Prediction of permeate flux during electric field enhanced cross-flow ultrafiltration—A neural network approach

Electric field enhanced cross flow ultrafiltration of aqueous solution of bovine serum albumin (BSA) has been carried out in a thin rectangular channel using 30,000 molecular weight cut-off membrane over a wide range of operating conditions. Detailed parametric study has been performed to observe the effect of feed concentration, electric field, cross flow velocity and pressure on permeate flux. An artificial neural network approach with single hidden layer has been systematically developed to predict permeate flux using some of the experimental results. The rest of the experimental results are successfully compared with the model predictions with less than 1% average error. The analysis reveals significant beneficial effect of electric field on permeate flux. In this work, emphasis has been given on random selection of training data and small network. The developed trained network is also able to capture the non-linear prediction of permeate flux with new operating conditions which has not been used in the training process and enhances the physical understanding of the process.