T r e n d s i n
C h e m i c a l
A comparative analysis of the performance of various
equations of state in thermodynamic property prediction
of reservoir fluid systems
Faith U. Babalola* and David O. Oduwole
Department of Chemical Engineering, University of Lagos, Lagos, Nigeria.
ABSTRACT
Six equations of state (including ours) were
compared for their thermodynamic property
prediction accuracy for reservoir fluid systems.
This was done using their percentage absolute
deviation to search for and establish a trend in
accuracy for each on conventional wells which
will give a clue to their applicability to HPHT
(High Pressure High Temperature) PVT modeling.
A group of three showed fairly good accuracy
with a trend of improvement up to about 1800 psi.
Another set of two interestingly showed very
similar trends with continuous improvement in
accuracy as pressure increased making them
potential candidates for high pressure reservoir
PVT modeling. One equation of state (EOS)
however showed an irregular error trend.
KEYWORDS: equation of state, prediction,
liquid volume, compressibility factor
1. Introduction
An equation of state is a mathematical expression
describing the relation between various
macroscopically measurable state properties of a
system. Many EOS have been developed and
modified to improve accuracy of prediction of
properties and are currently widely used in the oil
and gas as well as in other chemical process
industries.
*Corresponding author: fu_babalola@yahoo.com
Engineering
Vol. 15, 2014
The most prominent use of an equation of state is
to predict the state of gases and liquids. One of
the simplest and oldest equations of state for this
purpose is the ideal gas law, which is roughly
accurate for real gases at low pressures and
moderate temperatures. However, this equation
becomes increasingly inaccurate at higher pressures
and lower temperatures, and fails to predict
condensation from a gas to a liquid. Therefore, a
number of much more accurate equations of state
have been developed for gases and liquids. At
present, there is no single equation of state that
accurately predicts the properties of all fluids
under all conditions. This work is limited to EOS
that predict properties of reservoir fluids at
equilibrium. Since these EOS are semi-empirical,
in the sense that they have some theoretical basis,
their parameters must be adjusted. The number
of parameters determine the category/complexity
of the EOS. In general terms, the more complex
the EOS, the more accurate it is (with a few
exceptions). In some cases a rather simple EOS
can do a very good job.
Since the time of the ideal gas law (ideal gas
EOS), a great number of equations of state have
been proposed to describe real gas behaviour.
However, many of them have not passed the test
of time. Only few have persisted through the
years; this is because of their relative simplicity.
In the petroleum industry, the most common
modern EOS are the Peng-Robinson EOS (PR
EOS), Soave-Redlich-Kwong EOS (SRK EOS),
Patel-Teja EOS and a few more recent versions.
48
Most of these are cubic and hence are derivations
of the van der Waals EOS.
Fluid properties are essential factors to be
considered when dealing with equipment design
in petroleum reservoir engineering and chemical
engineering. Accurate values of these fluid
properties such as volumes, viscosities, densities,
etc. are required in solving design problems. In
most cases, experimental measurements of these
important fluid properties are difficult or
impossible to obtain directly. The alternative way
of obtaining the values of these properties is
empirical prediction through mathematical expression
of equations of state. The application of EOS
models in predicting fluid properties is preferred
because it is flexible, rigorous and useful for
describing complex systems such as hydrocarbon
systems.
There are various equations of state which are
useful in the prediction of fluid properties. The
accuracy of each equation of state depends on
the nature and condition of the fluid involved.
Adequate knowledge of this will definitely make
design work easy and more efficient.
The analysis of the performance of various
equations of state can be achieved if the available
experimental thermodynamic data have high
accuracy. The source of such experimental data
and the state of experimental equipment used to
obtain the data are of great concern. There are
little or no information, in most cases, on the
accuracy of the experimental equipment used to
obtain most available thermodynamic data. Also
the purity of fluid substances used in the
experiment of most thermodynamic data is not
always available with their data.
In this work, the performance of six equations of
state developed over a period of three decades for
use in the petroleum industry were analyzed and
compared using two thermodynamic properties -
volume and compressibility factor. The six
selected EOS are:
1. Peng-Robinson Equation of State (PR EOS)
of 1976.
2. Patel-Teja Equation of State (PT EOS) of
1982.
3. Elliott-Suresh-Donohue Equation of State
(ESD-EOS) of 1990.
Faith U. Babalola & David O. Oduwole
4. Erdogmus-Adewumi Equation of State
(EA-EOS) of 2000.
5. Esmaeilzadeh-Roshanfekr Equation of State
(ER-EOS) of 2004.
6. Babalola-Peng-Robinson Equation of State
(BPR-EOS) of 2005.
2. Brief literature review
The progress in the development of equations of
state has spanned a few centuries. Beginning with
the ideal gas laws by Boyle in 1662 and that by
Charles in 1787 after a period of more than a
century, Dalton’s law quickly followed in 1801
and then the most popular ideal gas law (still in
use today) by Emile Clapeyron in 1834. The
scope was expanded to include real gases by the
inclusion of parameters, the first of which was by
van der Waals [1]. His finite volume assumption
for real gases ended the work on EOS for ideal
gases and charted a new course for the application
of equations of state to real gases and liquids. A
well-improved EOS was developed by Redlich
and Kwong [2], modified by Soave [3] and was
very closely followed by the celebrated Peng-
Robinson EOS [4]. The following three decades
saw numerous modifications and development of
EOS, five of which have been selected in addition
to the Peng-Robinson EOS for performance
assessment in this work using percentage absolute
deviation analysis.
Peng-Robinson equation of state
A cubic three-parameter equation of state, developed
by Peng and Robinson is well established in
literature [4].
Patel-Teja equation of state
Patel and Teja [5] proposed an equation of state
that improves saturation properties prediction,
especially for heavy hydrocarbons and polar
substances, while maintaining the good features
of PR and SRK equations; it introduces a third
constant, ‘c’ into the van der Waals equation’s
attraction term. This equation is also well established
and available in literature.
Elliott-Suresh-Donohue equation of state
The Elliott-Suresh-Donohue (ESD) equation of
state was proposed in 1990 [6]. The equation
Performance of equations of state in reservoir fluids 49

seeks to correct a shortcoming in the Peng- Erdogmus-Adewumi equation of state

Robinson EOS due to the inaccuracy in the van
der Waals repulsive term. The EOS accounts for
the effect of the shape of a non-polar molecule
and can be extended to polymers with the addition
of an extra term. The EOS itself was developed
through modelling and computer simulations and
was meant to capture the essential physics of the
size, shape, and hydrogen bonding. It expresses
the compressibility factor (Z) as
The Erdogmus-Adewumi equation of state is an
extension of that of Patel and Teja. They modified
the Patel-Teja EOS to improve the prediction of
liquid phase density and vapor liquid equilibrium
[7]. The equation is written as
P=   (7)
where

(1) (8)
a = Ωaa( ) .α

where (9)
b = Ωb ( )

  (2)
c = Ωc ( ) (10)

  (3) α = {1 + m(1 - TR0.5)} 2 (11)

and m according to Erdogmus and Adewunmi is given

c is a “shape factor”, with c = 1 for spherical as:
molecules. m = KZc + n (12)
For non-spherical molecules the following
relation is suggested:
c = 1 + 3.535ω + 0.533ω2 where ω is the acentric   (13)
factor K= 1.466867 + 3.400456 + 3.776977 (14)
The reduced number density η is defined as
n = 0.283470 + 0.393438 –   (15)
  (4) Esmaeilzadeh-Roshanfekr equation of state
Here, there is an introduction of a substance-
where dependent critical compressibility factor which
v* is the characteristic size parameter allowed a more accurate reproduction of the
n is the number of molecules experimental saturated liquid density at a specified
V is the volume of the container temperature [8].
The characteristic size parameter is related to the Their work showed that the optimum value of
shape parameter c through substance critical compressibility was not equal to
the experimental critical compressibility of the
fluid of interest. Using the generalized van der
  (5) Waals theory, Esmaeilzadeh and Roshanfekr used
a mathematically simple cubic EOS to model
where attractive interactions between molecules. The
proposed equation of state in their work is as
  (6) follows:

 
and k is Boltzmann’s constant. (16)
50 Faith U. Babalola & David O. Oduwole

where “a” is a function of temperature and “b” predict two thermodynamics properties namely
and “c” are constants. In this EOS, a(T), ‘b’ are Compressibility factor and Specific Volume. In
given as: order to conveniently use a computer simulation
technique each equation had to be transformed
(17) into a machine-compliant form. The computer
simulations provide the required predictions.
  (18) The predicted values were compared with
experimental values (actual field data) and the
resulting deviations were critically analyzed and
 
(19) interpreted to illustrate the performances of these
equations of state.
Moreover, it is found that Esmaeilzadeh-
Roshanfekr EOS is most accurate for predicting 3. Calculations
gas-condensate properties, while the original SRK For a property prediction model, it is most
and PR equations remain reliable for oil samples convenient to transform each EOS into a
[9]. polynomial expression in terms of the property
to be predicted. This transformation involves
Babalola-Peng-Robinson equation of state algebraic mathematical manipulations and is no
Babalola [10] modified the Peng-Robinson equation simple task. The six EOS under investigation
of state as: were so transformed and are given below, firstly
in terms of molar volume (V) and secondly in
P=   (20) terms of compressibility factor (Z).

The attractive force parameter ‘a’, which was 3.1. Modeling

considered to be a constant with respect to Peng-Robinson (P-R) EOS
pressure in Peng-Robinson EOS as well as in
other EOS was modified by Babalola. The  V 
modification, in which parameter ‘a’ was treated
as a variable with respect to pressure for multi-   (22)
component multi-phase high density reservoir
systems was to improve accuracy in the prediction and
of properties of complex reservoir fluids for PVT
modeling. The variation was represented with a Z 
linear equation:
  (23)
a = α1 P + α2 (21)
where α1 and α2 represent the slope and the where
intercept respectively of the straight line obtained
when values of parameter ‘a’ are plotted against and
pressure.
This modification increases the accuracy of Peng- Patel-Teja (P-T) EOS
Robinson EOS for heavier fluids particularly at
high pressure ranges (> 30 MPa) and eliminates
the need for tuning the EOS before use thereby
minimizing the cost of using the EOS in the oil
industry.  
(24)
The comparison of the performance of these
equations was done by using each equation to and
Performance of equations of state in reservoir fluids 51

temperature-dependent parameters ‘α’ and ‘m’

which are contained in ‘A’.
(25)
Esmaeilzadeh-Roshanfekr (E-R) EOS
where

  and  
 
(32)
Elliot-Suresh-Donohue (E-S-D) EOS
and
Z = 1 +   +    (26)

where
  (33)
Z =   
where

 =    (27) and  

  (28) Babalola-Peng-Robinson (B-P-R) EOS

and  V 

  (34)

and

Z 
  (29)
  (35)
Erdogmus-Adewunmi (E-A) EOS where

and  

  (30)
These transformed EOS lend themselves easily to
and computer programs; they were solved using
MATLAB.
3.2. Mixing rules
  (31)
With various mixing rules available in literature,
where the choice here is arbitrary since the systems
whose properties are here predicted are normal
  and   reservoir fluids and whichever one is used will be
applied to all six EOS and mixing rules are not
Erdogmus-Adewunmi EOS in this form is EOS-specific. Therefore, the van der Waals I-fluid
identical to Patel-Teja EOS of which it is a mixing rules were used in this work and are given
modification. Their difference lies in the modified for parameter ‘a’, ‘b’ and ‘c’ as
52
  (36)
  (37)
  (38)
where
  (39)
The Mixing Rules for parameter ‘a’ were not
employed in the Babalola-Peng-Robinson EOS,
rather, the EOS uses a variable parameter ‘a’
approach (VPAA) as fully discussed by Babalola
and Susu [11].
3.3. Comparative analysis using percentage
absolute deviation trend
The predicted property values were compared
with experimental (field) data by calculating the
percentage absolute average deviation (%AAD)
for a set of N values. The standard %AAD is
defined mathematically as
(40)
N = number of data points used
For a closer observation and monitoring of the
curves generated using each of the six EOS
under study, the percentage absolute deviation
(%AD) instead of %AAD from the field data was
found for each pressure value (point) and these
deviations were plotted against pressure. This is
more revealing for our purpose than determining
the percentage absolute deviation %AAD which is
a lump value.
3.4. Computer simulation
The program, whose flow chart is shown in Figure 1
below was used along with our input data and
parameters for three reservoir fluid systems to
calculate the liquid molar volumes and the
compressibility factors for the three reservoir fluid
systems used in this work, namely Data 1, Data 2
and Data 3. Following the flow chart in Figure 1,
Faith U. Babalola & David O. Oduwole
the results generated by the program are presented
in Section 4.
4. Results and discussion
The predicted liquid volume for Data 1 as shown
in Figure 2 reveals that PR-EOS and PT-EOS
predicted almost exactly the same values with a
near match by EA-EOS throughout the range
spanned by the data. They deviated more from the
actual (experimental) values at lower pressures
than at higher ones. BPR-EOS and ESD-EOS
followed the same trend with ESD-EOS
predicting slightly higher values. Their deviations
from actual values are quite high at lower
pressures but reduced steadily with pressure
increase. ER-EOS deviated from experimental
values at higher and lower pressures but showed
its best prediction with lowest deviation at a
pressure of about 20 MPa. Very similar observations
are made in the performance of all six EOS in
predicting the compressibility factor of Data 1 as
shown in Figure 3.
In Figure 4, the volume prediction for Data 2
shows again that BPR-EOS and ESD-EOS
behaved alike, PR-EOS, PT-EOS and EA-EOS
behaved alike while ER-EOS showed large
deviations at low pressures but improved slightly
with pressure increase. This trend is close to that
seen in Figure 5 where compressibility factor was
predicted for Data 2. Data 3 however involves
temperature variation (an increase) with pressure
increase. Liquid volume and compressibility factor
predictions (Figures 6 and 7) still maintained
the grouping of the EOS earlier observed for
constant-temperature cases; PR-EOS, PT-EOS
and EA-EOS closely followed the experimental
values, BPR-EOS and ESD-EOS followed very
similar curves with BPR-EOS giving a better
prediction throughout the pressure range. ER-EOS
however showed a point of lowest accuracy at
about 20 MPa.
For our particular interest, rather than calculate
the commonly used %AAD (percentage average
absolute deviation), it was preferable to determine
the %AD (percentage absolute deviation) for each
predicted point and plot to reveal the trend. The
percentage absolute deviation from experiment for
the six EOS are shown in Figure 8 and Figure 9
Performance of equations of state in reservoir fluids 53

BEGIN

Read ω, Tc, Pc, R, Experimental V & Z

Calculate Zc, K & n

Calculate TR, m & α
Calculate Ωa, Ωb & Ωc
Calculate a, b & c
Calculate A, B & C

Calculate V
Calculate Z

V & Z = Real
Positive
Root

PRINT

Calculate AAD

PRINT

STOP

END

Figure 1. Flow chart for the program.

54 Faith U. Babalola & David O. Oduwole

Figure 2. Volume prediction for Data 1.

Figure 3. Compressibility factor prediction for Data 1.

Figure 4. Volume prediction for Data 2.

Performance of equations of state in reservoir fluids 55

Figure 5. Compressibility factor prediction for Data 2.

Pressure (MPa)

Figure 6. Volume prediction for Data 3.

Figure 7. Compressibility factor prediction for Data 3.

56 Faith U. Babalola & David O. Oduwole

(V)

Figure 8. Percentage absolute deviation for predicted volume for Data 1.

Figure 9. Percentage absolute deviation for predicted compressibility factor for Data 1.

Figure 10. Percentage absolute deviation for predicted volume in Data 2.

Performance of equations of state in reservoir fluids 57

Figure 11. Percentage absolute deviation for predicted compressibility factor for Data 2.

Figure 12. Percentage absolute deviation for predicted volume for Data 3.

25
Percentage Absolute Deviation

20

15

10

Figure 13. Percentage absolute deviation for Z values obtained from Data 3.
58 Faith U. Babalola & David O. Oduwole

for Data 1, Figure 10 and Figure 11 for Data 2 and m : parameter in the EOS for α
finally in Figure 12 and Figure 13 for Data 3. All n : number of molecules
EOS maintained their signatures even for Data 3 N : the number of data points
which had temperature variation except ER-EOS η : reduced number density
which had a minimum for Data 1, a maximum for Φ : shape parameter
Data 2 and no turning point for Data 3. PR-EOS, P : pressure
PT-EOS and EA-EOS as a group began with a R : ideal gas constant
%AD of about 8, dropped steadily to about 3.5 at T : temperature
about 10 MPa and remained almost constant with V : volume
pressure increase. The group of BPR-EOS and v* : the characteristic size parameter
ESD-EOS has an interesting feature. Their %AD ω : acentric factor
began though with a very high value of about 22 γ : adiabatic index
and 25, respectively, but a continuous decline was Zc : critical compressibility factor
maintained as pressure increased for Data 1 and Zrep : repulsive parameter
Data 2. This leads to an expectation that %AD Zatt : attractive parameter
will keep reducing even as pressure values Z : compressibility factor
increase beyond the range covered in this work. Ωa, Ωb, Ωc : parameter in the a, b, c
Even for the variable temperature case, Data 3, constants of the EOS
these two EOS also show a decline in %AD at Subscripts
high pressure. It is worth noting here that the c : critical condition
BPR-EOS was developed for use at moderate and r : reduce condition
possibly high pressures. m : molar value

5. Conclusion REFERENCES
An apparent grouping of the EOS into three can 1. van der Waals, J. D. 1873, doctoral dissertation.
be observed. PR-EOS, PT-EOS and EA-EOS 2. Redlich, O. and Kwong, J. N. S. 1949, Rev.,
predicted with reasonable accuracy at low pressures 44, 233-244.
but did not show any improvement with pressure 3. Soave, G. 1972, Chem. Eng. Sci., 27, 1197-
increase. BPR-EOS and ESD-EOS showed a very 1203.
similar and promising %AD trend as their 4. Peng, D. Y. and Robinson, D. B. 1976, Ind.
deviation from actual values maintained a steady Eng. Chem. Fundam., 15, No 1. 54, 59-64.
decline with pressure increase. This recommends 5. Patel, N. C. and Teja, A. S. 1982, Chem.
them for higher pressure PVT modeling as will be Eng. Sci., 77(3), 463-473.
demonstrated in a subsequent work. 6. Elliott Jr., R. J., Suresh, S. J. and Donohue,
M. D. 1990, Ind. Eng. Chem. Res., 29(7),
CONFLICT OF INTEREST STATEMENT 1476-1485.
There are no conflicts of interest in relation to this 7. Erdogmus M. and Adewumi, M. A. 2000,
manuscript. SPE Conf., Oct 17-19, Morgan Town.
8. Esmaeilzadeh, F. and Roshanfekr, M. 2004,
9th Iranian Chemical Engineering Congress.
ABBREVIATIONS
Nov. 23-25, Iran Univ. of Sci. & Tech., 522-533.
AAD% : absolute average deviation 9. Ibrahim, A., Nabeel, A., Amin, F. and
percentage Gholamreza, V. N. 2011, www.intechopen.com
a, b, c : parameters in the EOS 165-178.
A, B, C : parameters in the EOS 10. Babalola, F. U. 2005, PhD Thesis, Department
α : reduced temperature function of Chemical Engineering, University of
in the EOS Lagos, 171-193.
k : Boltzmann’s constant 11. Babalola, F. U. and Susu, A. A. 2008, Pet.
kij : the binary interaction parameter Sci. and Tech., 26(13), 1522-1544.