Prediction of Aeration Efficiency in Tilting Flume Equipped with Screens Using AI Techniques

Abstract

This study investigates the innovative use of perforated screens as an aeration method in an open channel wastewater treatment system. Although screens are typically used in primary treatment to eliminate solids, their use in aeration has not yet been investigated. To evaluate this potential, 189 experimental tests were conducted using acrylic screens with different perforations/jets, such as circular, triangular, and square. Input parameters such as number of jets, discharge, angle of inclination of tilting flume ( α ), hydraulic radius of each geometry perforated on screens, Froude number, and shape factor (SF) were used to predict aeration efficiency ( E 20 ) using three artificial intelligence techniques, including artificial neural network (ANN), random forest, and support vector machine with Pearson VII kernel. With a correlation coefficient of 0.9635, ANN outperformed the other developed models. Sensitivity analysis revealed that SF was the most important component affecting E 20 . The dependability of the results was further supported by the great accuracy ( R 2 = 0.9116 ) of a mathematical model that used the same variables. The results lay the groundwork for further study and field-scale validation of this innovative aeration technique.