Biologically Active Filtration Media Properties: Practical and Mechanistic Implications

Abstract

Biologically active filtration [BAF] can be used to concurrently remove particles and natural organic matter during drinking water treatment. The selection of a given media type for use in BAF can impact filter performance, capital costs, and operating costs. BAF performance using different media types has been previously compared; however, no single media type has been found to provide the best performance across all studies. Notably, no comparisons of BAF with various media types have been reported where the same grain size distribution was used for all media types; therefore, observed differences in performance cannot be attributed solely to the media types, but may have been impacted by differences in grain size distribution. Furthermore, mechanisms affecting BAF performance are not well understood and mechanistic implications of media selection on BAF have not been fully elucidated.

In this study, the performance provided by different media types and media-associated mechanisms that impact BAF were investigated through two phases of experiments.

In Phase I, a procedure for matching the grain size distribution of different media types was developed. Pilot-scale biologically active filters [biofilters] were filled with coal-based granular activated carbon [GAC], anthracite, rough engineered ceramic media [REC], or wood-based GAC; the media grain size distributions were closely matched. The biofilters were fed water that was flocculated, settled, and ozonated at a full-scale water treatment plant. One extra filter containing coal-based GAC was operated in a declining-rate mode, whereas all other filters were operated in a constant-rate mode. The biofilters were operated continuously for 660 days. Dissolved organic carbon [DOC] removal, assimilable organic carbon [AOC] removal, trihalomethane formation potential [THMFP] removal, turbidity removal, headloss, and filter run time were monitored and compared. Prior to this study, REC had not been tested for use in BAF.

The GACs provided better DOC removal than either REC or anthracite. This improved removal was observed even though the coal-based GAC had been used for seven years in full scale filters prior to these experiments. The GACs were adsorptive media types whereas the REC and anthracite were nonadsorptive. It was demonstrated that the adsorptive property of GAC is critical for enhancing DOC removal during biofiltration relative to other media over the long-term, even for GAC that has been used for many years. The results also implied that mechanisms related to a medium’s adsorptive properties (e.g. bioregeneration, adsorption of organic matter spikes) are significant to DOC removal during biofiltration in the long-term. It was also found that DOC removal improved when the filter was operated in declining-rate mode, as opposed to constant-rate mode. In some cases, operating a filter in declining rate mode helped to offset differences in DOC removal provided by different media types. Differences in AOC and THMFP removal provided by the media types were observed during some sampling events; however, no media type consistently provided the best AOC or THMFP removal. Interestingly, dibromochloromethane formation potential increased slightly because of biofiltration, especially in GAC as compared to anthracite or REC filters.

Turbidity removal was assessed in two ways: (1) by comparing the stable effluent turbidity between ripening and breakthrough and (2) by comparing the ability of the biofilters to dampen influent turbidity spikes. A kaolin clay suspension was injected into the biofilter influent to cause the influent turbidity spikes. Rough media types (i.e. wood-based GAC, coal-based GAC, and REC) provided better turbidity removal and better turbidity dampening than smooth media (i.e. anthracite). It was concluded that media roughness generally enhances turbidity removal and turbidity dampening during BAF. REC and wood-based GAC provided the best turbidity removal of all the media types. The media type that provided the best performance, between REC vs. wood-based GAC and between coal-based GAC vs. anthracite, was seasonally dependent.

REC and anthracite generally provided slower headloss development than GAC media during biofiltration. The specific media type that provided better (i.e. slower) headloss development within adsorptive (coal-based vs. wood-based GAC) and non-adsorptive (REC vs. anthracite) media was seasonally dependent. It was found that there may be a trade-off between choosing a media type that provides the greatest DOC removal and choosing a media type that provides the best headloss performance.

Finally, the media types that provided the longest filter run time were seasonally dependent, but, in general, REC provided longer filter run times than wood-based GAC and anthracite provided longer filter run times than coal-based GAC.

In Phase II, spikes of an acetate (a nonadsorptive compound) and maltose (an adsorptive compound) were injected into the influent of a biofilter located at the University of Waterloo [UW] and biofilters located in Toronto, Ontario [Toronto]. The UW biofilter contained coal-based GAC that had previously been used in a full scale biofilter for 25 months. The UW biofilter was fed synthetic water containing sodium acetate and nutrients. Two sets of spikes, consisting of one acetate spike and one maltose spike, were introduced to the UW biofilter. The removal of total organic carbon and the production of inorganic carbon were monitored before, during, and after the spikes to assess the fate of organic carbon in the biofilter. The Toronto biofilters consisted of GAC and anthracite biofilters that had been continuously operated for three years prior to the spike experiment. The biofilters were fed Lake Ontario water that was ozonated and flocculated. Two acetate spikes and one maltose spike were added to the filter influents.

The inorganic carbon produced by the UW biofilter exceeded the TOC removal in one of two spike experiments. This indicated that organic carbon adsorbed to the GAC or organic carbon present in the biomass was oxidized to CO2. It was concluded that either bioregeneration of adsorbed organic matter and/or net decay of accumulated biomass can occur in drinking water biofilters containing GAC media after spikes of organic matter have been attenuated. Further research is needed to differentiate between these two mechanisms and to elucidate the scenarios under which each of these mechanisms occurs during drinking water treatment.

Maltose spikes were adsorbed onto GAC at both UW and Toronto. This work demonstrated that organic matter spikes can adsorb onto GAC even after the GAC has been used in biofiltration for extended periods of time. Adsorption of spikes of organic matter is one mechanism that may explain how GAC biofilters can provide better removal of organic matter than biofilters containing nonadsorptive media (i.e. anthracite and REC) over the long-term.