6770 Ind. Eng. Chem. Res.

2006, 45, 6770-6777

Measurement and Modeling of the Solubility of Water in Supercritical Methane

and Ethane from 310 to 477 K and Pressures from 3.4 to 110 MPa
Matt Yarrison, Kenneth R. Cox, and Walter G. Chapman*
Department of Chemical and Biomolecular Engineering, Rice UniVersity, 6100 S. Main St. MS 362,
Houston, Texas 77005

We describe a new flow cell apparatus for measuring the water content of gases up to 120 MPa at 490 K that
uses a combination of gravimetric and electrical resistance techniques to determine the solubility of water in
the gaseous phase. The new experimental data for the solubility of water in supercritical methane and ethane
were obtained with our apparatus spanning pressures from 3.4 to 110 MPa covering a temperature range
from 310 to 477 K. We model the experimental results by combining two equations of state; vapor-phase
fugacities and fugacity coefficients are calculated with a modified Peng-Robinson equation of state, and
aqueous-phase fugacities are calculated using an equation by Wagner and Pruss (J. Phys. Chem. Ref. Data
2002, 31, 387-535) or by a modification of a correlation developed by Saul and Wagner (J. Phys. Chem.
Ref. Data 1987, 16, 893-901). We compare the model results with new and existing experimental data and
with commercially available simulators. Our model reproduces the experimental data within 2-6% using
one adjustable parameter, indicating that the predictions of the model are equal to or superior to the
commercially available simulators.

1. Introduction mental apparatus, which consists of a gas cylinder (S), a 1000

cm3 boost pump (P1), a 500 cm3 main pump (P2), a temperature
Accurate knowledge of the equilibrium water content of equilibration coil (EC), a saturation cell (SC), throttling valves
hydrocarbon gas mixtures is of vital importance to the energy (TV1 and TV2), an analysis train, and a flow meter (F). Both
industry. This information is necessary to prevent corrosion EC and SC are contained in a thermostatically controlled air
problems caused by moisture condensation in pipelines or bath, which is controlled to within 0.3 K using a 1.8 kW heater.
process equipment and to determine the dosage of chemical Gas enters the system from the cylinder and is compressed to
inhibitor to avoid hydrate crystal formation in subsea and the desired pressure using positive displacement pumps P1 and
terrestrial pipelines. Hydrates can block fluid transmission lines, P2. From pump P2, the gas enters the equilibration coil, where
can cause serious damage to plant equipment, and are a serious it is heated to bath temperature. All pressure and temperature
safety concern. data from the pumps, heat bath, and saturation cell are recorded
Despite the studies of Olds et al.,1 Reamer et al.,2 Deaton using a data logger.
and Frost,3 Kobayashi and Katz,4 McKetta and Katz,5 Culberson
A detailed drawing of the saturation cell is given in Figure
and McKetta,6-8 and Culberson et al.,9 data for methane and
2. The cell has a nominal internal volume of 1300 cm3, with an
ethane are available in the open literature only to pressures of
70.0 MPa. Only Dhima 10 has measured the solubility of gas in internal diameter of ∼7.5 cm. Gas enters the cell and flows
the aqueous phase (the other side of the phase diagram from through a glass dispersion frit (PF) which disperses the gas into
what we are measuring) between 70 and 100 MPa. Mohammadi the liquid water as bubbles of a nominal diameter of 0.01 mm.
et al.,11-13 Chapoy et al.,14 and Gillespie and Wilson15 have A stainless steel retaining ring (RR) holds the frit in place at
performed lower-pressure (∼0.1-34 MPa) and lower-temper- the bottom of the cell. The bubbles travel through the liquid
ature (273-340 K) measurements for similar systems. As oil water phase and into the headspace of the saturation cell. A
and gas exploration moves to greater depths both on- and stainless steel baffle and an antientrainment section (shown in
offshore, reservoir pressures are frequently on the order of 100 gray) prevent any water spray from leaving the saturation cell.
MPa, beyond the range of reasonable extrapolation from existing Gas exits the saturation cell through a throttle valve (TV1),
data. To gain a better understanding of the phase behavior of where the pressure is reduced to ∼1 bar before passing to the
these industrially important, highly nonideal systems at all analytical train. Three different pressure transducers (PT)
industrially relevant conditions, we systematically study the monitor cell pressure; pressures below 70 bar use a Heise
solubility of water in methane and ethane over a pressure and transducer, pressures from 70 to 700 bar use a Data Instruments
temperature range that covers reservoir conditions to platform transducer, and pressures from 700 to 1400 bar use a Senso-
conditions. metrics transducer. All transducers are calibrated yearly using
a Ruska dead-weight tester. A J-type thermocouple (TC) located
in a thermowell (TW, shown by crosshatching) monitors cell
2. Experimental Equipment
temperature. The J-type thermocouple was calibrated by OMEGA
The solubility of water in these supercritical hydrocarbon engineering against the ice and boiling points of water and the
gases is measured using a flow scheme similar to that of melting points of tin and zinc. An Autoclave Engineers rupture
Prausnitz and Benson16 or Rigby and Prausnitz17 or of Song et disk (RD) mounted in the antientrainment section provides
al.18 Figure 1 shows the important components of the experi- overpressure protection. The dashed line indicates the ap-
proximate water level in the cell.
* Corresponding author. Phone: (713) 348-4900. Fax: (713) 348- Analytical train A consists of the General Electric (GE)-
5478. E-mail: [email protected]. Panametrics moisture analyzer (PMA), three desiccant-charged
10.1021/ie0513752 CCC: $33.50 © 2006 American Chemical Society
Published on Web 09/02/2006
 Ind. Eng. Chem. Res., Vol. 45, No. 20, 2006 6771

Figure 1. Schematic of experimental apparatus. Data from all thermocouples, pressure transducers, and the PMA, if in use, are logged real-time using
LabVIEW. U-tubes are weighed using a Sartorius scale accurate to 0.01 mg.

and the analytical train is maintained 10-20 K above air bath

temperature using electrical heating tape to prevent condensa-
tion. When using the desiccant-charged u-tubes, the density of
the sample gas inside the u-tubes will generally be slightly
different from the density of the air inside the tubes when
charging with desiccant. This buoyancy difference can lead to
minor measurement errors if not taken into account. After
several experimental tests, we found that purging the u-tubes
with dried sample gas (dried using a combination of Dehydrite,
magnesium perchlorate, and molecular sieve 4a) before and after
the experimental run is sufficient to eliminate buoyancy effects.
A Sartorius CP-5000 series scale, with an accuracy of 0.01 mg,
is used to weigh the u-tubes. The scale is checked periodically
against a set of standard-grade Ohaus weights.

3. Materials
The methane and ethane used are ultrahigh-purity grade
(99.995% pure) from the Matheson gas company. Standard
laboratory-grade, deionized, UV-sterilized water with a maxi-
mum conductance of 0.25 micro siemens (µS) is used without
further distillation. The anhydrous ACS-grade magnesium
perchlorate used is from VWR.

4. Experimental Procedure
Deionized water (2 L) is vacuum boiled in a 4 L sample
cylinder to remove dissolved oxygen and nitrogen. Methane or
ethane (hereafter referred to as simply “gas”) is bubbled slowly
into the cylinder until the pressure in the sample cylinder reaches
∼0.2 MPa. After a 1 h equilibration period, the gas is vented
Figure 2. Schematic showing the saturation cell. The cell is made from and the process is repeated twice more to ensure that no
A-286 alloy stainless steel and has a maximum pressure rating of 2000 bar unwanted oxygen or nitrogen remain dissolved in the water.
at 204 °C. A standard NIST type high-pressure closure is used to seal the After degassing, the water is used to completely fill the 1.3 L
cell with a poly(tetrafluoroethylene) (PTFE) or Aflas O-ring. The equilibrium saturation cell. Gas is used to displace 300-400 cm3 of water
coil, closure nut, and baffles are omitted for clarity.
in order to create a gas cap in the saturation cell, and sufficient
stainless steel u-tubes (U), and a gas flow meter (F). The PMA gas is added to pressurize the cell and equilibration coil to ∼2
measures the water content by measuring the resistance across MPa. The heat bath is set to the desired temperature T and
an aluminum oxide sensor; as the water present in the gas stream allowed to equilibrate for ∼12 h until the cell internal temper-
changes, so does the resistance across the sensor, which is ature has stabilized.
converted by the PMA into a mole fraction and a mass of water Methane or ethane from gas source S fills pumps P1 and P2,
per volume reading. The unit is calibrated by GE using National which fill the saturation cell with gas at the desired pressure.
Institute of Standards and Technology (NIST) traceable stan- Since compression tends to raise the cell temperature ∼0.5 K,
dards and is recalibrated by GE yearly. Both PMA and the the cell is allowed to come to thermal equilibrium with the bath
desiccant-charged tubes are used at temperatures below 333 K, before an experiment starts. Once the cell is at the same
but temperature limitations of the PMA prevent its use above temperature as the bath, the throttle valves are opened so that
this temperature. All tubing between the throttle valve (TV1) gas flows out of the cell and into the analysis train at a rate of
6772 Ind. Eng. Chem. Res., Vol. 45, No. 20, 2006

Figure 4. Water content of supercritical ethane. Diamonds are data from

this investigation, squares are those of Reamer et al.,2 while lines are
Figure 3. Water content of supercritical methane. Circles are data from calculated using the model outlined in this paper. Note that y-axis is
this work, diamonds are those of Olds et al.,1 and squares are those of Rigby -log(yH2O).
and Prausnitz.17 Lines are the water contents predicted using the modeling
procedure outlined in this paper, using the mixing rules in eq 18. Please electrical resistance and the gravimetric measurements are listed
note that y-axis is -log(yH2O) and not yH2O. in Tables 3 and 4. Uncertainties are listed along with the data
in Tables 1 and 2, but generally range from 2 to 6%. Uncer-
0.014-0.028 L‚s-1 of expanded gas at ambient temperature and tainties reported are from a propagation-of-error analysis or from
pressure. Cell pressure is maintained through a continuous flow the observed deviation in measurements, whichever is greater.
of gas from pump P2. Approximately 17 L of gas is used to The reported results are the average of the experimental runs.
equilibrate the analysis train so it is at steady state. Upon equili-
bration, flow switches from the vent to the chemical absorption 6. Modeling
traps or PMA. An experiment will use between 17 and 56 L of
expanded gas, and a typical experiment will deposit in excess To make the data more useful for engineering practice, we
of 100 mg of water in the u-tube traps. Cell and pump pressures have developed an accurate engineering model for the methane
are constantly monitored, and the pump P2 flow rate is manually + water and ethane + water systems. The vapor phase is
adjusted to maintain cell pressure throughout the experiment. modeled using the Peng-Robinson (PR) equation of state (EoS),
Once the desired volume of gas passes through the analysis while the aqueous phase is modeled using an equation of state
train, the throttle and u-tube valves are closed and the traps are based on the work of Saul and Wagner.19 Using the Saul and
removed, sealed, and weighed. Gas volume is recorded using Wagner formulation allows quick and accurate water fugacity
the flow meter, and this volume is corrected from ambient calculations with a minimum of parameters.
temperature and pressure to 0.1013 MPa and 288.65 K. Each 6.1. Conditions for Equilibrium. For binary water +
(pressure, temperature) point is repeated a minimum of three hydrocarbon systems, the phase-equilibrium relationship for
times to ensure repeatability and accuracy. The u-tubes are made water is
from 3/8 in. o.d. stainless steel tubing and use Dow-Corning
yH2Oφ̂H2OP ) (1 - xHC)f H(T,P),pure
2O
(1)
glass wool to hold the solid desiccant in place. At the beginning
of each experiment, the tubes are packed with fresh desiccant
and flushed with dry sample gas, sealed using Suba-Seal where P is the pressure, f H(T,P),pure
2O
is the fugacity of pure liquid
stoppers, and weighed. Experiments have shown that the sealed water evaluated at temperature T and pressure P, yH2O is the
tubes do not pick up more than 0.01 mg of mass over a 24 h mole fraction of water in the vapor phase, φ̂H2O is the fugacity
period and that traps do not pick up more than 0.01 mg of mass coefficient for water in the vapor phase calculated using the
when the traps are connected and disconnected to/from the PR equation of state, and xHC is the mole fraction of the
expansion train. These results indicate that the measurement hydrocarbon dissolved in the liquid water. Here, we have
technique is adequate. Once enough gas has flowed through assumed that the liquid water phase is so rich in water that the
the expansion train so that it is at steady state, the tubes are activity coefficient of water is ∼1.
connected to the expansion train and the wet-test flow meter 6.2. Calculating Water Fugacity. We have used two
using Tygon tubing and clamps. After the experiment, the tubes approaches to calculate the fugacity of pure liquid water: the
are disconnected, sealed, and weighed. The tubes are then equation by Wagner and Pruss20 and a modification of the Saul
flushed with dry sample gas and reweighed. and Wagner19 formulation. Each water model gives similar
results over the range of conditions studied here. Here, we only
present our modification of the Saul and Wagner formulation.
5. Experimental Results The fugacity of water is written as
The data covering the temperature range spanning 310-477 ln(fH2O(P,T)pure) )

( )
K and the pressure range from 3.4 to 110 MPa are presented in
Figure 3 for methane and in Figure 4 for ethane. The data of VHL 2O(P,T)
Olds et al.1, Rigby and Prausnitz,17 and Reamer et al.2 are plotted ln(φHsat2O(T)PHsat2O(T)) + ∫P
P
sat
RT
dP (2)
for reference. The experimental data are shown in Table 1 for
methane and in Table 2 for ethane, while the raw data from the where φHsat2O(T) is the fugacity coefficient of water at saturation,
 Ind. Eng. Chem. Res., Vol. 45, No. 20, 2006 6773

Table 1. Experimentally Determined Values for Water Solubility in Supercritical Methanea

temperature, K
310.9 313.9 366.5 422.04 466.5 477.5
pressure, MPa 104 104 103 102 10 10
3.45 ( 0.0007 22.09 ( 1.1 23.54 ( 0.8 12.46 ( 0.42 4.15 ( 0.2 5.46 ( 0.17
6.89 ( 0.0007 12.48 ( 0.6 13.25 ( 0.4 7.46 ( 0.37 2.24 ( 0.11 2.95 ( 0.1
20.68 ( 0.086 6.997 ( 0.3 6.56 ( 0.2 3.16 ( 0.16 0.84 ( 0.04 1.13 ( 0.06
41.37 ( 0.086 5.636 ( 0.3 3.96 ( 0.2 2.04 ( 0.1 0.60 ( 0.03 0.70 ( 0.04
62.05 ( 0.68 4.90 ( 0.25 3.44 ( 0.15 1.65 ( 0.08 0.46 ( 0.02 0.57 ( 0.03
75.84 ( 0.69 4.72 ( 0.25 3.13 ( 0.15 1.43 ( 0.08 0.40 ( 0.02 0.51 ( 0.03
89.63 ( 0.69 4.67 ( 0.25 2.88 ( 0.15 1.37 ( 0.08
96.53 ( 0.69 0.35 ( 0.02 0.494 ( 0.03
110.32 ( 0.69 0.33 ( 0.02 0.455 ( 0.02
a Mole fractions and errors are reported multiplied by the power of 10 listed in that column. Uncertainties listed are calculated using a propagation of error

analysis or the observed variance, whichever was largest.

Table 2. Experimentally Determined Values for Solubility of Water using the following correlations:19

[ ]
in Supercritical Ethanea
temperature, K PHsat2O Tc
pressure, 314.8 366.5 422.0 466.5 ln ) (-7.858 23τ + 1.839 91τ1.5 - 11.781 1τ3 +
Pc T
MPa 104 103 102 10
3.45 ( 0.0007 22.46 ( 0.67 12.114( 0.36 4.154 ( 0.16 22.670 75τ3.5 - 15.939 3τ4 + 1.775 16τ7.5) (4)
3.59 ( 0.0007 21.88 ( 1.09
6.89 ( 0.086 12.77 ( 1.09 12.38 ( 0.37 6.957 ( 0.21 2.163 ( 0.11 Fsat(T)
20.68 ( 0.086 8.37 ( 0.41 6.43 ( 0.2 2.794 ( 0.08 0.920 ( 0.06 ) 1 + 1.992 06τ1/3 + 1.101 23τ2/3 -
41.37 ( 0.68 7.33 ( 0.36 4.06 ( 0.1 1.819 ( 0.05 0.658 ( 0.04
Fc
62.05 ( 0.68 5.82 ( 0.25 3.29 ( 0.1 1.583 ( 0.05 0.510 ( 0.03 0.512 506τ5/3 - 1.752 63τ16/3 - 45.448 5τ43/3 -
75.84 ( 0.69 4.93 ( 0.25 3.02 ( 0.1 1.498 ( 0.04 0.455 ( 0.03
89.63 ( 0.69 4.49 ( 0.25 2.87 ( 0.1 1.382 ( 0.04 675 615τ110/3 (5)
96.53 ( 0.69 0.401 ( 0.03
110.32 ( 0.69 0.372 ( 0.02 T
τ)1- (6)
a Mole fractions and errors are reported multiplied by the power of 10 Tc
listed in that column.
F(P,T) ) Fsat(T) + (P - PHsat2O)(8.48 × 10-6T2 - 5.91 ×
Table 3. Experimentally Determined Mole Fractions of Water in
Methane from Gravimetric and Electric Resistance Measurements
10-3T + 1.39) + b(T) (7)
for Methanea
274 < T < 510 K

( ( ))
pressure, mol fraction, mol fraction,
MPa gravimetric resistance reported Tc - T
Methane, 310.8 K b(T) ) 143 exp -14
Tc
3.45 2.228E-03 2.150E-03 2.209E-03
6.89 1.197E-03 1.400E-03 1.248E-03
20.68 7.076E-04 6.760E-04 6.997E-04 510 < T < 640 K (8)

( ( ))
41.37 5.821E-04 5.080E-04 5.636E-04
Methane, 313.9 K
Tc - T
62.05 4.914E-04 4.870E-04 4.903E-04
b(T) ) 143 exp -14 erf((P - PHsat2O)a(T))
Tc
75.84 4.733E-04 4.660E-04 4.715E-04
89.63 4.727E-04 4.510E-04 4.673E-04
a
a(T) ) 7.129 × 10-4T + 0.3692
The reported average was calculated using a weight average with a
weight of 0.75 assigned to the gravimetric measurement and 0.25 assigned
1
for the resistance measurement. VL ) (9)
F(P,T)

VHL 2O(P,T) is the liquid water volume, P and PHsat2O(T) are the Tc is the critical temperature of water, Fc is the critical density
system pressure and the water saturation pressure, respec- of water, and Pc (MPa) is the critical pressure of water.
tively, R is the universal gas constant, and T is the system Equations 3-9 are intended for use with water in the liquid
temperature. The fugacity from eq 2 is plotted against steam state in the range from 273.15 to 640 K and have been validated
table data in Figure 5. φHsat2O is calculated using the following against experimental data within that temperature range for
correlation pressures up to 200 MPa. Equation 7 is a modification of Saul
and Wagner’s19 saturation pressure correlation that makes the
274 < T < 647 K density and, consequently, the liquid volume more accurate.
Equation 7 is simply a best fit function using least-squares
274
φHsat2O ) 1 - 0.001 340 927 477 538 65 exp 9.7 1 - ( ( T
+ )) regression. Including the effect of pressure on the liquid density
is significant. For comparison, the approximation of using the
1.956 70 × 10-3 (3) saturated liquid density at all pressures is shown as a dashed
line in Figure 7. A few comparisons are shown with the 1995
which we fitted to the data of the 1995 steam tables;21 NIST/ASME steam tables in Figure 7. Both the NIST imple-
it is shown in Figure 6. The vapor pressure is calculated mentation21 of Wagner and Pruss’20 equation of state for water
6774 Ind. Eng. Chem. Res., Vol. 45, No. 20, 2006

Table 4. Experimentally Determined Mole Fractions of Water in

Ethane from Gravimetric and Electric Resistance Measurements for
Ethanea
pressure, mol fraction, mol fraction,
MPa gravimetric resistance reported
Ethane, 314.8 K
3.59 2.181E-03 2.209E-03 2.188E-03
6.89 1.239E-03 1.390E-03 1.277E-03
20.68 8.417E-04 8.220E-04 8.368E-04
41.37 7.310E-04 7.370E-04 7.325E-04
62.05 5.798E-04 5.900E-04 5.824E-04
75.84 4.952E-04 4.850E-04 4.926E-04 Figure 7. Comparisons of improved correlation from eq 7 vs NIST21 steam
89.63 4.508E-04 4.430E-04 4.488E-04 table data. The dashed lines are values of the saturated liquid density
a The reported average was calculated using a weight average with a calculated from eq 5 to show the effect of pressure on liquid density.
weight of 0.75 assigned to the gravimetric measurement and 0.25 assigned
for the resistance measurement.

Figure 8. Comparison of XHC as calculated by eq 10 with experimental

data of O’Sullivan and Smith.25 Error bars are 6%.

hydrocarbon in the aqueous phase, y and x are the mole fractions

of the hydrocarbon in the vapor and liquid phases, respectively,
and Hi,H2O(T,P) is the Henry’s constant for species i in water at
temperature T and pressure P. The amount of hydrocarbon gas
Figure 5. Fugacity of water as calculated by eq 2 vs fugacities from the in the liquid phase can be calculated by explicitly solving for
1995 steam tables.
xHC.
As the hydrocarbons used in this investigation have low
solubilities in the water-rich phase even at pressures up to 110
MPa, they are adequately represented10 using the Krichevsky-
Kasarnovsky22 equation,
∞
f LHC VHC,H2O
ln(HHC,H2O(T,P)) ) ln ) lnHi,0 + (P - PHsat2O)
xHC RT
(11)

where f LHC is the fugacity of the hydrocarbon component i in

water, xHC is the mole fraction of component i dissolved in the
water at temperature T and system pressure P, HHC,0 is the
Henry’s law constant for the hydrocarbon species in water at T
∞
and water saturation pressure, and VHC,H 2O
is the partial molar
volume of the hydrocarbon in water at infinite dilution. HHC,O
is calculated using a method developed by Harvey,23 while
∞
VHC,H 2O
is calculated using the corresponding states method of
Figure 6. Fugacity coefficient at saturation φsat is shown here from 274 to Lyckman et al.24 For reference, the solubility of methane in
640 K.
water is shown in Figure 8 in comparison with the data of
and the Saul and Wagner treatment as modified above give O’Sullivan and Smith.25 Model results based on eq 11 have been
similar results. found to reproduce the experimental data within 0.3-12% of
A relationship similar to eq 1 can be written for the the experimental value. However, this is not to say that using
hydrocarbons in the aqueous phase, eq 11 will always produce such results. Any error in the value
∞
of VHC,H 2O
will produce large deviations in the value of xHC at
high pressures, as P - PHsat2O becomes large at high pressures.
yHCφ̂HCP ) f VHC ) f LHC ) xHCHHC,H2O(T,P) (10)
For example, an error of 10% in the partial molar volume of
methane at infinite dilution will create a 0.1% error in the
where H is the Henry’s constant of hydrocarbon in the liquid Henry’s constant at 1 MPa, a 6% error at 50 MPa, and a 20%
water phase at temperature T and system pressure P, f VHC is the error at 150 MPa at 324 K. We can, therefore, expect the
fugacity in the vapor phase, f LHC is the fugacity of the maximum possible uncertainty in xHC to be on the order of 20-
 Ind. Eng. Chem. Res., Vol. 45, No. 20, 2006 6775
∞ Table 5. Parameters Used in Peng-Robinson EoSa
30%, as the uncertainties in VHC,H2O
are on the order of 5 cm3/
mol or ∼16%. component Pc (MPa) Tc (K) ω Fc (kg/m3)
6.3. Equation of State. The gas-phase fugacities of water water 22.12 647.3 0.344 322
and hydrocarbon components are calculated using the Peng- methane 4.604 190.6 0.011 163
Robinson (PR) equation of state.26 The PR26 equation of state ethane 4.88 305.4 0.099 207
can be written in dimensionless form as a Values taken from Poling et al.31

Z)
1
-
AMix
(
BMix/Z
(1 - BMix/Z) BMix 1 + 2BMix/Z - (BMix/Z)2 ) (12)
Table 6. Binary Interaction Parameters (kij) Used for Mixing Rule
in Equations 18 and 19
temperature, K
where 310.9 313.7 314.8 366.4 422 466.4 477.6

aiiP Lorentz-Berthelot Mixing Rule

P
biF ≡ Bi/ZZ ≡ Aii ≡ 2 2 (13) methane 0.48 0.48 0.47 0.45 0.41 0.37
FRT RT ethane 0.49 0.51 0.46 0.43 0.42
Linear Mixing Rule
R2Tc,i2 Tc,i methane 0.045 0.045 0.039 0.052 0.075
ai ≡ ζiRi; ζi ≡ 0.457 235 53 bi ≡ 0.077 960 7R
Pc,i Pc,i Table 7. Coefficients Used to Interpolate the Binary Interaction
Parameter (kij) for the Mixing Rules Used in This Work, Valid from
(14) 310 to 477 K
A × 106 B × 103 C
Rii ≡ [1 + κi(1 - xTr,i)]2
Lorentz-Berthelot Mixing Rule
κi ≡ 0.374 64 + 1.542 26ωi - 0.269 93ωi2 methane -3.5867 2.3165 0.1007
T ethane -5.5729 3.9974 -0.2206
Tr ) (15) Linear Mixing Rule
Tc,i
methane 3.0087 -2.1472 0.4221
and finally, taking the appropriate derivatives, the fugacity
of temperature. Equation 19 gives good agreement with the data
coefficient for the binary system is
using kij ) 0; however, better agreement is obtained on fitting
Bi kij. Since the best fitted value of A12 is the same regardless of
AMix
ln(φ̂Vi ) ) (ZV - 1) - ln(ZV - BMix) - ln whether eq 18 or eq 19 is used, a small kij in eq 19 for the
B x8
Mix Mix
B methane/water system corresponds to a large kij in eq 18. The

( ZV + (1 - x2)BMix )(
ZV + (1 + x2)BMix 2(y1Ai1 + y2Ai2)
AMix
-
Bi
BMix ) (16)
purpose of introducing eq 19 is to demonstrate that the large
kij's produced from the geometric mixing rule are reasonable in
this case. Both mixing rules reproduce the solubility data for
2 2 2
the methane + water and ethane + water systems to within
AMix ) ∑i ∑j yiyjAijBMix ) ∑i yiBi (17)
2-7% average absolute deviation (AAD) over the experimental
pressure and temperature range where

The form of the mixing rules used in eq 17 is sometimes

called van der Waals mixing rules; however, the form used here
is empirical. Numerous investigators26,27 have suggested the
AAD )
1
n
∑n( |yexp - ycalc|
y exp
) × 100 (20)

cross-interaction parameter, A12, should be of the form

In this case, n is the number of experimental data points and
A12 ) (1 - kij)xA11A22 (18) yexp is the observed mole fraction of water in the gas phase.
Table 5 lists the parameters used in the model, and Table 6
where kij is a fitted parameter, component 1 is water, and lists the binary interaction parameters (kij). It should be noted
component 2 is the hydrocarbon. This approximation is fairly that, while the mixing rule used in eq 19 gives values for the
accurate if the attractions between molecules are only due to binary interaction parameter for the methane/water system that
van der Waals interactions. For the water/hydrocarbon system, are closer to zero, we would not expect the model to work well
eq 18 results in large values of kij. This occurs since, in the for the ethane/water system. In this case, we might expect that
Peng-Robinson equation of state, the fitted A11 (pure water) replacing A22 in eq 19 with A22/2, where the 2 represents the
includes not only van der Waals attractions but also contributions carbon number of ethane, would give better results. Equation
from hydrogen bonding and strong dipole-dipole interactions. 19 was presented for illustrative purposes only. We propose to
A large kij is needed since the attraction between water and use the geometric mixing rule for multicomponent mixtures.
hydrocarbon molecules is only due to van der Waals interactions. 6.4. Model Results. Calculations made using the mixing rule
We expect that the van der Waals interaction between water in eq 18 are shown in Figures 3 and 4. AAD% results of 2-7%
and methane should be closer to the methane-methane van der are obtained by making kij a weak function of temperature.
Waals attraction than to A12 from eq 18. Therefore, for the Binary interaction parameters (kij) are shown in Table 6. For
methane/water system, we consider a simple, empirical mixing both sets of mixing rules, the binary interaction parameters can
rule of the form be interpolated using kij ) AT2 + BT + C, where T is in Kelvin
and A, B, and C are given in Table 7.
A12 ) [A11kij + A22(1 - kij)] (19) Comparisons with Multiflash cubic plus association (CpA),28
Colorado School of Mines Gibbs Energy Minimization29 (CSM
where component 1 is water, component 2 is the hydrocarbon, GeM), and CALSEP’s PVT SIM30 are presented in Figure 9.
and kij is a fitted binary interaction parameter, which is a function These calculations are presented using the default, recommended
6776 Ind. Eng. Chem. Res., Vol. 45, No. 20, 2006

Table 8. Experimental Mole Fraction Data and Model Predictions with AAD% for Methane + Water Binary System at 477.5 K
data CALSEP dev. % Multiflash dev. % CSM GeM dev. % this model dev. %
0.547 0.536 1.93 0.542 0.80 0.523 4.40 0.515 5.81
0.295 0.297 0.41 0.289 2.25 0.277 6.08 0.277 6.14
0.114 0.134 17.83 0.107 5.39 0.107 5.53 0.115 1.07
0.070 0.094 33.43 0.060 14.44 0.063 10.69 0.072 3.04
0.057 0.079 38.42 0.044 24.01 0.046 18.99 0.056 1.69
0.051 0.074 42.76 0.037 27.50 0.040 22.29 0.050 3.31
0.049 0.033 32.31 0.043 13.54
0.046 0.064 40.43 0.028 38.04 0.030 33.57 0.039 14.03
CALSEP AAD% 25.03 Multiflash AAD% 16.06 CSM GeM AAD% 16.73 this model AAD% 6.08

parameters for these systems. At low temperatures and low over- or underpredict the amount of water present in the vapor
pressures, all methods do an adequate job of predicting water phase at higher pressures.
solubility in the hydrocarbon-rich phase, but as temperatures
and pressures increase, they tend to diverge from the experi- Acknowledgment
mental measurements. Multiflash CpA and CSM GeM under-
estimate the amount of water in the vapor phase at high We gratefully acknowledge the financial support of the Gas
pressures, while CALSEP tends to overestimate the amount of Processors Association. We also thank Mr. Richard Chronister
water present in the vapor phase. CSM GeM gives the best for his invaluable mechanical assistance in fabricating the
equipment and Professor Derek Dyson for wise counsel. The
predictions next to the model presented here. It should be noted,
authors also thank Becky Yarrison for reading several drafts of
however, that the CSM GeM formalism was developed for use
this paper for coherency and Ryan Dunnavant and John Cliver
at or near hydrate-formation conditions, so the results presented for their help in screening experimental techniques and in
here are an extreme test. It is strongly suspected that the CpA apparatus fabrication. We would also like to thank Professor
model would provide much better results if fitted to the higher Wolfgang Arlt for his ideas in tuning the controller and Professor
pressure data, as it incorporates the statistical associating fluid Mark Thies for help in troubleshooting O-rings.
theory (SAFT) association term, which would greatly improve
calculations for the aqueous liquid phase. We would expect that, Literature Cited
for all models, if the model parameters were fitted to high-
pressure data, the calculated values would show improvement. (1) Olds, R. H.; Sage, B. H.; Lacey, W. N. Phase equilibria in
Table 8 lists the AAD% for methane + water at 477 K for the hydrocarbon systems. Composition of the dew-point gas of the methane-
water system. Ind. Eng. Chem. 1942, 34, 1223-7.
various models shown in Figure 9. (2) Reamer, H. H.; Olds, R. H.; Sage, B. H.; Lacey, W. N. Phase
equilibria in hydrocarbon series. Composition of dew-point gas in ethane-
water system. Ind. Eng. Chem. 1943, 35, 790-3.
(3) Deaton, W. M.; Frost, E. M., Jr. Gas hydrates and their relation to
the operation of natural-gas pipe lines. U.S. Bur. Mines, Monograph 1946,
8.
(4) Kobayashi, R.; Katz, D. L. Vapor-liquid equilibria for binary
hydrocarbon-water systems. Ind. Eng. Chem. 1953, 45, 440--51.
(5) McKetta, J. J., Jr.; Katz, D. L. Phase relations of hydrocarbon-
water systems. Am. Inst. Min. Metall. Eng., Petr. Technol. 1947, 10 (1);
Tech. Pub. No. 2123.
(6) Culberson, O. L.; McKetta, J. J., Jr. Phase equilibria in hydrocarbon-
water systems. II. The solubility of ethane in water at pressures to 10,000
lb. per sq. in. Trans. Am. Inst. Min. Metall. Eng., Tech. Pub. 1950, (2932).
(7) Culberson, O. L.; McKetta, J. J., Jr. Phase equilibria in hydrocarbon-
water systems. IV. Vapor-liquid equilibrium constants in the methane-
water and ethane-water systems. Trans. Am. Inst. Min., Metall. Pet. Eng.
1951, 192; Tech. Pub. No. 3201.
(8) Culberson, O. L.; McKetta, J. J., Jr. Phase equilibria in hydrocarbon-
water systems. III. The solubility of methane in water at pressures to 10,000
pounds per square inch absolute. Trans. Am. Inst. Min. Metall. Eng., Tech.
Pub. 1951, (3082).
Figure 9. Solubility of water in methane, including comparisons with (9) Culberson, O. L.; Horn, A. B.; McKetta, J. J., Jr. Phase equilibria in
CALSEP, CSM GeM, Multiflash CpA, and our model at 366 and 477 K. hydrocarbon-water systems. The solubility of ethane in water at pressures
to 1200 pounds per square inch. Am. Inst. Min. Metall. Eng., Tech. Pub.
7. Conclusions 1950, (2778), 1-6 (J. Pet. Technol. 2 (1)).
(10) Dhima, A. Solubilite des gaz naturels dans l’eau a pression elevee.
The water content in the hydrocarbon-rich phase of the binary Ph.D. Thesis, Claude Bernard, Lyon, France, 1998.
systems methane + water and ethane + water were measured (11) Mohammadi, A. H.; Chapoy, A.; Tohidi, B.; Richon, D. Water
from 310 to 477 K over a pressure range of 3.4-110 MPa. At Content Measurement and Modeling in the Nitrogen + Water System. J.
∼310 K, both a gravimetric and an electrical resistance method Chem. Eng. Data 2005, 50 (2), 541-5.
(12) Mohammadi, A. H.; Chapoy, A.; Richon, D.; Tohidi, B. Experi-
were used to measure the water content of the expanded gas mental Measurement and Thermodynamic Modeling of Water Content in
stream. Above 333 K, only gravimetric methods were employed. Methane and Ethane Systems. Ind. Eng. Chem. Res. 2004, 43 (22), 7148-
A calculation method using the Peng-Robinson equation of 62.
state and an accurate liquid water model was used to model the (13) Mohammadi, A. H.; Chapoy, A.; Tohidi, B.; Richon, D., Measure-
experimental data acquired in this investigation and that of other ments and Thermodynamic Modeling of Vapor-Liquid Equilibria in
Ethane-Water Systems from 274.26 to 343.08 K. Ind. Eng. Chem. Res.
investigators, showing good agreement with binary data. Similar 2004, 43 (17), 5418-24.
calculations with commercially available process (see Figure (14) Chapoy, A.; Mohammadi, A. H.; Richon, D.; Tohidi, B., Gas
9) simulators show that commercial models can substantially solubility measurement and modeling for methane-water and methane-
 Ind. Eng. Chem. Res., Vol. 45, No. 20, 2006 6777

ethane-n-butane-water systems at low-temperature conditions. Fluid Phase (24) Lyckman, E. W.; Eckert, C. A.; Prausnitz, J. M. Generalized
Equilib. 2004, 220 (1), 111. reference fugacities for phase equilibrium thermodynamics. Chem. Eng. Sci.
(15) Gillespie, P. C.; Wilson, G. M. GPA Research Report RR-48 1965, 20, 685-91.
Vapor-Liquid and Liquid-Liquid Equilibria: Water-Methane, Water- (25) O’Sullivan, T. D.; Smith, N. O. The Solubility and Partial Molar
Carbon Dioxide, Water-Hydrogen Sulfide, Water-n-Pentane, Water- Volume of Nitrogen and Methane in Water and in Aqueous Sodium Chloride
Methane-n-Pentane; Gas Processors Association: Tulsa, OK, 1982; pp from 50 to 125 °C and 100 to 600 Atm. J. Phys. Chem. 1970, 74 (7), 1460.
1-79. (26) Peng, D. Y.; Robinsion, D. B. A new two constant equation of
(16) Prausnitz, J. M.; Benson, P. R. Solubility of liquids in compressed state. Ind. Eng. Chem. Fundam. 1976, 15, 59-64.
hydrogen, nitrogen, and carbon dioxide. AIChE J. 1959, 5, 161-4. (27) Soave, G. Equilibrium constants from a modified Redlich-Kwong
(17) Rigby, M.; Prausnitz, J. M. Solubility of water in compressed equation of state. Chem. Eng. Sci. 1972, 27 (6), 1197-203.
nitrogen, argon, and methane. J. Phys. Chem. 1968, 72 (1), 330-4. (28) Yakoumis, I. V.; Kontogeorgis, G. M.; Voutsas, E. C.; Hendriks,
(18) Song, K. Y.; Yarrison, M.; Chapman, W. Experimental low- E. M.; Tassios, D. P. Prediction of Phase Equilibria in Binary Aqueous
temperature water content in gaseous methane, liquid ethane, and liquid Systems Containing Alkanes, Cycloalkanes, and Alkenes with the Cubic-
propane in equilibrium with hydrate at cryogenic conditions. Fluid Phase plus-Association Equation of State. Ind. Eng. Chem. Res. 1998, 37, 4175-
Equilib. 2004, 224 (2), 271-7. 82.
(19) Saul, A.; Wagner, W. International equations for the saturation (29) Ballard, A. L.; Sloan, E. D. The next generation of hydrate
properties of ordinary water substance. J. Phys. Chem. Ref. Data 1987, 16 prediction. Part III. Gibbs energy minimization formalism. Fluid Phase
(4), 893-901. Equilib. 2004, 218 (1), 15-31.
(20) Wagner, W.; Pruss, A. The IAPWS Formulation 1995 for the (30) CALSEP, CALSEP PVT SIM. In.
Thermodynamic Properties of Ordinary Water Substance for General and (31) Poling, B. E.; Prausnitz, J. M.; O’Connell, J. P. The properties of
Scientific Use. J. Phys. Chem. Ref. Data 2002, 31, (2), 387-535. gases and liquids, 5th ed.; McGraw-Hill: New York, 2001.
(21) Harvey, A. H.; Peskin, A. P.; Klein, S. A. NIST/ASME Steam
Properties, 2.2; U.S. Department of Commerce: Gaithersburg, MD, 2004.
ReceiVed for reView December 9, 2005
(22) Krichevsky, I. R.; Kasarnovsky, J. S. Thermodynamical calculations
of solubilities of nitrogen and hydrogen in water at high pressures. J. Am. ReVised manuscript receiVed June 15, 2006
Chem. Soc. 1935, 57, 2168-72. Accepted July 5, 2006
(23) Harvey, A. H. Semiempirical correlations for Henry’s constants
over large temperature ranges. AIChE J. 1996, 5, 1491-4. IE0513752