Predicting gas flux in silicalite-1 zeolite membrane using artificial neural networks

In this paper, artificial neural network (ANN) as a powerful tool for solving complicated problems is used to predict gas flux through silicalite-1 zeolite membrane. Network training was fulfilled using a collected database of the practiced operation including gas flux under various operating conditions (e.g. feed pressure and operating temperature) with different kinetic diameter of the permeating species (e.g. CO2, O2, N2 and CH4). Trying various types of the networks, a network with one hidden layer including 5 neurons was found to be optimum. Performance of the ANN model was compared with statistical analysis using datasets that were kept apart from the original database. The results showed that there is an excellent agreement between the experimental data and the predicted values, with high correlation (R2 = 0.9952) and less error (RMSE = 8.9E−4). In addition, sensitivity analysis revealed that the input feed pressure is the most sensitive parameter on the output gas flux.

► A higher correlation and lesser error between the predicted and measured fluxes.
► The model with architecture 3-5-1 gives the best correlation and minimum error.
► Feed pressure is the most influencing parameter on the flux through silicalite-1.