Artificial Neural Network Based Model for Reverse Osmosis Water Desalination Membrane

Availability of fresh drinking water continues to be a major concern around the globe. One of the most effective methods to provide fresh drinking water is through the use of desalination plants. Over the past years, reverse osmosis (RO) membrane technology has developed significantly, gained popularity, and become the most widely used desalination technology for both seawater and brackish water, due to advancements in membrane technology and improvements in energy efficiency resulting in higher performance and reduced cost. On the other hand, the modeling of RO membranes is challenging due to the complexity of the analytic models and the lack of the membrane’s characteristics. This work provides a multi-input-output Artificial Neural Network (ANN) model to predict the membrane’s characteristics at different flow conditions. ANN is a biologically inspired computer program that is constructed to mimic the way in which the human brain handles information, and it has been used widely to model different systems especially nonlinear and/ or complex systems. This model can be used to reduce the modeling complexity of the membrane and to predict the freshwater flow rate and salinity at different conditions by using a set of experimental data for the membrane. The model uses the feedwater pressure, flowrate, and salinity to predict the freshwater flowrate and salinity. The model shows correction value (R)of 0.985 for the testing data and it is capable to predict the membrane’s characteristics at pressure values from 90 to 160 psi.