Chemical coagulation-electro fenton as a superior combination process for treatment of dairy wastewater: performance and modelling

Publisher:
SPRINGER
Publication Type:
Journal Article
Citation:
International Journal of Environmental Science and Technology, 2021, 18, (12), pp. 3929-3942
Issue Date:
2021-12-01
Filename Description Size
Zakeri2021_Article_ChemicalCoagulation-electroFen.pdfPublished version1.57 MB
Adobe PDF
Full metadata record
Dairy wastewater contains high levels of organic matter that would be a serious threat if being discharged to the environment without any treatment. This study investigated the influence of various parameters including the concentration of poly aluminum chloride (PAC) on the chemical coagulation process and the concentration of H2O2, pH of the solution, reaction time, voltage and distance of electrodes in the electro-Fenton (EF) process. The optimum conditions obtained in this study included PAC concentration of 100 mg/L, pH of 3, reaction time of 60 min, voltage of 20 V, H2O2 concentration of 1.5 g/L and electrode distance of 2 cm. This resulted in the removal efficiency of 90.3%, 87.25%, and 87% for chemical oxygen demand (COD), biochemical oxygen demand (BOD5), and total suspended solids, respectively. The estimated cost of the process in terms of 1 kg COD removal was about $1.16. Two experimental models, called the feed-forward artificial neural network (FF-ANN) and partial least squares (PLS), were developed to predict the behavior of the process in COD removal. The results confirmed that both FF-ANN and PLS were powerful tools in predicting COD removal in the EF process, as the correlation coefficients between the predicted COD and the actual COD for the PLS and FF-ANN models were 0.854 and 0.971, respectively. Sensitivity analysis of both models showed that the reaction time and electrode distance had the most influence on the output of PLS and FF-ANN, respectively. Therefore, this combined process can be regarded as a thought-provoking one for dairy wastewater treatment, and the FF-ANN model was a more powerful tool than the PLS one.
Please use this identifier to cite or link to this item: