Evaluation of a mathematical model using experimental data and artificial neural network for prediction of gas separation

C3H8 separation from CH4 and H2 using a synthesized PDMS membrane: Experimental and neural network modeling

“…They reported a good consistency between experimental data and ANN results, but did not focus on selecting an optimum structure of ANN [47].…”

“…Modeling of submerged MBR treating cheese whey wastewater Curcio et al [40] 2006 UF FFNN-MLP SL-BP Reduction and control of flux decline in cross-flow UF Wang et al [41] 2006 GS FFNN-RBF SL-BP Modeling of hydrogen recovery from refinery gases Sahoo and Ray [42] 2006 FT FFNN-RBF SL-BP-LMT Prediction of flux decline in cross-flow membranes Shahsavand and Pourafshari Chenar [43] 2007 GS FFNN-MLP and RBF SL-BP Modeling the separation of CO2 from CH4 using hollow fiber module Al-Zoubi et al [44] 2007 NF FFNN-MLP SL-BP Modeling the rejection of sulphate and potassium salts by NF Liu and Kim [45] 2008 MF FFNN-MLP SL-BP-LMT Modeling of membrane fouling Sadrzadeh et al [46] 2008 ED FFNN-MLP SL-BP-LMT Modeling of lead ions separation from wastewater using ED Peer et al [47] 2008 from a gas stream using a glassy polymeric membrane (purchased from Tianbang membranes) was investigated at various operating conditions including operating temperature, the feed-side pressure, the permeate-side pressure, the residue-side pressure, the feed gas flux, and the feed-hydrogen concentration [41].…”

Prediction of Equilibrium Conditions for Hydrate Formation in Binary Gaseous Systems Using Artificial Neural Networks

“…[4] An approach using artificial neural networks (ANNs) meets the above requirements and is commonly used for solving complex, nonlinear problems without requiring knowledge of the relationships of the input-output data. [13][14][15][16][17] There are some reports on the use of ANNs to determine phase diagrams, [5,18] dissociation and/or equilibrium formation conditions, [19][20][21] stability zones, [22] and predict the inhibition of clathrate hydrates. [23] In one-step learning by optimization of the numbers of hidden neurons, the network can be trained for the existing correlative patterns of variables.…”

Modelling of multicomponent gas separation with asymmetric hollow fibre membranes—methane enrichment from biogas

“…Moreover, the calculation procedure of the iterative shooting method involves significant computational time and effort. To overcome mathematical complexities involved, different modifications and approaches of solving the model equations have been proposed (Kovvali et al, 1994;Coker et al, 1998;Kaldis et al, 2000;Marriott et al, 2001;Marriott and Sørensen, 2003;Jiang and Kumar, 2008;Peer et al, 2008;Makaruk and Harasek, 2009). Kovvali et al (1994) used a linear approximation to represent the feed and permeate compositions at certain intervals along the fibre length, and the solution accuracy thus entirely depends on the number of intervals along the fibre.…”