Thermodynamic modeling of CO2 mixtures

Martin Gamel Bjørner

Research output: Book/ReportPh.D. thesis

1314 Downloads (Pure)


Knowledge of the thermodynamic properties and phase equilibria of mixtures containing carbon dioxide (CO2) is important in several industrial processes such as enhanced oil recovery, carbon capture and storage, and supercritical extractions, where CO2 is used as a solvent. Despite this importance, accurate predictions of the thermodynamic properties and phase equilibria of mixtures containing CO2 are challenging with classical models such as the Soave-Redlich-Kwong (SRK) equation of state (EoS). This is believed to be due to the fact, that CO2 has a large quadrupole moment which the classical models do not explicitly account for.
In this thesis, in an attempt to obtain a physically more consistent model, the cubicplus association (CPA) EoS is extended to include quadrupolar interactions. The new quadrupolar CPA (qCPA) can be used with the experimental value of the quadrupolemoment and with or without introducing an additional pure compound parameter. In the absence of quadrupolar compounds qCPA reduces to CPA, which itself reduces toSRK in the absence of association.
As the number of adjustable parameters in thermodynamic models increase, the parameter estimation problem becomes increasingly complicated due to parameter identifiability issues. In an attempt to quantify and illustrate these issues, the uncertainties in the pure compound parameters of CO2 were investigated using qCPA as well as different CPA approaches. The approaches employ between three and five parameters.
The uncertainties in the parameters were propagated to physical properties,vapor liquid equilibria (VLE), and liquid-liquid equilibria (LLE) using Monte Carlosimulations.The uncertainties in the pure compound parameters were found to be negligible formodeling approaches which employed three adjustable parameters. For modeling approaches with more than three adjustable parameters, however, the uncertainties in the pure compound parameters were significant. As a result the propagated errors weresubstantial for certain output properties. The uncertainties in VLE were for instancemuch larger when qCPA was employed with four parameters rather than three. The uncertainty analysis indicated that the parametrization of multi-parameter models isat least as important as the specific model term.
The new qCPA and several CPA approaches were extensively evaluated for their ability to predict the thermodynamic properties of pure CO2. The predictions of these pure compound properties were satisfactory with qCPA, although similar predictions were achieved with the other CPA approaches. The model was subsequently evaluated for its ability to predict and correlate the binary VLE and LLE of mixtures containing CO2 and n-alkanes, water, alcohols, or quadrupolar compounds. For these binary mixtures qCPA appeared to other systematically improved predictions and correlations as compared to the cases where quadrupolar interactions were ignored. The improvements were particularly pronounced for mixtures of CO2 and hydrocarbons where the model is almost fully predictive.
Finally qCPA was evaluated for its ability to predict the phase equilibria of multicomponent mixtures containing CO2 and n-alkanes, water, and/or alcohols. A single binary interaction parameter was employed in qCPA for most binary combinations. Both qCPA and the best CPA approaches typically performed satisfactorily and predicted the general behavior of the systems, but qCPA used fewer adjustable parameters to achieve similar predictions.
It has been demonstrated that qCPA is a promising model which, compared to CPA, systematically improves the predictions of the experimentally determined phase equilibria between binary and ternary mixtures containing CO2 and other non-quadrupolar compounds. However, for mixtures containing two quadrupolar compounds, or aquadrupolar and polar compound, considerable uncertainty remains as to whether these mixtures are handled in the best possible way. When binary interaction parameters were employed to correlate experimental phase equilibria data, both qCPA andCPA yielded similar correlations - and predictions in the multicomponent case.
Original languageEnglish
Place of PublicationKgs. Lyngby
PublisherTechnical University of Denmark
Number of pages215
Publication statusPublished - 2016


Dive into the research topics of 'Thermodynamic modeling of CO2 mixtures'. Together they form a unique fingerprint.

Cite this