Neural network-based modelling of the adequate chlorine dosage for drinking water disinfection

A neural network modelling approach has been developed to estimate the disinfectant dose adjustments required during water re-chlorination in storage tanks. The approach is based on representative operational and water quality historical data which intrinsically characterize the operators' use of know-how in their routine tasks. The development of the model requires the elimination of the historical cases in which re-chlorination results were inadequate. The results obtained for the model demonstrate that neural networks are capable of satisfactorily identifying the knowledge patterns contained in data with regard to the re-chlorination process for both winter and summer conditions. The practical use of such a model may assist operators in adjusting re-chlorination doses and may favour chlorine economization and the improvement of the water quality in the distribution system.