PROCESS SYSTEMS ENGINEERING Prediction of Properties in Homologous Series with a Shortcut QS2PR Method Georgi St. Cholakov Dept. of Organic Synthesis and Fuels, University of Chemical Technology and Metallurgy, Sofia 1756, Bulgaria Roumiana P. Stateva Institute of Chemical Engineering, Bulgarian Academy of Sciences, Sofia 1113, Bulgaria Mordechai Shacham Dept. of Chemical Engineering, Ben-Gurion University, Beer-Sheva 84105, Israel Neima Brauner School of Engineering, Tel-Aviv University, Tel-Aviv 69978, Israel DOI 10.1002/aic.11059 Published online November 30, 2006 in Wiley InterScience (www.interscience.wiley.com). For prediction of properties of pure compounds in homologous series a shortcut ver- sion of the authors’ QS2PR technique is introduced. The structure–structure relation- ship between a target compound and three predictive compounds from its homologous series is described in the short version with the number of carbon atoms and one eas- ily calculated noncollinear molecular descriptor. This correlation provides predictions of high precision for properties of members of the homologous series, if property data for interpolation are available. It can also be used for short-range extrapolation with precision similar to that of asymptotic relationships. The advantages of the proposed method over the existing methods are that with the structural correlation parameters it predicts all structure-dependent properties of the target and enables estimation of its missing properties even if properties for only three predictive compounds are avail- able. The method can also use as a descriptor the normal boiling temperature. Ó 2006 American Institute of Chemical Engineers AIChE J, 53: 150–159, 2007 Keywords: property prediction, QSPR, molecular descriptors, homologous series Introduction Pure component properties are essential to chemical engi- neering calculations in such processes as steady-state and dynamic simulation, process and product design, environ- mental impact assessment, and hazard and operability analy- sis. 1 However, published experimental data are scarce and often contradictory. Moreover, there are compounds for which some properties cannot be measured (such as when the compound decomposes before reaching its critical proper- ties), and extrapolated values must be used. Therefore, it is very important to develop reliable prediction techniques for such compounds. Correlations using only the number of carbon atoms for predicting particular properties of compounds of some ho- mologous series have been suggested by many authors. 2,3 They are usually based on the Flory–Huggins ‘‘mers’’ approach. Because the carbon number–property relationships Correspondence concerning this article should be addressed to M. Shacham, at shacham@bgumail.bgu.ac.il. Ó 2006 American Institute of Chemical Engineers AIChE Journal January 2007 Vol. 53, No. 1 150