BOD Approximation for Common Effluent Treatment Plant Using ANN

Abstract

In developing countries like India, common effluent treatment plants (CETPs) are often suggested as cost-effective options for centralized treatment of effluents emerging from separate waste-streams. The treatment/operation cost of CETPs is met by individual waste generators in proportion to the strength of the waste generated by them. This necessitates the regular determination of parameters such as BOD, which often is cost intensive, requires specialized lab personnel, and is time consuming (3-5 days minimum). Approximation of BOD, therefore, presents a relevant strategy which can save upon cost, effort and time. This paper presents a framework that employs Artificial Neural Network (ANN) technique to approximate influent and effluent BOD for common effluent treatment process. The framework is applied to the case of a CETP at Bhopal city, India. In the present work, a three-layered feed forward ANN that compares two different learning algorithms has been applied, and suitable architecture of the neural network models has been ascertained after several steps of training and testing of the models. The results indicate accuracy above 90%, thereby ANN proves to be a promising tool in the field of modelling.