Development of corresponding states model for estimation of the surface tension of chemical compounds

Authors

  • Farhad Gharagheizi,

    1. Dept. of Chemical Engineering, Buinzahra Branch, Islamic Azad University, Buinzahra, Iran
    Search for more papers by this author
  • Ali Eslamimanesh,

    1. MINES ParisTech, CEP/TEP - Centre Énergétique et Procédés, Fontainebleau, France
    Search for more papers by this author
  • Mehdi Sattari,

    1. Saman Energy Giti Co., Tehran, Iran
    Search for more papers by this author
  • Amir H. Mohammadi,

    Corresponding author
    1. MINES ParisTech, CEP/TEP - Centre Énergétique et Procédés, Fontainebleau, France
    2. Thermodynamics Research Unit, School of Chemical Engineering, University of KwaZulu-Natal, Durban, South Africa
    • Dept. of Chemical Engineering, Buinzahra Branch, Islamic Azad University, Buinzahra, Iran
    Search for more papers by this author
  • Dominique Richon

    1. Thermodynamics Research Unit, School of Chemical Engineering, University of KwaZulu-Natal, Durban, South Africa
    2. Technical University of Denmark, Center for Energy Resources Engineering (CERE), Dept. of Chemical and Biochemical Engineering, Kgs. Lyngby, Denmark
    Search for more papers by this author

Correspondence concerning this article should be addressed to A. H. Mohammadi at amir-hossein.mohammadi@mines-paristech.fr.

Abstract

The gene expression programming (GEP) strategy is applied for presenting two corresponding states models to represent/predict the surface tension of about 1,700 compounds (mostly organic) from 75 chemical families at various temperatures collected from the DIPPR 801 database. The models parameters include critical temperature or temperature/critical volume/acentric factor/critical pressure/reduced temperature/reduced normal boiling point temperature/molecular weight of the compounds. Around 1,300 surface tension data of 118 random compounds are used for developing the first model (a four-parameter model) and about 20,000 data related to around 1,600 compounds are applied for checking its prediction capability. For the second one (a five-parameter model), about 10,000 random data are applied for its development, and 11,000 data are used for testing its prediction ability. The statistical parameters including average absolute relative deviations of the results form dataset values (25 and 18% for the first and second models, respectively) demonstrate the accuracy of the presented models. © 2012 American Institute of Chemical Engineers AIChE J, 59: 613–621, 2013

Ancillary