Multi-way modeling of wastewater data for performance evaluation of sewage treatment plant—A case study

The paper reports application of three-way regression modeling methods to evaluate the performance of an up-flow anaerobic sludge blanket (UASB) reactor based wastewater treatment plant (WWTP). The parallel factor analysis (PARAFAC) and multi-way partial least squares (N-PLS) regression models were applied to the three-way wastewater data set collected over a period of 48 weeks during the peak and non-peak flow regimes from the inlet and outlet of the UASB plant for predicting the treated wastewater quality. Both the two-component PARAFAC and tri-PLS models successfully predicted the levels of all the four selected chemical variables viz. biochemical oxygen demand (BOD), chemical oxygen demand (COD), total Kjeldahl nitrogen (TKN) and ammonical nitrogen (NH4–N) in treated wastewater of the UASB. The predictive capabilities of the models were evaluated in terms of the root mean square error of prediction (RMSEP) and correlations between the measured and model predicted values of the variables in treated wastewater. Although, both the methods yielded satisfactory results for the prediction of the treated wastewater quality, performance of the tri-PLS model was relatively better. The multi-way modeling methods can be used for the performance evaluation of WWTPs.