Prediction of sulfuric acid solution in the vacuum membrane distillation process using Artificial Neural Network

This paper presented an effective approach utilizing Artificial Neural Network (ANN) to predict the vacuum membrane distillation (VMD) process of sulfuric acid solution. Firstly, mathematical models have been developed and the influence mechanisms of critical parameters on the transfer resistance in the VMD process were explored via a simulation process. Furthermore, an ANN model from the experimental data under actual working conditions, was developed and applied to simulate the effects of critical parameters on the VMD process performance including membrane flux and gained output ratio (GOR). The simulation results revealed that decreasing feed concentration or vacuum side pressure and increasing feed temperature or feed velocity would contribute to the reduction of total transfer resistance. Then, the operation data at multiple working conditions using actual sulfuric acid solution was divided into three sets (training, validation and test) to develop the ANN model. The trained ANN model was subsequently observed to have good agreement between the predicted and experimental data. From the subsequent simulations with various variable controls, increasing feed concentration or vacuum side pressure led to the declines of membrane flux and GOR, while increasing feed temperature or feed flow rate were helpful to improve membrane flux and GOR. Obviously, the developed ANN model performed well to predict the VMD process performance with minimum error, which could be effectively utilized to predict and optimize the VMD system for industrial wastewater treatment without performing a complicated, expensive and dangerous experimental test.