Full-scale demonstration of a floating seal for enhanced biological nutrient removal in a sequencing batch reactor establishing chemical-free environment in wastewater treatment at low carbon source availability

ABSTRACT Due to the cyclical nature and changing water levels in the sequencing batch reactor (SBR), oxygen diffusion and utilization can be difficult to control particularly in light of the need to conserve the limited quantity of carbon source required to optimize biological nutrient removal. During the fill period, oxygen penetration may be undesirable since heterotrophic and autotrophic organisms cause a reduction in the readily biodegradable carbon source (rbCOD). This carbon source is essential and often limited in the anaerobic and anoxic periods. As a consequence, unwanted oxygen penetration can hinder efficient biological phosphorus and nitrogen removal. The purpose of the present research was to verify the advantage of a floating seal on the continuously moving surface of an SBR reactor to minimize undesirable oxygen penetration. In the floating seal-covered SBR both nitrification and denitrification efficiency proved to be higher due to insulation, and even during wintertime biological phosphorus removal met target removals without chemical dosing. The SVI values in the two SBR trains proved to be close to each other, despite the high difference in chemical dosing. Having experienced the higher efficiency of the seal-covered train, microbiome compositions of the two differently operated systems were investigated to determine potential differences via 16S rRNA gene amplicon sequencing experiments. In the samples taken from the seal-covered system, higher ratios of fermentative bacteria and phosphate accumulating organisms (PAOs) as well as glycogen accumulating organisms (GAOs) could be observed as compared to the samples deriving from the uncovered system. Highlights Seal-covering the periodically decreasing open water surface increased SBR efficiency Seal-covering the open water surface increased nitrification efficiency by insulation No chemical dosing was necessary for phosphorous removal in the Test system Metagenome investigations provided almost doubled amount of fermentative bacteria Production of GAOs indicated nutrient deficiency due to phosphorous removal. GRAPHICAL ABSTRACT


Introduction and purpose of the research
The sequencing batch reactor (SBR) technology includes reactors for basic biological wastewater treatment (generally combined with chemical dosing) and secondary clarifier in one basin, separated in time, and consequently requires less space than separate anaerobic, anoxic, and aerobic basins and secondary clarifier.Original SBR systems have five typical steps that continuously repeat during operation: fill, react, settle, decant and idle [1].In this original layout, at least two SBR reactors are needed for continuous flow operation, where one reactor is receiving influent wastewater while the other reactors are reacting or decanting.The early SBR reactors focused on decreasing the amount of substrate (carbon source).Several modifications have been created to achieve nutrient, i.e. nitrogen and phosphorus, removal.One of these modifications was carried out in Hungary by UTB Environtec Ltd., resulting in the Cyclator technology.
Due to the increasing water level during the fill period, there are continuously changing rates of dissolved oxygen supply and demand.In addition, the substrate concentration is the highest at the beginning of the fill period and decreases gradually as the carbon source is consumed by the biomass, resulting in changes in oxygen demand.Therefore, the initial substrate concentration is considerably higher than in continuous-flow systems [1].
Should filling be started under anaerobic conditions, denitrification and biological P-removal is possible if the concentration of the carbon source is high enough.However, when substrate availability is marginal or even deficient, oxygen penetration into nonaerated reactors through the open water surface may cause both metabolic and kinetic inhibition for biological P-removal as well as for denitrification.It has been pointed out that concentrations as low as 0.2 mg/l dissolved O 2 may decrease denitrification efficiency to half [2,3].It is well known that biological phosphorous removal requiring anaerobic conditions may be even more sensitive to oxygen penetration into non-aerated selectors.
The problem of marginal carbon source availability or even deficiency mainly attributable to lose in long sewer systems, has drown increasing attention in Hungary [4] and worldwide.Polish as well as international research groups, working successfully with Prof. Makinia, have investigated the possible applications of different external carbon sources to find alternatives for ethanol and methanol as carbon sources for denitrification.Experiments have been carried out as a comparative labscale batch or bench-scale studies.Promising initial results were obtained with the application of distillery, brewery and fish-pickling effluents [5].In two lab-scale sequencing batch reactors, the application of fusel oil, a complex distillery byproduct used as a complex carbon source, proved to be more efficient than ethanol being a clean carbon source [6].Acclimation of denitrifying activated sludge to the external carbon source proved to be approximately 3 weeks for both ethanol and fusel oil during the experiment.
Having experienced the advantages of fusel oil as an alternative external organic carbon source, it was investigated for its potential ability to enhance the growth of denitrifying polyphosphate accumulating organisms (DPAOs).Compared to non-acclimated biomass, both P-release and P-uptake as well as nitrate utilization increased remarkably in anaerobic-anoxic batch tests.However, the acclimation period of the original microbial community proved to be relatively long, > 50 days [7].Supported by simulation studies, the contribution of DPAOs to the total nitrate utilization rate could be estimated as 20% both for non-acclimated and acclimated biomass.
The impact of two different external carbon sources (acetate and ethanol) and the electron acceptors oxygen, nitrate and nitrite were investigated in two parallel operated lab-scale reactors using non-acclimated biomass from a full-scale biological nutrient removal activated sludge system.While dosing acetate as the external carbon source, phosphate release could be observed, with the smallest rate in the presence of dissolved oxygen, although a nitrification inhibitor was added [8].This result suggests that anaerobic processes can be severely hindered.
Another similar option for improving low carbon source availability has been the combined treatment of milk industrial wastewater lacking nutrients and domestic wastewater lacking sufficient carbon source [9].However, these fluctuating streams may not necessarily correspond to the requirements without ensuring storage for safety reasons.To increase the amount of carbon source having been significantly lost by dosing chlorine to combat odours from the sewer system, a new technology with the fermentation of a portion of mixed liquor suspended solids (MLSS) for reducing concentrations of nutrients to the required levels has been developed and presented at the 12th IWA (International Water Association) Conference of LWWTPs (Large Wastewater Treatment Plants) in Prague [10].The concept of mixed liquor fermentation (MLF) may also include intermittently switching off mixers to allow fermentation in the settled sludge.In addition to the fact that the settled sludge will have a higher MLSS concentration and better fermentation efficiency, avoiding high mixing velocity that could entrain too much oxygen through the open surface may also promote the growth of organisms characteristically requiring deeper anaerobic conditions [11].
The pioneer application of a floating seal for excluding oxygen penetration into non-aerated selectors through the open water surface was established first at the North-Budapest Wastewater Treatment Plant in Hungary, introduced for the 100th anniversary of the invention of activated sludge wastewater treatment [12], and detailed results were presented at the 12th IWA Conference of LWWTPs in Prague [13] and IWA 2017 Conference in Chongqing, China [14].
Seal-covering non-aerated selectors in commonly operated activated sludge wastewater treatment plants showed remarkable results in several respects [15].It assured avoiding oxygen penetration and thereby saved influent carbon source for biological P-removal and denitrification.Through avoiding microaerophile conditions and producing floc-forming bacteria good activated sludge structure could also be maintained.
Since in an SBR reactor the lifting water level can go significantly lower than those of common activated sludge basins, oxygen penetration through an equal open surface area can result in considerably higher oxygen concentrations, excluding thereby organisms like PAOs growing favourably under anaerobic conditions, especially when carbon source availability is not high enough (see Figure 1).
For biological nitrogen removal, there are three possible options in the typical SBR systems: mixed nonaerated filling period, cycling (on/off switching) of aeration during react period, as well as operating at low DO concentrations that is, however, likely to produce filamentous overgrowth [1], especially under low-S-low DO conditions [15].Cycling aeration is generally applied, as in the original Cyclator technology, where both the selector and the main reactor are alternately aerated, although denitrification during mixed nonaerated filling is considered to be the most efficient.This may also provide a selector-like operation to prevent filamentous sludge bulking.However, since influent quality can be highly different in different countries and even from plant to plant, sometimes reaching very high (up to 80-100 mg/l NH 4 N) influent ammonia concentrations, efficiency of commonly applied processes may not be satisfactory for strict effluent TN limitations, especially at marginal availability or deficiency of carbon source.Under these conditions, in SBR systems, appropriate P-removal, generally combined with reaching acceptable SVI (Sludge Volume Index) values are achieved by chemical dosing.
The purpose of the present research has been to verify the benefits of using a floating seal on the rising water surface of an SBR reactor.The activity focused on the comparison of the nitrogen and biological phosphorous removal of the alternately aerated uncovered Reference train (with alternately aerated selector and main reactor, and continuous inflow into the selector), and the fully seal-covered test train with basically nonaerated selector and alternately aerated main reactor (with continuous inflow into the non-aerated selector).

Experimental equipment and process
The experiment was carried out by applying two parallel operated SBR systems of company UTB Environtec Ltd. in Hungary.The original Cyclator system involved a staged reactor including an intermittently aerated selector and the main reactor also intermittently aerated (see Figure 2).The influent wastewater entered the selector continuously, even during sludge settling and decanting.In the Test train, the selector was operated under anaerobic conditions for phosphorous removal, whereas in the main reactor nitrification was established followed by denitrification intermittently.In the Reference train, the water surface was open, both nitrification and denitrification happened intermittently in the main reactor.
Before starting the experiment, the selector of the Test train had been converted into basically nonaerated operation, although sludge recycling was applied during both aerated and non-aerated operational phases of the main reactor, just like in the Reference train.The whole water surface of the Test train was covered by floating seal (see Supplementary Figure 1).The Floatseal (floating seal) carries HDPE (high-density polyethylene) material.It is a modular system, any size of a flat coverage can be built by connecting basic elements.Depending on operation conditions (e.g. over 100m 2 and windy environment) corrosion resistant material is normally recommended.Floatseal systems with connecting elements provide limited movement, but they are flexible enough to withstand turbulence caused by wind and water flow.
Based on the average conditions and characteristics to several wastewater treatment plants inside and outside Hungary [16], marginal C-source availability and deficiency could be observed in the influent of the experimental system as illustrated in Table 1.It is also important to note that the dissolved COD (Chemical Oxygen Demand) level proved to be much lower than the total COD value, and the influent NH 4 -N and TN values were very high, as also experienced in other areas [17].
The experiment was started on 26th October 2016, when the water surface of the Test train had been covered by the floating seal.To compare the efficiency of the uncovered and seal-covered systems, effluent concentrations of ammonia and TIN (Total Inorganic Nitrogen), influent and effluent TP (Total Phosphorous) concentrations, as well as SVI values and chemicals dosed were measured by standard processes until 30th March 2017.To investigate the insulation potential of the seal-covered Test system, effluent temperatures were followed until 5th April 2018.

Investigation of the microbiome
Having experienced the higher efficiency of the sealcovered Test train, metagenomes of the differently operated systems were investigated to detect potential differences in the microbiome compositions via 16S rRNA gene amplicon sequencing experiments.
Representative DNA samples were collected from the biomass of the seal-covered and uncovered reactors for 16S rRNA amplicon sequencing.Samples for DNA extraction were snap-frozen in liquid nitrogen and stored at −80°C.The DNA extraction was performed as previously described by [18].16S rRNA gene amplicon sequencing was performed by Novogene Genome Sequencing Company, beginning from DNA quality assessment and quantification, barcoded amplification, PCR product purification and library preparation.Sequencing primers were those used covering variable regions V3 through V4 on the 16S RNA gene.

Experimental results of onsite measurements
The experimental results supported the expected advantages of the seal-covered system appropriately.During wintertime, with ∼ 7-8°C influent wastewater temperature 1-1.5°C higher values could be detected in the effluent of the seal-covered Test system, as illustrated in Figure 3.The Test train could maintain relative stability during higher temperature conditions as well.
An important advantage of seal-covering the water surface in an SBR reactor is the insulation potential, as could be detected in the higher nitrification efficiency under cold winter conditions, despite the significantly decreased aeration capacity in the Test selector (as shown in Figure 4a).However, retaining oxygen in the seal-covered system could also contribute to the increased aeration efficiency.At 7-9 °C influent wastewater temperatures even 2-3 mg NH 4 N/l less ammonia concentrations could be detected in the effluent of the Test system.The difference in effluent temperatures was 1-1.5°C ensuring autotrophic organisms to consume ammonia with a higher μ value (specific growth rate) beside possibly better oxygen utilization in the seal-covered Test train.
The efficiency of denitrification also proved to be higher in the seal-covered Test train.Therefore, despite the higher nitrification efficiency, even 1-5 mg N/l lower effluent TIN (Total Inorganic Nitrogen) concentrations could be detected in the Test system (as shown in Figure 4b).This result can be attributed to both converting the alternately aerated selector of the Test train basically into non-aerated reactor and excluding both kinetic inhibition and metabolic advantage of oxygen through seal-covering the open water surface for reserving carbon source.
Besides saving chemicals, maintaining PAOs (Phosphorous Accumulating Organisms) in an activated sludge treatment system is extremely useful for creating good floc structure and thereby good sludge settleability, since these organisms are floc-formers.However, cultivating PAOs may generally have difficulties in wintertime due to the not available influent carbon source required.Figure 5(a) shows that in the sealcovered Test system a stable and continuous growth of PAOs could be achieved even in cold weather.As a consequence, the iron chloride dosing could be decreased progressively reaching zero in the experimental Test train, while the effluent phosphorous concentration remained practically zero even under a suddenly increasing influent peak.However, in the uncovered train iron dosing had to be increased at this point to decrease the effluent concentration appropriately.The floating seal basically excludes oxygen penetration into the Test selector and keeps the main reactor under hindered nitrification and appropriate denitrification.This behavior promotes the growth of PAOs.In the Reference reactor growth of PAOs might be hindered due to the open water surface, especially in the lack of carbon source.To reach the required effluent level dosing chemicals, especially iron chloride may be necessary.values measured throughout the experiment.SVI of the uncovered, Reference train was measured below 120 cm 3 /g, whereas in the Test train, at the beginning of starting with the seal-covered surface, the SVI was high, measuring 200 cm 3 /g.During the experiment, however, the settleability of the biomass of the Test system remained close to that of the Reference system, although the chemical dosed was much less, decreasing down to zero from 25th January 2017 until 30th March 2017, i.e. during the cold season until the end of the experiment.It can be assumed that the good settleability of the biomass at low water levels or even lack of chemical addition was based on the relatively high ratio of the PAOs in the seal-covered Test system.

Changes in microbial compositions of uncovered reference and seal-covered test systems revealing characteristics of the operations
The alterations in microbial community compositions in the floating seal covered and uncovered SBR systems were analysed using barcoded 16S rRNA gene amplicon sequencing.Samples for microbial community analysis were taken from both the uncovered Reference and seal-covered Test systems and stored at −80°C (see Materials and Methods for more details).DNA was extracted and sent for sequencing.Sequencing resulted in 66,837 (uncovered) and 59,696 (seal-covered) quality-filtered reads per sample, respectively.These were subjected to clustering according to Operational Taxonomic Unit (OTU) [19].Sequences with ≥97% similarity were assigned to the same OTUs.This clustering resulted in 1116 (uncovered) and 1084 (seal-covered) unique OTUs for the total sequence data set, respectively.For analysis and better comparison of samples, it was assumed that the read abundance, i.e. the number of reads in each OTU, corresponds to the actual abundance of the respective 16S rRNA phylotype in each activated sludge sample following the literature [20].
In samples taken both of the uncovered and sealcovered systems were found that the most abundant phylum corresponds to Proteobacteria (Figure 6).This observation is in line with typical community structure in wastewater treatment plants that have been investigated using metagenomic approaches [21,22].Other dominant phyla (relative abundance ≥ 5%) in the sealcovered samples were Actinobacteria, Bacteroidetes and Firmicutes, while in the uncovered samples were Actinobacteria, Bacteroidetes and Nitrospirae.Major differences between the seal-covered and uncovered samples could be observed in the case of the Nitrospirae, where the seal-covered sample showed 1.9-fold lower value as compared to the uncovered sample.It is also worthwhile to note that the abundance levels for Fusobacteria and Firmicutes showed a 2-fold or 1.5-fold increased value in seal-covered sample as compared to the uncovered sample.
Within Proteobacteria, Betaproteobacteria were the most dominant class found in the samples taken from the uncovered Reference system, followed by Alphaproteobacteria, Gammaproteobacteria and Deltaproteobacteria, while Gammaproteobacteria were the most dominant class found in samples taken from the sealcovered Test system followed by Alphaproteobacteria, Betaproteobacteria and Deltaproteobacteria.The abundance of Epsilonproteobacteria was the lowest in both cases.The abundance of Betaproteobacteria proved to be higher in the seal-covered sample (15.2% as compared to 8.1% in the uncovered sample), while the abundance of Gammaproteobacteria proved to be lower in the seal-covered sample (11.8% as compared to 15.8% in the uncovered sample).Differences in the abundance of microorganisms among Proteobacteria might be relevant for metabolism in activated sludge, as detailed below.
Within the Actinobacteria phylum, the Actinobacteria class was the most abundant followed by Acidimicrobiia.Approximately 1/3 increase in the relative abundance of Actionobacteria phylum could be observed in the sealcovered Test train, as compared to the uncovered sample.This observation is important in the light of some genus in the phylum Actinobacteria playing an important role in denitrification (e.g.Iamia) [23], P accumulation (e.g.Tetrasphaera) [24] and fermentation (e.g.Actinomyces) [25].Further, the relative abundance of Fusobacteria was observed to be twice as high and that of Firmicutes to be 1.5 times higher in the sealcovered Test system as compared to the uncovered one.Members of both phyla belong to the fermentative bacteria [26,27].
A major aim of the study was to identify the PAOs primarily responsible for biological phosphorus removal in the seal-covered and uncovered SBR trains used for the experiment.In this respect, only species with relative abundance >0.1% have been considered.We have constructed a list of potential PAOs possibly occurring in our samples based on literature [28][29][30].As illustrated in Figure 7(a), within the PAO group, bacteria from the genus Tetrasphaera were detected with the highest abundance in both systems (7.7% (uncovered) and 9.2% (sealcovered)).The relatively higher accumulation of Tetrasphaera in the seal-covered Test system may derive from the fact that it favours deeper anaerobic conditions, with ORP as low as −300 MV.It is important to emphasize that in the metabolism of activated sludge biomass, the ability of Tetrasphaera to produce volatile fatty acids (VFAs) is of high significance, since VFAs can also be further utilized as substrate for other PAOs as well.In addition, most species of Tetrasphaera genus are able to denitrify and to couple nitrite/nitrate reduction with phosphorus uptake [11,31].Thereby excluding oxygen penetration through seal-covering the open reactor surface may not just save but also produce appropriate substrate in accordance with the results of Jobbágy et al., (2019).This may help to overcome the problem of carbon source deficiency, especially in wintertime.
Moreover, it could also be observed that the abundance of Acinetobacter having been long believed to be the main PAO in EBPR plants [29], also showed considerably higher abundance in the seal-covered Test system than in the uncovered one (0.3% (uncovered) and 1.6% (seal-covered)) (see Figure 7a).The total amount of PAOs (>0.1%) proved to be 8.9% (in the uncovered) whereas 11.2% (in the seal-covered) system, respectively.
Besides PAOs, nitrifying microorganisms may also possess great relevance and significance in the metabolism of activated sludge biomass.With regard to this important microbial group, it was found, however, that although the abundance of both Nitrospiraceae and Nitrosomonadaceae was less in the seal-covered Test system (see Figure 7b), attributable to the non-aerated selector, this decrease did not affect the N-removal efficiency, moreover, it proved to be even better in wintertime on the contrary (see Figures 4a and b).As an explanation for this apparent discrepancy, it can be pointed out that the relatively higher temperature detected in the seal-covered Test system attributable to the insulation may maintain better conditions for nitrification.It cannot be excluded, however, that at least part of nitrogen can be removed through different ways, e.g. by the metabolism of Tetrasphaera [31].
Further, most types of Tetrasphaera can denitrify and to couple nitrite/nitrate reduction with phosphorus uptake.
Figure 8 provides a combined comparison for bacterial groups playing important roles in wastewater treatment.In addition to the previously discussed PAOs and nitrifying bacteria, considerable increases in relative abundance of both glycogen-accumulating organisms (GAOs) and fermenting bacteria could be observed in the seal-covered Test train.Among fermenting bacteria, species from the orders of Clostridiales [32] and Fusobacterialses [33,34] have been considered to represent major contributions to fermenting activities.We found that such species were present in higher abundances in the seal-covered Test system that could result in an elevated level of fermentation products as well, providing thereby sufficient substrates for both PAOs and GAOs as well as for denitrifying bacteria.The significant presence of GAOs as C. Competibacter may reflect to remaining excess carbon source having been possibly produced under deep anaerobic conditions at assumed P-deficiency being especially favouring conditions for the production of GAOs in the seal-covered Test system (see Figures 8 and 5(a) as well as [35][36][37]).

Conclusion
Sequencing batch reactors (SBRs) have been widely used for wastewater treatment including denitrification and biological P-removal, although oxygen penetration is well known to have both metabolic advantage and kinetic inhibition for these processes.
. With an abundance of readily biodegradable carbon source (rbCOD) in the influent, oxygen inhibition may be neglected.However, in an SBR as the water level drops during each cycle, the DO (Dissolved Oxygen) concentration can increase significantly. .Since decreasing temperature increases oxygen solubility, denitrification efficiency and even more biological P-removal can be decreased greatly and even lost completely in the wintertime, especially at low carbon source availability or deficiency. .During summertime in long sewer systems readily degradable carbon sources may be produced, however, in wintertime even marginally useful carbon sources may arrive in non-available form, especially for PAOs (Phosphorus Accumulating Organisms). .Excluding oxygen penetration from the anaerobic and anoxic reactors as well as keeping the wastewater insulated may allow not just denitrifiers but also PAOs to grow due to deeply anaerobic conditions even under cold circumstances. .Accordingly, important benefits have been achieved in the experimental Test system over the uncovered Reference system by seal-covering the periodically decreasing water surface. .This proved to be attributed to significantly higher amounts of fermenting bacteria producing appropriate carbon sources being essential for PAOs and could remain even for GAOs referring to nutrient deficiency having caused by excellent phosphorous removal in the seal-covered system. .Seal-covering non-aerated selectors can be effective in two simultaneous ways: (i) to maintain higher water temperature in winter and (ii) to provide deeper anaerobic condition for fermenting appropriate carbon sources.

Figure 1 .
Figure 1.1.1 Oxygen penetration into non-aerated selector and leaving the aerated reactor with free water surface at (a) high water level and (c) low water level; 1.2 Excluding oxygen penetration from non-aerated selector and retaining oxygen blown into aerated reactor at (b) high water level and (d) low water level.

Figure 2 .
Figure 2. The scheme of the original SBR reactor, investigated in the present study, operating with intermittently aerated selector and main reactor in the Reference train.

Figure 3 .
Figure 3. Effluent temperatures in the seal-covered Test train and uncovered Reference train.

Figure 5 (
Figure 5(b) illustrates the SVI (Sludge Volume Index)values measured throughout the experiment.SVI of the uncovered, Reference train was measured below 120 cm 3 /g, whereas in the Test train, at the beginning of starting with the seal-covered surface, the SVI was high, measuring 200 cm 3 /g.During the experiment, however, the settleability of the biomass of the Test system remained close to that of the Reference system, although the chemical dosed was much less, decreasing down to zero from 25th January 2017 until 30th March 2017, i.e. during the cold season until the end of the experiment.It can be assumed that the good settleability of the biomass at low water levels or even lack of chemical addition was based on the relatively high ratio of the PAOs in the seal-covered Test system.

Figure 5 .
Figure 5. (a) Influent and effluent TP concentrations and amounts of chemical dosed into the seal-covered and uncovered systems between 27th October 2016 and 30th March 2017.(b) Changes of the SVI values and amounts of chemical dosed into the seal-covered and uncovered systems between 27th October 2016 and 30th March 2017.

Figure 6 .
Figure 6.Relative abundance of phyla identified in the seal-covered and uncovered samples.The 10 phyla with the highest abundances are shown.'Others' represent a total relative abundance of the rest phyla besides the top 10 phyla.

Figure 7 .
Figure 7. (a) Differences in the relative abundance of phosphate-accumulating organisms (PAOs) in seal-covered Test vs uncovered Reference systems.(b) Differences in the relative abundance of nitrifying organisms in the seal-covered Test vs uncovered Reference system.

Figure 8 .
Figure 8. Differences in the relative abundance of major groups of microorganisms in seal-covered Test vs uncovered Reference systems.

Table 1 .
Original design parameters and average values measured during the experiment.(Differences between original design and calculated average influent COD, NH 4 N and TN values, and total and dissolved COD values, labelled in red).