Report on density estimation of pure Carbon dioxide and water at

subcritical, critical and supercritical regions and mixtures of carbon

dioxide and water at different compositions: Approach based on
comparative study through Van der Waal, Soave-Redlich-Kwong and
Peng Robinson EOS Modeling.

VIGNESH SATHYASEELAN ANKITHA RAMACHANDRAN

JUI JUNNARKAR MANN MANOHAR VERLEKAR
SAI BHARADWAJ

*THE LAST 3 PAGES HAS THE ANALYSIS THE OTHER PAGES INCLUDE DATA AND CODES*
INTRODUCTION

As one of the most popular materials used in supercritical technologies, carbon dioxide has been used in different
industries. More recently, CCS (Carbon Capture and Sequestration) has achieved increasing popularity all over the world
for mitigating the contribution of fossil fuel emissions to global warming (a.k.a. climate change). In order to
appropriately design and optimally operate a CCS project, the accurate representation of carbon dioxide density – is a
must. Density is one of the critical parameters required for accurately predicting wellbore and pipeline hydraulics and
well injectivity for CCS projects. Carbon dioxide density can be calculated through a black oil model, an equation of state
(EOS), or empirical correlations. The primary EOS that have been applied to carbon dioxide includes Peng-Robinson ,
Soave-Redlich-Kwong (a.k.a. SRK) , and Span & Wagner. Unfortunately, it has been found that these correlations may
not be very accurate in terms of predicting the carbon dioxide density under supercritical conditions, especially under
the conditions to be anticipated for CCS applications.
In the present report, the authors have modelled the Van der Waals (VDW), Soave-Redlich-Kwong (SRK) EOS and Peng
Robinon (PR), and calculated the energy parameter (a) and molecular volume (b) using appropriate expressions in terms
of critical properties. The EOS were then solved for the density roots and calculated the deviation of the values obtained
from the high accuracy carbon dioxide and water property data provided by the National Institute of Standards and
Technology (NIST).

APPROACH
A PYTHON code was written to calculate the volume roots of the VDW, SRK and PR EOS. They are cubic equations and
were solved by iterative Newton- Raphson Method. Through the NIST database carbon dioxide and water properties
have been generated under pressure and temperature ranges expected for CCS operation. The thermodynamic property
data were obtained from research papers.The parameters that were changed to simulate different states, such as sub-
critical, critical and supercritical regions, are Pressure and Temperature.
For VDW, whose energy parameter and molecular co-volume depend entirely on the critical constants, the values were
calculated and were modelled into the EOS as required.
For SRK EOS, there were a number of parameters that needed to be determined prior to solving the roots. This EOS
represents an attempt to take molecular geometry and polarity into account. The acentric factors were obtained from a
web search that led the authors to model that into the EOS as required. The critical temperature and pressure were
known from a thermodynamic datebook.
For the most part the Peng–Robinson equation exhibits performance similar to the Soave equation, although it is
generally superior in predicting the liquid densities of many materials, especially nonpolar ones
After obtaining the calculated values over different regions such as subcritical, critical and supercritical, these values
were compared to experimentally determined values from literature search. Statistics data such as Average Absolute
Deviation (AAD %), minimum error and maximum error are also provided. The accuracy of prediction of density values is
discussed in the results and discussion section.
CODE FOR PURE SUBSTANCES
Code for PR EOS for Pure Substances:
Tc=304.25
Pc=73.773
R=0.0831451
Tr=T/Tc
w=0.223
S=0.37464+1.54226*w-0.26992*(w**2)
M=(1+S*(1-Tr**0.5))**2
a=(0.45724*(R**2)*(Tc**2)*M)/Pc
b=(0.07780*R*Tc)/Pc
x=(R*T)/P;
for i in range (0,50):
f=x**3+x**2*((b*P-R*T)/P)-x*((3*P*b**2+2*b*R*T-a)/P)+((P*b**3+b**2*R*T-a*b)/P)
#this is the derivative of the above function
df=3*x**2+2*x*((b*P-R*T)/P)- ((3*P*b**2+2*b*R*T-a)/P)
x1=x-(f/df)
x=x1

sol=x/44

print('Approximate specific volume is ',sol)

final= 1/sol

print('Approximate density is ',final)

Code for VDW for Pure Substances:

a=5.536
b=0.03049
R=0.0831451
x=(R*T)/P
i=0
for i in range(0,50):
f=x**3-x**2*(b+(R*T)/P)+(a/P)*x - (a*b)/P
#this is the derivative of the above function
df=3*x**2 -2*x*(b-(R*T)/P)+(a/P)
x1=x-(f/df)
x=x1

sol=x/18

print('Approximate specific volume is ',sol)

final=1/sol

print('Approximate density is ',final)

Code for SRK EOS for Pure Substances

Tc=304.25
Pc=73.773
R=0.0831451
Tr=T/Tc
w=0.223
S=0.48+1.574*w-0.176*(w**2)
M=(1+S*(1-Tr**0.5))**2
a=(0.42748*(R**2)*(Tc**2)*M)/Pc
b=(0.08664*R*Tc)/Pc
x=(R*T)/P;
for i in range (0,50):
f=x**3-x**2*((R*T)/P)-x*(b**2+(R*T*b)/P-a/P)-(b*a)/P;
#this is the derivative of the above function
df=3*x**2-2*x*((R*T)/P)-(b**2+(R*T*b)/P-a/P);
x1=x-(f/df)
x=x1

sol=x/44

print('Approximate specific volume is ',sol)

final= 1/sol

print('Approximate density is ',final)

CODE FOR MIXTURES

c = 0.0086 #composition of mixture

Code for VDW for mixtures :

#x= specific volume (m^3/kg)

#Change here for different functions
a=((1-c)**2)*5.536+(c**2)*3.64+2*c*(1-c)*((5.536*3.64)**(1/2)*(1/2))# For CO2 a=3.64
b=(1-c)*0.03049+c*0.04267#For CO2 b=0.04267
R=0.0831451
x=0.001
i=0
for i in range(0,50):
f=x**3-x**2*(b+(R*T)/P)+(a/P)*x - (a*b)/P
#this is the derivative of the above function
df=3*x**2 -2*x*(b-(R*T)/P)+(a/P)
x1=x-(f/df)
x=x1

sol=x/((1-c)*18+(c*44))

print('Approximate Root is ',sol)

final=1/sol

print('Approximate Root is ',final)

Code for SRK EOS for Mixtures:

Tc=(1-c)*647.096+c*304.25 # For CO2 Tc=304.25 K

Pc=(1-c)*220.64+c*(73.773)# For CO2 Pc=73.773
Tcc = Tc=304.25
Pcc = 73.773
Tcw = 647.096
Pcw = 220.64
R=0.0831451
Tr=T/Tc
w= (1-c)*0.344+(c)*0.223 # For CO2 w=0.223
S=0.48+1.574*w-0.176*(w**2)
M=(1+S*(1-Tr**0.5))**2
aw = (0.45724*(R**2)*(Tcw**2)*M)/Pcw
ac = (0.45724*(R**2)*(Tcc**2)*M)/Pcc
a = (c**2)*ac+((1-c)**2)*aw+2*(1-c)*c*((aw*ac)**(1/2)*(1-0.5))
b=(1-c)*((0.08664*R*Tcw)/Pcw)+c*((0.08664*R*Tcc)/Pcc)

#Temperature and pressure are critical value

x=0.001

for i in range (0,50):

f=x**3-x**2*((R*T)/P)-x*(b**2+(R*T*b)/P-a/P)-(b*a)/P;
#this is the derivative of the above function
df=3*x**2-2*x*((R*T)/P)-(b**2+(R*T*b)/P-a/P);
x1=x-(f/df)
 x=x1

sol=x/((1-c)*18+c*44)

print('Approximate Root is ',sol)

final= 1/sol

print('Approximate density Root is ',final)

Code for PR EOS for Mixtures:

Tc=(1-c)*647.096+c*304.25 # For CO2 Tc=304.25 K
Pc=(1-c)*220.64+c*(73.773)# For CO2 Pc=73.773
Tcc = 304.25
Pcc = 73.773
Tcw = 647.096
Pcw = 220.64
R=0.0831451
Tr=T/Tc
w= (1-c)*0.344+(c)*0.223 # For CO2 w=0.223
S=0.37464+1.54226*w-0.26992*(w**2)
M=(1+S*(1-Tr**0.5))**2
aw = (0.45724*(R**2)*(Tcw**2)*M)/Pcw
ac = (0.45724*(R**2)*(Tcc**2)*M)/Pcc
a = (c**2)*aw+((1-c)**2)*ac+2*(1-c)*c*((aw*ac)**(1/2)*(1-0.5))
b=(1-c)*((0.07780*R*Tcw)/Pcw)+c*((0.07780*R*Tcc)/Pcc)
#Temperature and pressure are critical value

x=0.001

for i in range (0,50):

f=x**3+x**2*((b*P-R*T)/P)-x*((3*P*b**2+2*b*R*T-a)/P)+((P*b**3+b**2*R*T-a*b)/P)
#this is the derivative of the above function
df=3*x**2+2*x*((b*P-R*T)/P)- ((3*P*b**2+2*b*R*T-a)/P)
x1=x-(f/df)
x=x1

sol=x/((1-c)*18+c*44)

print('Approximate Root is ',sol)

final= 1/sol

print('Approximate density Root is ',final)

RESULTS AND DISCUSSIONS:

DATA FOR PURE:

*All errors are in percentage*

140.0000
120.0000
100.0000
80.0000 EXPERIMENTAL

60.0000 VDW

40.0000 SRK

20.0000 PR

0.0000
PRESSURE(BAR)
1200.0000

1000.0000

800.0000
EXPERIMENTAL
600.0000
VDW
400.0000
SRK
200.0000 PR

0.0000

PRESSURE(BAR)
700.0000
600.0000
500.0000
400.0000 EXPERIMENTAL
300.0000 VDW
200.0000 SRK
100.0000 PR

0.0000

PRESSURE(BAR)
1.4000
1.2000
1.0000
0.8000 EXPERIMENTAL
0.6000 VDW
0.4000 SRK
0.2000 PR
0.0000

PRESSURE(BAR)
700.0000

600.0000

500.0000

400.0000 EXPERIMENTAL
300.0000 VDW
SRK
200.0000
PR
100.0000

0.0000
 20.0000
40.0000
60.0000
80.0000

0.0000
100.0000
120.0000
140.0000
160.0000
20.0000
40.0000
60.0000
80.0000
100.0000
120.0000
140.0000
160.0000
180.0000
200.0000
220.0000
240.0000
260.0000
280.0000
300.0000

PR
SRK
VDW
EXPERIMENTAL
DATA FOR MIXTURES:
*All errors are in percentage*
ANALYSIS FOR PURE SUBSTANCES:

Analysis of Soave-Redlich-Kwong EOS Model

For CO2, at subcritical temperature of 280 K, the calculated data matches the experimental data with a maximum error
of 2.5%. The SRK EOS accounts for the energy parameter as function of temperature and acentric factor which
accurately predicts the density at such temperature with pressure ranging from 0bar to 40bar. It can be noted that the
experimental value of density is consistently higher than the calculated value.

For CO2, at critical temperature of 304.12 K, the calculated data matches the experimental data with a maximum error
of 19.33% in a pressure range of 0bar to 400bar. Within the pressure range of 50bar to 90bar, the rate of increase of
density with pressure increases significantly, and beyond the critical point with increase in pressure, there is a decrease
in the rate of increase of density with pressure. In the graph shown in the previous section the curve bends towards the
x-axis (pressure axis) beyond the critical point.

For CO2, at supercritical temperature of 450 K, the calculated data matches the experimental data with a maximum
error of 7.91%. From literature we know that, Span- Wagner EOS is the most accurate for density prediction of CO2
however we can conclude that the SRK EOS model is reasonably accurate within this pressure range.

For H2O, at subcritical temperature of 400 K, the calculated data is matches the experimentally determined value. The
max error is 1.07% or pressure range of 0bar to 2.25bar

For H2O, at critical temperature of 647.096 K, the calculated data matches for lower pressures but larger errors are
observed at higher pressures. In the pressure range of 0bar to 400bar the maximum error is around 37%.

For H20, at supercritical temperature of 750 K, the calculated data matches the experimentally determined value. The
values are more accurate than the critical range. The pressure range that the data has been calculated for is 0bar to
300bar. The maximum error is around 4.5%

The Soave-Redlich-Kwong Equation of State builds upon Redlich-Kwong EOS by introducing the energy parameter as
function of temperature and acentric factor- which takes into account the sphericity of the molecule. SRK EOS seems to
do well except at the critical regions of water. A possible explanation could be that the SRK EOS does not account for
the molecular interactions of water in the vapour phase, but rather accounts for the sphericity of molecules alone. These
interactions include dipole-dipole interactions, Van der Waals interactions, hydrogen bonding etc.

Analysis of Van der Waals EOS Model

For CO2, at subcritical temperature of 280 K, the calculated data matches the experimental data with a maximum error
of 11.3119% in a pressure range of 0bar to 40 bar. Although the error increases as high as 11.3119% at high pressure of
40bar, for a pressure range of 0-20 bar, the model predicts with greater accuracy. This is expected as this model builds
upon the Ideal Gas Law which predicts accurately for very high temperature and very low pressures- ideal gas
conditions.

For CO2, at critical temperature of 304.12 K, the calculated data varies significantly and rapidly throughout the pressure
range of 0 bar- 600 bar. The error is very low for very low pressure conditions but the model fails to account for high
pressure, roughly above 50bar.

For CO2, at supercritical temperature of 450 K, the calculated data matches the experimental data to a large extent
except for extremely high pressure like 500bar this is expected as this model builds upon the Ideal Gas Law which
predicts accurately for very high temperature and very low pressures- ideal gas conditions but is not accurate for higher
pressures and higher temperatures.

For H2O, At subcritical temperature of 400 K, the data as calculated is similar to the experimental data. The graphs are
almost linear. Data plotted is for pressure range from 0bar to 2.25 bar. The SRK EOS modelled the system more
accurately than VDW EOS the maximum error is 1.7%

At critical temperature of 647.096 K, the data calculated matches experimental values. Error in the values reaches a
maximum of around 45%. The SRK EOS modeled the system more accurately than VDW EOS.

At supercritical temperature of 750 K, the data calculated does match experimental values. The maximum error is
around 9.7%. The SRK EOS modeled the system more accurately than VDW EOS.

Carbon dioxide is a linear triatomic non-polar molecule, composed of one Carbon atom and two Oxygen atoms. Water is
a polar triatomic bent-shape molecule, composed of one Oxygen atom and two Hydrogen atoms.

The Van der Waals Equation of State has its drawback- it does not account for the intermolecular forces of attraction or
repulsion. The EOS do not account well enough for the interactions that exist such as Van der Waals forces, dipole-
induced forces, London forces etc.

Analysis of Peng-Robison EOS Model:

For CO2, at subcritical temperature of 280 K, the calculated data matches the experimental data with a maximum error
of 8% in a pressure range of 0bar to 40bar. Errors are low in the lower pressure ranges but increases at high pressure of
40bar.

For CO2, at critical temperature of 304.12 K, the EOS models the system accurately with a maximum error of 11%. The
model is the most accurate among the three EOS in consideration (VDW, SRK AND PR).

For CO2, at supercritical temperature of 450 K, the calculated data matches the experimental data to a large extent. But
high errors are observed at high pressures in the range of 300bar-500bar.

For H2O, at subcritical temperature of 400 K, the data as calculated is similar to the experimental data. The graphs are
almost linear. Data plotted is for pressure range from 0bar to 2.25 bar. The PR EOS modeled the system more accurately
than SRK EOS. The maximum error is 1.02%

For H2O, at critical temperature of 647.096 K, the data calculated matches experimental values. Error in the values
reaches a maximum of around 30.9%. The system was modeled accurately till a approximate pressure of 170bar above
this larger errors were observed.

For H2O, at supercritical temperature of 750 K, the data calculated does match experimental values. The maximum error
is around 19.7%. Low errors were observed small values of pressure but the errors increased with increase in pressure

Peng Robinson EOS was the most accurate for all regions except for supercritical H2O. From literature it is evident that
PR EOS is better at predicting liquid densities SRK EOS does a better job when it comes to the polar molecules. It can be
reasoned out that since water is a polar molecule SRK will do a better job at predicting densities.
ANALYSIS FOR MIXTURE:

The Peng Robinson equation of state seems to give the best results with an absolute average error of around 22%. SRK
EOS gives an average absolute error of around 33% while VDW gives an error of around 50%.

VDW’s inaccuracy is expected as it doesn’t account well for intermolecular interactions which are present between CO2
and H2O molecules in the mixtures. Peng Robinson is the most accurate as it accounts for the intermolecular interaction
between the 2 molecules that plays a significant role in determining the densities of the mixture.

REFERENCES:

 Viscosity and Density of Aqueous Solutions of Carbon Dioxide at Temperatures from (274 to 449) K and at
Pressures up to 100 MPa Mark McBride-Wright, Geoffrey C. Maitland, and J. P. Martin Trusler*Qatar Carbonates
and Carbon Storage Research Centre (QCCSRC), Department of Chemical Engineering, Imperial College London,
South Kensington Campus, London SW7 2AZ, United Kingdom

 Available from: http://webbook.nist.gov/chemistry/