The New Jersey Institute of Technology's
Electronic Theses & Dissertations Project

Title:	Prediction of heats of mixing by group contribution methods
Author:	Rupp, William Walter
View Online:	njit-etd1982-012 ([ix], 187 pages ~ 6.8 MB pdf)
Department:	Department of Chemical Engineering and Chemistry
Degree:	Master of Science
Program:	Chemical Engineering
Document Type:	Thesis
Advisory Committee:	Tassios, Dimitrios P. (Committee chair) McCormick, John E. (Committee member) Tomkins, R. P. T. (Committee member)
Date:	1982-05
Keywords:	Heat of mixing UNIFAC (Computer program)
Availability:	Unrestricted
Abstract:	Three group contribution methods used to predict heats of mixing were tested: AGSM, UNIFAC with temperature dependent parameters and the modified UNIFAC of Skjold-Jor-gensen et al. The modified UNIFAC provided the best results and was adopted in this study. Primary parameters were obtained for ten groups with alkanes. The simultaneous correlation of cyclic and straight-chain alkanes gave poor results. Attempts to improve the error by changing the group area parameter for cyclic CH2 failed. A major problem in correlating the experimental data was the multiplicity of roots. The best set of parameters was obtained by using a grid approach for the initial parameters. The model can be used in the temperature range 0° to 100°C. 104 systems from the literature were tested of which 78 had a prediction error of less than 20 percent. High errors were observed for some isomers, and long-chain molecules.