil 1
Crude oil 2
Crude oil 3
Crude oil 4
Crude oil 5
Crude oil 6
Crude oil 10

Crude oil 7
Malarge
Crude oil 8
24.5
11.7
D
Crude oil 9 (South Mendoza)
Crude oil 11*
* Crude oil 11 = 62% Crude oil 7 + 22% Crude oil 8 + 16% Crude oil 9

Asphaltene
content
(Wt %)

0.88

Table 1- Properties of each group of crude oils

Measured Viscosity (cP)

Group A

Crude 1
65%

Crude 2
35%

Specific Gravity
Temperature (C)
20
30
40
50
60
70

0.878

0.877

20,400
2,370
473
138
58.8
35.6

12,035
380
141
61.3
37
26

Calculated Viscosity (cP)

Mixture

Viscosity
Ratio

General
Mixing
Rule

18,500
1,580
227
81
43
29

1.69
6.2
3.35
2.25
1.59
1.37

17,472
1,674
356
111
51
32

Arrhenius

Bingham

Lederer

16,960
1,249
310
104
50
32

16,408
836
259
96
49
31

15,530
560
205
88
47
31

Table 2 Comparison of viscosity predictions for crude oils of group A

SPE 69420

OIL MIXTURE VISCOSITY BEHAVIOR: USE IN PIPELINE DESIGN

Group A
Specific Gravity
Temperature (C)
20
30
40
50
60
70

Crude 1
45%
0.8783
20,400
2,370
473
138
58.8
35.6

Measured Viscosity (cP)

Crude 2
Crude 3
Mixture
25%
35%
0.877
0.8836
12,035
380
141
61.3
37
26

11,400
1,440
103
43.3
16.5
11.3

18,500
804
151
78
41
25

Calculated Viscosity (cP)

General Mixing
Arrhenius Bingham
Rule

15,609
1,594
279
90
41
26

15,015
1,292
221
80
36
23

14,462
947
177
70
31
21

Table 3 - Comparison of viscosity predictions for crude oils of group A

Group B
Temperature (C)
10
20
30
40
50
60

Crude 4
43%
17400
5770
2230
917
474
276

Measured Viscosity (cP)

Crude 5
Crude 6
Mixture
23%
34%
16400
28700
18100
5430
18200
6430
2200
7650
2720
975
2340
1130
498
1590
589
259
886
307

Calculated Viscosity (cP)

General Mixing
Arrhenius Bingham
Rule
21012
20348
19769
9918
8409
7377
4066
3380
2926
1414
1279
1176
859
723
632
479
404
353

Table 4 - Comparison of viscosity predictions for crude oils of group B

Group D
Specific Gravity
Temperature (C)
40
50
60

Crude 7
62%
0.892
30.1
12.6
5.9

Measured Viscosity (cP)

Crude 8
Crude 9
Mixture
22%
16%
0.964
0.901
1070.0
216.7
66.2

33.9
16.8
7.7

51.9
20.0
10.8

Calculated Viscosity (cP)

General Mixing
Arrhenius Bingham
Rule

256.8
57.6
19.3

66.7
24.5
10.4

39.1
16.7
7.7

Table 5 - Comparison of viscosity predictions for crude oils of group D

Measured Viscosity (cP)

Group C
Group D
Specific Gravity
Temperature (C)
20
40
60

Crude 10
18%

Crude 11
82%

0.981

0.907

200,000
22,400
2,480

51.9
20.0
10.8

Calculated Viscosity (cP)

Mixture

Viscosity
Ratio

General
Mixing
Rule

155.0
48.6
22.3

3,853
1,120
230

35,642
4,003
450

Arrhenius

Bingham

Lederer

225
70
28

63
24
13

104
40
21

Table 6 - Comparison of viscosity predictions for crude oils of group C and D

G. L. DEGIORGIS, S. MATURANO, M. GARAY, G. R. GALLIANO, A. FORNES

Measured Viscosity
Group C

Crude 10
90%

Specific Gravity
Temperature (C)
30
40
50

0.981

0.8959

69,500
22,400
9,460

8.0
5.7
4.2

SPE 69420

Calculated Viscosity

Diesel Oil
Mixture
10%

6,670
2,640
1,200

Viscosity
Ratio

General
Mixing
Rule

Arrhenius

Bingham

Lederer

8,687
3,930
2,252

62,551
20,161
8,514

28,061
9,791
4,371

80
57
42

9,893
4,282
2,199

Table 7 - Comparison of viscosity predictions for crude oils of group C

Cases

Temperature Gradient
Using Measured
Viscosity [C]
Inlet

Case 1
66
Case 2
40
Case 3
55
Case 4
60
(1) Lederer Equation
(2) Arrhenius Equation
(3) Bingham Equation

Calculated Pressure Drop [kPa]

Outlet

Measured
Viscosity

53
34
45
60

1120
960
4090
2140

General
Mixing
Rule
1170
1240
5530
8360

Error (%)

Best Fitted
Mixing Rule

Error (%)

4.5
29
35
290

1139 (1)
1010 (2)
4200 (3)
32390 (1)

1.8
5.2
2.7
51.3

Table 8 Pressure drop predictions using different correlations