Regression Models for Determining the Fate of BOD5 under Biological Treatment Method in Polluted Rivers

Abstract

The estimation and prediction of BOD5 is an important criterion for wastewater management and biological treatment of polluted rivers. The biological treatment method has been identified as the most optimal and technologically efficient technology to treat polluted urban rivers and streams. This practice has seen tremendous acceptability and applicability in most parts of China. However, the high cost of measurements, laboratory tests and sampling uncertainties associated with water quality variables make monitoring and prediction of desired water quality variables during biological treatment campaigns difficult. This paper has developed empirical models to predict the fate of BOD5 during a biological treatment method. The developed ten models were evaluated using ten-stratified cross validation technique. The results indicate high R2 relationships between observed and computed values. Prediction accuracy of the models were also assessed and showed errors in the range of ± 26% ~ ± 37%. These errors seem acceptable according to previous work on BOD5 measurements and forecasting. It is presumed that the unexplained nature for empirical formulae to integrate all the natural processes underpinning BOD5 processes might have been the cause. This notwithstanding, the results show plausible application for prediction and management of biological treatment projects and replicable for wastewater treatment systems.