Equations of state for mixtures : an optimization based approach

To bring the costs of carbon capture and storage plants (CCS) down to more competitive prices, there have been developments to optimize the plant’s processes. As a consequence, the fluid’s thermodynamic properties need to be predicted with higher accuracy. This study analyzes one possible route for improving the accuracy of the prediction of thermodynamic properties: the mathematical optimization method used for the fitting of the experimental data. Using recent experimental data obtained by Mantovani et al. [1] for CO2 –based mixtures (CO2-N2, CO2-Ar, CO2-O2), both the Levenberg-Marquardt and Trust-Region methods were compared for the nonlinear fitting of the Peng-Robinson (PR) cubic equation of state. By applying the van der Waals mixing rules, the binary interaction parameters (BIP) were optimized using a Matlab code and the built-in function nonlincurvefit with these two different algorithms. As a result, there appears to be no influence of the optimization method in the accuracy of the predicted properties using the optimized BIP. Nonetheless, the use of the PR equation of state to predict the volumetric properties of the CO2-based mixtures cited above was carried out successfully and provided results with deviations in the range 1.2% to 2.8%.