The primary goal of a drinking water company is to produce safe drinking water that meets the quality standards defined in national and international guidelines. Depending on the source water, one or multiple treatment steps are required to produce safe drinking water. Generally, the drinking water treatment plants (WTP) are very robust and over-sized and, as a result, have not required stringent advanced control based on the incoming water quality. However, water companies are facing increased challenges due to changes in feed water qualities and increased (micro-) pollution loads. When the source water originates from surface water an extensive treatment system is required, since surface water can be characterized by seasonal variations influenced by temperature, algae blooms, rain-fall run-off containing pathogens, solids and pesticides, environmental spills upstream and, since recently, the increased threat by endocrine disrupting compounds. This continuous changing source water requires improved monitoring and operation of the WTP, anticipating on the disturbances in the process. THESIS OBJECTIVE: Improved monitoring and operation of ozonation and biological activated carbon (BAC) filtration for the removal of pathogens and organic matter, for the optimisation of drinking water production from surface water. Ozonation and BAC filtration processes are susceptible to changes in the feed water quality. Besides, these processes have several control options and interaction between the two processes exists. The main objective for ozonation is disinfection and oxidation of organic matter, which results in an increase in the biodegradability of the natural organic matter (NOM). The main objective for BAC filtration is the removal of organic micropollutants and biodegradation of NOM to ensure the production of biologically safe and stable drinking water. In this thesis, pilot plant research was carried out at Waternet, water cycle company for Amsterdam and surrounding areas, location Weesperkarspel, the Netherlands. Easily assimilable organic carbon (AOC) is frequently used for the assessment of biological stability of drinking water, which is an important consideration in the control of bacterial growth in distribution networks. The first AOC bioassay was developed in 1982 and is based on growth of two bacterial strains (Psuedomonas fluorescens P17 and Spirillum spp. NOX) in drinking water relative to their growth on acetate. Since the original developed method, several new methods for the determination of AOC have been published with the aim of being faster, more reliable and cheaper. Application of these assays raises legitimate questions about the comparison of AOC data from different studies. In this thesis, a round-robin test was performed to evaluate the correlation between three established AOC methods. A total of 14 water samples covering a wide range of AOC concentrations were analysed with the original “van der Kooij” method, the “Werner & Hambsch” method and “Eawag” method. Good correlations were found between AOC concentrations measured with the various methods. The data suggest an acceptable compatibility between different AOC methods, although deviations between the methods call for careful interpretation and reporting of AOC data. The results from the round robin test emphasized the need to understand which measurement method was used to obtain the reported concentration, since the method gives insight in the actual meaning of the results. Sampling of the drinking water is carried out on a regular (almost daily) basis, to ensure the produced drinking water meets the quality standards. There is a trade-off between having a high probability of detecting a deviation while minimizing the measuring effort. In order to determine which measurements should be put in place, in this thesis, a seven step design methodology was developed. This enabled the determination of the required water quality monitoring strategy around ozonation and BAC filtration. It was shown how the previous on-line monitoring program of the treatment plant Weesperkarspel was optimised. Evaluation of on-line water quality sensors showed that the parameters typically measured to show compliance with the WHO standards were commonly available. Direct measurement of the more complex parameters such as AOC and bromate were not available on-line. It was shown that real-time information on the actual Ct value, the bromate and AOC concentration was necessary for continuous optimization of the applied ozone dosage. To address this gap, algorithms were developed for the on-line estimation of the Ct value and the formation of bromate and AOC during ozonation, based on the measured change in UV-Vis spectrum before and after ozonation. It was shown that these algorithms allow for the calculation of the optimal ozone dosage and provide a reliable indication of the amount of bromate and AOC formed during ozonation. Besides using these soft-sensors as surrogate sensors for parameters currently not available on-line, they also provide a cost effective alternative when used to determine multiple parameters through one single instrument. BAC filters are frequently used in the production of drinking water for the removal of organic micro- pollutants and organic matter, especially when produced from (humic rich) surface water. Differences in filter feed water quality are the result of differences in pre-treatment steps (coagulation/flocculation, ozonation and phosphate addition) commonly applied in the production of drinking water. Understanding of how BAC filters react to a change in feed water quality helps to identify where the focus in operating the BAC filters should be on. In this thesis, the immediate response of the BAC filters to a rapid change in feed water quality was investigated as well as the long term effects. The immediate response showed that all filters were able to mitigate a sudden change in feed water quality, either through improved adsorption or increased activity of the biomass on the filter. As a result of this resilience against sudden changes, it was therefore concluded that there is no direct need for very stringent on-line monitoring and continuous adjustments of the feed water quality of the BAC filters. Only the pressure drop and the pH and oxygen concentration in the effluent should be measured. The long term effects of changes in feed water quality were compared to previously published research and confirmed the need for sufficient nutrients (readily available carbon and phosphate) in the feed water for optimal performance. The addition of phosphate resulted in the lowest dissolved organic carbon (DOC) concentration in the effluent of the BAC filters. In this study the influence of intact cells in the feed water on the performance of the BAC filters was shown to be limited. One parameter in the BAC filters that requires enhanced control and understanding is the clogging of the filters, especially when the water temperature starts to increase in spring and summer time. In the BAC filters clogging takes place as a result of the physical, chemical and biological processes occurring in the filters. A simplified model, based on retention and mass balances, to predict the filter run times was developed. Results showed that clogging in the BAC filters was a combination of chemical and biological mechanisms. Already small concentrations of AOC resulted in the development of a biofilm. This biofilm caused the formation of a cake layer and accounted for the majority of the pressure build up in the BAC filters. The simplified model was able to accurately predict the head loss development. The model subsequently allowed for optimising the process control of the full scale BAC filters in terms of predicting the pressure drop and subsequently optimising the backwash sequence, saving 32% backwash water. In this thesis it was shown that the current monitoring strategy of ozonation-BAC filtration could be improved through implementation of the established design methodology in combination with the developed algorithms allowing for on-line estimation of AOC, bromate and Ct value around ozonation, based on the measured change in UV-Vis spectrum. The operation of BAC filters could be improved through a better understanding of the direct response of the BAC filters to a change in feed water quality and the use of simplified models to optimise the operational strategy around backwashing of BAC filters. At Waternet, location Weesperkarspel, some of the findings from this and previous studies have been taken and tested at full scale. As a result, on-line monitoring of ozonation was extended with UV/Vis sensors before and after ozonation and with ozone in water sensors in the influent of the first ozonation contact chambers.
Original languageEnglish
QualificationDoctor of Philosophy
Awarding Institution
Thesis sponsors
  • European Commisssion
Award date12 Jun 2019
Print ISBNs978-94-6323-655-3
Publication statusPublished - 12 Jun 2019

ID: 54336096