F Farshad - Computer Aided Process Engineering (CAPE) Lab, Department of Chemical Engineering, Iran University of Science and Technology (IUST),Narmak, Tehran, Iran
N Kasiri - Computer Aided Process Engineering (CAPE) Lab, Department of Chemical Engineering, Iran University of Science and Technology (IUST),Narmak, Tehran, Iran
T Mohammadi - Research Lab for Separation Processes, Department of Chemical Engineering, Iran University of Science and Technology (IUST),Tehran, Iran.
J Ivakpour - Petroleum Refining Division, RIPI (Research Institute of Petroleum Industry),

in this paper, the results of Mathematical Model (MM) and an Artificial Neural Network (ANN) model for separating dilute alcoholic mixture by pervaporation processusing PDMS membrane have been compared. Mathematical model is developed based on thermodynamic phase equilibrium between solvent bulk and solvent in polymerphase. Another model is established based on black box modeling. MLP feed forward neural network with one hiddenlayer and three neurons is selected as an optimized network to model this process. For the evaluation purpose and comparison between two models, a set of experimental data has been used.This data set has been collected on dilute aqueous Methanol, Ethanol and Phenol solution. Result indicates that ANN model ismore capable than MM to predict permeation flux especially in higher concentrations and temperature.