Effects of coal ash supplementation on aerobic granular sludge cultivated in a simultaneous fill/draw sequencing batch reactor

ABSTRACT This study aimed to verify if coal ash, a residue from thermal power plants, could act as a granulation nucleus, cations source, and abrasive element to favor granules formation and stability in aerobic granular sludge (AGS) systems. Two simultaneous fill/draw sequencing batch reactors (SBRs) (R1 and R2) were operated with 6-h cycles, i.e., the filling and drawing phases occurred simultaneously, followed by the reaction and settling phases. R1 was maintained as control, while R2 was supplemented with coal ash (1 g·L-1) on the first day of operation. Granulation was achieved in both reactors, and no significant differences were observed in terms of settleability, biomass retention, morphology, resistance to shear, and composition of the EPS matrix. However, the ash addition did not change the settleability, biomass retention, granule morphology, shear resistance, and extracellular polymeric substances (EPS) content significantly. COD removal was high (≥ 90%), while nitrogen (~50%) and phosphorus (~40%) removals were low, possibly due to the presence of nitrate during the anaerobic phase. With granulation, microbial population profile was altered, mainly at the genus level. In general, the operational conditions had a more considerable influence over granulation than the ash addition. The possible reasons are because the ash supplementation was performed in a single step, the low sedimentation rate of this particular residue, and the weak interaction between the ash and the EPS formed in the granular sludge. These factors appear to have decreased or prevented the action of the ash as granulation nucleus, source of cations, and abrasive element.


INTRODUCTION
Aerobic granulation is a process through which microorganisms (mainly bacteria) self-immobilize due to the influence of several selection pressures, such as short settling times and high aeration intensity (ROLLEMBERG et al., 2018). The granules formed usually present a compact and robust structure, as well as a good settling ability, being also capable of withstanding high organic loadings and simultaneously removing carbon, nitrogen, and phosphorus (ADAV et al., 2008;LAI, 2009). Furthermore, when compared to conventional activated sludge (CAS), the aerobic granular sludge (AGS) technology reduces operational costs (20-25%), electricity demand (23-40%), and space requirements (50-75%) (ADAV et al., 2008;BENGTSSON et al., 2019;NEREDA, 2017).
AGS is commonly cultivated in sequencing batch reactors (SBR), which are operated in cycles, including the phases of filling, aeration, settling and decanting (BEUN et al., 1999). Recently, researches have been conducted on simultaneous fill/draw SBR, also known as constant-volume SBR (DERLON et al., 2016;. These studies are still incipient, even though the majority of full-scale AGS systems are operated in such a manner, e.g., the Nereda ® technology (NEREDA, 2017).
The biggest challenges of aerobic granulation are the long start-up period required and the maintenance of long-term granule stability.
In order to solve these issues, studies have demonstrated that the addition of calcium (JIANG et al., 2003), magnesium (LI et al., 2009), and polyaluminum chloride (PAC) (LIU et al., 2016) can lead to a faster granulation, thus improving sludge settleability. Supplementation with dried sludge micropowder also proved beneficial to granule stability, having eliminated extended filaments through several mechanisms, such as collision and friction against granules, which stimulate extracellular polymeric substances (EPS) secretion (LIU et al., 2019).
In this context, the present research aims to investigate whether the addition of coal ash (a residue from thermal power plants) can shorten granulation time in AGS systems. Firstly, coal ash might behave as a granulation nucleus (ZHANG et al., 2017). Secondly, it could function as a source of divalent cations. Finally, it is possible that the friction between coal ash and granules would lead to an increase in granule resistance to shear, making them more stable, especially when a long-term operational stability is considered. Furthermore, no similar studies are known to have been conducted in simultaneous fill/draw SBR with a low liquid upflow velocity.

Sequencing batch reactors configuration and operation
The experiments were conducted in two cylindrical acrylic SBR with a diameter of 100 mm, a total height of 1 m and a working volume of 7.2 L.
The reactors were operated at a simultaneous fill/draw regime  of 6-h operating cycles, which were divided into 30 min of simultaneous filling and drawing, 90 min of anaerobic/ anoxic phase, 210-235 min of aeration, 30-5 min of settling. To stimulate granulation, the sedimentation time was gradually decreased from 30 (Stage I) to 15 (Stage II), 10 (Stage III), and 5 (Stage IV) min. The time subtracted from the settling phase was added to the aeration phase. Each stage was maintained for approximately six weeks.
The fill/draw was performed using a Masterflex peristaltic pump model BTG 2344, which produced a volumetric exchange rate of 50% and a liquid upflow velocity of 0.92 m·h -1 . Aerators were positioned at the bases of the reactors, providing an air upflow velocity of 2.12 cm·s -1 .

Inoculum and feeding solution
The reactors were inoculated with aerobic sludge from a domestic wastewater treatment plant (WWTP) (Fortaleza, Ceará, Brazil) at an initial concentration of mixed liquor volatile suspended solids (MLVSS) of approximately 2 g·L -1 , whose sludge volume index at 30 min (SVI 30 ) was 110 mL·g -1 .
To investigate the influence of coal ash addition on the granulation, a reactor (R1) was maintained as control, while the other (R2) was supplemented with coal ash at a concentration of 1 g·L -1 (~3.1 mg·L -1 of calcium and ~0.2 mg·L -1 of magnesium). It is important to emphasize that the coal ash application was performed only once, at the beginning of the operation. This decision was made in the light of the experiment conducted by Liu et al. (2016), which demonstrated that the long-term application of PAC in a conventional SBR did not produce significant differences in relation to the short-term application, although both accelerated the granulation in comparison with the control reactor.
In addition, such a frequency of supplementation was chosen to avoid an excess of inert solids inside the reactor. In order to characterize the sludge settleability, a dynamic SVI, which is a modified version of the SVI, was performed for 5, 10, and Effects of coal ash on AGS in simultaneous fill/draw SBR 30 min (SCHWARZENBECK; BORGES; WILDERER, 2005). The frequency of the analyses was also twice a week.

Analytical methods
To characterize the coal ash added to R2, a solubilization test was performed according to NBR 10006:2004(ABNT, 2004. During this analysis, 250 g of coal ash was added to 1 L of distilled water. The solution was stirred for 5 min and then left idling for 7 days. After this period, the sample was filtered with a membrane of porosity equal to 0.45 μm. The filtered solution was analyzed by inductively coupled plasma optical emission spectrometry (ICP-OES) (Thermo Fisher iCAP 6000) to determine the concentrations of calcium and magnesium.

Characterization of mature granule
At the end of Stage IV, the mature granules were subjected to a physical resistance analysis (shear test) (NOR-ANUAR et al., 2012). Samples of the granules (> 0.2 mm) were subjected to a shear force caused by a stirrer at approximately 200 rpm for 10 min. The fragmented fraction was expressed in terms of a stability coefficient (S), obtained by the ratio of the amount of total solids after and before the stirring of the sludge sample.
This coefficient was classified into three categories: very stable (S < 5%), stable (5% ≤ S ≤ 20%) and not stable (S > 20%). Therefore, the lower the value of S, the greater the resistance of the aerobic granules to shear.
A scanning electron microscope (SEM) (Inspect S50 -FEI model) with a nominal resolution of 3 nm was utilized to obtain detailed images of the granules and to carry out semi-quantitative chemical analysis by energy-dispersive X-ray spectroscopy (EDX). The mature granules were also observed under an optical microscope (Opton). The preparation for such microscopy analyses was performed following the methodology proposed by Motteran, Pereira and Campos (2013).
The analyses were performed at the Analytical Central of Microscopy of Universidade Federal do Ceará, Brazil. The granule size profiles of the mature granules were obtained. For this, sieving was carried out with sieves of 0.2, 0.6, and 1 mm openings. The depth of oxygen penetration in the granules was also estimated, according to Equations 1 and 2 (DERLON et al., 2016;HENZE et al., 2008).

RESULTS AND DISCUSSION
Settleability and sludge retention It is also important to notice that coal ash addition promoted a decrease in the MLVSS/MLSS ratio in R2, reaching a value of 0.5.
This result is compatible with the fact that the supplemented material behaves as fixed solids. However, the ratio starts to increase at day 20, and, at day 34, the same value is observed in both reactors. Considering that the two SBR had similar MLVSS at this point, this seems to indicate that the coal ash was eliminated from R2. Sedimentation tests conducted showed that, after 30 min, the coal ash would settle only partially. Since after its addition, the settling time was kept at 30 min, a considerable fraction of it would have been washed out in each cycle.
After a certain number of cycles, its concentration would have decreased considerably, resulting in an increase of the MLVSS/MLSS ratio.

Characterization of mature granules
At the end of Stage IV, both reactors had similar granule size profiles, with granules with diameters larger than 0.2 mm representing 98% of the biomass, and granules larger than 1 mm accounting for 90%.
These results confirm the tendency observed in SVI and MLVSS values.
Concerning the morphology of mature granules, Figure 2 presents the images obtained at the end of Stage IV by optical microscopy and SEM. SEM images show an absence of inorganic crystals or organic material other than the amorphous material which covered the granules' surfaces entirety. This is compatible with well-delineated surfaces, with only a few cavities. It is also important to notice that no trace of coal ash has been identified in the microscopic images.
A different outcome was obtained by Liu et al. (2019), whose optical microscopy images clearly showed the presence of micropowder, and even identify its interaction with the granules. Zhang et al. (2017) worked with the addition of biochar and obtained photographs in which the compound was identified as granule nucleus. The fact that SEM analysis did not indicate the presence of coal ashes suggests that they were washed out from R2. This evidence is in accordance with the MLVSS/MLSS ratios observed as well as with the results of the sedimentation tests conducted (see section "Settleability and sludge retention").
On the subject of resistance to shear, mature granules of reactors R1 and R2 demonstrated stability coefficients of 24.3 and 27.4%, respectively.
According to the guidelines developed by Nor-Anuar et al. (2012), the granules can be classified as not strong. Xavier (2017), working with a pilot-scale conventional SBR for the treatment of real wastewater, also found stability coefficients around 25%. However, the author emphasizes that, since the shear force applied during the resistance test is usually much higher than the one actually maintained inside the reactors,

Mechanisms of granule formation and maintenance in the presence of coal ash
Regarding the role played by coal ash in the formation and maintenance of granules, one possibility would be that coal ash might behave In the present study, the increase in COD and particulate organic matter could not be a possible mechanism of granule stabilization, since coal ash was not constituted of organic compounds. Therefore, three mechanisms of granule formation and/or maintenance by the coal ash are proposed for further discussion. In essence, the ash might behave as granulation nucleus (hypothesis 1), it would function as a source of cations and anions (hypothesis 2), and it would be a friction source against the granules (hypothesis 3).
Regarding the first hypothesis, since none of the microscopy images seem to suggest it, coal ash did not work as a granulation nucleus.
The composition of the EPS matrix obtained by EDX reinforces this conclusion.
Regarding the second hypothesis, the estimated concentrations in the reactor were ~3.2 mg·L -1 of calcium and ~0.2 mg·L -1 of magnesium.
These concentrations are likely insufficient to produce significant changes in sludge, since they are much lower than those developed by researchers who reported meaningful results. For example, Jiang et al. (2003) worked with a conventional SBR with the addition of 100 mg Ca 2+ ·L -1 in every cycle and obtained lower SVI 30 (control reactor: 150 mL·g -1 , calcium reactor: 100 mL·g -1 ), greater sludge retention (control reactor: 2 g SS·L -1 , calcium reactor: 7.9 g SS·L -1 ) and increased amounts of PS in the EPS (control reactor: 41 mg L -1 , calcium reactor: 92 mg L -1 ). Li et al. (2009) also worked with a conventional SBR but supplemented it with 10 mg Mg 2+ ·L -1 in every cycle. They obtained greater sludge retention (control reactor: 6.8 g MLSS·L -1 , magnesium reactor: 7.6 g MLSS·L -1 ), mature granules with larger sizes (control reactor: 1.8 mm, magnesium reactor: 2.9 mm) and increased PS production (control reactor: 35 mg L -1 , magnesium reactor: 70 mg L -1 ). However, the SVI 30 remained the same (between 20 and 25 mL·g -1 ).
Finally, regarding the third hypothesis, since extended filaments were not observed in the granules of the control reactor (R1), it is reasonable to assume that the granules cultivated in R2 would also not have developed this feature. In such an environment, the effects of friction would not be pronounced, since there might not have been filaments to be broken from the granule surface. This might explain why very few differences between the settling properties and retention ability of the sludges were found. Similar behavior was observed by Liu et al. (2019) when they applied micropowder over granules without extended filaments. The authors reported that SVI 30 was kept nearly the same (decrease from around 55 to 45 mL·g -1 ), and stabilized biomass retention (increase from around 3.8 to 4 g MLVSS·L -1 ).
Furthermore, the ash washout evidenced by MLVSS/MLSS ratios and SEM analysis would have lowered the possibility of contact between the coal ash and the granules, reducing friction. Such a reduction, combined with the absence of extended filaments, would have made friction an ineffective mechanism of granule stabilization. In the absence of the friction promoted by coal ashes, EPS production would not be stimulated, which may explain the similar EPS composition obtained in the reactors. The coal ash washout becomes even more relevant because the last two mechanisms through which the ashes were considered to influence granulation (friction and ion supplementation) are highly dependent on the concentration of the agents used. However, since the coal ashes were present in the reactor for at least 20 days (see section "Settleability and sludge retention"), they would have still been able to act as a granulation nucleus. The fact that the coal ash could not act as such seems to indicate that this property is not inside their scope of capabilities. Table 1 summarizes the results regarding the removal of COD, nitrogen, and phosphorus, showing that the reactors performed similarly throughout the operation stages. It can be seen that COD removals remained above 90% for both reactors. These values correspond to those reported in the literature regarding the operation of conventional and simultaneous fill/draw SBR, for which COD removal is equal to or higher than 80% (DERLON et al., 2016;LIU et al., 2010;LONG et al., 2014).

Organic matter and nutrient removal
Concerning the nitrogen removal (Table 1), the small concentration of nitrite over nitrate in the effluent indicates the occurrence of complete nitrification. However, nitrate was accumulated, indicating that simultaneous nitrification and denitrification did not occur at considerable levels.
The phosphorus removal (below 60%) is considered low, especially for reactors operated with cycles containing an anaerobic phase, since the inclusion of this period is aimed at optimizing phosphorus removal (BASSIN, 2011). H. , for example, worked with a conventional SBR with an anaerobic phase included (total cycle: 6 h, anaerobic phase: 2 h) and obtained phosphorus removals around 98%.
However, in many cases, the high efficiencies of phosphorus removals reported in the literature are achieved by applying low concentrations of this compound. Therefore, its removal is due to the assimilative rather than dissimilative metabolism. Stage   R1  R2   I  II  III  IV  I  II  III  IV COD Influent (mg·L -1 ) 985 (123) 776 (107) (130) 631 (130) 723 (68) 694 (204) 783 (130) 631 (130) 723 (68) Effluent (mg CaCO 3 ·L -1 ) 275 (172) 458 (157) 312 (116) 367 (165)  The accumulation of nitrate could be explained by the absence of a large enough anoxic zone inside the granules. However, oxygen penetration depth was estimated at 194 μm. Since most granules had at least 1 mm of diameter (see section "Characterization of mature granules"), and considering that nitrification occurred at high levels, it is reasonable to assume that, in the aerobic phase, the anoxic zone would have comprised the granule volume without oxygen, which corresponds to a 306 μm radius. This invalidates the hypothesis of an non-existent or small anoxic zone. Therefore, it is likely that the low occurrence of denitrification is being driven by another factor: the scarcity of substrate at the end of the aeration phase (ZHONG et al., 2013).

Effects of coal ash on AGS in simultaneous fill/draw SBR
Furthermore, the low phosphorus removal may be associated with the accumulation of nitrate. With the nitrate accumulated inside the reactor, the filling would occur in an anoxic environment.
This would allow substrate competition between denitrifying heterotrophic bacteria and polyphosphate-accumulating organisms (PAO).
In this case, due to the kinetics of the removal reactions involved, PAO would lose the competition . Therefore, complete denitrification of the nitrate accumulated during the previous cycle would be performed after feeding, impairing the action of PAO. Then, with the start of aeration, new amounts of nitrate would be produced. However, this portion would accumulate due to substrate scarcity to allow denitrification to take place at this point in the cycle. Thus, both denitrification and phosphorus removal efficiencies would be lowered.
Lastly, regarding other operational parameters, both the influent feed and the treated wastewater pH were close to neutral, varying between 6.5 and 8. Additionally, influent alkalinity (~700 mg CaCO 3 ·L -1 ) was lowered by half due to nitrification.

Microbial community diversity
The metagenomic analysis found 1,731 individuals in the inoculum sludge, 2,739 in the control sludge, and 2,790 in the sludge exposed to coal ash. The diversity and richness of the microbial community ( presented similar values to one another. This indicates that the granulation process in simultaneous fill/draw SBR influenced the microbial community selection more than the use of coal ash. Figure 3 shows that among microorganisms with an abundance greater than 10%, the most prominent at the phylum level were Proteobacteria, Chlorobi, Bacteroidetes, Planctomycetes, Chloroflexi, and Verrucomucrobia. These phyla have already been shown to be relatively abundant in both granular aerobic and activated sludge systems (HE et al., 2016;ZHANG et al., 2017).
Proteobacteria was the main phylum of the microbial communities, with no significant difference between the communities developed in R1 and R2. These results indicate that coal ash did not promote a strong selection of microorganisms at the phylum level.
Rhodocyclaceae was one of the families present in the three sludges analyzed (inoculum = 4.5%, R1 = 6.9%, R2 = 5.7%). Some studies associate this family with the production of EPS (SZABÓ et al., 2017). The presence of this type of microorganism is fundamental for the granulation process since the EPS formed mainly by PS, PN, and soluble microbial products (SMP) provides the aggregation of the various populations, especially when the selection pressure is limited.
In addition, some bacteria, not only utilize the carbohydrates and proteins present in the medium but also are responsible for the degrada- At the genus level, the granulation altered the population profile of the bacteria involved in the removal of nitrogen. In the inoculum sludge, Candidatus Nitrososphera (63%) was responsible for the oxidation of NH 4 + to NO 2 -, and Nitrospira (4.5%), by the oxidation of NO 2 to NO 3 -. However, after granulation, the first group did not adapt. The conversion of NH 4 + to NO 2 was performed by Prosthecobacter (R1 = 34% and R2 = 32%) , while Nitrospira continued to perform the conversion of NO 2 to NO 3 -(R1 = 16% and R2 = 11%).

CONCLUSIONS
Granulation was achieved in simultaneous fill/draw SBR operated with low upflow velocity. With the addition of coal ash, a residue from power plants, no significant differences were observed in terms of settleability, biomass retention, morphology, resistance to shear, and composition of the EPS matrix. COD removals were high (≥ 90%), while removals of nitrogen (~50%) and phosphorus (~40%) were low, possibly due to the presence of nitrate during the anaerobic phase. With granulation, the population profile of the microbial community was altered, mainly at the genus level. In general, it is verified that the operational conditions had a more considerable influence over granulation than the addition of coal ash. The possible reasons are because coal ash  Effects of coal ash on AGS in simultaneous fill/draw SBR supplementation was performed in a single step, the low sedimentation rate of this particular residue, and the weak interaction between the coal ash and the EPS formed in the granular sludge. These factors appear to have decreased or prevented the action of the ash as granulation nucleus, source of cations, and abrasive element.