Use of near-infrared spectroscopy on predicting wastewater constituents to facilitate the operation of a membrane bioreactor

The implementation of near-infrared (NIR) spectroscopy in the wastewater treatment has been continuously expanding. As an alternative to the conventional analytical methods for monitoring constituents in wastewater treatment processes, the use of NIR spectroscopy is considered cost-effective and less time- consuming. NIR spectroscopy does not in any way adulterate measured sample as no prior treatment is needed thus making it a waste-free technique. On the negative side, one must be very well acquainted with chemometric techniques to interpret results. In this study, filtered and centrifuged wastewater and sludge samples obtained from a lab-scale membrane bioreactor (MBR) were analysed. Essential wastewater constituents were determined and compared, employing two analytical methods (conventional and NIR spectroscopy). Special attention was given to the soluble microbial products (SMP) and the extracellular polymeric substances (EPS) which are known to promote membrane fouling. The measured parameters through NIR spectroscopy were analysed and processed by partial least squares regression (PLSR) and artificial neural networks (ANN) models to assess whether the evaluated wastewater constituents could be monitored by NIR spectroscopy. Very good results were obtained with PLSR models except for the determination of SMP thus limiting the model for their monitoring to qualitative rather than quantitative. ANN exhibited a better performance in terms of correlating NIR spectra with all the measured parameters, resulting in correlation coefficients in most cases higher than 0.97 for training, test, and validation. Based on the results achieved by this research, the combination of NIR spectra and chemometric modelling offers advantages compared to conventional analytical methods.