781 Q IWA Publishing 2008 Water Science & Technology—WST | 58.4 | 2008 Dissolved oxygen as a key parameter to aerobic granule formation B. S. McSwain Sturm and R. L. Irvine ABSTRACT Much research has asserted that high shear forces are necessary for the formation of aerobic B. S. McSwain Sturm Department of Civil, Environmental & Architectural granular sludge in Sequencing Batch Reactors (SBRs). In order to distinguish the role of shear and Engineering, University of Kansas, dissolved oxygen on granule formation, two separate experiments were conducted with three Lawrence, KS 66045, bench-scale SBRs. In the ﬁrst experiment, an SBR was operated with ﬁve sequentially decreasing USA E-mail: firstname.lastname@example.org superﬁcial upﬂow gas velocities ranging from 1.2 to 0.4 cm s21. When less than 1 cm s21 shear R. L. Irvine was applied to the reactor, aerobic granules disintegrated into ﬂocs, with corresponding SBR Technologies, Inc., Chicago, IL, increases in SVI and efﬂuent suspended solids. However, the dissolved oxygen also decreased USA from 8 mg L21 to 5 mg L21, affecting the Feast/Famine regime in the SBR and the substrate E-mail: Robert.Irvine@sbrtechnologies.com removal kinetics. A second experiment operated two SBRs with an identical shear force of 1.2 cm s21, but two dissolved oxygen concentrations. Even when supplied a high shear force, aerobic granules could not form at a dissolved oxygen less than 5 mg L21, with a Static Fill. These results indicate that the substrate removal kinetics and dissolved oxygen are more signiﬁcant to granule formation than shear force. Key words | aerobic granules, dissolved oxygen, feast and famine, shear force, SBR INTRODUCTION In recent years, the Sequencing Batch Reactor (SBR) has loading rate on the formation and structure of granules. been used to form compact sludge structures, termed A high shear force has been shown to be an important aerobic granules. Granules can be described as a collection factor in the formation of aerobic granular sludge, and an of self-immobilized cells into a spherical form. This special extensive literature review presented the effect of superﬁcial case of bioﬁlm growth occurs without the addition of carrier gas velocity in a variety of reactor systems including UASB, material. Granular sludge have a wide range of beneﬁcial ﬂuidized-bed reactors, and aerobic granular reactors (Liu & properties compared to activated sludge ﬂocs, most notably Tay 2002). their strong structure and good settling property. A brief Shear caused by upﬂow aeration is the main detach- review of both anaerobic and aerobic granulation shows ment force in column-type SBRs during the formation of that a variety of microbial species are able to form granules, granules. Once granules are formed, interaction between leading researchers to hypothesize that granulation is not a granules may also serve as a detachment force. Most papers function of microbiological groups but of reactor operating published for aerobic granulation describe the shear force in conditions. terms of the superﬁcial upﬂow gas velocity, which is the Research on the necessary factors for aerobic granule aeration rate applied over the surface area of the reactor. formation has focused on operational parameters, such as Although the superﬁcial upﬂow gas velocity ignores shear the effect of settling time, shear force, and volumetric force caused by particle interaction, it is a constant and doi: 10.2166/wst.2008.393 782 B. S. McSwain Sturm and R. L. Irvine | Dissolved oxygen in aerobic granule formation Water Science & Technology—WST | 58.4 | 2008 easily calculated description of shear force caused by a municipal wastewater treatment plant. All SBRs were fed aeration. with a synthetic wastewater of glucose and peptone with In one study of the impact of shear force on granule nutrients (similar to that used by (Moy et al. 2002)) at a formation, Beun et al. utilized three different superﬁcial gas volumetric loading rate of 2.4 kg COD m23 day21 over 90 velocities in the operation of one column-type SBR. At the minutes of FILL for 4 cycles per day (30 min Static Fill, 21 highest superﬁcial gas velocity of 4.1 cm s , smooth 180 min React, 2 min Settle, 15 min Draw, 13 min Idle). granules formed. At lower superﬁcial gas velocities of 1.4 For the ﬁrst experiment, granules were formed in an 21 and 2.0 cm s , stable granules did not form (Beun et al. open SBR (Reactor 1) under a high superﬁcial upﬂow gas 1999). Tay et al. then studied the effect of aeration rate using velocity of 1.2 cm s21. After steady-state operation was four different column-type SBRs with superﬁcial gas reached and stable granules were formed, the aeration velocities of 0.3, 1.2, 2.4, and 3.6 cm s21, respectively. rate was systematically decreased to test the effect of shear Granules did not form in the reactor with the lowest force on the maintenance of granular sludge. After each superﬁcial gas velocity of 0.3 cm s21. The authors also change, the reactor was allowed to stabilize for at least two showed that SVI decreased as the superﬁcial gas velocity weeks before ﬁnal measurements were taken. Table 1 increased (Tay et al. 2001). presents the days of operation and aeration rates used, Although these papers suggest that a shear force is together with the corresponding superﬁcial upﬂow gas necessary for aerobic granulation to occur, the shear force is velocities. calculated from the aeration rate to the reactor. In these Except for the source of compressed air, the two experiments, the superﬁcial upﬂow gas velocity is decreased reactors used for the second experiment were operated by decreasing the aeration rate, which inherently changes identically: both an open SBR (Reactor 2) and a closed the dissolved oxygen supply and concentration in the SBR (Reactor 2_lowDO) had a superﬁcial upﬂow 21 reactors. However, the dissolved oxygen concentration is gas velocity of 1.2 cm s . Compressed air was supplied often times not reported, so it is hard to discern the constantly to Reactor 2 with the DO always near saturation respective effects of a decreased shear force versus a (8– 8.5 mg L21) during React. Reactor 2_lowDO received a decreased DO on granule formation. mixture of recycled air, nitrogen gas, and compressed air, The purpose of this study is to distinguish the effect of with the DO always controlled to be less than 5 mg L21. shear force and dissolved oxygen concentration on granule Standard wastewater measurements were taken regu- structure and formation in SBRs with a Static Fill. To larly according to Standard Methods (Clesceri et al. 1998). accomplish this objective, two reactor experiments were The development of ﬂocs and granules were observed using conducted. In the ﬁrst, one SBR was operated with a stereomicroscope (Leica Wild MPS 46/52), and images sequentially reduced aeration rates, and the aeration rate were taken of mixed liquor. Image analysis was performed and DO were coupled. In a second experiment, two SBRs to measure the average diameter and aspect ratio of all were operated with identical aeration rates, but one reactor particles greater than 300 mm in diameter. For each sample, was fed a nitrogen/air mixture to maintain a lower DO than 300 –500 particles were measured using image analysis. the SBR receiving air only. Table 1 | SBR Aeration rates applied in experiment 1 Aeration rate Superﬁcial upﬂow gas velocity METHODS Days of operation (L hr21) (cm s21) 114 275 1.2 Experiments were performed in two open and one closed 14 230 1.0 column-type SBRs (Figure 1). All SBRs were aerated at a 22 185 0.8 rate of 275 L h21 unless otherwise noted, with a 50% 38 140 0.6 volumetric exchange ratio and 4 L total volume. The 21 95 0.4 reactors were inoculated with 4 L of activated sludge from 783 B. S. McSwain Sturm and R. L. Irvine | Dissolved oxygen in aerobic granule formation Water Science & Technology—WST | 58.4 | 2008 Figure 1 | (Left) Schematic of reactor 1 and reactor 2 with compressed air supplied to create shear force; (right) schematic of reactor 2_lowDO with compressed air and nitrogen gas supplied to maintain a high shear force with a reduced dissolved oxygen concentration. The aspect ratio (min diameter/max diameter of a given SVI30 measurement (height of sludge bed after 30 minutes) particle; 0 ¼ line, 1 ¼ circle) reﬂects the roundness of a does not adequately describe the settling property of particle, with an aspect ratio of one being a perfect circle. granular sludge, which settles in reactors within a few minutes. In order for the SVI measurement to reﬂect settling velocity as well as sludge bed compactness, the SVI should be measured after shorter settling times RESULTS (Schwarzenbeck et al. 2004). In Figure 3, the SVI is Experiment 1—different shear forces in one reactor reported after 2 and 30 minutes settling; two minutes At the initial superﬁcial gas velocity of 1.2 cm s 21 , Reactor 1 was chosen since it is the time of the Settle phase in the had completely granular sludge with an average MLSS of SBR. Except for the SVI at the lowest aeration rate, there 6.0 ^ 0.9 g L 21 , an SVI of 33 ^ 5 mL g 21 , an efﬂuent SS of was no difference between the SVI2 and SVI30 values, 61 ^ 20 mL g21, and an efﬂuent COD of 27 ^ 9 mg L21. As indicating that the maximum settling and compactness shown in Figure 2, the average aspect ratio of sludge in granular reactors occurs within the ﬁrst few minutes of decreased linearly below a superﬁcial gas velocity of Settle. At the lowest aeration rate, the biomass was 0.8 cm s 21 . The distribution of ﬂocculent versus granular predominantly ﬂocculent, and there was a difference sludge was analyzed using the total percentage of particles between SVI30 and SVI2. with aspect ratios less than 0.5 (ﬂocculent) and greater then The overall COD removal efﬁciency in Reactor 1 did 0.7 (granular). Although the values of 0.5 and 0.7 are not change as the aeration rate was altered and granules arbitrary, they are useful in categorizing particles into non- disappeared from the reactor. However, the slope of spherical (,0.5) and spherical (. 0.7) categories. At substrate uptake at the beginning of React decreased as superﬁcial gas velocities above 1.0 cm s 21 , over 60% of all the shear force decreased. The COD removal curves during particles had an aspect ratio greater than 0.7, indicating that one SBR cycle at the highest and lowest aeration rates are the majority of aggregates were granular. Alternatively, only presented in Figure 4. At the highest superﬁcial upﬂow gas 36% of aggregates at the lowest aeration rate were spherical. velocity of 1.2 cm s21, the duration of the feast period was The sludge volume index (SVI) in Reactor 1 began to shortest (tfeast < 45 min) with the magnitude and concomi- 21 tant duration of the famine period the longest (tfamine < 135 increase at gas velocities below 0.8 cm s . Similarly, the efﬂuent suspended solids (ESS) concentration began to min). At the lowest superﬁcial upﬂow gas velocity of increase at gas velocities below 1.0 cm s 21 . The average 0.4 cm s21, the duration of the feast period was longer, values for each aeration rate are plotted in Figure 3. For the extending over the majority of React, as the COD did not SVI measurement, it has been suggested that the standard reach background levels until 120 minutes of React. 784 B. S. McSwain Sturm and R. L. Irvine | Dissolved oxygen in aerobic granule formation Water Science & Technology—WST | 58.4 | 2008 extended period of low DO (below 5 mg/L), and the DO in the SBR never reached saturation. Experiment 2—same shear forces and different dissolved oxygen in two reactors In the ﬁrst experiment, DO and shear force were coupled and incrementally reduced. In the second experiment, the shear force was held constant (1.2 cm s21 superﬁcial upﬂow gas velocity) in Reactor 2 and Reactor 2_lowDO, but the latter had a DO less than 5 mg L21 throughout the React phase. In this experiment, Reactor 2 formed granules in the ﬁrst few weeks after the initial biomass washout that is typically seen with short setting times. This reactor behaved similarly to Reactor 1 in the ﬁrst experiment. However, in Reactor 2_lowDO, the MLSS never reached 1 mg L21, and only ﬁlamentous ﬂoc structures were observed in the reactor. Figure 5 shows two images of sludge formed in each reactor at steady-state (Day 90 Operation). Due to the nature of biomass in Reactor 2_lowDO, it was impossible to perform image analysis. In order to have a representative sludge sample for image analysis, 300– 500 particles were measured per day. Since Reactor 2_lowDO had such low biomass content, a quantitatively meaningful sample could not be analyzed. In terms of COD removal, both reactors had similar performance, despite the differences in MLSS content and sludge properties. The inﬂuent COD varied between 350 and 500 mg L21 at the beginning of React, and the efﬂuent COD for both reactors was consistently less than 80 mg L21. The Figure 2 | Aspect ratio versus the superﬁcial upﬂow gas velocity in reactor 1. observation of such a consistently low efﬂuent COD for (Top) below a superﬁcial gas velocity of 0.8 cm/sec, the average aspect Reactor 2_lowDO was interesting since the reactor had such ratio decreases linearly. (middle) The % of total particles with an aspect ratio less than 0.5 increases, and (bottom) the % of particles with an a low mixed liquor sludge concentration. If COD was aspect ratio greater than 0.7 decreases. consistently being removed by the mixed liquor, this should result in a growth of biomass—which was not observed in the To understand why substrate uptake rates changed with mixed liquor. An MLSS measurement represents only the reduced aeration rates, the DO in the mixed liquor was suspended biomass in the reactor volume, and it omits any measured during React. With a superﬁcial upﬂow gas biomass accumulated on the reactor walls. Bioﬁlm wall velocity of 1.2 cm s21, the DO dropped to 2 mg L21 during growth is generally a signiﬁcant problem during the ﬁrst few the ﬁrst 20 minutes of React while the majority of weeks of reactor operation, when the biomass has suffered biodegradable COD was consumed. The DO in the SBR signiﬁcant washout. For Reactor 2 and other granule was maintained at saturation for the remainder of React. In reactors, there was very little bioﬁlm growth on the walls of contrast, the SBR with the lowest shear force had an the reactor at steady-state. However, for Reactor 2_lowDO, 785 B. S. McSwain Sturm and R. L. Irvine | Dissolved oxygen in aerobic granule formation Water Science & Technology—WST | 58.4 | 2008 Figure 3 | SVI and efﬂuent SS as a function of superﬁcial upﬂow gas velocity in reactor 2. Error bars represent a 95% conﬁdence interval. bioﬁlm wall growth was a problem over the entire operation shear forces function to detach ﬁlamentous organisms period. Wall growth in both SBRs was estimated over a week from the surface of aggregates and form spherical aggregates using inserted glass slides. For Reactor 2_lowDO, approxi- (Liu & Tay 2002). This would have major implications for mately 4 g biomass grew attached to the reactor walls per the application of granular reactors at full-scale. If this week, while Reactor 2 accumulated only 0.25 g biomass per hypothesis were true, airlift reactors would be desired over week. In the SBR with low DO, biomass grew preferentially bubble-column reactors since they create more localized attached to the wall, with only ﬁlamentous ﬂocs accumulat- shear, and the reactor would have a high energy demand ing in the mixed liquor. to supply mixing via aeration (Beun et al. 2002; de Bruin et al. 2004). However, when the shear force is controlled only by the aeration rate, reactors with high versus low aeration rates DISCUSSION do not only differ in terms of shear force, and this fact has The results from experiment 1 correlate with previous been previously ignored in literature. The DO proﬁle in the experiments that conclude a sufﬁciently high shear force is reactor under the lowest aeration rate shows that the DO necessary for aerobic granule formation (Tay et al. 2001). was below 5 mg L21 during the period of maximum Using only this data, one might extend this observation to substrate uptake at the beginning of React. At high say that a high shear force is also necessary for maintenance superﬁcial gas velocities (when the DO was always near of an established granule reactor. This would support a saturation), biodegradable substrate disappeared quickly shear-based granule formation hypothesis that states high and completely. For Reactor 1 operated with a superﬁcial Figure 4 | (Left) COD removal and (right) DO during the react phase of an SBR cycle with the highest and lowest aeration conditions in reactor 1. 786 B. S. McSwain Sturm and R. L. Irvine | Dissolved oxygen in aerobic granule formation Water Science & Technology—WST | 58.4 | 2008 Figure 5 | Sludge structure at day 90 operation for reactor 2_lowDO (left) and reactor 2 (right). (scale bar ¼ 5 mm). upﬂow gas velocities of 1.0 or 1.2 cm s21, the feast period In separate studies, aerobic granulation has been was always short (tfeast , 45 min) with a long famine period reported in Mixed Anaerobic Fill SBRs with a range of that extended the duration of React. A decreased aeration DO concentrations in the subsequent React phase. In rate resulted in a longer feast period and the absence of a Mixed Fill reactors, phosphate accumulating organisms famine period. In turn, smooth, compact granular sludge (PAOs) and glycogen accumulating organisms (GAOs) are became ﬁlamentous and increasingly ﬂocculent. selected, and researchers have argued that granule stability The importance of feast and famine periods to the improves with the selection of slow-growing organisms (de formation of aerobic granules has been previously reported Kreuk & van Loosdrecht 2004). The results presented in this (McSwain et al. 2004; Liu & Tay 2006). In the study by study are not comparable to granules formed under McSwain et al. granular sludge became increasingly anaerobic or Anaerobic, Mixed Fill conditions since the ﬁlamentous as the intensity of feast to famine in an SBR microbial species and substrate removal kinetics are decreased. As granules disappeared, the SVI and ESS values different. In Static Fill reactors, biological reactions do not increased. In experiment 1 reported herein, the limitation of begin until the Aerobic, React phase, and aerobic hetero- DO during React altered the feast and famine periods trophs are expected to dominate the reactor. For Anaerobic, within the reactor, limiting the substrate uptake at the Mixed Fill SBRs, such as those used by de Kreuk et al. beginning of React and causing granules to disappear. Liu biological reactions involved in nutrient removal occur et al. conﬁrmed the importance of this feast to famine during the Fill phase. Therefore, the selection of microbial transition in granule reactors. In their study, Liu et al. kept species depends on different environments in Static Fill and Mixed Fill SBRs. aeration rates high during the ﬁrst 110 minutes of React, and decreased the aerations rates once the famine period began. Since this operation did not disrupt the substrate uptake kinetics, the granular structure was not affected (Liu CONCLUSIONS & Tay 2006). The ﬁrst experiment conﬁrmed previous reports that In experiment 2, a DO below 5 mg L21 prevented the granular sludge disintegrates under low shear forces. formation of aerobic granular sludge at high shear forces. Aerobic granular sludge is not stable at superﬁcial The two separate experiments indicate that the dissolved upﬂow gas velocities below 1.0 cm s21 when the shear oxygen concentration is more important to substrate uptake force is controlled only by the aeration rate. In such cases, kinetics and granule formation than the shear force. This low aeration rates create an oxygen limitation during conclusion differs from previous reports, but it is based on a React, decreasing the intensity and extending the duration more complete data set. of the feast period and shortening the famine period. 787 B. S. McSwain Sturm and R. L. Irvine | Dissolved oxygen in aerobic granule formation Water Science & Technology—WST | 58.4 | 2008 Although a correlation can be made between the decrease Clesceri, L. S., Greenberg, A. E., Eaton, A. D. (eds) 1998 Standard of shear force and loss of granules in a reactor, this Methods for the Examination of Water and Wastewater. American Public Health Association/American Water conclusion ignores the inﬂuence of shear on DO and Works Association/Water Environment Federation, substrate removal kinetics. The current results show that Washington, DC, USA. DO and substrate removal kinetics is more important to de Bruin, L. M. M., de Kreuk, M. K., van der Roest, H. F., van Loosdrecht, M. C. M. & Uijterlinde, C. 2004 Aerobic granular aerobic granule formation and maintenance than the sludge technology, alternative for activated sludge. Water Sci. shear force. Technol. 49(11 –12), 1–7. de Kreuk, M. K. & van Loosdrecht, M. C. M. 2004 Selection of slow growing organisms as a means for improving aerobic granular sludge stability. Water Sci. Technol. ACKNOWLEDGEMENTS 49(11 – 12), 9 –17. Liu, Y. & Tay, J. H. 2002 The essential role of hydrodynamic shear This research was funded by the US Department of force in the formation of bioﬁlm and granular sludge. Water Education and the German Research Foundation. The Res. 36, 1653 –1665. authors thank Dr. Peter Wilderer and the Technical Liu, Y. Q. & Tay, J. H. 2006 Variable aeration in sequencing batch reactor with aerobic granular sludge. J. Biotechnol. 124(2), University of Munich Institute for Water Quality and 338– 346. Waste Management for hosting this research. McSwain, B. S., Irvine, R. L. & Wilderer, P. A. 2004 The effect of intermittent feeding on aerobic granule structure. Water Sci. Technol. 49(11 –12), 19 –25. REFERENCES Moy, B. Y. P., Tay, J. H., Toh, S. K., Liu, Y. & Tay, S. T. L. 2002 High organic loading inﬂuences the physical characteristics of Beun, J. J., Hendriks, A., van Loosdrecht, M. C. M., Morgenroth, aerobic sludge granules. Lett. Appl. Microbiol. 34, 407 –412. E., Wilderer, P. A. & Heijnen, J. J. 1999 Aerobic granulation Schwarzenbeck, N., Erley, R. & Wilderer, P. A. 2004 Aerobic in a sequencing batch reactor. Water Res. 33(10), granular sludge in an sbr-system treating wastewater rich in 2283 –2290. particulate matter. Water Sci. Technol. 49(11 –12), 41 –46. Beun, J. J., van Loosdrecht, M. C. M. & Heijnen, J. J. 2002 Aerobic Tay, J. H., Liu, Q. S. & Liu, Y. 2001 The effects of shear force on granulation in a sequencing batch airlift reactor. Water Res. the formation, structure and metabolism of aerobic granules. 36, 702 –712. Appl. Microbiol. Biotechnol. 57, 227– 233.