A Simple Group Contribution Model to Predict Thermal Conductivity of Pure Ionic Liquids
Soares, Luan Henrique
Guirardello, Reginaldo
Rolemberg, Marlus
Download PDF

How to Cite

Soares L.H., Guirardello R., Rolemberg M., 2019, A Simple Group Contribution Model to Predict Thermal Conductivity of Pure Ionic Liquids, Chemical Engineering Transactions, 74, 1195-1200.
Download PDF


Ionic liquids (ILs) have been attracted considerable attention in separation and purification processes as green solvents. Researchers use ILs in many areas, such as micro-extraction and catalysis in biodiesel production. However, there are still few studies about it. A comprehension of ILs thermophysical properties, for example, heat capacity and thermal conductivity may improve new technological processes and minimize energy costs. Since these fluids can be composed of different ions, it is harder to obtain these properties for several ILs. Therefore, this paper evaluated a group contribution (GC) model from the literature due to be accurate and straightforward for thermal conductivity (k) prediction of pure ionic liquids in a wide range of temperatures at atmospheric pressure. Once ILs containing C(CN)3-, B(CN)4-, DCA-, CH3COO-, HOPO2-, SER-, LIS-, CYS-, PRO-, TAU-, THR-, VAL- and FAP- groups are commercially used, and the authors did not use an experimental database of ILs containing these anions, the model is unable to predict k for them. In this manner, from ILThermo, a larger experimental database including these anions was used to propose GC parameters for them and reestimate the others. This was done by minimizing the sum of the square of the residues comparing calculated and experimental value to obtain each group contribution parameter, using the generalized reduced gradient algorithm in Excel® and VBA programming. The revised model obtained results with mean deviation of 1.16 % for k prediction, including 13 more GC parameters. Both models were compared to predict k for other data set, not used in the parameters estimation. The proposed model was better in all evaluated cases and increased the amount of ILs to predict k.
Download PDF