The primary objective of conventional drinking water treatment and distribution is to deliver to the consumer water that is both aesthetically pleasing and does not constitute a human health risk. To achieve this, water utilities employ a range of physical (i.e. sand and membrane filtration) and chemical (i.e. flocculation and disinfection) barriers in order to reduce the numbers of microorganisms as well as the nutrients that may support their growth within biofilms. In this thesis, biofilms and microbial barriers in water treatment and distribution were therefore examined. The development of biofilms within artificial recharge was investigated in pilot column at Norsborg waterworks in Stockholm. The proportion of active bacteria, measured as numbers of EUB338-positive cells relative to the total number of bacteria enumerated by total direct counts, decreased with time. Through the addition of nutrients however, two to three times more bacteria were able to be active (measured by increase in activity after activation with additional nutrients). By extracting the recalcitrant hydrophilic and hydrophobic fractions of humic substances it was possible to assess the microbiological response to those compounds. It was shown that bacteria more firmly attached to the sand grains preferred the hydrophobic fraction whilst more loosely-associated bacteria preferred the hydrophilic one. The amount of easily degradable matter in raw water (measured as assimilable organic carbon) was generally low. Biofilms were investigated by two different methods for extraction and analysis of microorganisms. Glass slides introduced into the sand material were dominated by α-Proteobacteria, and underestimated loosely-associated bacteria whilst extracts from sand were dominated by γ-Proteobacteria, and also caused variations due to the extraction method employed
The barrier function of biofilms was investigated in biofilters, also fed with raw water from Gothenburg. The focus here was on particle removal in size-intervals of 1-15 µm (protozoa) and 0.4 - 1 µm (bacteria). In both size fractions, autofluorescent microalgae, which were naturally-occurring in raw water, were also enumerated in parallel. Their removal was 60-90%. In parallel, defined amounts of fluorescent hydrophilic and hydrophobic microspheres (1 µm) were added. They showed a reduction of hydrophobic spheres by 98% and hydrophilic ones by 86%. Removal of viruses was determined by adding a defined dose of bacteriophages and gave lower reduction values of 40 - 61%. Both naturally-occurring particles in defined size intervals and added particles or organisms were shown to provide a clearer picture of barrier function than usually performed measurements of turbidity.
The efficiency of chemical treatment against viruses was also measured in a pilot-plant in Gothenburg. It was shown that commonly-used MS-2 bacteriophages were much more sensitive than φX174 bacteriophages. Reduction of MS-2 over the entire chemical step (when added after dosing of chemicals) was 5-log10 whilst φX174 was reduced by 1-log10. The latter was shown to be a more conservative model for virus removal. The effects of different steps in the chemical precipitation showed that the primary dosage of chemicals and the development of flocs had great importance for the assessment of removal efficacy. When added before the dosing of chemicals, reduction of φX174 and MS-2 was 3.8-log10 and 6.2-log10, respectively.
The establishment of biofilm within a distribution system was followed in a 1000 metre long pilot-plant (with parallel lines) at Lovö waterworks as well as in two of Stockholm's main distribution systems (Nockeby and Hässelby). The pilot-plant was shown to satisfactorily represent processes within the distribution systems. The development of biofilms was slow, producing thin biofilms over a one to two month periods. Numbers of bacteria were generally in the range of 104 - 105 per cm2, which is lower than shown in other earlier investigations. The implementation of primary ultra viloet (UV)-treatment in place of chlorination (both being chloraminated prior to distribution) did not considerably change the numbers of bacteria in biofilms. No significant difference could be seen between the system that had UV-treatment as a primary treatment step, and the system that was chlorinated over the whole period. Chlorine residuals were generally low at the distal parts of the distribution systems. Naturally-occurring protozoa were present in distribution systems in numbers ranging from 280 - 3500 protozoa per cm-2. Protozoa may play a significant role as predators of biofilm bacteria, however they can also act as protection for bacteria against external influences i.e. disinfection. Should sudden contamination of a distribution system occur, biofilm can provide protection and act as a site for potential regrowth of introduced microorganisms. Biofilms developed in the pilot-scale that represented water from different distances from waterworks were exposed to fluorescent microspheres, (hydrophobic and hydrophilic, 1 µm) legionellae (as a model for opportunistic bacterial pathogens) and bacteriophages (human enteric virus model) in order to determine their accumulation and persistence within the biofilm, and release to the bulk water. It was shown that introduced model organisms were released continuously, primarily through desorption, and additionally through the influence of disinfection and activity of protozoa.
Desorption was also assessed in a laboratory experiment under laminar and turbulent flow. Laminar flow conditions that were representative of a distribution system gave a slow and continual release of individual cells, whilst turbulent conditions detached larger aggregates. In conclusion, based on this work an increased understanding was gained both of barrier functions at the different steps of water treatment, their effects on overall biofilm dynamics and structure and the role that biofilm plays within the drinking water system itself.
Stockholm: Mark och vatten , 2004. , xii, 67 p.