Extended Flexibility for GE-Models and Simultaneous Description of VLE and LLE Using a Non-Linear Transformation of the Concentration Dependence
Jürgen Rarey
DDBST GmbH, 26121 Oldenburg, Germany
Industrial Chemistry, Institute for Pure and Applied Chemistry, University of Oldenburg, 26111 Oldenburg, Germany
Abstract
A method to increase the flexibility of the composition dependence of GE-models or equations of state is proposed. The formalism is based on a non-linear transformation of the concentration space and can be applied to any mixture model without re-deriving the model equations. Any number of additional binary parameters can be introduced, which show nearly no intercorrelation with the original model parameters. The reason is that the additional parameters only effect the form, but not the size and symmetry of e.g. GE.
The mathematical formalism for the introduction into GE-models was derived and applied to the UNIQUAC model. The modified model (FlexQUAC) was then used to regress a large number of VLE data sets (approx. 4000), whereby significant improvements were observed for systems with medium to large positive deviation from Raoult’s law. Due to the very small intercorrelation with the original model parameters it was assumed that the additional parameters have no negative effect on the prediction of multicomponent behavior from binary data. This was verified for 13 carefully selected systems.
The motivation for this development was not primarily the improvement of the regression capability but the simultaneous description of VLE and LLE data. Typically models like NRTL or UNIQUAC predict way too large immiscible regions from VLE data or too low VLE separation factors from LLE. This shortcoming of the original models is corrected by the concentration space transformation. Examples for the simultaneous description of VLE and LLE for binary and ternary systems are presented and compared to the corresponding results of the UNIQUAC equation.