Support vector machine and CPA EoS for the prediction of high-pressure liquid densities of normal alkanols

Abstract

The already numerous industrial applications of alcohols is continuously growing to even higher significance. Accurate estimations of the physical properties of various alcohols within wide ranges of pressures and temperatures are necessary in the design and optimum operation of such processes. In this study, high-pressure liquid densities of the first six members of the normal alkanol family were predicted as a function of temperature and pressure. This was done using two different approaches: The computer method of support vector machine (SVM) and the thermodynamic model of Cubic Plus Association Equation of State (CPA EoS). A total of 2979 experimental data on alcohol density, also covering high pressure and high temperature conditions, were obtained from literature. The results of the calculations show that the value of AARD% for all of the 2979 data was 0.06% for the SVM computer model and 1.36% for the thermodynamic CPA model. © 2014 Taiwan Institute of Chemical Engineers.