Journal of Water Process Engineering 45 (2022) 102505 Contents lists available at ScienceDirect Journal of Water Process Engineering journal homepage: www.elsevier.com/locate/jwpe Ammonium oxidation activity promotes stable nitritation and granulation of ammonium oxidizing bacteria Xènia Juan-Díaz, Julián Carrera *, Julio Pérez GENOCOV Research Group, Department of Chemical, Biological and Environmental Engineering, School of Engineering, Universitat Autònoma de Barcelona, Ed. Q- Campus UAB, 08193 Bellaterra, Barcelona, Spain A R T I C L E I N F O A B S T R A C T Keywords: Two-stage partial nitritation/anammox (PN/AMX) processes have been pointed out as a feasible configuration Operational start-up strategy for achieving mainstream anammox. For two-stage configurations, stable partial nitritation has been reported Superficial gas flow velocity feasible in granular sludge reactors. This study aimed to explore the operating conditions involved in the Conventional activated sludge development of an autotrophic aerobic granular sludge using floccular sludge as inoculum. The influence of Autotrophic aerobic granular sludge Ammonium oxidation rate different parameters such as free ammonia concentration, settling time, superficial gas flow velocity and ammonium oxidation rate was investigated. Enhancing ammonium oxidation activity since the early phase of the operation (i.e. using conventional activated sludge as inoculum enriched with a fraction of a floccular nitrifying biomass) promoted a fast development (ca. 30 days) of an autotrophic aerobic granular sludge performing stable nitritation. When the seeded sludge presented a low nitrifying activity (lower than 0.1 g N L− 1 d− 1), the increase of the air-flow rate triggered the formation of an autotrophic aerobic granular sludge since ammonium oxidation activity was promoted. Contrarily, imposing low settling times (10 min) or strong free ammonia inhibitory conditions (FA concentrations higher than 50 mg N L− 1) were shown to negatively influence the achievement of high ammonium oxidation rates, hampering the development of an autotrophic aerobic granular sludge. This study demonstrated the importance of ensuring high ammonium oxidation rates (higher than 0.2 g N L− 1 d− 1) for the proper development of an autotrophic partial nitritation granular sludge. 1. Introduction study revealed that AOB presented a better oxygen affinity compared to NOB simply because the position of cell clusters within a stratified The partial nitritation/anammox (PN/AMX) process has been pro­ granule [6]. Most of lab- and pilot-scale studies that attained stable posed as an efficient and cost-effective technology for nitrogen removal partial nitritation in granular sludge reactors at mainstream conditions in urban wastewater treatment plants (WWTPs) [1]. Under this scenario, used a pre-formed granular sludge [2,4] or supplied a substratum (i.e. two-stage PN/AMX systems have been pointed out as a feasible config­ activated carbon) with the inoculum [7,8]. Designing a start-up phase to uration due to the reported successful results at temperatures as low as form a nitrifying granular sludge in which conventional activated sludge 11 ◦ C [2,3]. One of the main challenges of the partial nitritation stage is (CAS) is used as inoculum seems to be an appealing strategy for the the repression of nitrite oxidizing bacteria (NOB) during long-term scale-up of this technology. However, seeding CAS as inoculum prevents operation, this is currently limiting the full-scale implementation of the initial suppression of nitrite oxidation activity through the same PN/AMX in the mainstream of WWTPs. However, successful partial strategy designed for autotrophic aerobic granular sludge (described nitritation stability has been reported in granular sludge reactors at lab- and applied in [2,7]). The nitrite oxidation can be reduced or even scale for long-term mainstream conditions [2,3]. Granule morphology suppressed through inhibition of NOB by free nitrous acid. However, allows to stablish oxygen and substrate gradients, which enhances this technology requests the availability of a nitrite concentrated stratification of ammonium oxidizing bacteria (AOB) over NOB [4]. By wastewater [31]. An alternative start-up strategy (i.e. in the short term) operating at certain residual ammonium concentrations, AOB increase is to avoid nitrite oxidation by transient inhibition of NOB by free their oxygen consumption rate, meaning that less oxygen is available for ammonia (FA) within floccular biomass [9,10], and simultaneously NOB, relegating them to deeper layers [4,5]. Accordingly, a modelling develop a granular sludge. To this end, a nitrogen concentrated * Corresponding author. E-mail address: julian.carrera@uab.cat (J. Carrera). https://doi.org/10.1016/j.jwpe.2021.102505 Received 29 October 2021; Received in revised form 18 November 2021; Accepted 6 December 2021 2214-7144/© 2021 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). X. Juan-Díaz et al. Journal of Water Process Engineering 45 (2022) 102505 wastewater, such as that coming from the digested sludge dewatering 2. Materials and methods system (i.e. reject water) could be used. Once the granular sludge has been developed, FA inhibition will not be longer required and, thus, 2.1. Reactor configuration and operation operation at mainstream conditions after the start-up phase could be safely imposed by maintaining suitable Dissolved Oxygen/Total An air-lift reactor with a working volume of 16.3 L operated in a SBR Ammonia Nitrogen ([DO]/[TAN]) concentration ratio in the bulk mode was used (Fig. 1). The reactor presented a downcomer-to- reactor liquid [2,3]. separator diameter ratio of 0.36, a total length-to-downcomer diam­ Despite some studies focused on the achievement of an autotrophic eter ratio of 11 and a riser height-to-diameter (Hr/Dr) ratio of 19.4. The aerobic granular sludge performing partial nitritation [11,12], the riser diameter was of 0.05 m and it was located at ca. 10 cm above the particular set of operating conditions leading to an effective start-up of reactor bottom. The SBR operation consisted of four phases: a static this reactor type remains to date not well defined. In contrast, the feeding phase (16 min), an aeration phase (with variable time length) development of heterotrophic aerobic granular sludge for the simulta­ neously removal of Chemical Oxygen Demand (COD), nitrogen (nitrifi­ Table 1 cation and denitrification) and phosphorus has been widely studied (e.g. Experimental conditions for each operational strategy. NEREDA®) [13]. Heterotrophic aerobic granular sludge has been mostly cultivated from CAS in sequential batch reactor (SBR) operations [14]. FA (mg N L− 1) and pH Settling time Superficial gas flow (min) velocity (cm s− 1) The heterotrophic aerobic granulation process is known to be enhanced under short settling times, high shear stress and feast-famine conditions, Strategy-I 7 ± 2 (pH = 8.0 ± 0.2) 30 to 10 0.06 (day 1 to day 18) whereas the exposure to insoluble and particulate COD can comprise the in 18 daysc 0.37 (day 20 to day 40) granulation process as filamentous bacteria tend to proliferate [14,15]. Strategy- 57 (day 0 to 8, pH = 8.5) 30 to 10 0.06 (day 1 to day 56) Nevertheless, it is unknown whether these conditions could also favour II 17 ± 4 (day 8 to end of in 44 daysc 0.23 (day 57 to day autotrophic aerobic granulation. Indeed, although autotrophic nitrify­ operation, pH = 8.0 ± 0.2) 80) ing granules for nitrification have been developed, the information on Strategy- 15 ± 4 (day 0 to 7, pH = 8.1 30 to 10 0.26 ± 0.07d III ± 0.1)a in 50 daysc the nitrifying bacteria granulation mechanisms remains limited [16,17]. 88 (day 8, pH = 8.6)a The aim of this study is to investigate the operating conditions that 113 (day 14, pH = 8.6)b can contribute into the development of an autotrophic aerobic granular 47 (day 18, pH = 8.3)a sludge achieving stable nitritation. For this purpose, an air-lift reactor 21 ± 7 (day 19 to end of seeded with CAS was used by working under a SBR mode. Among the operation, pH = 8.1 ± 0.2)a Strategy- 14 ± 4 (pH = 8.0 ± 0.1) 30 0.28 (day 1 to day 20) different operational parameters that can influence the formation of a IV 0.13 ± 0.04 (day 21 to granular biomass performing partial nitritation, this study has focused in day 60) the following: (i) FA concentrations in the bulk liquid, (ii) settling time, a Volumetric exchange ratio was maintained at a17 and b50%, respectively. (iii) shear stress forces (in terms of superficial gas flow velocity) and (iv) b Volumetric exchange ratio was maintained at a17 and b50%, respectively. ammonium oxidation activity (in terms of ammonium oxidation rates, c The length period refers to the needed time to progressively decrease settling AORs). Four operational strategies, with different combinations of the time from 30 to 10 min. above mentioned parameters, have been planned and executed to d Air and nitrogen gas were supplied in strategy-III. The corresponding air and determine which of these parameters present more influence on the nitrogen gas contributions into the superficial gas flow velocity are shown in autotrophic granulation. Fig. A.2. A B TAN probe pH probe PLC DO probe Sodium Effluent bicarbonate dispenser Jacketed system Reject water Air/Nitrogen Inflow pump Fig. 1. (A) Picture of the air-lift reactor. (B) Schematic diagram of the reactor set-up with the corresponding peripheral instrumentation. The aeration phase lasted until the total ammonia nitrogen (TAN=N-NH4+ + N-NH3) decreased to the given TAN set-point. When the aeration was stopped, a settling phase (with a variable time length) occurred followed by the final discharge step (16 min). 2 X. Juan-Díaz et al. Journal of Water Process Engineering 45 (2022) 102505 and a settling phase (with variable time length) (see Table 1) and a size was measured by a laser diffraction analysis system (Malvern discharge phase (16 min). Cycle length phase was automatically regu­ Mastersizer Series 2600, Malvern instruments Ltd., UK). The percentage lated by using the on-line measurement of the total ammonia nitrogen of granular sludge was determined as the volumetric fraction of particles (TAN = N-NH4+ + N-NH3) concentration in the bulk liquid (Fig. 1). The with a diameter higher than 200 μm [20]. aeration phase lasted until TAN decreased to the given TAN set-point. Experiments were classified in numbered strategies, as detailed in Sec­ 2.4. Calculations tion 2.5 (Table 1). Limited research effort has been given to the importance of the volumetric exchange ratio in the development of an The nitrogen loading rate (NLR) and ammonium oxidation rate autotrophic aerobic granular sludge, as most studies focused on the (AOR) were calculated based on Eqs. (1) and (2), respectively. Using development of a heterotrophic aerobic granular sludge [18]. Indeed, nitrite and nitrate concentrations when treating reject water to estimate most of the studies that focused on the development of an aerobic the AOR it is not very precise since the high concentrations of nitrite granular sludge performing partial nitration set the volume exchange (and/or nitrate) together with potential interferences in the analysis ratio at 40–50% [11,12]. In this study, the volumetric exchange ratio would result in a low accuracy measurement. Therefore, AOR was was set at 40%, except for strategy-III, where it was correspondingly calculated based on influent and effluent TAN concentrations. Addi­ adjusted according to the operational needs (Table 1). The feeding was tionally, mass balances were periodically checked to evaluate the degree directly supplied from the bottom of the reactor. During the aeration of denitrification, which was minimal as there was no biodegradable phase, air (and nitrogen gas, if applicable) were supplied through a COD in the inflow and because no anammox was inoculated. Since TAN, porous membrane diffusor located at 30 mm above of the reactor bot­ TNN and nitrate concentrations in the effluent were precisely moni­ tom. Air and nitrogen flow rates were controlled by two different rota­ tored, no specific respirometric assays were performed as they would meters. DO was measured on-line by a means of a DO electrode (DO only provide and indirect measurement of activity and they could affect 60–50, Crison Instruments, Spain). The pH was measured on-line with a the granulation process since they should be performed in a mechanical pH-electrode (pH 53–33, Crison Instruments, Spain) with a variable set- stirred respirometer. point by adding dissolved NaHCO3. Ammonium concentrations and TANINF temperature in the bulk liquid were measured by using an on-line probe NLR = (1) (AISE sc with a CARTRICAL cartridge, Hach Lange, Germany). Tem­ HRT perature was maintained at 20 ± 1 ◦ C by means of a cooling system TANINF − TANEFF connected to the reactor jacket. All sensors and actuators were moni­ AOR = (2) HRT tored and connected to the PLC system. where 2.2. Inoculum and wastewater characteristics VR HRT = (3) VR ⋅VER⋅(number of cycles/day) The air-lift reactor was inoculated with CAS from a municipal WWTP in each start-up strategy. For the last experiment (strategy-IV), the air- and TANINF corresponds to influent TAN concentrations and TANEFF to lift reactor was also inoculated with flocculent sludge coming from a effluent TAN concentrations (g N L− 1). HRT, VR and VER correspond to pilot-scale reactor treating a sidestream water successfully performing the hydraulic retention time (day), effective reactor volume (L) and to partial nitritation. The inoculated ratio of volatile suspended solids the volume exchange ratio, respectively. (VSS)/ total suspended solids (TSS) was 0.8 for all strategies. By using the corresponding TAN concentrations, FA concentrations The air-lift reactor was fed in all operational strategies with a side­ at the beginning of the cycle (FABC) and at the end of the cycle (FAEFF) stream water coming from the dewatering of the digested sludge of a were calculated based on acid-base equilibrium [21], municipal WWTP (i.e. reject water). The used reject water was stored in a tank at 13 ± 1 ◦ C. During strategy-I, the reactor was fed with a side­ FA = TAN⋅10pH (4) stream water coming from the Manresa WWTP (Catalonia, Spain). The 6344 eT+273 + 10pH composition of the sidestream presented the following characteristics: TAN: 238–309 mg N L− 1; total nitrite nitrogen (TNN = N-NO− 2 + where FA corresponds to the FA concentration (mg N L− 1), TAN corre­ HNO2): 1 to 28 mg N L− 1; nitrate: 0.5–17 mg N L− 1; pH: 7.9 ± 0.1. To sponds to ammonium concentrations at the end of the cycle (TANEFF) (i. work at higher FA concentrations, a sidestream water coming from a e. effluent) or to the ammonium concentrations at the beginning of the different municipal WWTP (Rubí – Valldoreix WWTP, Catalonia, Spain) cycle (TANBC) (mg N L− 1). TANBC concentrations were calculated was used in strategies II, III and IV. The characteristics of this sidestream considering the volumetric exchange ratio and the effective reactor wastewater were: TAN: 512–963 mg N L− 1; TNN: 0–97 mg N L− 1; ni­ volume. pH and T (◦ C) correspond to the values of pH and temperature trate: 0–21 mg N L− 1; pH: 8.1 ± 0.1. The wide range of TAN concen­ in the air-lift reactor. trations depended on the anaerobic digester performance. The The superficial gas flow velocity (ug, cm s− 1) was calculated using variability of TNN and nitrate concentrations were because nitrification the following equation, occurred within the stored wastewater tank. qair + qN2 ug = (5) A 2.3. Analytical methods where qair and qN2 correspond to the air and nitrogen gas flow rates Influent and effluent samples of the air-lift reactor were regularly (cm3 s− 1), respectively and A corresponds to the area of the riser (cm2). analysed to determine TAN, TNN and nitrate concentrations. Samples The superficial gas flow velocity was used as an indirect form of quan­ were previously filtered (0.22 μm) before analysis. TAN concentrations tifying the shear stress conditions applied to the biomass of the airlift were analysed off-line by means of a gas selective electrode (GSE) reactor. (AMTAX sc, Hach Lange, Germany) or by using colorimetric Hach Lange kits (LCK303 or LCK302, Hach Lange, Germany). TNN and nitrate 2.5. Operational strategies concentrations were analysed off-line with ionic chromatography using ICS-2000 Integrated Reagent-Free IC system (DIONEX Corporation, The following parameters: (i) FA concentrations in the bulk liquid, USA). TSS and VSS concentrations and sludge volume index (SVI) were (ii) settling time, (iii) shear stress forces (in terms of superficial gas flow analysed according to Standard Methods [19]. Average biomass particle velocity), and (iv) ammonium oxidation activity (in terms of AOR) were 3 X. Juan-Díaz et al. Journal of Water Process Engineering 45 (2022) 102505 combined in four different operational start-up strategies to achieve an velocity was kept at the same initial value than in strategy-III, though autotrophic granular sludge performing partial nitritation (Table 1). by exclusively supplying air. Each one of the experiments (listed below as strategies) lasted the period of time required to (i) develop a granular sludge and (ii) achieve a stable 3. Results end product (either nitrite, i.e. partial nitritation or nitrate i.e., nitrifi­ cation). The NLR, and thus, the HRT, were regulated in each strategy as The time course concentrations of TAN, TNN and nitrate in the the cycle length was automatically controlled by the on-line TAN mea­ effluent as well as TAN influent concentrations for each operational surement (see Section 2.1 and Fig. 1 for details). Detailed information of strategy can be found in Fig. 2. Particle size together with settling time each strategy can be found below. and superficial gas flow velocity, as well as AORs have been plotted in Fig. 3. The time course concentrations of FA, nitrate (influent and • Strategy-I. The aim of strategy-I was to operate at moderate FA effluent) and solids can be found in Fig. 4. NLR, together with AOR, and concentrations with a fast decrease of settling time and a low initial the DO concentrations are depicted in Supporting Information (Figs. SI.1 superficial gas flow velocity. The applied FA concentrations were 7 and SI.3, respectively). ± 2 mg N L− 1 (Table 1), in the range reported for an effectively NOB repression within floccular sludge [22,23]. Settling time was 3.1. Strategy-I: operating at moderate FA concentrations with a fast decreased from 30 to 10 min (Table 1). Superficial gas flow velocity decrease of settling time and low initial superficial gas flow velocity was maintained constant in the first part of the experiment by exclusively supplying air. Then, superficial gas velocity was stepwise Initial inoculum size was 93 ± 5 μm and slightly increased to 139 ± increased (Table 1) when the granulation process was still not 15 μm after 17 days of operation, with a settling time decrease from 30 detected after decreasing settling time below 20 min. The reactor to 15 min (Fig. 3). Imposing a FA concentration (7 ± 2 mg N L− 1, was seeded with CAS with an initial biomass concentration of 1 g VSS Table 1) in the reactor bulk liquid resulted in a fast achievement of L− 1 . partial nitritation as observed by the gradually nitrate depletion from 28 • Strategy-II. During strategy-II, high initial FA concentrations were to 1 ± 1 mg N L− 1 in 13 days of operation (Fig. 2). Around day-20, imposed with a progressive decrease of settling time and a low initial particle size increased up to 200 μm (Fig. 3), a threshold usually uti­ superficial gas flow velocity. During the first 8 days of operation, FA lized to identify the achievement of granular sludge. Some days before concentrations were maintained at 57 mg N L− 1 (Table 1). From day- reaching this milestone, both superficial gas flow velocity and AOR 8 onwards, FA concentrations were maintained at 17 ± 4 mg N L− 1 increased. Thereafter, particle size gradually increased, reaching a until the end of the operation. Settling time was decreased from 30 to diameter of 270 ± 21 μm after 38 days of operation (Fig. 3). At the end of 10 min at slower pace than in strategy-I (Table 1). Superficial gas the operation (i.e. day 38), 58% of particles were greater than 200 μm flow velocity was controlled by exclusively supplying air as in with good settleability properties (SVI5/SVI30 = 1), indicating a shift strategy-I (Table 1). The reactor was seeded with CAS with an initial towards granular sludge (Fig. 5). However, from day 22-onwards, par­ biomass concentration of 2 g VSS L− 1. tial nitritation was progressively deteriorated, achieving up to a • Strategy-III. The objective of strategy-III was to start the operation maximum nitrate concentration of 121 mg N L− 1 in only 11 days (Fig. 2), at low FA concentrations followed by a sharp increase during a short despite FA concentrations remained stable throughout the operation period to try to knock down NOB, when granule development was (Table 1). still incipient (Table 1). After this initial marked increase, FA con­ In the next experiment (strategy-II), high FA inhibitory concentra­ centrations were decreased and maintained at 21 ± 7 mg N L− 1 until tions were applied in the early stage of the operation to ensure a proper the end of the operation (Table 1). Settling time was decreased as in NOB inhibition. After the initial inhibitory period, FA concentrations strategy-II (Table 1). Additionally, nitrogen gas was supplied were subsequently maintained at higher levels (compared to strategy-I, together with air to operate at higher superficial gas flow velocity see Fig. 4A) to strengthen NOB inhibition throughout the operation. (compared to strategies I and II), and thus, to explore the effects of Settling time decrease and superficial gas flow velocity followed the shear stress on granulation (see Fig. SI.2 in Supporting Information same pattern than in strategy-I. for details). The reactor was seeded with CAS with an initial biomass concentration of 1 g VSS L− 1. 3.2. Strategy-II: operating at initial inhibitory FA concentrations with a • Strategy-IV. The aim of strategy-IV was to enhance ammonium progressive decrease of settling time and low initial superficial gas flow oxidation activity since the early phase of the operation. The airlift velocity reactor was seeded with CAS (26.1 g VSS) but also with a fraction of floccular biomass performing partial nitritation (12.1 g VSS) (ratio Initial particle size was 98 ± 4 μm and only increased to 149 ± 25 μm ca. 2:1, respectively). To reduce the influence of other parameters, in the first 51 days of operation (Fig. 3), despite decreasing settling time FABC concentrations were maintained similar as in strategy-I. from 30 to 20 min. To avoid nitrate production throughout the opera­ Settling time was not decreased but maintained high and roughly tion, as occurred in strategy-I, high inhibitory FA conditions were constant throughout the operation (Table 1). The superficial gas flow applied (57 mg N L− 1, Table 1). However, during the first 8 days of operation, neither nitrite nor nitrate concentrations were detected, Fig. 2. Operational parameters and reactor performance for the four (I, II, III and IV) start-up strategies. 4 X. Juan-Díaz et al. Journal of Water Process Engineering 45 (2022) 102505 35 0.4 0.6 800 I II IIIa 30 Diameter size (µm) Superficial gas flow Settling time (min) AOR (g N L-1 d ) 0.3 600 velocity (cm s-1) -1 25 0.4 20 0.2 400 15 0.2 0.1 200 10 5 0.0 0.0 0 0 10 20 30 40 0 20 40 60 80 0 20 40 60 80 100 120 Time (days) Time (days) Time (days) 35 0.6 0.6 800 IV Superficial gas flow 30 0.5 Diameter size (µm) Settling time (min) velocity (cm s-1) AOR (g N L-1 d ) -1 600 25 0.4 0.4 Settling time Superficial gas flow velocity 20 0.3 400 AOR Diameter size 15 0.2 0.2 200 10 0.1 5 0.0 0.0 0 0 10 20 30 40 50 60 Time (days) Fig. 3. Time course superficial gas-flow velocity (ug), ammonium oxidation rates (AORs), diameter size, settling time for the four (I, II, III and IV) start-up strategies. (a) The contribution of air and nitrogen flow rates of strategy-III can be found in Table 1 and Fig. SI. 2 in Supporting Information. Fig. 4. Operational parameters and reactor performance for the four (I, II, III and IV) start-up strategies. (A) Time course free ammonia concentrations at the end of the cycle (FAEFF) (i.e. effluent) and at the beginning of the cycle (FABC); (B) Time course of produced nitrate and influent nitrate concentrations and (C) Time course solids concentrations. indicating an inhibition of both AOB and NOB (Fig. 3). To reduce the FA first cycle after 13 days of operation. Despite the initial nitrate pro­ inhibitory conditions, a cycle was manually performed at day 8 by filling duction, partial nitritation was successfully achieved and maintained for the 40% of the reactor volume with tap water. This operation resulted 44 days as nitrate concentrations were maintained at 1.9 ± 1.1 mg N L− 1 into a decrease of FA concentrations from 57 to 9 mg N L− 1. After (Fig. 2). On day 56, particle size increased up to ca. 300 μm. Just before alleviating the imposed conditions, FA concentrations were maintained reaching this goal, the superficial gas flow velocity incremented while roughly constant throughout the operation (17 ± 4 mg N L− 1, Table 1). the AOR experienced a threefold increase (Fig. 3). A size of 535 ± 25 μm Consequently, nitrifying activity was detected as observed by the was reached at the end of the operation (Fig. 3). Further, 60% of sludge gradual production of TNN and nitrate (up to 12 mg N L− 1). Neverthe­ particles surpassed the 200 μm threshold μm and the SVI5/SVI30 ratio less, due to the low initial nitrifying activity, the system completed its was 1. The difference in size distribution at the initial and at the end of 5 X. Juan-Díaz et al. Journal of Water Process Engineering 45 (2022) 102505 10 I II III IV Volume fraction (%) day 0 day 0 day 0 day 0 8 day 97 day 42 day 38 day 113 6 4 2 0 0 500 1000 1500 2000 0 500 1000 1500 2000 0 500 1000 1500 2000 0 500 1000 1500 2000 Particle size (µm) Particle size (µm) Particle size (µm) Particle size (µm) Fig. 5. Size distribution of the sludge for the four (I, II, III and IV) start-up strategies at the initial (solid line) and at the end-phase (dashed line) of the operation. the operation confirmed the achievement of an autotrophic aerobic According to the obtained results, a last experiment (i.e. strategy-IV) granular sludge, more developed than in strategy-I (Fig. 5). However, was planned to work at higher AORs compared to previous experiments after attaining a granular sludge, nitrite built-up became compromised and to assess its effects on the development of an autotrophic aerobic after 60 days of operation, as observed for the unceasing nitrate pro­ granular sludge. The reactor was inoculated with CAS but also with a duction, resulting into a complete failure of the partial nitritation fraction of an enriched AOB floccular biomass. FA concentrations and process. settling time were maintained roughly constant throughout the opera­ Based on the obtained results in previous strategies, strategy-III was tion. The superficial gas flow velocity was set to maintain the same shear designed to evaluate the effect of shear stress forces on the development stress forces as in strategy-III but supplying only air. of an autotrophic aerobic granular sludge. On this sense, nitrogen gas was supplied together with air to work at higher initial shear stress 3.4. Strategy-IV: enhancing ammonium oxidation activity by working at conditions than those of previous strategies. Higher air flow rates were moderate FA concentrations and constant settling time and superficial gas required compared to previous strategies to maintain adequate bulk flow velocity oxygen concentrations (i.e. as oxygen mass transfer rate was reduced because nitrogen gas addition) (Figs. SI.2 and SI.3 in Supporting Infor­ Initial particle size was 107 ± 1 μm and rapidly increased up to 465 mation). To reduce the influence of other parameters, settling time was ± 11 μm in 27 days of operation, crossing the threshold of 200 μm in less decreased as in strategy-II, while initial FA concentrations were main­ than 20 days (Fig. 3). During this period, settling time was maintained tained to guarantee an initial ammonium oxidation activity, they were high (30 min) while the initial superficial gas flow velocity was in the subsequently increased to high inhibitory levels to ensure a proper NOB range of strategy-III (Table 1). Superficial gas flow velocity uninten­ inhibition and they were finally maintained at those levels of strategy-II tionally decreased during the operation as reactor diffusor became to strengthen NOB inhibition throughout the operation (see Fig. 4A). progressively clogged. The main difference compared to other strategies was that the achieved AOR was, at least, one order of magnitude higher 3.3. Strategy-III: progressive increase of FA concentration from moderate since the beginning of the operation (Fig. 3). Nevertheless, AOR to high inhibitory concentrations with a progressive decrease of settling decreased during reactor operation, along with a progressive biomass time and higher initial superficial gas flow velocity loss, yet it remained higher than in previous strategies (Figs. 3 and 4C, respectively). Particle size was maintained at 470 ± 61 μm (SVI5/SVI30 The main difference of strategy-III compared to previous strategies = 1), with essentially 50% of aggregates surpassing the 200 μm was the higher superficial gas flow velocity applied from the beginning threshold (Figs. 3 and 5, respectively). Partial nitritation was attained of the experiment, as settling time was decreased at the same pace than since the early stage of the operation as nitrate concentrations remained in strategy-II (Fig. 3). Initial particle size of CAS inoculum was 129 ± 3 low (2 ± 1 mg N L− 1) (Fig. 2). It should be stressed that nitrate gradually μm and progressively increased, reaching an average particle size of 257 increased to 7 mg N L− 1 (from day 27 to 42), though it was properly ± 15 μm after 50 days of operation (Fig. 3). Nevertheless, it decreased to reversed within next days (Fig. 2). 155 ± 47 μm from day 50 to 79 (Fig. 3) with a simultaneous biomass loss (Fig. 4C). FA concentrations were initially maintained at 15 ± 4 mg N 4. Discussion L− 1 (Table 1). Full nitrification was detected during the first 4 days of the operation, as observed by the increasing nitrate production. To reverse 4.1. Effect of free ammonia and settling time decrease in the development nitrate production, FA concentrations were gradually increased to avoid of an autotrophic aerobic granular sludge performing partial nitritation AOB inhibition until reaching a maximum of 113 mg N L− 1 (Table 1). This resulted into a progressive nitrate rate depletion. To avoid negative During the first 13 days of strategy-I, nitrate was produced (Fig. 2) effects within AOB activity, FA concentrations were progressively despite working at the reported inhibitory FA concentrations for NOB decreased and maintained at 21 ± 7 mg N L− 1 until the end of the inhibition within CAS (7 ± 2 mg N L− 1) [22,23]. The initial NOB operation (Table 1). After this, nitrite built-up took place, while nitrate persistence was reduced in strategy-II by applying a higher FA inhibitory concentration maintained low (3 ± 2 mg N L− 1), resulting into stable concentration (i.e. 57 mg N L− 1) (Fig. 4A). However, these FA concen­ partial nitritation (Fig. 2). From day 78-onwards, nitrogen gas supply trations severely reduced ammonium oxidation activity (i.e. null AORs) was almost zero (Fig. SI.2 in Supporting Information). This change (Fig. 3). Indeed, ammonium oxidation activity was only detected some affected the oxygen transfer in the bulk liquid, increasing the DO con­ days after alleviating the applied FA inhibitory conditions (Fig. 3). centration (Fig. SI.3 in Supporting Information) and causing a momen­ Regardless of the applied FA concentrations, partial nitritation desta­ tary increase of the AOR (Fig. 3). From then on, granule sludge size bilization occurred at the end of both strategies I and II as observed by increased up to 409 ± 76 μm presenting good settleability properties the unceasing nitrate production (see Section 4.4 for a detailed expla­ (SVI5/SVI30 = 1) (Fig. 3). The concentration of granular sludge nation). During the initial phase of strategy-III, FA concentrations in the increased in a 30% within the last 30 days of operation, confirming the airlift reactor were maintained at values ensuring that the ammonium achievement of a significant amount of granular sludge at the end of the oxidation activity was not inhibited, in spite of nitrate production (see operation (ca. 60% of particles with size above 200 μm) (Fig. 5). In Fig. 4A and Table 1). These FA concentrations were regulated through contrast to previous strategies, stable partial nitritation was successfully the TAN concentration imposed at the beginning of the cycle and the maintained until the end of the operation (Fig. 2). duration of each cycle. Then, FA was progressively increased up to high 6 X. Juan-Díaz et al. Journal of Water Process Engineering 45 (2022) 102505 inhibitory conditions (i.e. 113 mg N L− 1), which resulted into a layer and into a smaller granular sludge size, incrementing oxygen consecutively nitrite built-up (Fig. 2). However, as CAS was not adapted availability to the inner layers and, thus, allowing nitrite oxidation [27]. to tolerate high FA concentrations, these were relaxed to avoid negative effects on AOB metabolism [10,24] (Fig. 4A). Along with the suppres­ 4.3. Linking the effects of ammonium oxidation activity to the sion of NOB within the seeding sludge, settling time was simultaneously development of an autotrophic aerobic granular sludge for partial decreased in strategies I, II and III. The reduction of settling time has nitritation been described as a mechanism contributing to the achievement of heterotrophic aerobic granulation, mainly due to the selection of slow- The role of ammonium oxidation activity on the development of an growing heterotrophic bacterial species [14]. However, no granule autotrophic aerobic granular sludge from CAS was further investigated development was observed when settling time was progressively in strategy-IV. Compared to previous strategies, higher AORs were ob­ decreased from 30 to 15 min for strategies-I, -II and -III (Fig. 3). Indeed, tained since the beginning of strategy-IV, as the inoculum was enriched the reduction of settling time acting as a selection pressure was not with a fraction of nitrifying biomass (Fig. 3). The high ammonium beneficial for the obtention of an autotrophic aerobic granular sludge as oxidation activity derived into a high oxygen consumption (i.e. low bulk it resulted into a progressively biomass wash out (ca. 0.5 g VSS L− 1, DO concentrations, Fig. SI. 3 in Supporting Information). Under higher Fig. 4C). This wash out required a long time to be compensated due to AORs (compared to previous strategies), granular sludge was rapidly the low nitrifying growth rate and, thus, hindered the possibility of developed while partial nitritation was successfully achieved and attaining high ammonium oxidization activities (Fig. 3). maintained stable throughout the operation (Figs. 3 and 2, respectively). A fast decrease in settling time negatively influenced the ability of This is consistent with a study that allowed for a fast start-up of the retaining a sufficient concentration of nitrifying bacteria. Also, applying nitritation process within an aerobic granular sludge by enhancing AOB high FA shocks to a non-acclimated sludge showed to present disad­ at high DO concentrations (3–5 mg O2 L− 1), resulting into even higher vantages in terms of attaining a high ammonium oxidation activity. and more compact granules [28]. As the applied shear stress conditions Consequently, the inhibitory FA concentrations with the concomitant were maintained similar than in strategy-III and settling time remained reduction of settling time resulted into low AORs, which were not high, the development of an autotrophic aerobic granular sludge was beneficial for the development of an autotrophic aerobic granular shown to be promoted by the enhancement of ammonium oxidation sludge. activity in the early phase of the operation. A period of nitrate production (up to 7 mg N L− 1) was detected from 4.2. Is the enhancement of superficial gas flow velocity promoting day 27 to 42 (Fig. 2) related to an increase in bulk DO concentrations (up autotrophic aerobic granulation? to 8 mg O2 L− 1), which could not be monitored due to a failure of the bulk DO monitoring system that lasted for 15 days (Fig. SI.3 in Sup­ Operating at high superficial gas flow velocities allows for higher porting Information). The higher oxygen availability derived into a shear stress conditions, which can promote EPS secretion, contributing transient nitrite oxidation, as nitrate production successfully decreased to the attachment and self-immobilization process of bacteria [25,26]. (≤ 2 mg N L− 1, Fig. 2) once bulk DO concentration was lowered to 1.5 This mechanism is well-established for heterotrophic aerobic granular mg O2 L− 1, and further maintained at ca. 6 mg O2 L− 1 (day 40-onwards, sludge, but it is unclear for autotrophic aerobic biomass. Also, working see Fig. SI.3 in Supporting Information). The bulk DO concentration at high air flow rates grants for greater bulk oxygen concentrations, affects the oxygen penetration depth since when bulk DO concentration allowing AOB to increase their activity. The obtained results showed increases, oxygen can reach deeper layers of the granule. This fast shift that granule development was achieved in first strategies (I and II) just from nitrification to partial nitritation when DO is manipulated has been after increasing the air flow rate, in turn resulting in an increase of the identified as a strong indication of the stratification of nitrifier guilds in AOR (Fig. 3). To evaluate which mechanisms derived from incrementing granular sludge [4–6]. the air flow rate triggered the development of an autotrophic aerobic The first three strategies tested in this study (strategies I, II and III) granular sludge (i.e. shear stress conditions vs. ammonium oxidation demonstrated that the reduction of settling time was not promoting the activity), a third experiment (strategy-III) was performed. In strategy-III, development of an autotrophic granular sludge. Also, the effects of shear a higher superficial gas flow velocity was applied since the early stage of stress forces derived from incrementing the superficial gas flow velocity the operation by supplying both air and nitrogen gas. During this (by adding nitrogen gas) showed to be not fully determinant if a high strategy, a faster increase of particle size was detected, from 128 ± 3 μm ammonium oxidation activity was not simultaneously promoted. The to 257 ± 15 μm in 50 days, in accordance with the higher superficial gas obtained results in strategy-IV showed that the high initial ammonium flow velocity (Fig. 3). However, particle size subsequently decreased oxidation activity, measured as AOR, accounted for the development of from day 50 to 79 (Fig. 3) due to detachment, as a significant increase in an autotrophic aerobic granular sludge successfully performing partial the effluent VSS concentration was measured (see Fig. 4.C.III). The high nitritation. The representative fraction corresponding to nitrifiers in CAS detachment rates declined after day 80, consistent with the interruption is rather low (i.e. 3%) [12,29]. Thus, promoting high AORs in the early of nitrogen gas supply. When only air was supplied (day 78, Fig. SI.2 in stage of the operation can be challenging. Under these conditions, the Supporting Information), an increase in granule size was measured positive effects derived from increasing air flow rate were demonstrated (from 155 ± 47 μm on day 79 to 339 ± 46 μm on day 85, see Fig. 3). This to be effective in promoting autotrophic aerobic granular sludge for­ change also resulted into a marked increase in the ammonium oxidation mation as ammonium oxidation activity was enhanced. However, the activity (i.e. AOR) (Fig. 3) and higher bulk DO concentrations (Fig. SI.3 increase of the air flow rate in a full-scale reactor presents several in Supporting Information). Likewise, granular sludge development of challenges: (i) the increase in energy demand and (ii) the potential previous strategies (I and II) occurred just after incrementing the su­ limitation of aeration devices. These challenges would require assess­ perficial gas flow velocity and, consequently, observing an increase in ment in future research plus development studies. the AORs (Figs. 3 and SI.3, respectively). Experiments indicate that the development of an autotrophic aerobic granular sludge was promoted 4.4. NOB influent inoculation deteriorates partial nitritation stability by the higher ammonium oxidation activity derived from boosting air flow rate (i.e. higher DO), whereas the effects of shear stress forces alone Partial nitritation was not successfully maintained in strategies I and were not enough, because they could also trigger significant detachment II (Fig. 2), even after attaining a significant fraction of granular sludge at events. Indeed, high shear stress conditions caused AOB detachment and the end of reactor operation (ca. 50–60%). In contrast, nitrite oxidation derived into a failure of the nitritation process in a stratified nitrifying was almost completely suppressed in strategy-III (Fig. 2). The difference granule structure; the high detachment rates resulted into a thinner AOB among these strategies was not the FA concentrations maintained 7 X. Juan-Díaz et al. Journal of Water Process Engineering 45 (2022) 102505 throughout the operation (Fig. 4A), neither the [DO]/[TAN] ratio Industria y Competitividad (MINECO), Agencia Estatal de Investigación (0.014 ± 0.006). Indeed, this last parameter was lower than the reported (AEI) and Fondo Europeo de Desarrollo Regional (FEDER, EU), to maintain a successfully stable nitritation in a granular sludge reactor CTQ2017-82404-R. [7]. The reason that could explain the recovery of nitrite oxidation ac­ tivity in strategies I and II is the continuous NOB reinoculation with the Appendix A. Supplementary data inflow. The existence of this reinoculation can be inferred from the presence or absence of nitrate influent concentrations (Fig. 4B). The Supplementary data to this article can be found online at https://doi. used reject water was stored in a tank several weeks and part of the TAN org/10.1016/j.jwpe.2021.102505. concentration was progressively oxidized to nitrite, and, further to ni­ trate (Fig. 4B). Hence, the continuous TAN oxidation clearly indicated References the existence of AOB and NOB populations within the storage waste­ water and, consequently, both were continuously seeded to the reactor. [1] B. Kartal, J.G. Kuenen, M.C.M. van Loosdrecht, Sewage treatment with anammox, In contrast, the presence of nitrate in the influent of strategy-III was Science 328 (2010) 702–703, https://doi.org/10.1126/science.1185941. [2] E. Isanta, C. Reino, J. Carrera, J. Pérez, Stable partial nitritation for low-strength almost not detected (Fig. 4B). The NOB population developed in the wastewater at low temperature in an aerobic granular reactor, Water Res. 80 storing tank probably possessed a high tolerance to FA, which resulted in (2015) 149–158, https://doi.org/10.1016/j.watres.2015.04.028. failure of the partial nitritation process [30]. Interestingly, in strategy-IV [3] C. Reino, M.E. Suárez-Ojeda, J. Pérez, J. Carrera, Kinetic and microbiological characterization of aerobic granules performing partial nitritation of a low-strength there was a significant amount of nitrate in the influent from day 40-on­ wastewater at 10 ◦ C, Water Res. 101 (2016) 147–156, https://doi.org/10.1016/j. wards (Fig. 4B), yet this did not result into nitrate production within the watres.2016.05.059. reactor. Compared to previous strategies, the achieved AORs were, at [4] V. Poot, M. Hoekstra, M.A.A. Geleijnse, M.C.M. van Loosdrecht, J. Pérez, Effects of the residual ammonium concentration on NOB repression during partial nitritation least, one order of magnitude higher (Fig. 3). When NOB inoculation with granular sludge, Water Res. 106 (2016) 518–530, https://doi.org/10.1016/j. took place in strategy-IV, a significant fraction of granular sludge was watres.2016.10.028. already obtained (ca. 50% of particles being greater than 200 μm), [5] A. Soler-Jofra, R. Wang, R. Kleerebezem, M.C.M. van Loosdrecht, J. Pérez, Stratification of nitrifier guilds in granular sludge in relation to nitritation, Water which decreased the impact of the NOB seeding, because the number of Res. 148 (2019) 479–491, https://doi.org/10.1016/j.watres.2018.10.064. cycles per day was higher at higher AORs (i.e. shorter HRT and higher [6] C. Picioreanu, J. Pérez, M.C.M. van Loosdrecht, Impact of cell cluster size on washout of the reinoculated NOB). The wastewater storage during the apparent half-saturation coefficients for oxygen in nitrifying sludge and biofilms, Water Res. 106 (2016) 371–382, https://doi.org/10.1016/j.watres.2016.10.017. development of a granular sludge performing partial nitritation should [7] A. Bartrolí, J. Pérez, J. Carrera, Applying ratio control in a continuous granular be done in conditions to avoid NOB proliferation. For eventual full-scale reactor to achieve full nitritation under stable operating conditions, Environ. Sci. applications of the partial nitritation process, wastewater storage is not Technol. 44 (2010) 8930–8935, https://doi.org/10.1021/es1019405. required, therefore this issue would not require particular attention. [8] A. Bartrolí, J. Carrera, J. Pérez, Bioaugmentation as a tool for improving the start- up and stability of a pilot-scale partial nitrification biofilm airlift reactor, Bioresour. Technol. 102 (2011) 4370–4375, https://doi.org/10.1016/j. 5. Conclusions biortech.2010.12.084. [9] I. Jubany, J. Lafuente, J.A. Baeza, J. Carrera, Total and stable washout of nitrite oxidizing bacteria from a nitrifying continuous activated sludge system using In this study, a specific operational start-up strategy based on automatic control based on oxygen uptake rate measurements, Water Res. 43 enhancing ammonium oxidation activity was shown to promote the (2009) 2761–2772, https://doi.org/10.1016/j.watres.2009.03.022. development of an autotrophic partial nitritation granular sludge by [10] V.M. Vadivelu, J. Keller, Z. Yuan, Free ammonia and free nitrous acid inhibition on the anabolic and catabolic processes of nitrosomonas and nitrobacter, Water Sci. using conventional activated sludge as inoculum in a short period of Technol. 56 (7) (2007) 89–97, https://doi.org/10.2166/wst.2007.612. time (ca. 30 days). When the inoculum presented a low activity, the [11] X. Li, R. Du, Y. Peng, Q. Zhang, J. Wang, Characteristics of sludge granulation and ammonium oxidation activity of the seeding sludge was enhanced by EPS production in development of stable partial nitrification, Bioresour. Technol. 303 (2020), 122937, https://doi.org/10.1016/j.biortech.2020.122937. increasing the air flow rate, which allowed for a higher bulk oxygen [12] J.A. Torà, E. Moliné, J. Carrera, J. Pérez, Efficient and automated start-up of a pilot concentration. Against expectation –and based on the reported condi­ reactor for nitritation of reject water: from batch granulation to high rate tions for granulation of heterotrophic aerobic sludge– settling time continuous operation, Chem. Eng. J. 226 (2013) 319–325, https://doi.org/ 10.1016/j.cej.2013.04.013. decrease was not linked to autotrophic aerobic granulation. In contrast, [13] M. Pronk, M.K. de Kreuk, B. de Bruin, P. Kamminga, R. Kleerebezem, M.C.M. van high settling times allowed to retain a suitable biomass concentration, Loosdrecht, Full scale performance of the aerobic granular sludge process for beneficial for attaining higher ammonium oxidation activities. Finally, sewage treatment, Water Res. 84 (2015) 207–217, https://doi.org/10.1016/j. applying strong free ammonia inhibitory conditions in the early start-up watres.2015.07.011. [14] Y. Liu, J.H. Tay, State of the art of biogranulation technology for wastewater phase (i.e. seeding sludge) resulted into low ammonium oxidation rates, treatment, Biotechnol. Adv. 22 (2004) 533–563, https://doi.org/10.1016/j. which also hindered the development of an autotrophic aerobic granular biotechadv.2004.05.001. sludge. [15] Y. Liu, Q.S. Liu, Causes and control of filamentous growth in aerobic granular sludge sequencing batch reactors, Biotechnol. Adv. 24 (2006) 115–127, https:// When starting-up the cultivation of aerobic granular sludge for doi.org/10.1016/j.biotechadv.2005.08.001. partial nitritation (for two-stage PN/AMX in mainstream) the findings [16] Y.Q. Liu, W.W. Wu, J.H. Tay, J.L. Wang, Formation and long-term stability of highlighted that it is possible to inoculate activated sludge and use reject nitrifying granules in a sequencing batch reactor, Bioresour. Technol. 99 (2008) 3919–3922, https://doi.org/10.1016/j.biortech.2007.07.041. water in an SBR configuration by applying the following set of experi­ [17] S. Tsuneda, T. Nagano, T. Hoshino, Y. Ejiri, N. Noda, A. Hirata, Characterization of mental conditions: high aeration flow rates, long settling times and nitrifying granules produced in an aerobic upflow fluidized bed reactor, Water Res. applying a loading rate that avoids FA inhibitory conditions. Once 37 (2003) 4965–4973, https://doi.org/10.1016/j.watres.2003.08.017. [18] Y. Liu, Z.W. Wang, L. Qin, Y.Q. Liu, J.H. Tay, Selection pressure-driven aerobic enough biomass concentration is achieved with this experimental start- granulation in a sequencing batch reactor, Appl. Microbiol. Biotechnol. 67 (2005) up strategy, the reactor operation could be switched to continuous mode 26–32, https://doi.org/10.1007/s00253-004-1820-2. of operation for partial nitritation of pretreated municipal wastewater. [19] APHA, Standard methods for the examination of water and wastewater, 21st ed., American Public Health Association, Washington, DC, USA, 2005. [20] M.K. de Kreuk, N. Kishida, M.C.M. van Loosdrecht, Aerobic granular sludge - state Declaration of competing interest of the art, Water Sci. Technol. 55 (8–9) (2007) 75–81, https://doi.org/10.2166/ wst.2007.244. [21] A.C. Anthonisen, R.C. Loehr, T.B.S. Prakasam, E.G. Srinath, Inhibition of The authors of this study declare there is not any conflict of interest nitrification by ammonia and nitrous acid, Water Pollut. Control Fed. 48 (1976) regarding this publication. 835–852. [22] U. Abeling, C.F. Seyfried, Anaerobic-aerobic treatment of high-strength ammonium Acknowledgments wastewater - nitrogen removal via nitrite, Water Sci. Technol. 26 (5–6) (1992) 1007–1015, https://doi.org/10.2166/wst.1992.0542. [23] J. Chung, H. Shim, S.J. Park, S.J. Kim, W. Bae, Optimization of free ammonia This research was funded by the Spanish Ministerio de Economía, concentration for nitrite accumulation in shortcut biological nitrogen removal 8 X. Juan-Díaz et al. Journal of Water Process Engineering 45 (2022) 102505 process, Bioprocess Biosyst. Eng. 28 (2006) 275–282, https://doi.org/10.1007/ [28] J. Wang, J. Song, F. Yin, Y. Shen, D. Yang, W. Liu, Insight into how high dissolved s00449-005-0035-y. oxygen favors the startup of nitritation with aerobic granules, Chemosphere 270 [24] Y. Liu, H.H. Ngo, W. Guo, L. Peng, D. Wang, B. Ni, The roles of free ammonia (FA) (2021), 128643, https://doi.org/10.1016/j.chemosphere.2020.128643. in biological wastewater treatment processes: a review, Environ. Int. 123 (2019) [29] G. Harms, A.C. Layton, H.M. Dionisi, I.R. Gregory, V.M. Garrett, S.A. Hawkins, K. 10–19, https://doi.org/10.1016/j.envint.2018.11.039. G. Robinson, G.S. Sayler, Real-time PCR quantification of nitrifying bacteria in a [25] C. Di Iaconi, R. Ramadori, A. Lopez, R. Passino, Hydraulic shear stress calculation municipal wastewater treatment plant, Environ. Sci. Technol. 37 (2003) 343–351, in a sequencing batch biofilm reactor with granular biomass, Environ. Sci. Technol. https://doi.org/10.1021/es0257164. 39 (2005) 889–894, https://doi.org/10.1021/es0400483. [30] H. Duan, L. Ye, Q. Wang, M. Zheng, X. Lu, Z. Wang, Z. Yuan, Nitrite oxidizing [26] J. Tay, Q.S. Liu, Y. Liu, Microscopic observation of aerobic granulation in bacteria (NOB) contained in influent deteriorate mainstream NOB suppression by sequential aerobic sludge blanket reactor, J. Appl. Microbiol. 91 (2001) 168–175, sidestream inactivation, Water Res. 162 (2019) 331–338, https://doi.org/ https://doi.org/10.1046/j.1365-2672.2001.01374.x. 10.1016/j.watres.2019.07.002. [27] W. Liu, F. Yin, D. Yang, Granules abrasion cause deterioration of nitritation in a [31] D. Wang, Q. Wang, A. Laloo, Y. Xu, P.L. Bond, Z. Yuan, Achieving stable nitritation mainstream granular sludge reactor with high loading rate, Chemosphere 243 for mainstream deammonification by combining free nitrous acid-based sludge (2020), 125433, https://doi.org/10.1016/j.chemosphere.2019.125433. treatment and oxygen limitation, Sci. Reports 6 (2016) 25547, https://doi.org/ 10.1038/srep25547. 9