Artificial neural network modelling approach for the prediction of turbidity removal efficiency of PACl and Moringa Oleifera in water treatment plants

The present study uses Artificial Neural Network for predicting the water turbidity removal after coagulation–flocculation–sedimentation process in water treatment plants. The coagulants namely, poly aluminium chloride and Moringa Oleifera have been used for modelling ANN. Conventional jar test experiments at various pH, coagulant dosage and settling time at different initial turbidity values were carried out to generate the data set for model development. The ANN architecture with a structure of 4:10:1 yielded high predictability of turbidity removal efficiency. Sensitivity analysis revealed that an R2 of 0.99 was achieved between predicted and observed turbidity removal efficiency using the model for both the coagulants, thereby indicating that coagulation performance depends on pH, initial turbidity, dosage and settling time. Further, floc characteristics of PACl and MO flocs analysed using an image capturing and processing technique revealed that spherical flocs settle at a faster rate and occurs during the initial 10 min of settling for PACl flocs and between 10 and 20 min for MO flocs, thus depicting the role of settling time in turbidity removal efficiency.