Pilot Study of Oxic–Anoxic Process under Low Dissolved Oxygen for Nitrogen Removal from Low COD/N Tropical Wastewater

: Conventionally, nitriﬁcation in biological nitrogen removal (BNR) requires high dissolved oxygen (DO) concentrations (>2 mg L − 1 ), making the process energy intensive. Recent studies have shown that efﬁcient ammonium removal and energy reduction can be realized by operating the nitriﬁcation at low DO concentrations (<1 mg L − 1 ). In this study, the low-DO oxic anoxic (low-DO OA) process was operated in a pilot-scale sequencing batch reactor (SBR) over 218 days to evaluate the feasibility of nitrogen removal from low chemical oxygen demand-to-nitrogen ratio (COD/N) tropical municipal wastewater. The results revealed that the low-DO OA process attained high removal efﬁciency for ammonium (97%) and total nitrogen (TN) (80%) under an average DO concentration of 0.6 mg L − 1 . The effective TN removal efﬁciency is attributed to the occurrence of simultaneous nitriﬁcation–denitriﬁcation (SND) under low DO conditions. Further batch tests revealed that slowly biodegradable COD (sbCOD) in tropical wastewater can support denitriﬁcation in the post-anoxic phase, resulting in a high TN removal rate. Compared with high DO concentrations (2 mg L − 1 ), low DO conditions achieved 10% higher TN removal efﬁciency, with similar ammonium and COD removal efﬁciency. This study is crucial in promoting the energy efﬁciency and sustainability of wastewater treatment plants treating low COD/N wastewater.


Introduction
To prevent eutrophication in water environments, biological nitrogen removal (BNR) processes have been applied extensively in many wastewater treatment plants (WWTPs) worldwide [1]. Conventional BNR is often energy-and chemical-intensive due to the high aeration demand in nitrification and external carbon dosage in denitrification. Dissolved oxygen (DO) in a nitrifying system is usually maintained at a high concentration (>2 mg L −1 ) to ensure complete nitrification and prevent sludge bulking [2]. Such a high DO concentration from aeration consequently makes up nearly half of the entire WWTP's energy cost [3,4]. Another challenge is that municipal wastewater often contains a low chemical oxygen demandto-nitrogen ratio (COD/N) (<4), thus limiting the heterotrophic denitrification [5][6][7][8]. To enhance total nitrogen (TN) removal, external carbon sources such as acetate, ethanol, and methanol are typically added to improve denitrification performance [1,2]. Incorporating external carbon sources in the nitrogen removal processes of wastewater treatment further increases the operating costs and poses a risk of residual chemicals in the effluent if supplied in excess [9][10][11]. Therefore, nitrogen removal from wastewater using the conventional BNR process may not be economical and sustainable.
Several technologies have been developed to address the high energy input for aeration and external carbon supplements in BNR such as anaerobic ammonia oxidation (anammox), partial nitritation combined with anammox (PN-anammox) and low DO nitrification. Although anammox and PN-anammox have attracted considerable attention in recent years, these processes are more complex and are less suitable for WWTPs in developing countries due to the expensive plant retrofitting and process control [12][13][14]. In recent years, studies have demonstrated that low DO nitrification (<1 mg L −1 ) is capable of reducing by 23 to 25% energy consumption compared to a conventional high DO system [15,16] without compromising the nitrification treatment performance [17][18][19]. Wang et al. [17] studied the performance of a lab-scale anaerobic-anoxic-oxic system treating municipal wastewater using high DO (2 mg L −1 ) and low DO (0.5 mg L −1 ) and discovered that reducing DO levels did not negatively impact the ammonium and COD removal efficiency. Likewise, high nitrification efficiency with an average of 95% and 93% were attained under low DO (0.9 mg L −1 ) and high DO (1.7 mg L −1 ), respectively, in an anoxic-oxic (AO) reactor treating tropical wastewater [18]. Although earlier studies have shown that operating at low DO leads to greenhouse gases release, specifically in nitrous oxide (N 2 O) gas [20,21], Liu et al. [19] proved that prolonged (141 days) low-DO (0.4 mg L −1 ) operation in a labscale membrane reactor resulted in a reduced N 2 O emission (0.11%) compared to high DO (2 mg L −1 ) operation with 0.24% N 2 O emission that was fed with synthetic municipal wastewater. In another study, Li et al. [22] deduced that the enriched complete ammonia oxidizers (comammox) could be the reason for the N 2 O gas reduction (20%) under longterm low DO in a weakly acidic (pH 6.3 to 6.8) bioreactor. Hence, low-DO nitrification could be a potential strategy to enhance the energy sustainability of the BNR process in developing countries.
Most municipal WWTPs adopt the pre-anoxic BNR process (AO configuration), where the anoxic zone precedes the aerobic zone. A high mixed liquor recycle (MLR) flow rate is necessary to supply nitrate from the aerobic zone to the anoxic zone [23,24]. While enhanced TN removal can be achieved through this setup, some drawbacks exist, such as increased MLR pumping energy costs, the potential for DO carryover from the aerobic zone, and a dilution of the influent carbon. The most significant limitation is that the removal of oxidized nitrogen (NO x -N; nitrate and nitrite) is ultimately constrained by the MLR rate, making complete NO x -N removal unachievable (3-5 mg L −1 TN remains in effluent) [24,25]. Conversely, the post-anoxic BNR process (OA configuration) eliminates the need for MLR pumping and improves the TN removal efficiency as the anoxic zone is located downstream of the aerobic zone [26]. Many studies attained remarkable TN removal efficiency ranging from 90-98% in treating low COD/N ratio (1. 3-4.4) wastewater in the OA configuration [5,18,27,28]. A recent study carried out by How et al. [6] successfully achieved complete NH 4 -N removal and a significant reduction in effluent NO 3 -N concentrations (<0.3 mg L −1 ) in the low-DO oxic-anoxic (low-DO OA) when treating low COD/N tropical wastewater. As a result, the low-DO OA process could offer WWTPs a cost-effective alternative for WWTPs to reducing both their capital and operating expenses.
Several studies examining wastewater characteristics in tropical regions revealed that soluble COD/N ratios (3)(4)(5) are insufficient for denitrification, as they fall below the desired range of 6-11 [5,27,29,30]. Lack of carbon sources in the wastewater can impede heterotrophic denitrification during the post-anoxic phase in the OA system, necessitating the addition of external carbon sources that increase operational expenses and potentially result in excess COD in the effluent [31][32][33]. Despite concerns about the availability of carbon sources from low-strength wastewater, several studies have reported that external carbon dosage may not be necessary for the post-anoxic BNR process [6,29,29,34]. Excellent TN removal efficiency (90-98%) was still achieved when treating low COD/N wastewater (1. 3-4.4) in post-anoxic denitrification without carbon dosage. Apart from the endogenous decay of biomass, the particulate settleable solids (PSS) hydrolysis also provides sufficient readily biodegradable COD (rbCOD) to support the post-anoxic denitrification [8]. rbCOD is an easily available COD component that can be readily taken up by heterotrophic mi- croorganisms for biosynthesis and energy production. The availability of rbCOD directly influences the performance of the nitrogen removal process. On the other hand, slowly biodegradable COD (sbCOD) is typically composed of PSS that require extracellular enzymes to break them into simpler forms before they can be utilized [35]. Denitrifying bacteria cannot directly use sbCOD as an organic source for denitrification. Instead, PSS needs to be first hydrolyzed into rbCOD, which can be easily consumed by denitrifying bacteria [36,37]. Studies have shown that the high PSS hydrolysis rate in warm wastewater (30 • C) provides sufficient rbCOD for denitrification in low COD/N tropical wastewater, making external carbon dosage unnecessary in the OA BNR process [18,38]. Hence, this can lead to savings in chemical costs for wastewater treatment plants in tropical regions.
Studies on the low-DO OA process for nitrogen removal in municipal wastewater have thus far been limited at the laboratory scale and its long-term stability has yet to be validated at a larger scale. To address the gaps, a pilot-scale sequencing batch reactor (SBR) operating in low-DO OA configuration was set up in this study to treat low COD/N tropical wastewater over a period of 218 days. The pilot study aimed to (1) evaluate the long-term nitrogen removal performance of the low-DO OA process, (2) compare the nitrogen removal performance of low-DO OA with the high-DO OA operation and (3) determine the utilization of sbCOD from tropical wastewater for denitrification. The findings of this study provide crucial insights into the potential of the low-DO OA process for a cost-effective and efficient BNR process in WWTPs.

Inoculum and Wastewater Characteristics
The initial mixed liquor suspended solids (MLSS) concentration in the reactor was set at approximately 2800 mg L −1 using seed sludge collected from the return activated sludge point of a municipal WWTP (hereafter referred to as WWTP X). The WWTP X, located in Kuala Lumpur, Malaysia, is responsible for treating municipal wastewater and serves 430,000 population equivalent. The raw wastewater after preliminary treatment (bar screening and grit removal) from WWTP X was supplied and stored in a 400-L buffer tank under mixing conditions to keep the particulate matter in suspension ( Figure 1). The wastewater from the buffer tank was pumped into the reactor during the feeding phase using a Husky 515 diaphragm pump (Graco, Minneapolis, MN, USA). The raw wastewater was sampled weekly, and the characteristics are presented in Table 1. The temperature was in the range of 27 to 33 • C, similar to previous studies dealing with tropical wastewater [39,40]. The soluble COD/N ratio of the raw wastewater was in the low range, between 1.9 to 3.5. No nitrite nitrogen (NO 2 -N) and nitrate nitrogen (NO 3 -N) were detected in the raw wastewater.

Pilot-Scale Sequencing Batch Reactor Operation
Pilot-scale sequencing batch reactor (Figure 1), hereafter called the pilot SBR, was a cylindrical-shaped vessel made of fiber-reinforced plastic. The pilot SBR had a working volume of 150 L, a height of 94 cm, and an inner diameter of 56 cm. To ensure homogeneous mixing in the reactor, an overhead mechanical mixer was installed and operated at 300 rpm during the oxic and anoxic phases. The DO and pH probes were installed on the C-D103 process Clark DO sensor (LEADTEC, Petaling Jaya, Malaysia) and pH sensor (LEADTEC, Malaysia) coupled with a PD6000 dual-channel analyzer (LEADTEC, Petaling Jaya, Malaysia). The DO concentration was controlled using a solenoid valve connected to the programmable logic controller (OMRON, Tokyo, Japan), which automatically switches on/off the compressor to maintain pre-selected minimum and maximum DO set points. The air supply from the compressor was sparged through ceramic disc diffusers placed at the base of the reactor to maintain a low DO concentration in Phase I and a high DO concentration in Phase II during the oxic phase (Table 2). The pilot SBR was operated from March to November 2022 under an oxic-anoxic configuration in a 6 h cycle, with each cycle consisting of the following five phases: filling (10 min), oxic (120 min), anoxic (180 min), settling (30 min) and withdrawing (10 min). The hydraulic retention time (HRT), sludge retention time (SRT), and volume exchange ratio of the pilot SBR operation are stated in Table 2. Mixed liquor in the pilot SBR was sampled at regular intervals to monitor the evolution of nitrogen compounds (NH 4 -N, NO 2 -N, NO 3 -N, and TN). The raw wastewater and mixed liquor samplings were performed twice a week on selected days, such as Tuesday and Friday, ensuring sufficient time gap between each sampling day and avoiding any significant variation in wastewater characteristics that could arise from changes in human activity over the weekend. All the samples were stored at 4 • C until further analysis.

Denitrification Batch Experiment
The denitrification batch tests were performed following the procedure outlined by van Loosdrecht et al. [41]. Three sets of denitrification batch tests (Set A, B, and C) simulating anoxic conditions (DO < 0.1 mg L −1 ) were carried out in 2-L beakers using sludge samples taken from the pilot SBR. Set A, B, and C were conducted using filtered (0.45 µm) raw wastewater, and particulate settleable solids (PSS) in nutrient solution and blank nutrient solution, respectively. The PSS portion that was obtained from the resulting settled fraction of raw wastewater had undergone three successive washing cycles. The nutrient solution concentrations were as follows (per liter): 1460 mg KH 2 PO 4 , 20 mg N-Allylthiourea, 1070 mg NH 4 Cl, 660 mg MgSO 4 ·7H 2 O and 3.33 mL trace element solution. The nutrient and trace element solutions were formulated by Ong et al. [42]. All three sets were seeded with pre-washed sludge samples to achieve the mixed liquor volatile suspended solids (MLVSS) of 2500 mg L −1 . Prior to performing the experiment, nitrogen gas sparging was activated in the sludge mixture for 10 min to ensure oxygen-free conditions. Then, all three sets were sealed and aerated at the lowest speed to ensure homogeneous mixing. The initial NO 3 -N concentration of the reaction mixture was maintained between 50 to 60 mg L −1 . The DO and pH of each set were measured using an InPro6850i DO probe (Mettler-Toledo, Columbus, OH, USA) and 405-DPAS-SC K851200 pH probe (Mettler-Toledo, Columbus, OH, USA). Mixed liquor samples were collected at 15 to 30 min intervals for NO 2 -N, NO 3 -N, and sCOD analyses to monitor the denitrification performance.

Analytical Methods
All the analyses were performed according to the standard methods [43]. The anion (NO 2 -N, NO 3 -N, PO 4 -P) and cation (NH 4 -N) concentrations were analyzed using 861 Advanced Compact Ion Chromatography (Metrohm, Herisau, Switzerland) from filtered (0.2 µm) mixed liquor samples. TN was determined using a TNM-1 total nitrogen measuring unit (Shimadzu, Kyoto, Japan). The unfiltered and filtered (0.45 µm) wastewater samples were measured for TCOD and sCOD, respectively, using the dichromate method with a DRB 200 COD digester (Hach, Loveland, CO, USA). TP was analyzed using Test 'N Tube High Range Phosphorus Reagent Set (Hach, Loveland, CO, USA)' following the molybdovanadate with acid persulfate digestion method 10127. Mixed liquor samples were analyzed for MLSS and MLVSS. The DO and pH in the pilot SBR were recorded every 2 min. The pH and temperature of raw wastewater samples were measured using HQ440d digital multimeter (Hach, Loveland, CO, USA).

Performance of Low-DO OA in Pilot SBR
The overall profiles of soluble (sCOD), NH 4 -N, NO 3 -N, and TN concentrations over 260 days of pilot SBR operation are depicted in (Figure 2a-c). The representative pH and DO profiles under low DO and high DO conditions are illustrated in Figure A1 of Appendix A. In Phase I, the pilot SBR was operated with a low DO level (0.6 mg L −1 ) during the oxic phase. After 37 days of SBR operation, the removal of sCOD, NH 4 -N, and TN was stable with average removal efficiencies of 90%, 97%, and 80%, respectively (Figure 2a-c). The low-DO OA pilot study complied with Malaysia's discharge standard (A) in the Environmental Quality (Sewage) Regulations 2009 for NH 4 -N (5 mg L −1 ) and NO 3 -N (10 mg L −1 ) with low effluent NH 4 -N (0.7 mg L −1 ) and NO 3 -N (3 mg L −1 ) concentrations [44]. The NO 2 -N concentration in the effluent was also consistently low (below 0.5 mg L −1 ) in low DO conditions. The findings were comparable with How et al. [6] whereby the lab-scale low-DO OA attained complete NH 4 -N removal and low effluent NO 3 -N (0.3 mg L −1 ) at a low DO (0.2-0.6 mg L −1 ) with SRT of 20 days.
In addition to the nitrogen removal performance, variations in biomass concentration and sludge volume index (SVI) were also monitored (Figure 2d,e). Both MLSS and MLVSS concentrations were maintained at 2950 mg L −1 and 2320 mg L −1 during the stable period from day 37 to 98 in Phase I, respectively (Figure 2d). Early research reported that low DO concentrations in the activated sludge process may lead to sludge bulking, preventing them from settling and reducing treatment efficiency [45]. This is primarily caused by excessive filamentous bacteria growth and porous flocs formation. SVI is an important parameter to assess the settleability of activated sludge in the WWTPs. SVI values below 120 mL g −1 are desired for good settling characteristics. Conversely, SVI above 150 mL g −1 reflects sludge bulking and poor settling [46]. In the stable period in Phase I, SVI values remained at an average of 109 mL g −1 despite a low DO concentration of 0.6 mg L −1 (Figure 2e). Although sludge bulking commonly happens in long-term low-DO operations, no sign of sludge bulking was observed in this study. The low SVI obtained indicated that activated sludge has good settling characteristics. contribute to better TN removal efficiency by allowing slowly biodegradable COD (sbCOD) to be utilized for denitrification. How et al. [18] found that the PSS hydrolysis rate in warm wastewater temperature (30 °C) is 2.5 times higher than the reported values at 20 °C. The higher PSS hydrolysis rate in warm wastewater could potentially supply sbCOD for denitrification for low soluble COD/N wastewater. To understand if raw wastewater was providing the sbCOD for post-anoxic denitrification in this study, a denitrification batch test was performed (as described in Section 3.3). The energy reduction to operate at a low DO concentration of 0.6 mg L −1 was estimated by considering the oxygen transfer rate to maintain biological activities and bulk DO concentration in tropical wastewater, with an average temperature of 30 °C [41]. The calculation assumes surface aerator was used as the aeration system. Based on the calculation, the estimated energy saving accounts for 18% of the energy needed to operate at a high DO (2 mg L −1 ) concentration. How et al. [16] study also found a comparable range (around 23%) of energy reduction when operating the nitrification process at low DO (0.5 mg L −1 ) compared to high DO concentration (2 mg L −1 ). In a study by Keene et al. [15], a nearly 25% decrease in aeration energy savings could be achieved when operating at DO of 0.33 mg L −1 relative to conventional BNR operation at a higher DO level (0.9-4.3 mg L −1 ). Therefore, operating the BNR process at a low DO indeed enhances the energy efficiency of WWTPs.  Figure 3a,b, in Phase I. For instance, only 5 mg L −1 of NO3-N was produced (in hour 2) and 10 mg L −1 of NH4-N was oxidized (in hour 0) under low DO conditions (Figure 3b), suggesting that the nitrification and denitrification occur concurrently during the oxic phase. How et al. [18] also observed a similar trend where the amount of NH4-N oxidized is higher than the NO3-N production by 4 mg L −1 during the oxic phase in a low-DO (0.9 mg L −1 ) anaerobic-OA reactor. The low DO concentrations The pilot SBR experienced an unscheduled downtime (UDT) on day 99 due to the malfunction of a compressor, during which the settled sludge was stored at 4 • C. Following the UDT phase, the average MLSS and MLVSS fluctuated drastically during the first 22 days but remained stable from day 155 onwards at an average MLSS of 3039 mg L −1 and MLVSS of 2389 mg L −1 (Figure 2d). The SVI value exhibited an abrupt increase from 588 to 167 mg L −1 during the first 22 days after the UDT phase (Figure 2e). The sudden increase in SVI is likely attributed to the technical interruption, which could have impacted the sludge settleability. As a result, the sludge may require some time to adapt and restore its settleability after the operation was resumed. From day 156 onwards, the SVI value remained stable at 145 mg L −1 . On the other hand, the sCOD, NH 4 -N, and TN removal also remained effective after 31 days of UDT with removal efficiencies of 89%, 98%, and 80%, respectively. Despite the UDT, the low-DO OA sludge showed good resilience, recovering its nitrification and denitrification activities as the operation resumed.

Occurrence of SND under Low DO
The DO concentration was increased from 0.6 to 2 mg L −1 on day 221 as a means of comparing the effectiveness of low-DO OA with the high-DO counterpart. During the high DO operation in Phase II, complete nitrification was attained as expected. In comparison to the low-DO operation, it can be inferred that a high-DO operation is unnecessary, given that 97% nitrification efficiency was achieved at low-DO. The TN removal efficiency was lower at 70% in Phase II than in Phase I. Similarly, Wang et al. [17] found that elevating the DO level from 0.6 to 2 mg L −1 in an anaerobic-AO system decreased TN removal efficiency from 79% to 69%. Fan et al. [47] also reported that increasing the DO level from 0.5 to 1 mg L −1 in the oxic reactor of an AO system resulted in a slight decrease in TN removal efficiency from 88% to 85.5%. The low DO conditions positively affected TN removal performance, which is most likely due to enhanced simultaneous nitrificationdenitrification (SND). The limited oxygen diffusion within the sludge floc may have led to enhanced SND in a low DO environment during the oxic phase. Section 3.2 will examine the potential SND occurrence in low and high DO conditions. Another possible reason could be that the high DO operation causes the complete utilization of soluble COD in the oxic phase, resulting in inadequate carbon sources for the subsequent anoxic phase. Higher particulate settleable solids (PSS) hydrolysis rate in tropical wastewater may contribute to better TN removal efficiency by allowing slowly biodegradable COD (sbCOD) to be utilized for denitrification. How et al. [18] found that the PSS hydrolysis rate in warm wastewater temperature (30 • C) is 2.5 times higher than the reported values at 20 • C. The higher PSS hydrolysis rate in warm wastewater could potentially supply sbCOD for denitrification for low soluble COD/N wastewater. To understand if raw wastewater was providing the sbCOD for post-anoxic denitrification in this study, a denitrification batch test was performed (as described in Section 3.3).
The energy reduction to operate at a low DO concentration of 0.6 mg L −1 was estimated by considering the oxygen transfer rate to maintain biological activities and bulk DO concentration in tropical wastewater, with an average temperature of 30 • C [41]. The calculation assumes surface aerator was used as the aeration system. Based on the calculation, the estimated energy saving accounts for 18% of the energy needed to operate at a high DO (2 mg L −1 ) concentration. How et al. [16] study also found a comparable range (around 23%) of energy reduction when operating the nitrification process at low DO (0.5 mg L −1 ) compared to high DO concentration (2 mg L −1 ). In a study by Keene et al. [15], a nearly 25% decrease in aeration energy savings could be achieved when operating at DO of 0.33 mg L −1 relative to conventional BNR operation at a higher DO level (0.9-4.3 mg L −1 ). Therefore, operating the BNR process at a low DO indeed enhances the energy efficiency of WWTPs.  Figure 3a,b, in Phase I. For instance, only 5 mg L −1 of NO 3 -N was produced (in hour 2) and 10 mg L −1 of NH 4 -N was oxidized (in hour 0) under low DO conditions (Figure 3b), suggesting that the nitrification and denitrification occur concurrently during the oxic phase. How et al. [18] also observed a similar trend where the amount of NH 4 -N oxidized is higher than the NO 3 -N production by 4 mg L −1 during the oxic phase in a low-DO (0.9 mg L −1 ) anaerobic-OA reactor. The low DO concentrations and organic matter breakdown during the oxic phase may have further facilitated SND [15,18,48,49]. Several studies have shown that a low DO concentration of 0.5-0.9 mg L −1 can favor SND [15,17,18,48]. The occurrence of SND is influenced by specific hydraulic and oxygen conditions that affect the formation of large sludge flocs with a modified internal structure. The size and structure of flocs result in DO concentration gradients, creating anoxic micro-zones in the center for heterotrophic denitrification, while the outer layers of flocs promote an efficient nitrification [50]. Operating the oxic phase under an oxygen-limited environment could sustain the co-existence of nitrifiers and denitrifiers, thereby improving nitrogen removal performance through SND occurrence [51]. As a result, less energy, chemicals, and supplementary carbon are required.

Occurrence of SND under Low DO
Meanwhile, in Phase II, the produced NO 3 -N (12.8 mg L −1 ) is almost equivalent to the oxidized NH 4 -N (12 mg L −1 ) during the oxic phase (Figure 3c). This indicates that SND was not enhanced under high DO conditions. Ma et al. [52] confirmed that SND efficiency increased (39 to 81%) with decreasing DO concentration (4.6 to 0.35 mg L −1 ) in a sequencing batch biofilm reactor treating coal gasification wastewater. Likewise, Yan et al. [53] reported that in an SBR system, higher SND efficiency (74%) is observed at a low DO concentration (0.7 mg L −1 ) compared to a high DO concentration (1.2 mg L −1 ) with an SND efficiency of 66%. A high DO of 2 mg L −1 limits the SND occurrence by inhibiting denitrification, consequently lowering TN removal efficiency. oxidized NH4-N (12 mg L −1 ) during the oxic phase (Figure 3c). This indicates that SND was not enhanced under high DO conditions. Ma et al. [52] confirmed that SND efficiency increased (39 to 81%) with decreasing DO concentration (4.6 to 0.35 mg L −1 ) in a sequencing batch biofilm reactor treating coal gasification wastewater. Likewise, Yan et al. [53] reported that in an SBR system, higher SND efficiency (74%) is observed at a low DO concentration (0.7 mg L −1 ) compared to a high DO concentration (1.2 mg L −1 ) with an SND efficiency of 66%. A high DO of 2 mg L −1 limits the SND occurrence by inhibiting denitrification, consequently lowering TN removal efficiency.

Utilization of sbCOD for Denitrification
In tropical regions, the amount of soluble biodegradable organic compounds, referred to as readily biodegradable COD (rbCOD), present in wastewater is generally lim-

Utilization of sbCOD for Denitrification
In tropical regions, the amount of soluble biodegradable organic compounds, referred to as readily biodegradable COD (rbCOD), present in wastewater is generally limited and may not be adequate for complete nitrification [29,54,55]. As stated in the literature, the municipal wastewater in the warm region (24-34 • C) has a lower sCOD concentration (66 to 170 mg L −1 ) compared to the cold region (15-21 • C), where sCOD concentration is (182-232 mg L −1 ) [29,54,55]. Sophonsiri and Morgenroth [56] and How et al. [18] demonstrated that 45 to 50% of the total COD (TCOD) was present in PSS as slowly biodegradable COD (sbCOD). The warm temperature of tropical wastewater can potentially increase the PSS hydrolysis rate [18]. The higher PSS hydrolysis rate may facilitate the sbCOD utilization for denitrification, leading to enhanced nitrogen removal performance. Hence, a denitrification batch test was performed to validate the utilization of sbCOD for denitrification using raw wastewater with activated sludge taken from the low-DO OA pilot study.
The  Figure 4 a-c, respectively. In Set A, the NO3-N concentrations were reduced from 52 mg L −1 to 43 mg L −1 within the first 8 h and remained stable around 42 mg L −1 until the end of the experiment. During the first 8 h of Set A experiment, the rbCOD in the filtered wastewater was consumed by the denitrifying bacteria for denitrification. However, the soluble COD was used up in the subsequent hours, resulting in a minimal degree of denitrification in the subsequent hours.
In Set B, where PSS was used as the biodegradable organic source, 19 mg L −1 of NO3-N concentration was denitrified during the first 8 h of the experiment. Most likely, the PSS hydrolysis occurred actively during the first 8 h of the experiment. The NO3-N concentration continued to reduce by 4 mg L −1 in the following hours. The reduction in NO3-N was attributed to the rbCOD that was made available through the hydrolysis of PSS. In Set C, no NO3-N reduction was observed since no carbon sources were added. Therefore, it can be deduced that PSS in low COD/N tropical wastewater is feasible to be utilized for the biological nitrogen removal process. These findings validated that the low-DO OA process can achieve efficient and steady nitrogen removal from wastewater without requiring the addition of an external carbon source. In Set B, where PSS was used as the biodegradable organic source, 19 mg L −1 of NO 3 -N concentration was denitrified during the first 8 h of the experiment. Most likely, the PSS hydrolysis occurred actively during the first 8 h of the experiment. The NO 3 -N concentration continued to reduce by 4 mg L −1 in the following hours. The reduction in NO 3 -N was attributed to the rbCOD that was made available through the hydrolysis of PSS. In Set C, no NO 3 -N reduction was observed since no carbon sources were added. Therefore, it can be deduced that PSS in low COD/N tropical wastewater is feasible to be utilized for the biological nitrogen removal process. These findings validated that the low-DO OA process can achieve efficient and steady nitrogen removal from wastewater without requiring the addition of an external carbon source.

Conclusions
The low-DO OA pilot-scale SBR has been shown to be feasible and highly efficient for nitrogen removal in low COD/N tropical wastewater. The process was established within 3 to 4 weeks and performed stably over 218 days under 18 h HRT and 20 days SRT. The low-DO OA process achieved a high NH 4 -N removal efficiency (97%), comparable to that of a high-DO OA operation (99%). The experimental results also showed that a low DO concentration of 0.6 mg L −1 was sufficient for nitrification as well as promoting SND, leading to a 10% increase in TN removal compared to the high-DO OA (2 mg L −1 ) operation without the addition of an external carbon source. The BNR process operating at a low-DO concentration (0.6 mg L −1 ) was estimated to reduce 18% of the aeration energy requirement compared to the conventional high-DO BNR process. Although the reactor operation experienced 31 days of operational breakdown, the sludge was robust enough to recover BNR performance quickly. Nitrogen removal can be facilitated by the high rate of PSS hydrolysis in tropical wastewater that provides sufficient biodegradable organic substances for denitrification in the anoxic phase. Overall, the low-DO OA process is a simple and low-cost retrofit option for the conventional BNR system, offering an attractive solution for improving process sustainability by reducing aeration energy and chemical costs for BNR in WWTPs.

Acknowledgments:
The authors would like to thank Tan Keng Pek and Tan Wee Wee for consistently providing technical support for the pilot SBR. We also greatly appreciate the assistance from Loi Jia Xing, Ng Wai Lun, and Sea Yi Fen with laboratory analysis and reactor start-up.

Conflicts of Interest:
The authors declare no conflict of interest.