MODELLING OF SUPERCRITICAL FRACTIONATION COLUMNS: RATE-BASED MODEL OR EQUILIBRIUM STAGE MODEL?

GAÑAN, NICOLAS ALBERTO; CAMY, SEVERINE; CONDORET, JEAN-STEPHANE; MORCHAIN, JEROME

Niza

Congreso; 10th European Conference of Chemical Engineering (ECCE 2015); 2015

Fractionation of liquid mixtures using a supercritical solvent (mostly CO2) is an underexploited technology although its potential for industrial applications is significant. Indeed, a better understanding of the process is still needed for its full development, as stated by Brunner. Most of engineering studies have utilized conventional chemical engineering tools, similar to those for liquid-liquid extraction, i.e, the use of ternary diagrams for graphical design. But, when thermodynamic behavior of the mixture can be satisfactorily described using standard thermodynamic models, such as cubic equations of state, commercial process simulation softwares are useful tools to model a fractionation column and/or the entire process, including solute recovery and solvent recycling. Typically, with these softwares, counter-current contactors are represented by a succession of equilibrium theoretical stages. From comparison with experiments, it was possible to obtain values of Height Equivalent to Theoretical Stage, which were found rather high, indicating that mass transfer has to be better understood to be improved. Although convenient, the equilibrium stage approach is somewhat limited for that purpose, giving the low generalization capability of this concept. A more reliable approach is to model the multicomponent mass transfer inside the contactor thanks to the so-called ?rate-based? or ?non-equilibrium? approach, taking into account mass-transfer kinetics between phases. This was first proposed for supercritical fractionation by Martin and Cocero with a model based on Maxwell-Stephan diffusion2. A similar work is proposed in this work but using the more conventional approach of liquid and fluid mass transfer coefficients. A dynamic numerical model, including axial dispersion in both phases, has been elaborated to provide fast numerical computation. The model, developed using Matlab, was based on the numerical resolution of differential mass balances for each component over the column height. The thermodynamic equilibrium at the interface was calculated using the Soave-Redlich-Kwong equation of state (SRK) modified by Boston Mathias with PSRK mixing rules and UNIQUAC activity coefficient models. This modelling was validated by comparison with experimental fractionation of aqueous solutions of isopropanol, from our recent works and compared with results of the equilibrium stage model. The effect of different operating conditions and parameters (such as temperature, pressure, flow rates and feed composition) on purity and recovery in extract and raffinate phase was studied. The rate-governed model was proved to be more informative and more accurate, especially for the low height laboratory column (17.5 mm ID and 2 m height) which was used in our case.