Forecasting of chlorophenols removing with advanced oxidation processes: an artificial neural networks application

Aysun Altikat1, Zeynep Ceylan2, Alper Gulbe3

1 Igdir University Engineering Faculty Department of Environmental Engineering, Igdir, 76000, Turkey
2 Ataturk University Engineering Faculty Department of Environmental Engineering, Erzurum, 25240, Turkey
3 Igdir University, Technical Sciences Vocational School, Department of Computer Programming, Igdir, 76000, Turkey


Advanced oxidation processes (AOPs) are used with high efficiency in wastewater treatment due to the formation of hydroxyl radicals with high oxidation capacity. However, there exists many different variables which affect the efficiency of theses processes as it affects the amount of hydroxyl radicals formed; such as reaction time, pH, temperature, reactant concentrations and catalyst amount. In this study, the removal of 2-chlorophenol (2-CP) and 2,4-dichlorophenol (2,4-DCP), which are highly toxic organic pollutants, with Fenton oxidation was investigated. This research has two aims. The first is modelling of substituted phenols removing with AOPs utilizing artificial neural networks (ANN), and the second is determination of the relative important among the input parameters for removing. In the research, 18 different ANN structures were used. The best ANN model for prediction of 2-CP and chemical oxygen demand from 2-CP (COD2-CP) were determined at the ANN10 structure. In this structure, 15 neurons, Levenberg-Marquardt and logarithmic sigmoid-symmetric sigmoid were used as learning and transfer functions, respectively. In addition, at the ANN16 and ANN2 structures, which were used for the prediction of removing efficiency for 2,4-DCP and COD2,4DCP, have obtained better results. In both models, Levenberg-Marquardt learning function was used. In ANN16, transfer functions were logarithmic sigmoid-symmetric sigmoid with 20 neurons and, in ANN2, symmetric sigmoid-symmetric sigmoid transfer functions with 10 neurons. When examined the sensitivity analysis results, all the input parameters have a significant effect on the substituted phenols removing with AOPs.


2-chlorophenol; 24-dichlorophenol; artificial neural network; Fenton oxidation; modeling

Full Text:

 Subscribers Only