Title: Artificial neural network modeling and response surface methodology of desalination by reverse osmosis
 
Journal: Desalination
Authors: M. Khayet , C. Cojocaru and M. Essalhi
Institute: Department of Applied Physics I, Faculty of Physics, University Complutense of Madrid, Av. Complutense s/n, 28040 Madrid, Spain

The original and creativity of paper: The study focused on applying response surface methodology (RSM) and artificial neural networks (ANN) techniques to predict the performance index of RO desalination process.  Also, the efficiency of RMS and ANN techniques was compared.
 
Summary: 

Predictive models (i.e. RSM and ANN) were develop and applied to simulate and optimize RO desalination process. All experiments were run using sodium chloride solution. Pilot scale RO unit with polyamide thin film composite membrane, in spiral wound configuration were used for filtration test. The results found that RSM models valid for different ranges of feed salt concentrations, while ANN model was valid over the whole range of feed salt concentration demonstrating its ability to overcome the limitation of the quadratic polynomial model obtained by RSM.

Specification of ANN
-    ANN approach provides a global model to explain RO performance of the pilot plant in a wide range of feed salt concentration.
-    ANN does not need a standard experimental design to build the model.
-    ANN model is flexible to add new experimental data to develop a better and trustable model.
-    A greater number of experiments are necessary for ANN model development.
-    ANN provides better optimum conditions than RSM and also represents the global optimal solution for the tested RO pilot plant. 
-    Performance index of RO desalination is likely independent from feed flow rate due to it has smallest non-linear effect on the performance index. 
-    The effect of the temperature of feed solution is higher when operating pressures is higher. 

Specification of RSM
-    RSM unable to develop a global model to predict the RO performance over a wide range of salt concentration in feed solution.
-    The most important effects on the RO performance index were found to be: (1) salt concentration in feed; (2) operating pressure; and (3) feed temperature.
-    Feed flow rate showed insignificant effect on the RO performance index at high salt concentrations and its effect can be negligible at low salt concentrations.

 
Application: The paper provides the useful information that ANN model presented the better performance than RSM model to simulate and optimize RO desalination process. Thus, ANN can be considered for further study.
  
By Monruedee Moonkhum
Email: moon@gist.ac.kr