Rapid Start-Up of the Aerobic Granular Reactor under Low Temperature and the Nutriment Removal Performance of Granules with Different Particle Sizes

: The start-up of the aerobic granular sludge (AGS) process under low temperature is challenging. In this study, the sequencing batch reactor (SBR) was fed with synthetic wastewater and the temperature was controlled at 15 °C. The main components in the synthetic wastewater were sodium acetate and ammonium chloride. The inﬂuent chemical oxygen demand (COD) and NH 4 + -N concentrations were 300 and 60 mg/L, respectively. The AGS was successfully cultivated in 60 days by gradually shortening the settling time. During the stable operation stage (61–100 d), the average efﬂuent COD, NH 4+ -N, NO 2 − -N, and NO 3 − -N concentrations were 47.2, 1.0, 47.2, and 5.1 mg/L, respectively. Meanwhile, the nitrite accumulation rate (NAR) reached 90.6%. Batch test showed that the smaller AGS had higher NH 4+ -N removal rate while the larger AGS performed higher NAR. The NH 4+ -N removal rates of R1 (1.0–2.0 mm), R2 (2.0–3.0 mm), and R3 (>3 mm) granules were 0.85, 0.61, and 0.45 g N/(kg VSS · h), respectively. Meanwhile, the NAR of R1, R2, and R3 were 36.2%, 77.2%, and 94.9%, respectively. The obtained results could provide important guidance for the cultivation of AGS in low-temperature wastewater treatment.


Introduction
With the increasingly serious water environment pollution problem, many countries have put forward higher requirements on the nitrogen and phosphorus discharge standards [1]. Therefore, it is necessary to upgrade the existing sewage treatment process. Moreover, carbon neutrality is a key indicator towards sustainable wastewater treatment [2]. It is important to reduce energy consumption in wastewater treatment plants and develop energy-saving technologies. The traditional biological nitrogen removal was achieved through nitrification and denitrification process. Recently, the partial nitrification/anammox (PN/A) process has been developed to remove nitrogen, which requires less aeration and carbon source [3]. In the PN/A process, the ammonium was first oxidized to nitrite by ammonia-oxidizing bacteria (AOB). Then the anammox bacteria converted the ammonium and nitrite to nitrogen gas. The inhibition of nitrite-oxidizing bacteria (NOB) is one of the biggest challenges in the application of PN/A process, which needs to be further investigated.
The aerobic granular sludge (AGS) process is considered to be one of the most promising wastewater treatment technologies [4]. In the past few decades, researchers have applied the AGS process to the treatment of mainstream wastewater and industrial wastewater [5]. Aerobic granule is composed of microorganisms and extracellular polymeric substances. The AGS process has the advantages of high biomass concentration, good sedimentation performance, and energy saving. The AGS is composed of multiple layers, which makes it possible to enrich aerobic bacteria, facultative bacteria, and anaerobic microorganisms, simultaneously. Previous studies reported that the AGS reactor achieved excellent partial nitrification performance under the conditions of high temperature (30)(31)(32)(33)(34)(35) • C) and high ammonium concentration (>500 mg/L) [6]. As the growth rate of AOB is greater than NOB under high operating temperature (30 • C), NOB could be washed out by controlling short sludge age [7]. For high ammonium wastewater treatment, the NOB could be severely inhibited by high free ammonia (FA) concentration (>1 mg/L), while the AOB activity was slightly suppressed [8]. In some areas, especially northern China, the mainstream waste water has a low temperature and low ammonium concentration. Temperature has a significant impact on the sludge activity and microbial population structure. Under low temperature, the growth and biological activity of microorganisms is seriously inhibited [1]. Meanwhile, the excessive reproduction of filamentous bacteria and poor sludge sedimentation performance have also been observed [9,10]. According to prior research, the AGS has been cultivated at 7 • C by inoculating cold-adapt sludge [11]. When dealing with low ammonium wastewater, low dissolved oxygen (DO) and intermittent aeration strategies were proposed to suppress the NOB activity [12,13]. Vazquez-Padin et al., (2010) reported that stable nitrite accumulation could be achieved in the AGS reactor at room temperature (24 • C) [14]. As described by Zhang et al., (2018), the NOB activity of the AGS reactor was successfully inhibited by operating anaerocic/aerobic mode [15]. The nitrite accumulation ratio reached 97.3% when the influent NH 4 + -N concentration was 39.3-78.7 mg/L. So far, the cultivation of partial nitrification AGS under low temperature and low ammonium concentration was rarely reported. Moreover, the high mass transfer resistance of granules led to low DO concentration in AGS, which was considered as the major reason for NOB inhibition [16]. However, due to the technical limitation, the earlier research mainly analyzed the DO distribution in AGS through model fitting [17,18]. Later, the DO distribution in granules under room temperature operation was explored with microelectrode device [16,19]. The DO distribution in AGS was closely related to the temperature and granule size [20]. Therefore, it is necessary to investigate the impact of granule size on the partial nitrification performance. So far, the integrated relationship between the granule size, DO distribution, and partial nitrification performance has rarely been analyzed.
The aim of this study was to cultivate AGS for mainstream wastewater treatment under low temperature. The nutrient removal performance and sludge characteristics were investigated. Batch experiments were conducted to evaluate the substrate oxidization rate and nitrite accumulation rate (NAR) of mature AGS with different particle sizes. The DO distribution in AGS was further analyzed with microelectrode device. The results could provide important guidance for the rapid start-up of AGS reactors and the performance of high NAR in mainstream wastewater treatment.

Reactor Setup and Operation
As shown in Figure 1, the sequencing batch reactor (SBR) was used in the experiment. The reactor was of cylindrical structure, with an inner diameter of 8 cm, a height of 170 cm, an effective volume of 8 L, and a height diameter ratio of 21.2. The exchange ratio of the reactor was 50%. Aeration is carried out by the blast aeration pump, and the aeration rate was controlled to 2-3 L/min by the rotameter. In the aeration period, the DO concentration in the reactor reached 6 mg/L. The reactor temperature was controlled to 15 • C by water bath through the constant temperature cooling tank. The SBR operation procedures were: feeding (5 min), aeration (690 min), settling (adjusted as needed), and decant (10 min). The experiment was divided into two stages. In stage 1 (1-60 d), the aerobic granules were formed by reducing the settling time from 5 to 1 min. In stage 2 (61-100 d), the aerobic granular reactor achieved stable performance and the settling time was 1 min. The specific operating conditions are shown in Table 1.

Batch Experiment and Microelectrode Analysis
At the end of stage 2, mature aerobic granules were taken out from the reactor and screened to three categories: R1 (1.0-2.0 mm), R2 (2.0-3.0 mm), and R3 (larger than 3.0 mm). The batch experiment was carried out in the 500 mL serum bottle, and the mass transfer performance of the substrate was enhanced by magnetic stirring [21]. The synthetic wastewater used in the batch test was the same as the influent of the SBR reactor. The obtained granules of R1, R2, and R3 were added to three different serum bottles. The initial MLSS concentration was 4500 mg/L. The DO concentration was maintained at 5-6 mg/L and the temperature was controlled at 15 • C. Samples were taken at regular intervals to analyze COD and nitrogen concentrations.
The DO concentrations in R1, R2, and R3 granules were measured by microelectrodes. The microelectrode measurement device was the same as described by Meyer et al., (2003) [22] ( Figure S1). The AGS were placed on the mesh in the flow tank. The synthetic wastewater used in the flow tank was the same as the influent of the SBR reactor. The DO concentration was controlled at 6-7 mg/Land the temperature was controlled at 15 • C.

Analytical Method
The NH 4 + -N, NO 2 − -N, NO 3 − -N, COD, and MLSS concentrations and SVI were determined with the standard methods [23]. The DO and temperature were measured by using the WTW 3420 device (Munich, Germany). SVI 30 (mL/g TSS) was calculated after the activated sludge was settled for 30 min in a graduated cylinder. The total concentration of NH 4 + -N, NO 2 − -N, and NO 3 − -N concentrations was defined as the total nitrogen (TN) concentration. The DO concentration distribution in granules was measured by using commercial microsensors operated with a motorized micromanipulator (Unisense A.S., Aarhus, Denmark). The NAR was calculated by using the following equation:

Microbial Population Analysis
At the end of stage 2, the AGS were taken out from the reactor for microbial community analysis. High-throughput experiment was conducted by the Shanghai Shenggong Biological Engineering Co., Ltd. (Shanghai, China). The results of high-throughput sequencing were compared to the ribosomal database project (RDP) database and the taxonomic levels were assigned. Figure 2 shows the nutrient removal performance of the AGS reactor. In stage 1 (1~60 d), the DO concentration in the reactor reached 6 mg/L, and the settling time was gradually reduced from 5 to 1 min. On the first day, the reactor had good ammonium removal performance. The effluent NH 4 + -N, NO 2 − -N, and NO 3 − -N concentrations were 0, 1.4, and 54.5 mg/L, respectively. However, from the second day to the 15th day, the effluent NH 4 + -N concentration gradually increased, indicating the decrease in NH 4 + -N removal performance. From the 16th day to the 30th day, theNH 4 + -N removal performance of the reactor was gradually recovered. On the 30th day, the effluent NH   Table 2 shows the changes in sludge characteristics in the reactor. From the first day to the 15th day, serious loss of sludge was observed due to the poor sedimentation performance of floc sludge. The MLSS concentration decreased from 4630to 1824 mg/L, and the SVI 30 increased from 68 to 104 mL/g TSS. From the 16th day to the 60th day, the sludge concentration gradually increased and the sludge sedimentation performance was improved. On the 60th day, the MLSS concentration increased to 4356 mg/L, and the SVI 30 decreased to 36 mL/g TSS. In stage 2 (61-100 d), the sludge concentration and SVI 30 remained stable. On the 100th day, the sludge concentration and SVI 30 were 4437 mg/L and 33 mL/g TSS, respectively. Figure 3 shows the morphology of AGS during the granular formation process. On the 30th, 60th, and 100th day, the average particle sizes of sludge reached 1.3, 3.2, and 3.5 mm, respectively.  In low-strength wastewater treatment, previous studies showed that the low DO concentration in granules played an important role in the nitrite accumulation [17]. There are two main reasons for the low DO concentration in AGS. First, the DO concentration could be largely consumed by AOB and heterotrophic bacteria on the surface of granules [24,25]. In addition, the penetration depth of oxygen in granules was hindered due to the dense surface structure and large diffusion resistance of oxygen [16,26]. It has been proposed that the oxygen saturation constants of AOB and NOB were 0.2-0.4 mg/L and 1.2-1.5 mg/L, respectively [17,27]. As AOB had stronger oxygen affinity than NOB, the nitrite accumulation could occur at low DO concentration. In addition, the growth rate of AOB was accelerated at low DO concentration, which compensated for the decline in metabolic activity [28]. However, the growth rate and metabolic activity of NOB decreased significantly under low DO condition, favoring the accumulation of nitrite. In this study, the AOB activity would be greatly suppressed by low temperature. In order to improve the AOB activity, high DO concentration (6 mg/L) was maintained. In addition, the high air flow rate led to a higher water shear force, which was important to the fast cultivation of AGS. This was because the higher water shear force could stimulate the production of extracellular polymeric substances.

The AGS Reactor Performance and Sludge Characteristics
Under normal working conditions, the SVI value of activated sludge in wastewater treatment plant was between 70 and 100 mL/g TSS. The high SVI indicates poor sludge sedimentation, which will lead to sludge bulking. In this study, the SVI of mature AGS was much lower than the inoculated sludge. The start-up of the AGS reactor and sludge characteristics have been explored in early studies by using synthetic wastewater with sodium acetate and ammonia nitrogen. At room temperature, Corsino et al., (2016) have cultivated AGS in 120 days when the influent COD and NH 4 + -N concentrations were maintained at 900 and 90 mg/L, respectively [29]. By controlling the settling time at 5 min, the average particle size and SVI 30 of the granules reached 1.3 mm and 30-40 mL/g TSS, respectively. Meanwhile, Deng et al., (2016) have developed the AGS reactor by gradually increasing the influent COD and NH 4 + -N concentrations [30]. The AGS was successfully cultivated in 30 days. Meanwhile, the particle size and SVI 30 of the granules reached 0.8-1.1 mm and 30 mL/g TSS, respectively. As for low temperature, the proliferation of filamentous bacteria was observed in the AGS reactor [11]. By controlling the settling time to 5-3 min, the filamentous bacteria were successfully washed out. The particle size of the AGS increased to 4.5 mm after 60 days. Moreover, the AGS process was applied to treat industrial wastewater with high phenol and ammonium concentrations [31]. The influent phenol, NH 4 + -N, and SCN − concentrations were 400, 100, and 100 mg/L, respectively. After 35 days, the particle size and SVI 30 of AGS reached 2938 µm and 35.25 mL/g TSS, respectively. As discussed above, the particle size and the sedimentation performance of AGS in this study were comparable to previous studies. Figure 4 shows the nitrification performance of AGS with different particle sizes in batch experiments. The initial NH 4 + -N concentration was 60 mg/L. From 0 to 12 h, the NH 4 + -N concentration gradually decreased while the NO 2 − -N concentration gradually increased. After 12 h, the NH 4 + -N concentrations of R1, R2, and R3 reduced to 0, 15.1, and 28.0 mg/L, respectively. Meanwhile, the NO 2 − -N concentrations increased to 18.5, 24.8, and 18.6 mg/L, respectively. The NH 4 + -N removal efficiencies were 100.0%, 73.4%, and 52.7%, respectively. The NH 4 + -N removal rates were 0.85, 0.61, and 0.45 g N/(kg VSS·h), respectively. The results of batch experiment implied that the particle size had a significant impact on the ammonium oxidation rate. As the particle size increased, the NH 4 + -N removal performance of granular sludge gradually decreased. In addition, during the batch experiment, the NO 3 − -N concentrations of R1 and R2 began to increase rapidly after 2 and 6 h, respectively. However, the NO 3 − -N concentration of R3 was kept at a low concentration before the NH 4 + -N was entirely oxidized. At the end of the batch experiment, the NO 3 − -N concentrations of R1, R2, and R3 reached 32.7, 7.3, and 1.0 mg/L, respectively. Meanwhile, the NARs were 36.2%, 77.2%, and 94.9%, respectively. The larger granules performed higher nitritation performance than the smaller granules.  Figure 5 shows the DO distribution in AGS with different particle sizes. The DO concentration in the water phase was 6.6 mg/L. From the surface of the AGS to the inner part of the AGS, the DO concentration gradually decreased. The particle size of R1 granule was 1066 µm, and the lowest DO concentration was 1.32 mg/L. The particle size of R2 granule was 2292 µm. The DO concentrations reduced to 0.5and 0 mg/L at the distances of 582 and 700 µm from the surface of the granules, respectively. The particle size of R3 granule was 3048 µm. The DO concentrations reduced to 0.5 and 0 mg/L at the distances of 606 and 700 µm from the surface of the granules, respectively. In R1, R2, and R3 granules, the low DO concentration area (DO < 0.5 mg/L) accounted for 0%, 11.9%, and 21.8% of the total volume of granular sludge, respectively. Kishida et al., (2006) found that the oxygen penetration depth in granules was 100 µm when the temperature and DO concentration in the bulk liquid were 20 • C and 5.5 mg/L, respectively [32]. Meanwhile, Rathnayake et al., (2013) reported that the oxygen penetration depth in granules was 300 µm when the temperature and DO concentration in the bulk liquid were 30 • C and 2 mg/L, respectively [16]. Morales et al., (2015) proposed that the DO concentration in the bulk liquid had a significant impact on the DO concentration distribution in granules [19]. In their study, the SNAD granular sludge reactor operated at 25 • C. The microelectrode analysis results showed that the oxygen penetration depths in granules were 150 and 250 µm when the DO concentrations in the bulk liquid were controlled at 4 and 8 mg/L, respectively. In this study, the AGS reactor was operated at 15 • C, and the oxygen penetration depth in granules was much larger than previous studies. De kreuk et al., (2005) also proposed that the low temperature led to a larger oxygen penetration depth [20]. As the low DO concentration area (DO < 0.5 mg/L) in R3 granules wan only 21.8%, the low DO concentration might not be the main factor for the NOB activity inhibition. Apart from the low DO condition, some studies have shown that the NOB activity could be seriously inhibited by NO [33,34]. Starkenburg et al., (2008) reported that the Nitrobacter activity could be inhibited when the NO concentration was maintained at 7-448 µg N/L [33]. It was found that the nitrite affinity of NOB was closely related to the inhibitory effect of NO [34]. By maintaining the NO concentration at 2 µg N/L, the NOB activity was inhibited by 30-50% and 60-80% when the nitrite saturation constants of NOB were 0.36 and 0.06 mg/L, respectively. In sewage treatment, there are three main ways to produce NO: (1) the hydroxylamine pathway; (2) the nitrifier denitrification; and (3) the heterotrophic denitrification. In the nitrification process, ammonium was first oxidized to hydroxylamine, and the hydroxylamine was further oxidized to NO [35]. According to previous studies, the copper-containing nitrite reductase (NirK) found in AOB was important in converting NO 2 − -N to NO, and the NirK gene was actively expressed under low DO and high nitrite concentrations [36,37]. During the denitrification process, the NO 2 − -N was reduced to NO by NirK, and the NO was further reduced to N 2 O by the NO reductase (NOR). The NirK gene of denitrifiers was over expressed under high nitrite concentrations and the NOR could be severely inhibited by oxygen, which might lead to NO accumulation [35]. Based on the discussion above, the low DO and high nitrite concentrations were favorable for NO accumulation. In this study, the low DO concentration area (DO < 0.5 mg/L) in R3 granules was much larger than the R1 granules. The NOB activity in larger granules might be inhibited by the production of high NO concentration. Therefore, it is necessary to further investigate the NO distribution in granular sludge. Figure 6 shows the microbial community analysis results of aerobic granular sludge. The dominant phyla in granular sludge were Proteobacteria (54.61%), Bacteroidetes (17.43%), Saccharibacteria (5.28%), Planctomycetes (2.87%), Verrucomicrobia (2.36%), Chloraflexi (1.04%), Firmicutes (0.53%), and Acidobacteria (0.26%). The most abundant phyla were Proteobacteria and Bacteroidetes, which was similar to previous study [1]. The dominant genera were Paracoccus (8.51%), Bacillus (6.11%), Solibacillus (5.98%), Thaurea (3.66%), Azoarcus (2.83%), Gemmobacter (1.57%), Acinetobactor (1.22%), Pesudoxanthomonas (1.01%), Pesudomonas (0.79%), Meganema (0.76%), and Nitrosomonas (0.57%). Table 3 shows the abundance of AOB and NOB in granular sludge. The abundance of AOB in granular sludge was much higher than that of NOB. The AOB genera were Nitrosomonas (0.57%) and Nitrosospira (0.24%), while the NOB genera were Nitrobacter (0.07%) and Nitrospira (0.04%), respectively.

Conclusions
The AGS reactor was developed under low temperature in the SBR. The SVI 30 and particle size of granular sludge reached 33 mL/g TSS and 3.5 mm, respectively. The particle size of AGS had a significant impact on the NH 4 + -N removal rate and nitritation performance. When the DO concentration in the bulk liquid was 6-7 mg/L, the microelectrode analysis results showed that the penetration depth of oxygen was 700 µm. The larger granules performed higher NAR than the smaller granules. The low DO concentration in larger AGS might not be the direct reason for NOB inhibition. The low DO concentration area in larger AGS and high nitrite concentration in bulk liquid were favorable for NO accumulation, which might play an important role in inhibiting NOB activity. The abundance of AOB and NOB in mature AGS were 0.81% and 0.11%, respectively.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author. The data are not publicly available due to privacy.