Modeling of Electrical Energy in Industrial Wastewater Treatment Plant with Traditional and Artificial Neural Network Approaches

Document Type : Original Article

Authors

1 Assisstant Professor, Faculty of Marine and Oceanic Sciences, University of Mazandaran, Mazandaran, Iran

2 Faculty of civil Engineering, Shahrood University of Technology, Shahrood, Iran

3 Faculty of Civil Engineering, Shahrood University of Technology, Shahrood, Iran.

10.22044/jhwe.2024.15328.1044

Abstract

The rapid development of industries and the establishment of plenty of industrial parks have initiated several environmental issues during recent decades. The environmental standards and rules issued by the environmental organization for increasing the quality of the treated wastewater on the one hand and increasing the energy price on the other hand, have caused the energy management debate to be of particular importance. The main aim of energy management is to minimize the high energy consumption in industrial wastewater treatment plants (IWTP). In this paper, the electric power consumption of IWTP in Amol’s industrial park was measured by implementing both traditional and advanced methods (using artificial neural networks). In the first step, total energy consumption, involving energy used by flow or aeration pumps and mixers was calculated through an energy activity diagram, mathematical equations, and mass balances. In addition, linear regression equations for electrical energy consumption were estimated based on the amount of oxygen needed with an appropriate correlation coefficient. In the next step, a three-layer artificial neural network (ANN) with the Leonberg-Marquard training algorithm was employed. Various parameters, including COD, BOD, total phosphorus, total nitrogen, mixed liquor suspended solids (MLSS), and the flow rate (Q) were employed in 4 models to predict the electrical energy consumption of IWTP. Results showed that COD, MLSS, and Q can be considered as the most important selective indices for the determination of energy consumption by which the highest correlation coefficient and the lowest error rate of 0.928 and 0.0098 were obtained, respectively.

Keywords