Loading...

Prediction of liquid-liquid equilibrium behavior for aliphatic+aromatic+ionic liquid using two different neural network-based models

Hakim, M ; Sharif University of Technology | 2015

763 Viewed
  1. Type of Document: Article
  2. DOI: 10.1016/j.fluid.2015.03.018
  3. Publisher: Elsevier , 2015
  4. Abstract:
  5. In this study, the liquid-liquid phase behavior of aromatic compound. +. aliphatic compound. +. ionic liquid (IL) ternary systems was estimated by using two artificial neural networks (ANN) developed based on back propagation (BP) and hybrid group method of data handling (GMDH). Molar ratio of aliphatic compound, aromatic compound, and IL as well as temperature, molecular weight ratio of aliphatic compound to IL, and molecular weight ratio of aromatic compound to IL were chosen as the inputs to the networks. Additionally, the mole fraction of components in final alkane-rich phase and IL-rich phase was considered as desired outputs. The best topology of the BP-ANN model was found as (6-8-4). Besides, corresponding quadratic expression for each neuron of the hybrid GMDH model was determined. The obtained results by the hybrid GMDH model provided a near to accurate prediction of phase behavior with average relative deviation (ARD%) of 7.27%. However, the BP-ANN model presented a slightly more accurate prediction with ARD% of 4.49%. Furthermore, the comparison of the results with those obtained by the famous thermodynamic models, proved the preciseness of the proposed networks
  6. Keywords:
  7. Artificial neural network ; Aromatic compounds ; Aromatization ; Backpropagation ; Data handling ; Forecasting ; Ionic liquids ; Molecular weight ; Neural networks ; Phase equilibria ; Ternary systems ; Topology ; Accurate prediction ; Aliphatic compound ; Group method of data handling ; Liquid liquid equilibrium ; Liquid-liquid phase ; LLE ; Relative deviations ; Thermodynamic model ; Liquids
  8. Source: Fluid Phase Equilibria ; Volume 394 , May , 2015 , Pages 140-147 ; 03783812 (ISSN)
  9. URL: http://www.sciencedirect.com/science/article/pii/S0378381215001302