Artificial neural network modelling of reactive red 45 Azo dye removal by peroxi-electrocoagulation

In the present study, the Peroxi-electrocoagulation process was employed to remove chemical oxygen demand and the decolorization of reactive red 45 dye wastewater. This process offers several advantages, including the oxidation of dye molecules by hydroxyl radicals and coagulation through iron hydroxide precipitation. The reactor was equipped with six iron electrodes, providing a total surface area of 546 cm2. The effects of applied current (0.3–0.18 A), hydrogen peroxide concentration (100–700 mg/L), initial pH (2.5–5.0), and dye concentration (100–500 mg/L) on chemical oxygen demand and color removal were investigated. Optimum operational conditions were determined to be applied current of 0.150 (corresponding to current density of 0.27 mA/cm2), pH of 3.0, H2O2 concentration of 600 mg/L, and dye concentration of 200 mg/L. Under these conditions, COD and color efficiencies reached 83% and > 99%, respectively. The specific energy consumption under optimal conditions was 33.4 kWh/kg COD with 82.8% COD removal. The findings indicate that increasing the applied current and H2O2 concentration enhances removal performance up to a certain threshold, beyond which no significant improvement is observed. This suggests that the availability of electrogenerated reagents governs the overall reaction efficiency. Additionally, an artificial neural network model was developed to predict COD and color removal efficiencies. The network employed a 4:10:2 architecture and was trained using a backpropagation algorithm. Input variables included applied current, pH, H2O2 concentration, and dye concentration. The model exhibited high predictive accuracy, with R2 of 0.9782 for COD removal and 0.9579 for Color removal, confirming the effectiveness of the ANN in modeling.