Get access

Ternary gas permeation through synthesized pdms membranes: Experimental and CFD simulation basedon sorption-dependent system using neural network model

Authors

  • Ehsan Farno,

    1. Computer Aided Process Engineering Lab (CAPE), Faculty of Chemical Engineering, Iran University of Science and Technology (IUST), Narmak, Tehran, Iran
    Search for more papers by this author
  • Mashallah Rezakazemi,

    1. Research Centre for Membrane Separation Processes, Faculty of Chemical Engineering, Iran University of Science and Technology (IUST), Narmak, Tehran, Iran
    Search for more papers by this author
  • Toraj Mohammadi,

    Corresponding author
    1. Department of Chemical Engineering, Islamic Azad University, Tehran, Iran
    • Department of Chemical Engineering, Islamic Azad University, South Tehran Branch, 11365-4435, Tehran, Iran. E-mail: torajmohammadi@iust.ac.ir

    Search for more papers by this author
  • Norollah Kasiri

    1. Computer Aided Process Engineering Lab (CAPE), Faculty of Chemical Engineering, Iran University of Science and Technology (IUST), Narmak, Tehran, Iran
    Search for more papers by this author

Abstract

In this study, a predictive model for the separation of gases via a polydimethylsiloxane (PDMS) membrane has been developed. This model takes into account the effects of gas composition and pressure at the membrane surfaces on the gas sorption and diffusion coefficients in the membrane. Computational fluid dynamics (CFD) modeling has been employed in order to predict the behavior of a gas mixture containing C3H8, CH4, and H2 at various operating conditions and three zones (upstream, downstream, and membrane body). Artificial neural network (ANN) modeling has been applied to predict sorption and diffusion coefficients of each component of the gas mixture in the membrane. A procedure of calculation has been applied to combine the CFD modeling and the ANN modeling in order to predict sorption, diffusion, and composition of each component at various sites of the membrane. The results determined using the developed prediction model have been found to be in agreement with those determined using experimental investigations with an average error of 10.21%. POLYM. ENG. SCI., 54:215–226, 2014. © 2013 Society of Plastics Engineers

Get access to the full text of this article

Ancillary