PAPER 2003-148

PETROLEUM SOCIETY
CANADIAN INSTITUTE OF MINING, METALLURGY & PETROLEUM

New Compositional Models for Calculating

Viscosity of Crude Oils
A.M. Elsharkawy, S.A. Hassan, Y.S. Kh. Hashim, M.A. Fahim
Kuwait University

This paper is to be presented at the Petroleum Society’s Canadian International Petroleum Conference 2003, Calgary, Alberta,
Canada, June 10 – 12, 2003. Discussion of this paper is invited and may be presented at the meeting if filed in writing with the
technical program chairman prior to the conclusion of the meeting. This paper and any discussion filed will be considered for
publication in Petroleum Society journals. Publication rights are reserved. This is a pre-print and subject to correction.

ABSTRACT of state based viscosity models. The comparison shows

Crude oil viscosity is an important physical property
that controls and influences the flow of oil through
porous media and pipelines. Hence, it is the basis of
many reservoir engineering and production system
calculations. In the crude oil recovery and processing,
viscosity is significant parameter as a large amount of
time and money are spent in experimental measurements
of viscosity. This necessitates the development of reliable
viscosity models capable of predicting crude oil viscosity
and eliminates the expensive and cumbersome
experimental measurements.
This paper presents new compositional models for
estimating crude oil viscosity. The new models use fluid
composition, temperature and pressure to predict oil
viscosity. The models were derived from viscosity
measurements of several crudes from the Middle East,
North Sea and others. Accuracy of the proposed models
has been compared to that for several empirical
correlations, corresponding state models, and equation
the superiority of the new models over the other methods.
INTRODUCTION
The mechanism or theory of gas viscosity has been
reasonably well defined by the application of the kinetic
theory of the gases. The theory of liquid viscosity is
poorly developed due to the intermolecular forces
between the molecules, which consist of short-range
effects such as repulsion and hydrocarbon bonding, wide
range effects such as electrostatic effects, and long-range
effects such as attractions. A further complication is the
structure and degree of disorder between the molecules.
Thus, there is no widely accepted simple theoretical
method to calculate viscosity of liquids1.
Petroleum crude oils and fractions are typically
complex mixtures, and the physical and chemical
properties vary considerably depending on the
composition of the constituents. For instance, the
viscosity of a crude oil or a fraction composed mostly of
aromatics might have quite different temperature
dependence than that composed mostly of saturates.
Useful viscosity prediction methods are most
conveniently based on parameters such as boiling
temperatures and specific gravities that are commonly
used to characterize each fraction2. The relatively
successful viscosity models available in the literature can
be classified as:
ß Empirical methods
ß Corresponding states methods
ß EOS-based viscosity models
ß Group contribution methods
Although numerous viscosity correlations for
hydrocarbon liquids and gases are available in the
literature, there are three main drawbacks in their
applications3:
1. Application range and accuracy are limited.
2. As the viscosity of liquid phase and gas phase is
calculated by using different graphs or
correlations, a smooth transition in the near-critical
region cannot be achieved.
3. Density is involved in evaluating the fluid
viscosity, and hence, separate density correlation is
required.
Recently, the efforts focused toward developing a
theoretical viscosity models based on an equation of state
(EOS), the major advantages of such models are:
1. A single model, achieving a smooth transition of
liquid/gas viscosity in the near-critical region can
describe the viscosity of both gas and liquid
phases.
2. Both high pressure and low-pressure data can be
correlated, and density is not involved in
evaluating the fluid viscosity.
3. Using a single EOS can perform P-V-T, V-L-E
and fluid viscosity calculations; the
thermodynamic consistency in process
simulation/reservoir simulation is thus improved.
2
The state of Kuwait contains about 10% of the world
proven reserve. Precise viscosity calculations are very
vital to the production and transport of crude oil.
Therefore, the current study is aimed at: 1) investigating
the accuracy of previously published methods for
estimating the viscosity of Kuwaiti crude oils as well as
others using experimentally measured viscosity. 2)
Modifying one of the best methods available in literature
or proposing a new method for estimating Kuwaiti crude
oil viscosity. 3) Studying the effect of including the
binary interaction parameter (BIP) on EOS-based
viscosity model for estimating viscosity of Kuwaiti
crudes.
METHODS OF VISCOSITY ESTIMATION
In general, there is no simple rule or theory for
predicting the rate of change of crude oil viscosity with
temperature and pressure. The only alternative is to
obtain the necessary viscosity measurements and model
their behavior.
Empirical Correlations
Previous published correlations are restricted to predict
the viscosity of gas-free (dead or stabilized) crude oils at
atmospheric pressure, on the other hand, there are other
models which covers all kinds of crude oils: dead,
saturated, and undersaturated crudes. The most popular
empirical models presently used are those developed by
Beggs and Robinson4, Labedi5, Kartoatmodjo and
Schmidt6, and Elsharkawy and Alikhan7. These
correlations are a function of field measurable parameters
such as temperature (T), pressure (P), separated gas
gravity (g g), and tank oil gravity (API). The dead oil
viscosity (mod) was expressed as a function of both oil
API gravity and reservoir temperature. The saturated oil
viscosity (mob) was correlated to dead oil viscosity and
solution gas-oil ratio (Rs). It is important to note that
solution gas-oil ratio is a function of oil API gravity, gas
gravity, reservoir pressure and temperature. Thus
accuracy of these correlations in estimating saturated oil
viscosity is largely dependent of the success of estimating
the gas-oil ratio. As no gas is being dissolved into oil
above the bubble point, pressure becomes the primary
independent variable for predicting the viscosity of
undersaturated oil (mo).
Corresponding State Models
The principle of corresponding states is one of the
most useful methods for predicting properties of fluids.
The principle states that dimensionless properties of all
fluids have the same numerical values at the same
reduced conditions. This provides the most important
basis for the development of correlations and estimation
methods. These methods require knowledge of the
properties of, at least, one reference fluid. Hanley8
developed an extended corresponding state to calculate
transport properties of mixtures, which additionally
requires the acentric factor and critical compressibility of
the reference fluid and each component. Hanely et al.9
Used the viscosity of methane as reference fluid viscosity
was correlated as a function of temperature and density.
For fluids that do not correspond with the reference fluid,
Hanley derived a correction factor. The major difficulty
in using methane as reference fluid is that its normal
freezing point is at a reduced temperature of 0.476, which
is above that for many other hydrocarbons. Pedersen et
al.10 used a new parameter (a ) to account for the
molecular size and density effects. Pedersen and
Fredenslund11 proposed an improvement of the model by
extending the method of Pedersen et al. to Tr < 0.4, which
is below the freezing point of methane by modifying the
equations with additional experimental viscosity data.
Teja and Rice12 and Teja et al.13 proposed an alternative
formulation of the corresponding state principle, based on
the properties of two-reference fluid. However, all these
methods use the acentric factor as an interpolation
parameter of reference fluid properties. Petersen et al.14
proposed a method based on the Teja-Rice method with
the molecular weight as an interpolation parameter
instead of the acentric factor since, for heavy petroleum
fractions, the acentric factor decreases with increasing
molecular weight. Since the method was developed for
hydrocarbons, methane and n-decane were used as
reference components. The model proposed by Petersen
et al. yields accurate viscosity predictions of light
petroleum fractions. However, its application to heavy
fractions may give inaccurate predictions. In addition, the
prediction by this model is restricted to the condition of
3
reduced temperature Tr > 0.476. Moharam et al.1 used
decane and eicosane as the reference fluids in predicting
the viscosity of heavy crude oils and petroleum fractions
of average molecular weight higher than 142. This model
relaxes the limitation imposed on reduced temperature Tr
> 0.476 to Tr > 0.4.
EOS-Based Viscosity Models
Two widely used procedures have been developed to
calculate the viscosities of in situ reservoir gases and
liquids from their composition at the desired temperature
and pressure. Given a composition expressed in methane
through heptane-plus, hydrogen sulfide, nitrogen and
carbon dioxide together with the molecular weight and
specific gravity of the heptane-plus fractions. The first
viscosity prediction method that requires the composition
of crude oil was presented by Lohrenz et al. 15. This
procedure based on the residual viscosity concept and the
low of corresponding states. Little and Kennedy16 is the
second well known method which applies reliably to both
liquids and gases, and the equation is similar in form to
Van der Waal’s EOS. The disadvantage of the above two
methods is that the predicted viscosity is very sensitive to
the density, which is normally determined by a separate
correlation and may be very inaccurate for high viscosity
fluids. The similarity between pressure-volume-
temperature (P-V-T) and temperature-viscosity-pressure
(T-m-P) relationships was pointed out by Philips17 early
in 1912. Lawal18 developed a viscosity model based on
the four-parameter Lawal-Lake-Silberberg EOS.
However, it is not a predictive model since specific
constants are required for each substance, and poor
results were obtained when applied to C7+ fraction
containing reservoir oils. Guo et al.2 proposed two
different viscosity models for pure hydrocarbons (mainly
for n-alkanes). The models developed from the three-
parameter Patel-Teja (PT) (eq. 1) and the two-parameter
Peng-Robinson (PR) (eq. 2) cubic equations of state.
These models are applicable to both liquid and gas
phases, and has been extended to hydrocarbon mixtures.
rP a
T= - ............................(1)
m - b' m(m + b) + c(m - b)
 rP a Methods for Viscosity calculations considered
T= - ........................... (2)
m - b' m(m + b) + b(m - b)
Where the parameters a, b, and c are EOS parameters.
Guo reported an average absolute deviation (AAD) of
6.18% for pure substances with PR-viscosity model, but
the accuracy was reduced when both models (PR and PT)
were used to estimate the viscosity of hydrocarbon
mixtures. PR-viscosity model gives AAD of 15%,
whereas PT-viscosity model gives 12.6% when tested
against nine gas condensates taken from Lawal18. Also,
PT-viscosity model was tested on reservoir crude oils and
AAD of 15.07% was obtained. Because PR has limited
accuracy, it was modified by Guo et al.19 to improve the
viscosity prediction accuracy for reservoir fluids,
including the highly a symmetric CO2 injected enhanced
oil recovery (EOR) systems. The overall AAD for
reservoir crude oils was 14%, whereas AAD for the
previous nine gas condensates was reduced from 15% to
9.8%.
CURRENT WORK
Viscosity Data
A data bank of 89 fluid samples has been used in this
study to check the accuracy of the published methods and
develop new compositional model. The samples have
been described according to their types and sources in
Table 1. Viscosity data for Kuwaiti crude-oil samples are
measured for each sample as a part of differential
liberation (DL) test. The DL test provides measurements
of crude oil viscosity, solution gas-oil ratio, and oil
density at reservoir temperature for several pressures. The
test also provides compositional analysis of reservoir oil
sample (methane through hexane, non-hydrocarbon, and
heptane plus fraction), molecular weight and specific
gravity of the heavy hydrocarbon plus fraction. The
composition and viscosity measurements of the Kuwaiti
crude oils as well as the various crudes used in this study
are reported in Tables 2 through 5.
4
in the study
The following methods are considered for
investigating their accuracy in predicting the viscosity of
crude oils.
ß Empirical methods: Beggs and Robinson, Labedi,
Kartoatmodjo and Schmidt, and Elsharkawy and
Alikhan.
ß Corresponding state methods: Pedersen et al. and
Petersen et al.
ß EOS-based viscosity models: Lohrenz et al.,
Little and Kennedy, and Guo et al.
Characterization of C7+ -fraction
In order to calculate the viscosity of reservoir fluids
from their composition, the critical properties of the
fluids: critical pressure and temperature, and acentric
factor are required. The critical properties of pure
components are well documented. However, the critical
properties of heavy hydrocarbon plus fraction are usually
estimated from correlations. Several correlations are
available in the literature for estimation of properties of
undefined fractions. The selection of a particular
correlation has significant influence on calculation of
crude oil viscosity, as the heavy hydrocarbon fraction
comprises large portion of the composition of most crude
oils. For this reason, two sets of correlations are
considered in this study for characterizing the heavy plus
fraction of crude oils and to observe the impact of the
selection of Tc, Pc-correlations on the viscosity
prediction. The critical properties (Tc, Pc) and acentric
factor (w) of heptane plus fraction were estimated in this
study by the following two sets of empirical correlations:
Set 1: The critical temperature (Tc) and critical
pressure (Pc) of plus fraction are estimated from Lee-
Kesler20 correlation and the acentric factor (w ) is
estimated from Edmister21 correlation.
Set 2: The critical temperature (Tc) and critical
pressure (Pc) of plus fraction are estimated from Twu22
correlation and the acentric factor (w) is estimated from
Lee-Kesler correlation.
Viscosity Models
Evaluation of the various methods, which use
composition of reservoir fluids as input parameter,
considered in this study indicated that Guo et al. method
has the lowest average absolute deviation (32%) for all
the viscosity data. Therefore, many attempts have been
made to improve or develop a new compositional model
for predicting the viscosity of crude oils with reasonable
degree of accuracy. These attempts resulted into two
models. The first is an empirical model that relates
viscosity to temperature, pressure and composition in a
simple form for all crudes considered in this study. The
second is a modification of viscosity model based on
Peng- Robinson EOS, which was originally proposed by
Guo et al. This model is recommended for Kuwaiti
crudes.
Empirical Model
The proposed empirical model is developed by
regression technique to fit the measured viscosity data.
The database used in developing the model consists of
361 data points comprising 77 world crude oils from
Kuwait, UAE, North Sea, and others. Analysis of these
data showed that the viscosity of crude oil could be
expressed as a function of temperature, pressure and
several groups related to oil composition. In the proposed
model, the mole fractions of the light components G L
(methane and ethane) were taken as one group. The
second group G m represents the intermediate pure
components (propane through hexane). The third group
Gh comprises of components heavier than heptane, C7+
fraction. The fourth group Gnon consists of non-
hydrocarbon components (N2, CO2 & H2S). Several
attempts have been made to correlate oil viscosity to
compositional groups in addition to temperature and
pressure. These attempts are summarized in Table 6.
These various attempts resulted in estimating crude oil
viscosity with average absolute deviation (AAD) that is
less than that for all published methods considered in this
study. The first attempt in developing a new
compositional model considers the crude oil to be formed
of three groups: light, medium and heavy. This result in a
model that is able to predict the crude oil viscosity with
AAD of 31.28%. Introducing the non-hydrocarbon group
and the properties of heavy hydrocarbon plus fraction,
5
molecular weight and specific gravity, have substantially
improved accuracy of the proposed model. This result in
an empirical model that is capable of predicting the crude
oil viscosity with AAD of 24%. The proposed empirical
model has the following form:
m o = a1 T a2 P a3 G La 4 Gma 5 Gnon
a6
Gca77+ g ca78+ Mw ca79+ .......(3a)
where:
a1= 13148.23752 a2= -1.29044 a3= 0.112653
a4= -0.33936 a5 = 0.058341 a6= 0.018099
a7= 0.638023 a8= 14.01134 a9= -0.27881
The model is improved further by reducing the number
of variable by eliminating the non-hydrocarbon group
and the molecular weight of the heptane plus fraction.
The data bank used in this study, and crude oils in
general, contains low concentration of non-hydrocarbon
components. This attempt resulted in reducing the
number of variables in the proposed empirical model as
follows:
m o = a1 T a 2 P a 3 g ca74 + G La5 Gma6 Gha7 ......................(3b)
where:
a1= 2248.089447 a2= -1.27846 a3 = 0.117425
a4 =13.19727 a5 = -0.32428 a6 = 0.066623
a7 = 0.655418
Hence, the model described by equation (3a) is
accounting for the presence of non-hydrocarbon
components in the crude oils. However the second form
(3b) is used only where the amount of non-hydrocarbon
components is nil. In the above equations (3a) and (3b)
temperature is in oF, pressure in psia, and oil viscosity in
cp.
Modified Viscosity Model Based on EOS for
Kuwaiti Crudes
Guo et al. presented a viscosity model based on Peng-
Robison EOS for calculating viscosity of hydrocarbon
mixtures. The model when used to calculate viscosity of
Kuwaiti crudes resulted in AAD of 45.7 %. Therefore the
model has been modified to improve its accuracy for
viscosity calculation of the Kuwaiti crudes. The model is Q 2 = 1.94546 - 3.19777w + 2.80193w 2 ................... (14)
described in the following equations.

Q 3 = 0.299757 - 2.20855w + 6.64959w 2 ................... (15)

rP a
T= - ........................... (4)
m - b' m(m + b) + b(m - b) It was noticed that the functional forms of Q1, Q2, and
Q3 for C7+-fraction (i.e. w ≥ 0.30) play a critical role in
where P in atm, T in oK and m in micropoise.
the prediction of hydrocarbon mixture viscosity. This is
because C7+-fraction comprises a major part of the crude
rc2 Pc2 oil. Hence, the following correlations were proposed in
a = 0.45724 ....................................................... (5) the present study for Kuwaiti crudes:
Tc

Q1 = -0.5395w3 + 0.9335w 2 - 0.4535w + 18.71 ........... (16)

rP
b = 0.07780 c c ............................................................... (6)
Tc

Q 2 = -180.71w3 + 324.01w 2 - 175.96w + 7.8352 ........... (17)

r = rct (Tr , Pr ) ................................................................... (7)

Q 3 = 23.96 w3 - 46.804w 2 + 28.846w - 2.456 ....... (18)

m c Tc ........................................................................... (8)
rc =
Pc Z c The EOS based viscosity models was extended to
mixtures by applying the following conventional mixing
where m c is the critical viscosity is calculated by the rules:
empirical correlation proposed by Uyehara and Watson23:

a m = Â x i a i ..................................................................... (19)
-1 / 6 1/ 2 2/3 i
m c = 7.7T c Mw P
c
................................................ (9)

bm = bm  xi bi ................................................................. (20)
b' = bj (Tr , Pr ) ................................................................. (10) i

[
t (Tr , Pr ) = {1 + Q1 (Tr Pr ) 0.5 - 1 ]} -2
................................11) b 'm = ÂÂ x i x j b i' b 'j (1 - k ij ) ...................................... (21)
i j

j(Tr , Pr ) = exp Q 2[ ( Tr - 1)]+ Q 3 ( Pr - 1)2 ...................... (12) rm = Â x i ri ..................................................................... (22)

For hydrocarbons, carbon dioxide and nitrogen, the
three coefficients, Q1, Q2 and Q3, were generalized in
terms of acentric factor w as follows, respectively:
When w < 0.3:
Q1 = 0.829599 + 0.350857w - 0.747680w 2 ........... (13)
i
The parameters of EOS are considered to represent the
attractive and repulsive forces between the molecules of
different substances forming the mixture. For molecules,
which do not differ greatly in size or chemical structure
the binary constant kij can be set equal to zero. For
binaries where both components fall into one of these
categories (hydrocarbons, rare gases, permanent gases,
carbon monoxide) kij may be estimated by:

6
 8(Vci Vcj ) 0.5
k ij = 1 - ..............................................(23)
0.33
(V +V 0.33 3
) al. (1964), Little and Kennedy (1968), Guo et al. (1997)
ci cj
RESULTS AND DISCUSSION
Evaluation of Previously Developed Models
Figure 1 shows a summary of the accuracy of 4
different empirical models in predicting viscosity of
Kuwaiti crudes. These models were tested using 910
experimental points from 49 Kuwaiti crudes. They are:
Beggs and Robinson (1975), Labedi (1992), and
Kartoatmodjo and Schmidt correlations (1994), and
Elsharkawy and Alikhan (1999). In these models,
Composition of crude oil is characterized by API gravity
and composition of gas in solution is characterized by gas
gravity. Figure 1 indicates that Elsharkawy and Alikhan
model has AAD of 20.86%, which is the smallest among
the other empirical models. Beggs and Robinson model
has AAD of 26.4%, Kartoatmodjo and Schmidt model
has 32.11% and Labedi model shows the highest AAD
with 33.45%. This is because Elsharkawy and Alikhan
used Middle East crudes to develop their model. Those
crudes have some physical and chemical similarity to the
Kuwaiti crudes. Kartoatmodjo and Schmidt and Beggs
and Robinson have used data bank to develop their
models and Labedi used light crudes, which are quite
different in physical and chemical properties from
Kuwaiti crude oils. The strong weakness of these models
is that paraffinic oil and aromatic oil having the same API
gravity would have the same viscosity regardless their
composition. Therefore these models cannot be used for
calculations of viscosity in compositional reservoir
simulation and oils production by gas injection.
Figure 2 shows a comparison between measured and
predicted viscosity by the previously mentioned models
for Kuwaiti crude oil, sample # 16. This figure also
indicates that Elsharkawy and Alikhan model closely
matches the experimentally measured viscosity. None of
the available models could predict the crude oil viscosity
at all pressures with satisfying accuracy.
Six different models, which use composition of
reservoir fluids rather than gravity, have also been tested
to predict viscosity of 370 data points from the data bank.
7
These models are: EOS-based models such as: Lohrenz et
and Guo et al. (2001), and the corresponding-states
models by Pedersen et al. (1987), and Petersen (1991).
Figure 3 shows the accuracy in estimated viscosity for 49
Kuwaiti crude oils at the bubble point pressure using
corresponding states models. Because Pedersen model
does not predict the true bubble point pressure, it was
adjusted to predict the actual bubble point pressure.
Surprisingly, the AAD by the adjusted model is higher
than the original one. Figure 3 shows that the
corresponding state models have relatively large AAD
ranging from 39% to 51%.
Table 7 shows a comparison between Guo’ s work and
calculation of gas condensates and crude oils viscosity by
programs made for this study. Our results agree with
those reported by Guo et al. that modified PR-EOS based
viscosity model (MPR) is superior to all other methods in
viscosity calculation. However, our results disagree with
Guo and Lawal of viscosity calculations by Lohrenz and
Little and Kennedy. Based on our experience and
previous studies, it is believed that our viscosity
calculations by Lohrenz and Little and Kennedy is
reasonable. Guo et al. and Lawal calculation of viscosity
calculation using Lohrenz and Little and Kennedy
method is questionable as they reported AAD of several
hundreds. It is well known that Lorenz and Little and
Kennedy methods are widely used for viscosity
calculations in most reservoir simulation packages. The
comparison between Guo’s work and present study
indicates that the predicted results obtained by MPR
model in both studies are comparable to each other even
though the characterization procedures for C7+-fraction in
both studies are different. Guo’s work uses Tc, Pc-
correlations of Twu (1984) and the w-correlation of
Kesler and Lee (1976) whereas; this study uses
characterization method #1. It is also noticed that BIP has
insignificant impact on the viscosity prediction of crude
oils, as the AAD is 17.7% when BIP was set to zero and
the AAD is 17.73% when BIP was considered.
Table 8 shows summary of the results of viscosity
calculations of 361 data points representing to 77 crude
oils from different parts of the world. These results
indicates that MPR-EOS based viscosity is more accurate
than other methods. However, the model as it stands
predicts viscosity of Kuwaiti crudes with AAD of 46%.
Therefore, the model needs to be modified to predict
viscosity of Kuwaiti crude with reasonable accuracy.
Results in table 8 also indicated that the binary
interaction parameters (BIP) have little effect on viscosity
calculations by Guo et al. model.
Evaluation of Proposed Models
The calculation presented in this section is using
characterization method #2 for describing the C7+-
fraction of crude oils to compare between the
performance of the proposed models presented in this
study and that by Guo et al.
Table 9 shows a comparison between the empirical
model proposed in this study and Guo et al. model for
viscosity calculation of Kuwaiti crudes as well as others.
This table indicates that the proposed empirical model
has overall AAD of 24% for all the crudes in the data
bank compared with 32% given by Guo et al. viscosity
model. . Also, the empirical model is simple and reduces
the tedious calculation involved in EOS-based viscosity
correlations.
On the other hand, the original MPR-viscosity model
suggested by Guo et al. produces satisfactory results for
gases, but produces unsatisfactory results for the 77
world crude oils (Kuwait, UAE, North Sea, and others).
The reason for good viscosity predictions by the original
model is that the concentration of heavy fraction in gases
is rarely exceeding 10%. However, the hydrocarbon plus
fraction forms the major portion in most crude oils (20 to
80%) and it is not properly defined in the model for
Kuwaiti crudes. Therefore, the model was considered for
modification. Calculation of viscosity from EOS-based
viscosity model greatly depends on critical properties of
components forming the reservoir fluids. The critical
properties of pure components are well documented.
However, the critical properties of heavy hydrocarbon
fraction are usually estimated from correlations. The
fitting parameters describing the C7+ fraction play a
critical role in viscosity prediction by EOS-based
viscosity model. Therefore these parameters have been
modified to properly describe the behavior of the Kuwaiti
crudes. The modification has been achieved using
8
optimization routine. The proposed modifications
resulted in significant improvement in viscosity
prediction of Kuwaiti crudes.
Table 9 also shows a comparison between the accuracy
of the original EOS-based viscosity model by Guo et al.
and that of the proposed modification in the model for 49
Kuwaiti crude. The new fitting parameters have resulted
in reducing the AAD from 46 % to 23 %.
Figure 4 shows cross plots of measured and estimated
viscosity by empirical model presented in this study. It
shows that predictions made by the proposed empirical is
close to the unit slope line, however, Guo et al. model is
underestimating the viscosity of Kuwaiti crudes and has a
scatter below the unit slope line. Figure 5 shows another
cross plot between measured and calculated viscosity by
original Guo et al. model and modification suggested in
this study. This figure indicate that cross plot of points by
the proposed modification is close to the 45 degree line
and Guo et al. model shows scatter and underestimate.
Figure 6 shows comparison between measured and
calculated viscosity as a function of pressure for one of
the oil samples produced from oil fields in Kuwait. This
figure depicts that the proposed empirical model and
modification for Guo et al. are closely matching the
experimentally measured viscosity.
CONCLUSIONS
1. Several correlations, which use oil API gravity as
input parameter and widely used for prediction of
crude oil viscosity have been evaluated using a large
data bank. The results indicate that Elsharkawy and
Alikhan correlation has the best accuracy among the
other correlation. However, these correlations cannot
be used for compositional simulation and for oil
recovery by gas injection.
2. The current study also evaluated various EOS-based
viscosity methods and corresponding state methods
to calculate viscosity of various crude oils. The
results indicate that Guo et al. model properly
predicts viscosity of gas condensates and
unsatisfactory predicts viscosity of crude oils.
3. The original Guo et al. model has been improved in
this study by modifying its fitting parameters to
 properly describe the heptane plus fraction of the rc = Parameter in equation (8)
Kuwaiti crudes. The proposed modification results in Rs = Gas-oil ratio
reducing the AAD from 46% by the original model T = Temperature
to 23% by the proposed modification. Tc = Critical temperature
Tr = Reduced-temperature = T/Tc
4. A simple compositional empirical model is also
V = Molar volume
presented in this study. The new empirical model
Vc = Critical volume
successfully predicts 361 measurements of crude oil
Zc = Critical compressibility
viscosity from data bank with AAD of 23% for
Kuwaiti crudes. The proposed model eliminates Greek
characterization of heptane plus, splitting the heavy w = Acentric factor
fraction, complex mixing rule or inclusion of binary t = Functional form, eq. 11
interaction parameters, which are needed for EOS- j = Functional form, eq. 12
base viscosity model. a = Parameter for molecular size and density
5. Two methods for characterizing the hydrocarbon effect
plus fraction of crude oil have been considered in mC = Critical viscosity
this study. It was found that viscosity estimation by gc7+ = Specific gravity of C7+
various methods strongly dependent on the method mo = Viscosity of the crude oil, cp.
of characterizing the heptane plus fraction. m = Viscosity

6. The effect of including the binary interaction Abbreviations

parameters (BIP) in viscosity calculation by EOS AAD Average absolute deviation
based models has been investigated. The current API API gravity (=141.5 / g - 131.5)
work indicates that BIP has insignificant effect on ARD Average relative error
viscosity calculation by EOS based viscosity models. BIP Binary interaction parameter
DL Differential Liberation
LIST OF SYMBOLS EOS Equation of State
MPR Modified Peng-Robinson EOS
a,b,c,b' = Equation of state parameters of eq. (1)
PR Peng-Robinson
and (2)
PT Patel-Teja
a m ,b m ,rm ,b = Mixture coefficient parameters of eq. (1)
'
m

† & (2) ACKNOWLEDGEMENTS

Gh = Mole fraction of C7+
The authors thank Dr. Karen Pedersen for providing
GL † = Mole fraction of light components,
the viscosity data of North Sea Crudes used in this study.
C1+ C 2
Gm = Mole fraction of medium components C3
LITERATURE CITED
through C6
Gnon = Mole fraction of non-hydrocarbons, N2, 1. Moharam H.M., and Fahim M.A., ”Prediction of viscosity
H2S, and CO2 of heavy petroleum fractions and crude oils using a
corresponding states method”, Ind. Eng. Chem. Res. –
kij = Binary interaction parameter
1995, Vol. 34, 4140-4144.
Mwc7+ = Molecular weight of heptane plus fraction
P = Pressure 2. Monnery W.D., Svrcek W.Y. and Mehrotra A.K.,
Pc = Critical pressure “Viscosity: A critical review of practical predictive and
correlative methods”, The Canadian Journal of Chemical
Pr = Reduced-temperature = P/Pc
Engineering- Feb 1995, Vol. 73, 3-40.
Qi = Coefficient, i=1,2,3
r = Parameter in equations (1) and (2)

9
3. Guo X.Q., Wang L.S., Rong S.X and Guo T.M., process design”, Ind. Eng. Chem. Process Develop.-1985,
“Viscosity model based on equations of state for Vol. 24, 344-394.
hydrocarbon liquids and gases”, Fluid Phase Equilibria-
14. Petersen K.A., Knudsen K., and Fredenslund A.,
1997, Vol. 139, 405-421.
“Prediction of Viscosities of Hydrocarbon Mixtures”,
4. Beggs H.D. and Robinson J. R., “Estimating the viscosity Fluid Phase Equilibria – 1991, Vol. 70, 293 – 308.
of crude Oil system”, J. Pet. Technol. – 1975, Vol. 27,
15. Lohrenz J., Bray B.G. and Clark C.R., “Calculating
1140-1149.
viscosities of reservoir fluids from their compositions”,
5. Labedi R., “ Improved correlations for predicting the Journal of Petroleum Technology-1964, 1171-1176.
viscosity of light crudes”, Journal of Petroleum Science
16. Little J.E. and Kennedy H.T., “A correlation of the
and Engineering – 1992, Vol. 8, 221-234.
viscosity of hydrocarbon systems with pressure,
6. Kartoatmodjo T. and Schmidt Z., “Large data bank temperature and composition”, Society of Petroleum
improves crude physical property correlations”, Oil and Engineers Journal-1968, 157-162.
Gas Journal – July 1994, Vol. 4, 51-55.
17. Philips P, Proc. R. Soc. 87A (1912) 48.
7. Elsharkawy A.M. and Alikhan A.A., “Models for
18. Lawal A.S., “Prediction of vapor and liquid viscosities
predicting the viscosity of Middle East crude oils”, Fuel –
from the Lawal-Lake-Silberberg equation of state”,
1999, Vol. 78, 891-903.
SPE/DOE Paper No. 14926, Presented at the 5th
8. Hanley H.J.M., “Prediction of viscosity and thermal Symposium on Enhanced Oil Recovery- April 1986, Tulsa.
conductivity coefficients of mixtures, Cryogenics –1976,
19. Guo, X.Q., Sun C.Y., Rong S.X., Chen G.J. and Guo T.M.,
Vol. 16, 643-651.
“Equation of state analog correlations for the viscosity and
9. Hanley H.J.M., McCarty R.D., and Haynes W.M., thermal conductivity of hydrocarbons and reservoir fluids”,
“Equations for the viscosity and thermal conductivity Journal of Petroleum Science and Engineering, March
coefficients of methane”, Cryogenics – July 1975, 413- 2001, 15-27.
415.
20. Kesler M.G. and Lee B. I., “Improve prediction of
10. Pedersen K.S. and Fredenslund A., “Viscosity of crude enthalpy of fractions”, Hydrocarbon Processing-1976, Vol.
oil”, Chemical Engineering Science-1984, Vol. 39, No. 6, 55, 153-158.
1011-1016.
21. Edmister W.C., “Applied hydrocarbons thermodynamics
11. Pedersen K.S. and Fredenslund Aa., “An improved Part 4. Compressibility factors and equations of state”,
corresponding states model for the prediction of oil and gas Petroleum Refiner-1958, Vol. 37, 173-179.
viscosities and thermal conductivities’, Chemical
22. Twu C.H., “An internally consistent correlation for
Engineering Science – 1987, Vol. 42, 182-186.
predicting the critical properties and molecular weights of
12. Teja A.S, and Rice P., “Generalized corresponding states petroleum and coal-tar liquids”, Fluid Phase Equilibria –
method for the viscosities of liquid mixtures”, Ind. Eng. 1984, Vol. 16, 137-150.
Chem. Fundam.-1981, Vol. 20, 77-81.
23. Uyehara O. A., and Watson K.M., “A Universal Viscosity
13. (13)Teja A.S., Thurner P.A. and Pasumarti B., “The Correlation”, National Petroleum News – Oct. 1944, 714-
calculation of transport properties of mixtures for synfuels 723.

10
 Table 1: Data sources

No. of Samples Type Origin Reference

9 Natural Gas - Lawal (1986)
14 Crude Oil - Lawal (1986)
7 Crud Oil Abu Dhabi Moharam (1995)
7 Crude Oil North Sea Pedersen
49 Crude Oil Kuwait Present study

Table 2: Composition and viscosity measurements of Kuwaiti crude oils

Crude T P mexp. Composition
No. (F) ( psi ) ( cp ) N2 CO2 H 2S C1 C2 C3 I-C4 C4 I-C5 C5 C6 C 7+ gC7+ MWC7+

1 130 1365 0.51 0.0036 0.0017 0 0.2723 0.0893 0.086 0.0168 0.0439 0.0175 0.0196 0.0255 0.4238 0.88 271
2 133 1632 1.12 0.0029 0.0048 0 0.2836 0.0829 0.0738 0.0134 0.0372 0.0156 0.0186 0.0441 0.4231 0.88 252
4 133 1595 1.25 0.0033 0.0022 0 0.2556 0.0687 0.0639 0.0158 0.0403 0.0167 0.0301 0.0433 0.4601 0.88 222
5 135 1500 1.35 0.0035 0.0047 0 0.2652 0.0771 0.0605 0.0102 0.0301 0.0157 0.0251 0.0422 0.4657 0.89 253
6 134 1615 1.24 0.0012 0.0051 0 0.2881 0.0791 0.0646 0.0091 0.0226 0.0084 0.0123 0.0325 0.477 0.88 250
7 134 1400 1.95 0.0016 0.003 0 0.2466 0.0751 0.0692 0.0129 0.0347 0.0102 0.0163 0.0343 0.4961 0.88 239
8 134 1590 1.63 0.0021 0.0015 0 0.2777 0.0768 0.0619 0.0097 0.0244 0.0071 0.011 0.0284 0.4994 0.86 228
9 135 1540 1.16 0.0011 0.0028 0 0.2753 0.069 0.0587 0.0122 0.032 0.0112 0.0135 0.0352 0.489 0.88 245
10 134 1399 1.6 0.0079 0.001 0 0.2479 0.0684 0.065 0.015 0.0387 0.0179 0.023 0.0367 0.4785 0.88 227
11 135 1690 2.81 0.0004 0.0017 0 0.3122 0.065 0.0475 0.0074 0.0179 0.0068 0.0103 0.0264 0.5045 0.89 264
12 134 1548 1.15 0 0.0017 0 0.2856 0.0863 0.0743 0.012 0.0261 0.0096 0.0165 0.0411 0.4468 0.86 249
13 134 1705 1.3 0.0045 0.0008 0 0.2935 0.0785 0.0728 0.017 0.0433 0.0172 0.0309 0.0453 0.3962 0.86 227
14 133 1653 1.5 0.0056 0.0007 0 0.299 0.0784 0.0698 0.0198 0.0407 0.0136 0.0148 0.0266 0.431 0.88 242
15 134 1645 1.01 0.0003 0.003 0 0.278 0.0826 0.0673 0.0115 0.0257 0.0128 0.028 0.042 0.4488 0.88 254
16 132 1751 1.61 0.0006 0.0081 0 0.3134 0.0742 0.0568 0.009 0.0192 0.0065 0.0098 0.0264 0.476 0.89 270
18 132 1025 1.28 0 0.0015 0 0.195 0.0817 0.0843 0.0154 0.0446 0.0121 0.0171 0.0361 0.5122 0.85 225
19 136 1640 1 0.0048 0.0041 0 0.3325 0.0755 0.0621 0.0107 0.0304 0.0102 0.0118 0.0288 0.4291 0.88 236
20 135 1650 0.66 0.0015 0.0019 0 0.3115 0.0892 0.0738 0.0115 0.0292 0.0128 0.022 0.0434 0.4032 0.85 221
21 134 1634 1.25 0.005 0.0087 0 0.2985 0.0733 0.0598 0.0104 0.0244 0.0089 0.0107 0.0428 0.462 0.88 271

11
 Table 2 : (Continued)
Crude T P mexp. Composition
No. (F) ( psi ) ( cp ) N2 CO2 H 2S C1 C2 C3 I-C4 C4 I-C5 C5 C6 C 7+ gC7+ MWC7+
22 134 1533 2.3 0.0068 0.0102 0.0004 0.2549 0.0637 0.0653 0.0147 0.0368 0.0162 0.0326 0.0381 0.4603 0.9 264
23 135 1548 1.74 0.0022 0.0137 0 0.2792 0.0651 0.0516 0.0086 0.0166 0.0074 0.0084 0.0259 0.5213 0.89 255
24 134 1825 3.52 0.0019 0.0251 0 0.3189 0.0751 0.0659 0.0124 0.0353 0.0135 0.0216 0.0329 0.3974 0.92 324
25 134 1825 3.39 0.0018 0.0246 0 0.3026 0.073 0.0636 0.0119 0.0342 0.014 0.0214 0.0322 0.4208 0.91 299
27 133 1565 1.56 0.0044 0.0083 0 0.2775 0.0754 0.069 0.0142 0.0319 0.0148 0.0162 0.0311 0.4572 0.9 272
28 134 1580 1.91 0.0014 0.0054 0 0.2944 0.0796 0.0619 0.0107 0.0289 0.0125 0.02 0.0384 0.4468 0.9 290
30 132 1380 2.01 0.0015 0.0039 0 0.2755 0.0786 0.063 0.0102 0.0257 0.0099 0.0159 0.0368 0.479 0.9 152
31 134 1680 1.9 0.0026 0.0126 0 0.2827 0.06 0.0485 0.0087 0.0273 0.0153 0.0247 0.0426 0.475 0.91 274
32 132 1738 2.02 0.002 0.008 0 0.3142 0.076 0.0723 0.0142 0.0346 0.0089 0.0108 0.0099 0.4491 0.89 251
33 132 1920 1.77 0 0.013 0 0.3453 0.0724 0.0531 0.0087 0.0192 0.0073 0.0118 0.0348 0.4344 0.89 281
34 133 1655 1.86 0.0054 0.005 0 0.2779 0.0689 0.0634 0.0125 0.032 0.0134 0.0195 0.035 0.467 0.89 214
35 135 1880 1 0.0041 0.0065 0 0.3021 0.0689 0.066 0.0163 0.0425 0.0177 0.0285 0.0427 0.4047 0.88 247
36 133 1788 1.54 0.002 0.0119 0 0.3226 0.0757 0.0597 0.0111 0.0274 0.0122 0.0195 0.0363 0.4216 0.88 134
37 133 1805 1.52 0.001 0.0116 0 0.3293 0.0778 0.0606 0.0108 0.0265 0.0114 0.0178 0.037 0.4162 0.88 279
38 135 1500 1.74 0.002 0.0049 0 0.2946 0.0851 0.0663 0.0107 0.029 0.0123 0.0214 0.0379 0.4359 0.9 146
39 135 1500 1.91 0.0015 0.0048 0 0.2851 0.0817 0.0636 0.0107 0.0308 0.0143 0.023 0.0387 0.4459 0.9 142
40 133 1715 1.05 0.0049 0.0047 0 0.3155 0.0866 0.0792 0.0147 0.0401 0.0155 0.0192 0.0445 0.3751 0.9 289
41 130 1670 1.55 0.0015 0.0065 0 0.3048 0.0727 0.0515 0.0069 0.0135 0.0078 0.0118 0.0276 0.4954 0.87 239
42 135 1735 1.48 0.0045 0.0051 0 0.3056 0.0725 0.0728 0.0144 0.038 0.0155 0.0221 0.0385 0.411 0.89 268
43 134 1530 2.1 0.0023 0.0106 0.0001 0.2475 0.0678 0.068 0.0175 0.0426 0.016 0.0275 0.0389 0.4612 0.91 256
44 135 1766 1.99 0.0019 0.0012 0 0.2862 0.0866 0.0652 0.0114 0.0364 0.0114 0.0157 0.0362 0.4478 0.93 274
45 140 1730 3.01 0.0021 0.0072 0 0.3312 0.0744 0.0607 0.0094 0.0228 0.0067 0.0112 0.0305 0.4438 0.9 271
46 168 2505 1.3 0.0008 0.0137 0.0069 0.3597 0.0867 0.0594 0.0085 0.0212 0.0058 0.0092 0.031 0.3971 0.9 274
48 208 1877 0.79 0 0.0112 0.0021 0.2695 0.1076 0.0923 0.0096 0.0401 0.009 0.0172 0.0333 0.4081 0.88 249
49 230 2110 0.49 0 0.0125 0 0.3335 0.0924 0.0607 0.0063 0.0177 0.0053 0.0109 0.0278 0.4329 0.85 252
50 230 3730 0.25 0 0.0068 0 0.4879 0.1057 0.0621 0.0073 0.0273 0.0284 0.0172 0.028 0.248 0.85 218
51 241 3335 0.44 0.0005 0.0085 0 0.4105 0.1107 0.0707 0.0096 0.0386 0.0104 0.0185 0.0319 0.2901 0.85 198
52 240 3630 0.43 0.0006 0.0094 0 0.4444 0.1076 0.0618 0.008 0.0324 0.0091 0.0172 0.0257 0.2838 0.85 195
53 241 3120 0.45 0.0004 0.0078 0 0.4091 0.1072 0.0664 0.0087 0.0357 0.0095 0.0167 0.0251 0.3134 0.85 202
55 243 3180 0.46 0.0006 0.0085 0 0.407 0.1054 0.0655 0.0087 0.0354 0.0094 0.0178 0.0281 0.3136 0.85 195

12
 Table 3: Composition and viscosity of crude oils from UAE

Crude T P mexp. Composition

No. ( F ) ( psi ) ( cp ) N2 CO2 H 2S C1 C2 C3 I-C4 C4 I-C5 C5 C6 C 7+ gC7+ MWC7+

1 215 1261 0.897 0.0025 0.0219 0.0116 0.1633 0.0629 0.0748 0.0156 0.0453 0.0163 0.0273 0.0358 0.5227 0.8803 249
2 190 1140 0.668 0.0024 0.0153 0.0060 0.1316 0.0638 0.0762 0.0169 0.0508 0.0246 0.0319 0.0637 0.5168 0.8764 275
3 239 1490 0.673 0.0017 0.0065 0.0193 0.1259 0.0605 0.0651 0.0070 0.0356 0.0146 0.0306 0.0114 0.6218 0.8766 230
4 239 1592 0.579 0.0032 0.0369 0.0068 0.2155 0.0860 0.0766 0.0114 0.0526 0.0219 0.0288 0.0262 0.4341 0.8687 243
5 230 993 0.750 0.0043 0.0347 0.0368 0.1949 0.0828 0.0685 0.0116 0.0314 0.0191 0.0227 0.0242 0.4690 0.8764 246
6 235 901 0.813 0.0021 0.0034 0.0000 0.2004 0.0793 0.0800 0.0193 0.0467 0.0252 0.0335 0.0508 0.4593 0.8606 230
7 234 1190 0.675 0.0077 0.0199 0.0140 0.1738 0.0642 0.0762 0.0125 0.0437 0.0219 0.0234 0.0514 0.4913 0.8912 267

Table 4: Composition and viscosity of crude oils from North Sea

Crude T P mexp. Composition

No. ( F ) ( psi ) ( cp ) N2 CO2 H 2S C1 C2 C3 I-C4 C4 I-C5 C5 C6 C 7+ gC7+ MWC7+

1 238 2753 0.380 0.0067 0.0211 0.0000 0.3493 0.0700 0.0782 0.0122 0.0426 0.0145 0.0235 0.0304 0.3515 0.8528 226
2 200 3981 0.404 0.0034 0.0084 0.0000 0.4923 0.0632 0.0446 0.0086 0.0218 0.0093 0.0133 0.0206 0.3145 0.8650 230
3 199 3926 0.320 0.0044 0.0038 0.0000 0.4908 0.0760 0.0613 0.0097 0.0287 0.0102 0.0150 0.0200 0.2803 0.8564 231
4 166 3394 0.425 0.0090 0.0016 0.0000 0.4712 0.0597 0.0462 0.0099 0.0250 0.0109 0.0146 0.0219 0.3300 0.8497 217
5 156 3813 1.120 0.0036 0.0106 0.0000 0.5050 0.0454 0.0090 0.0055 0.0060 0.0062 0.0027 0.0160 0.3900 0.8972 291
6 160 2305 2.100 0.0033 0.0019 0.0000 0.3542 0.0336 0.0090 0.0069 0.0026 0.0026 0.0014 0.0072 0.5773 0.9165 255
7 208 2952 0.299 0.0041 0.0044 0.0000 0.4024 0.0769 0.0815 0.0122 0.0420 0.0142 0.0220 0.0281 0.3123 0.8448 210

13
 *
Table 5: Composition and viscosity of other crudes

Crude T P mexp. Composition

No. ( F ) ( psi ) ( cp ) N2 CO2 H 2S C1 C2 C3 I-C4 C4 I-C5 C5 C6 C 7+ gC7+ MWC7+

1 145 1297 0.660 0.0000 0.0037 0.0000 0.2289 0.0059 0.0141 0.0077 0.0071 0.0094 0.0012 0.0241 0.6979 0.8660 167
2 146 1162 0.550 0.0000 0.0013 0.0000 0.2218 0.0045 0.0061 0.0067 0.0088 0.0091 0.0067 0.0429 0.6921 0.8378 162
3 94 696 2.620 0.0104 0.0000 0.0000 0.1303 0.0834 0.0899 0.0118 0.0598 0.0188 0.0332 0.0539 0.5085 0.8628 274
4 86 1275 1.620 0.0147 0.0019 0.0000 0.2546 0.0362 0.0358 0.0098 0.0345 0.0242 0.0228 0.0735 0.4920 0.8639 230
5 156 1650 0.660 0.0004 0.0003 0.0000 0.3018 0.0264 0.0123 0.0085 0.0064 0.0060 0.0046 0.0299 0.6034 0.8363 188
8 282 8590 0.110 0.0038 0.0164 0.0000 0.7257 0.0541 0.0307 0.0066 0.0134 0.0062 0.0054 0.0108 0.1269 0.8778 273
9 218 4040 0.530 0.0009 0.0022 0.0000 0.5105 0.0345 0.0139 0.0053 0.0067 0.0041 0.0023 0.0218 0.3978 0.8692 246
10 250 80 1.490 0.0010 0.0170 0.0000 0.0290 0.0060 0.0030 0.0020 0.0040 0.0040 0.0040 0.0280 0.9020 0.9042 229
11 112 335 1.130 0.0050 0.0150 0.0000 0.0690 0.0770 0.0820 0.0200 0.0630 0.0210 0.0260 0.0830 0.5410 0.8789 209
14 110 900 1.530 0.0560 0.0040 0.0000 0.1350 0.0670 0.0630 0.0150 0.0340 0.0230 0.0230 0.0670 0.5130 0.8877 231
15 164 945 0.750 0.0060 0.0060 0.0000 0.1480 0.0610 0.0920 0.0190 0.0650 0.0320 0.0320 0.0650 0.4750 0.8670 216
16 194 1090 1.370 0.0060 0.0040 0.0000 0.2400 0.0140 0.0280 0.0060 0.0240 0.0120 0.0160 0.0790 0.5700 0.8602 231
17 274 2180 0.440 0.0160 0.0160 0.0000 0.3510 0.0110 0.0080 0.0060 0.0140 0.0130 0.0170 0.0560 0.4920 0.8246 219
18 223 3665 0.450 0.0000 0.0010 0.0000 0.5070 0.0550 0.0320 0.0100 0.0150 0.0070 0.0070 0.0200 0.3470 0.8493 252

* Lawal (1986)

Table 6: Summary of the main attempts in the empirical approach

Attempt N v* Function form COR** ARD% AAD%

No.
1 5 m G,P,T(f =LG,mG,h) 0.75114 8.42 31.28

2 8 m G,P,T(f =LG,mG,hG,non, 0.863282 4.84 24.13

g+7C, wM+7c)
3 6 m , ,P,T(f =g+7C, GLG ,mG ,h) 0.859674 5.05 25.09

* Nv : No of Independent variables
** COR: Coefficient of regression

14
Table 7: Comparison of the present study to Guo’s work for gas condensates and crude oils*

Gas
Crude oils
Condensates
Data ranges: T – range, oF, P- range, psimexp, cp
† 110 – 262 86- 282
† 2000 – 5750 80 – 8590
† 0.023 – 0.07 0.11 – 2.62
PR (Kij = 0) 15.04 -
MPR (Kij = 0) 9.81 16.32
Pedersen - 33.37
Guo
Lohrenz - 167.93
Little & Kennedy - 47.25

PR (Kij = 0) 18.5 58.95

AAD% MPR (Kij = 0) 16.66 17.7
MPR (Kij#0) 16.77 17.73
Present
Study Pedersen - 27.96
Lohrenz 15.9 26
Little & Kennedy - 29.43

* Data source: Lawal (1986)

15
 *
Table 8: Summary of results in estimated oil viscosity using various models

o
Data Source T- range ( F) P-range Np AAD%
(Psi) PR(Kij=0) MPR(Kij=0) MPR(Kij#0) Pedersen Little Lohrenz
a
Unknown (86-282) (80-8590) 14 58.95 17.7 17.73 27.96 26 29.43
b
UAE (Abu Dhabi) (190-239) (901-5000) 53 67.84 24 24.16 36.4 35.13 72.9
c
North Sea Crudes (156-238) (115-5829) 99 62.96 16.93 17.06 36.74 174.34 67.33
d
Kuwaiti Crudes (130-243) (14.7-5000) 146 70.68 45.71 45.8 42.31 103.78 51.37
Kuwaiti Crudes @B.P. (30-243) (1365-3730) 49 73.41 35.76 36.1 38.71 30.83 43.83
Overall 361 68.06 32.2 32.34 38.87 100.13 57.03

* Set 1 Tc – Pc-w correlation was used for C7+ - fraction

Data references: a. Lawal (1986)
b. Moharam (1995)
c. Pedersen (1984)
d. This study
Data references: a. Lawal (1986)
b. Moharam (1995)
c. Pedersen (1984)
d. This study

Table 9: Comparison Between Proposed Models and Guo’s Work

Data Np Guo’s Model (%AAD) Present Study (%AAD)

Empirical correlation Modified EOS-based viscosity model
Kuwait 49 45.71 23.02 23.73
All 361 32.2 24.13 ---

* Pc , Tc, w for C7+ are calculated by characterization set #2

16
 40

35

30

25
Overall % AAD

20

15

10

0
Elsharkawy & Alikhan Beggs&Robinson Labedi Kartoatmodjo&Schmidt

Fig.1. Error distribution by various correlations for estimating the viscosity of

Kuwaiti Crudes

7 exp
ELSHARKAWY

6 Beggs&Robinson
Labedi
Oil Viscosity, (cp)

Kartoatmodjo&Schmidt
5

0
0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000
Pressure, (psi)

Fig.2. Measured and predicted viscosity for Kuwaiti crude sample #16

17
 60

50

40
% AAD

30

20

10

0
Pedersen Adjusted Pedersen Petersen

Fig.3. Error in viscosity calculation by corresponding-states models for Kuwaiti Crudes @

the bubble point

4
Guo MPR model(2001)

Proposed Empirical Correlation

3
Estimated viscosity, cp

0
0 1 2 3 4
Measured viscosity, cp
Fig. 4 Comparison between measured and estimated viscosity for Kuwaiti crudes.

18
 4
Guo MPR model(2001)

Proposed Modified MPR

Estimated viscosity, cp
2

0
0 1 2 3 4
Measured viscosity, cp
Fig.5. Comparison between measured and estimated viscosity for Kuwaiti crudes.

2.5

EXP.

2.0 GUO

Proposed Modified MPR

Oil viscosity, (cp)

Proposed Empirical Correlation

1.5

1.0

0.5

0.0
0 1000 2000 3000 4000 5000 6000
Pressure, (psi)
Fig. 6. Measured and predicted viscosity for Kuwaiti crude sample #55

19