Академический Документы
Профессиональный Документы
Культура Документы
Desalination
j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / d e s a l
A simple model for the prediction of CO2 solubility in H2O–NaCl system at geological
sequestration conditions
N.A. Darwish a,⁎, N. Hilal b
a
American University of Sharjah, College of Engineering, Department of Chemical Engineering, Sharjah, P.O. Box 26666, United Arab Emirates
b
The Centre for Clean Water Technologies, School of Engineering, Swansea University, Singleton Park, Swansea SA2 8PP, UK
a r t i c l e i n f o a b s t r a c t
Article history: Carbon dioxide (CO2) released in the atmosphere is the principal cause of the so-called global warming or
Received 16 February 2010 greenhouse phenomenon. One procedure for the mitigation of CO2 consists of the capture, transport and
Received in revised form 11 April 2010 geological storage (sequestration) of the gas in deep saline aquifers. In this work, a model for predicting CO2
Accepted 27 April 2010
solubility in H2O–NaCl system covering conditions typically encountered in geological sequestration (e.g.,
Available online 31 May 2010
300–500 K, 50–2000 bar, and 1–4 mol salt/kg) has been developed and validated by comparison with available
Keywords:
experimental and theoretical data. The model is a simple one with four adjustable parameters that is based on
Setschenov model an extension of the well-known Setschenov model [Kgs / Kg0 = (P / P0)f(Tr,m) with f(Tr,m) = (0.461 + 0.078 m)
Solubility (− 1 + Tr / (1.0 + 0.1267 m)]. The model is developed by exploiting interesting features in the behavior of
Sequestration solubility data of CO2 in saline water over the above-mentioned ranges of conditions. The model is capable of
Salting out predicting CO2 solubility in saline water with accuracy comparable to that made by multi-parameter models
CO2–NaCl–H2O system (having 20 parameters or more). It is also in very good agreement with the available experimental data with
root-mean-squared-error (RMSE) in the range of 0.03–0.225 mol CO2/kg H2O. The extrapolative power of the
model is acceptable at both ends of its range of applicability.
© 2010 Elsevier B.V. All rights reserved.
1. Introduction studied, i.e., CO2 dissolving in formation water and CO2 forming
carbonate mineral precipitates. In view of findings of this study
The emission of greenhouse gases such as carbon dioxide, which geological mineral fixation is found to be a minor CO2 trapping
results from the combustion of fossil fuels, is a major source of mechanism in natural gas fields. The study suggests that long-term
environmental concern in terms of its relation to the global warming anthropogenic CO2 storage models in similar geological systems
problem [1]. Regardless of the increasing world energy demand and should focus on the potential mobility of CO2 dissolved in water.
how well “zero emission” renewable energy is developed, it is Carbon sequestration, therefore, requires accurate evaluation of the
expected that carbon producing fossil fuels will continue to be the capacity of a saline aquifer for CO2 addition. Accurate prediction of
major energy contributor over the coming decades [2]. CO2 released in CO2 solubility over a wide range of temperature (T), pressure (P) and
the atmosphere, therefore, must be reduced. One procedure towards ionic strength (m) is also of paramount importance for designing
this end consists of the capture, transport, and geological storage separation equipment in chemical or oil-related industries [7]. The
(sequestration) of the gas in deep saline aquifers [3]. Because of the precise knowledge of CO2 solubility in pure water is an essential
available massive underground capacity, as in depleted oil and gas prerequisite for any modeling activity.
reservoirs and deep saline formations, geological sequestration is Usually, equations of state (EOS) are used as thermodynamic
anticipated to be an important choice for carbon capture and storage. models to calculate the phase equilibrium properties of fluid mixtures
A minimum depth of 800 m is recommended to ensure the safe [8]. The disadvantage of such models when used for large-scale
potable water aquifers [4]. At this depth CO2 will reach supercritical storage of CO2 in aquifers is that they involve computational
state (for CO2, the critical temperature (Tc) is 304.2 K, and the critical complexities and are relatively expensive [9]. There is a voluminous
pressure (Pc) is 73.8 bar). amount of experimental equilibrium data on the CO2-brine systems
It is well known that H2O, CO2 and NaCl are the most common that have been used extensively in tuning EOS adjustable parameters.
species in geological fluids [5]. In a recent study by Gilfillan et al. [6] There has also been extensive theoretical effort on modeling CO2
two mechanisms of CO2 trapping in geological storage sites were solubility in aqueous solutions. Although several models on CO2
solubility have been published, few can reliably predict CO2 solubility.
Furthermore, in most cases these are extremely complicated models
⁎ Corresponding author. that require many adjustable parameters that have no physical
E-mail address: ndarwish@aus.edu (N.A. Darwish). significance. For example, Bahadori et al. presented a model that
0011-9164/$ – see front matter © 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.desal.2010.04.056
N.A. Darwish, N. Hilal / Desalination 260 (2010) 114–118 115
requires a total of 48 adjustable parameters to predict the solubility of acceptance criterion have been correlated by a thermodynamic model
CO2 in NaCl–H2O system over only three levels of salt concentration based on Henry's law and an equation of state. According to this study,
[10]. This drastically limits the extrapolative and interpolative power the assumption that the activity coefficients of aqueous CO2 are equal
of such models. to unity is found to be valid up to a solubility of approximately 2 mol%.
For many T–P–m conditions, CO2 solubility values are still Portier and Rochelle [20] presented experimental data on the
unknown and, in general, predicting these data is not accurate. For solubility of CO2 in a mixed salt solution at CO2 sequestration pressure
example, Li and Ngheim [11] presented a model to calculate the and temperature conditions. The data were compared with predic-
solubility of CO2 in pure water of up to 473 K and in NaCl solutions tions made by a thermodynamic model for CO2 solubility in pure
with up to 4.0 M salt. Another example is presented by Harvey and water and in aqueous NaCl solutions of ionic strengths up to 3 m from
Prausnitz [12], who developed an EOS that overestimates CO2 273 to 573 K and from 1 to 300 bar. Sieder and Maurer [21] extended
solubility in NaCl aqueous solutions at elevated pressures. Specifically Peng–Robinson EOS by employing a Debye–Huckel term and a
oriented towards CO2 sequestration purposes, Yuanhui et al., [13] Margules term to allow for the consideration of salts. The extended
reviewed and analyzed studies on the relevant phase equilibria and Peng–Robinson equation of state is then used to predict the phase
densities on the system CO2–H2O–NaCl in the temperature and equilibrium of quaternary systems of the type CO2–H2O–organic
pressure range for disposal of CO2 in deep saline aquifers and solvent-salt.
suggested models in order to obtain more reliable predictions in a Duan and Sun [8] developed a thermodynamic model for the
wide temperature and pressure range. In this review three different solubility of carbon dioxide (CO2) in pure water and in aqueous NaCl
models (the electrolyte nonrandom two-liquid (ELEC-NRTL), the solutions for temperatures from 273 to 533 K, pressures from 0 to
electrolyte NRTL combining with Helgeson model (ENRTL-HG), and 2000 bar, and ionic strength from 0 to 4.3 m. The model is based on a
Pitzer activity coefficient model combining with Helgeson model specific particle interaction theory for the liquid phase and a highly
(PITZ-HG)) are used to calculate the vapor-liquid phase equilibrium of accurate equation of state for the vapor phase. The model is also
CO2–H2O and CO2–H2O–NaCI systems. According to this review, more claimed to predict CO2 solubility in other systems, such as CO2–H2O–
experimental data and more accurate thermodynamic calculations are CaCl2 with an accuracy of about 7%. In the development of their model,
needed in the high temperature and pressure ranges (above 400 K Duan and Sun used an extensive data base on the system CO2–H2O–
and 300 bar). NaCl and documented merits and limitations of some important
To study the effect of T, P, and salinity (m) on CO2 solubility in brine theoretical models, such as that of Li and Ngheim [11], Harvey and
solutions, Al-Anezi et al., [14] determined the solubility of CO2 in Prausnitz [12], and the model of Zuo and Guo [15].
different mixtures of aqueous solutions (normal water and different In this work we develop and employ a simple, Setschenov-type
composition of saline water) at temperature from 298 to 363 K and model to predict the solubility of CO2 in NaCl–H2O system over
pressure from 1 to 2 bar. Zuo and Guo [15] extended Patel-Teja EOS to conditions of interest to geological sequestration of carbon dioxide,
phase equilibrium calculations for electrolyte solutions. This model is i.e., from 300 to 500 K, 50–2000 bar, and 1–4 salt molality. The model
reported to be no more accurate than the Harvey–Prausnitz model for is a simple generalization of Setschenov constant which has only four
CO2 solubility, where it underestimates CO2 solubility by more than adjustable parameters. More importantly, the model is explicit in
12% in 20 wt.% NaCl solutions at high pressures (N1000 bar) and solubility and doesn't require iterative algorithms for its use. The
overestimates CO2 solubility by more than 10% in 6 wt.% NaCl solution model is developed following careful analysis of CO2 Setschenov
at moderate pressures [10]. These models are usually associated with constants over the above-mentioned conditions of temperature,
a large number of tuning parameters and require complex and long pressure and salt concentration. Being a generalization of the simple
computations. A great amount of work has been performed to Setschenov model, which has been proven to be a successful model at
determine the solubility of CO2 in aqueous NaCl solutions based on low salt concentrations, the model developed here requires the
the Peng–Robinson EOS, Henry's law and the scaled-particle theory availability of CO2 solubility in pure water.
[8]. Jaiwen Hu et al., [16] presented a comprehensive review of the
experimental PVTx properties and the thermodynamic models of 2. The proposed mathematical model
CO2–H2O and CO2–H2O–NaCl systems. Among other important
conclusions, this review shows that accuracy of the available data The mathematical model employed in this study is a simple one,
and theoretical models is not sufficient for assessing CO2 sequestra- which uses the framework advanced in 1889 by Setschenov [22], i.e.,
tion. Evans and Powell [17] developed a method for activity
calculations in saline, mixed solvent, and supercritical aqueous !
Cg0
solutions. Non-ideal interactions are represented by an activity Log = Kgs m ð1Þ
Cgs
coefficient term that combines a limited form of Debye–Huckel and
a van Laar formulation, which is reported to be thermodynamically
valid over a wide range of P, T and fluid composition. The model has where Cg0 stands for the solubility of the gas (CO2) in pure water, Cgs is
small number of calibration parameters. Conditions modeled include the solubility of the gas in saline water, Kgs is the Setschenov constant
P, T, m ranges of 200–1400 bar, 773–1173 K, 0–5.56 molality, in kg/mol and m is the molal concentration (mol/kg) of the salt in the
respectively. Koschel et al., [18] investigated the solubility of CO2 in aqueous phase. Originally, this relation was intended to represent
water and in aqueous solution of NaCl by measuring the heat of (salting-out) effects of dissolved salts on gas solubility in water. The
mixing of supercritical CO2 (and H2S) with the liquid phase over a Setschenov relation (Eq. (1)) found good success at low-to-moderate
range of 323.1–373.1 K, at pressures of up to 200 bar, and a salt conditions of temperature (T), pressure (P), and salt concentration
concentration of up to 3 m. The salting-out effect, as measured by (m), where the constant Kgs becomes constant independent of these
Setschenov constant, does not strongly vary with the T, P, or m. This variables. This independence, however, is expected to break down at
work presents an excellent literature review for the subject. In the high values of T, P and m encountered in CO2 sequestration. To
particular, it cites many valuable review studies about both investigate this, Fig. 1 is generated using the model developed in this
experimental measurements and theoretical modeling for the topic study for two levels of temperatures (363 and 513 K) and two levels of
under consideration. salt concentration (1 and 4 m). Shown also in the same figure, are
Diamond and Akinfiev [19] compiled and evaluated experimental predictions made by the Duan and Sun model [8] (referred to
measurements from 25 literature sources of the solubility of CO2 in hereafter as D–S model). It is quite evident from this figure that a
pure water at pressures above 10 bar. Data that passed a certain single value of the Setschenov constant cannot describe gas solubility
116 N.A. Darwish, N. Hilal / Desalination 260 (2010) 114–118
Tr
f ðTr ; mÞ = ð0:461 + 0:078 × mÞ × −1 1≤m≤4
ð1:0 + 0:1267 × mÞ
ð6Þ
Fig. 2. Comparison of CO2 solubility in aqueous NaCl solution at 323 and 373 K, Fig. 4. Comparison of CO2 solubility in aqueous NaCl solution at 523 K, pressures of up
pressures of up to 200 bar and salt concentration of 1 and 3 m. to 2000 bar and salt molality of up to 4 m.
D–S [8] model, which is based on a specific particle interaction theory for model predicts higher values of CO2 solubility. Both models also are in
the liquid phase and an equation of state for the vapor phase. The simple excellent agreement with the experimental data of Takenouchi and
model of this work is in perfect agreement with the D–S model over the Kennedy [23]. The RMSE values of CO2 solubility at 523 K as predicted
whole pressure range at 473 and 1.09 salt molality. Both models equally by this model for both salt concentrations of 1.0 and the 4.0 M are 0.12
reproduce the experimental data at 473 K and m = 1.09. For the 423 K and 0.22 mol/kg, respectively. At 363 K, where there is no experi-
isotherm, the figure shows a perfect match between the two models of mental data to compare with, Fig. 5 shows excellent agreement
up to a pressure of 1200 bar, after which the model in this work deviates between the two models at salt concentrations of 1 and 2 m over the
positively and progressively from the D–S model. Both models also start whole pressure range of up to 2000 bar. For the 4-m isotherm,
overestimating CO2 solubility after a pressure of around 800 bar. however, the two models start deviating from each other after a
Another comparison of CO2 solubility in aqueous NaCl solution at pressure of 200 bar. For pressures of up to 1700 bar, the difference in
523 and 363 K, up to 2000 bar and salt concentration from 1 to 4 m are the predicted CO2 solubility from the two models lies within ±
displayed in Figs. 4 and 5, respectively. At 523 K, as shown in Fig. 4, the 0.1 mol/kg. Actually, the trend taken by this 4-m isotherm in Fig. 5 as
two models are in excellent agreement with each other of up to a predicted by the D–S [8] model looks abnormal as it tends to make an
pressure of 1400 bar over all salt concentrations. After 1400 bar, this intersection with the 2-m isotherm predicted by the same model.
Fig. 3. Comparison of CO2 solubility in 1.09 m aqeuous NaCl solution at 423 K and 473 K Fig. 5. Comparison of CO2 solubility in aqueous NaCl solution at 363 K, pressures of up
over a pressure range of 50 to 2000 bar. to 2000 bar and salt concentration of 1, 2, and 4 m.
118 N.A. Darwish, N. Hilal / Desalination 260 (2010) 114–118
Fig. 6. Performance of the simple model used in this study in predicting CO2 solubility in
1.09 m NaCl–H2O system.