Consistent Prediction of Properties of Systems with Lipids

Equilibria between vapour, liquid and/or solid phases, pure component properties and also the mixture-phase properties are necessary for synthesis, design and analysis of different unit operations found in the production of edible oils, fats and biodiesel. A systematic numerical analysis is employed to determine the needs of phase equilibria and related properties in processes such as Deodorization, Dry Fractionation, Solvent Extraction and Biodiesel Production.

Other important use for the data and analysis is in property model development for correct and consistent property prediction. Lipids are found in almost all mixtures involving edible oils, fats and biodiesel. They are also being extracted for use in the pharma-industry. A database for pure components (lipids) present in these processes and mixtures properties has been developed and made available for different applications (model development, property verification, property prediction, etc.).

The database has verified data for fatty acids, acylglycerols, fatty esters, fatty alcohols, vegetable oils, biodiesel and minor compounds as phospholipids, tocopherols, sterols, carotene and squalene, together with a user friendly interface that allows an easy search/retrieve of the data (pure component and mixture properties).

In total, there are 240 different pure compounds, 229 different phase equilibrium systems (72 VLE, 85 LLE, 72 SLE), 63 solid solubility systems, and around 4500 data points on phase equilibria. Critical properties, enthalpy of formation, heats of combustion, heats of fusion, normal melting point, normal boiling point, and also temperature dependent properties such as vapour pressure, liquid enthalpy, heats of vaporization, liquid density, liquid viscosity and surface tension are present.

Also the parameters for different well-known thermodynamic models (UNIFAC, UNIQUAC, NRTL) have been added to the database. There is a parameter regression option for those systems for which model parameters have not been reported but experimental data can be found in the database. The available data in literature and the well-known property models have been further extended with group-contribution combined with atom connectivity based models.

Some of these models are UNIFAC-CI, PC-SAFT (for pure component properties only) and the NRTL-SAC. In this paper, details of the data available in the database together with an analysis of the data will be presented. The analysis checks for consistency of pure component properties and their use in mixture properties.

Thermodynamic consistency tests such as the Van Ness Test, Herrington Test (Area Test), Point Test (Differential Test) and Infinite Dilution Test have also been performed. Based on the analysis, an uncertainty score is given, which is then used to define the weights of data-sets to be used for property model parameter regression.

The performance of selected models for pure component properties as well as mixture properties when applied to systems containing lipids will be highlighted.