Enhancing UV-Vis Spectrophotometry Technology with Data Analytics for Real-time Water Quality Monitoring and Treatment Process Operation

Abstract

Water quality monitoring is an essential element of the water quality management system and water treatment process. Conventional water quality monitoring relies on grab sampling and laboratory analysis, which is unable to provide quick responses to water quality events as it often takes hours and even days to transport and analyse water samples. Online water quality monitoring measures water quality continuously and allows quick responses to water quality events by providing real-time water quality data. In recent years, the online UV-Vis spectrophotometer has been reported as a promising technology for continuous water quality monitoring and process control. It reveals the real-time water quality changes and enables the development of surrogate parameters for online water quality monitoring and process control. However, there are some technic and data processing issues with using the online instruments for water quality monitoring. Besides, limited knowledge and research were reported on the utilisation of the online UV-Vis spectrophotometers for water quality management. This thesis project uses advanced data analytics to enhance the UV-Vis spectrophotometer for real-time water quality monitoring and treatment process control. Laboratory investigations were conducted to explore the impact of water matrix and suspended particles on the online water quality measurements using a submersible UV-Vis spectrophotometer, and to assess the water quality monitoring performance for different water sources. Both particle types and particle concentrations were found to have significant impacts on the UV254 measurements, showing that water quality data measured by the submersible UV-Vis spectrophotometer varied when the water matrix changes. These findings provide evidence that the particle influence on the UV-Vis measurements is source-water dependent. Surrogate models were developed as software techniques to eliminate particle impact from the measurements. Various software particle compensation techniques (surrogate models) including single wavelength compensation, linear regression compensation and multiplicative scatter correction methods were developed for online UV-Vis measurements of water quality. Moreover, cost-effective simple UV-Vis instruments could be employed in the field to monitor water quality instead of using sophisticated full-spectrum UV-Vis instruments. The real-time water quality measurement technology, UV-Vis spectrophotometer, was used for water treatment process control. Surrogate modelling approaches were used for the first to build coagulant dose determination models using only online UV-Vis spectra of raw water quality combined with chemometrics to determine coagulant doses and control the coagulation process for a drinking water treatment plant. The results revealed that an online UV-Vis spectrophotometer combined with a software surrogate model is a promising technology that determinates coagulant doses for real-time process control.