The mineral media composition used in this study is as given in Table 1 and is intended to represent a typical COD, NH₄⁺-N, and PO₄³⁻-P composition of domestic wastewaters. The seed inoculum was procured from an aeration tank treating domestic wastewater. It consisted of activated sludge flocs with an MLSS (Mixed Liquor Suspended Solids) of 0.26 g/L and SVI₃₀ (Sludge Volume Index) of 380 mL/g.

SBR studies were carried out for 328 days using two reactors with 1 L capacity each. An imhoff cone (BOROSIL) was used as the experimental reactor, while a measuring cylinder (BOROSIL) was used as a control reactor. During start-up, both the reactors were added with 1 L of mineral media inoculated with seed sludge. The initial MLSS was 0.26 g/L. An air pump was used to provide aeration and the diffuser stone was placed at the bottom portion of the reactors.

The SBRs were operated at room temperature (30 ± 2 ºC) in four phases by varying the COD and cycle time. In the start-up phase and phase III of reactor operation, each reactor was fed once a day (24 h intervals), and in phase I and phase II reactor was fed twice a day (unequal cycle times of 9 h and 15 h). The SBRs were operated as fill and draw method comprising four stages during every cycle which consist of feeding mineral media (15 min), reaction (8 h/14 h/23 h; varied in each phase), settle (30 min), and decant (15 min). The operating parameters during various phases of reactor operation are given in Table. 2. The samples after each cycle were quantified for dissolved oxygen (DO), pH, NH₄⁺-N, NO₂⁻-N, NO₃⁻-N, PO₄³⁻, and COD every day, while MLSS, EPS (carbohydrates and proteins content), and SVI were analyzed once every week.

Kinetic studies of COD, NH₄⁺-N, and PO₄³⁻-P removals were carried out in steady state phase III of reactor operation. The reactor was operated similarly, and samples were collected at every 2 h intervals from 0 h to 8 h and one sample after 24 h. All the samples were quantified for NH₄⁺-N, NO₂⁻-N, NO₃⁻-N, COD, PO₄³⁻-P, and pH. The order and rate coefficients were determined graphically using integrated rate laws. The differential rate law for zero-order reaction is:

The integrated rate law is:

Where,

The plot of [S] versus t is a straight line with slope of –k and intercept of [S]₀.

The differential rate law for first-order reaction is:

The integrated rate law is:

The plot of ln[S] versus t is a straight line with slope of –k and intercept of ln[S]₀.

The differential rate law for second-order reaction is:

The integrated rate law is:

The plot of 1/[S] versus t is a straight line with slope of k and intercept of 1/[S]₀.

All the physico-chemical analyses were conducted as per the protocols of standard methods for the examination of water and wastewater [29]. The instruments used to measure DO and pH were DO meter (YSI 55 model, USA) and pH meter (Mettler Toledo-Five Easy Plus) respectively. The nitrogenous species (NH₄⁺-N, NO₂⁻-N, and NO₃⁻-N) were analyzed using Ion Chromatography (IC) (883 Basic IC Plus, Metrohm) with incorporated MagicNet program to process the data. NH₄⁺-N analysis was performed using cationic column (Metrosep C4-150) and eluent comprising 1.7 mM of HNO₃, 0.7 mM diplocolinic acid, and 100 ml/L of acetonitrile as organic carbon suppressor. NO₂⁻-N, NO₃⁻-N, and PO₄³⁻-P analyses were performed with an anionic column (Metrosep A Supp5-250) and eluent comprising of Na₂CO₃ (0.3487 g/L) and NaHCO₃ (0.084 g/L), and 100 ml/L of acetonitrile as organic carbon suppressor. Closed reflux dichromate digestion method was used to analyze COD using WTW-CR-3200 digester, (Germany). Microbial growth was measured by OD₅₅₀ using a UV-Vis spectrophotometer (Evolution 300, Thermo Scientific). Later, the MLSS was quantified with a standard calibration curve that was established by relating MLSS with corresponding OD₅₅₀.

EPS extraction was done by ethanol extraction technique. To begin with, the sludge samples were centrifuged at 20,000 rpm for 20 min at 4 ºC. To the obtained supernatant three measurements of ice-cold ethanol were supplemented and incubation was done at 4 ºC for 12 h. Later the EPS as precipitates were separated by centrifugation of the incubated samples at 4000 rpm for 15 min. The proteins and carbohydrates contents in EPS were quantified by the Lowry’s technique (BSA as standard) [30] and the phenol sulfuric acid technique (with glucose as standard) [31], respectively.

The performance profile of conical SBR is shown in Fig. 1. The percentage nutrient removals at various phases of reactor operation are shown in Table 3 and the values of SVI₃₀, MLSS, and MLVSS are given in Table S1. In the start-up phase effective nitrification was observed with an average NH₄⁺-N removal of 78%, while the denitrification (42% of total nitrogen (TN) removal) was not effective due to higher DO levels in the reactor. Ineffective PO₄³⁻-P (43%) removal in the start-up phase could also be due to high DO levels. Also, the MLSS was very less and sludge lysis of un-acclimated biomass could be attributed to inefficient removals. This is evident from the decreased MLSS from 0.26 g/L at start-up to 0.14 g/L by day 14 of the reactor operation. DO in the reactor during start-up phase was in the range of 3.21 mg/L to 4.26 mg/L. In such DO levels, aerobic digestion of MLSS is possible on long starvation and the obtained results support aerobic digestion of biomass.

In phase I, the reactor was fed twice a day with cycle times of 9 h and 15 h respectively. Though the cycle time was decreased the NH₄⁺-N and the TN removals remained almost stable which could be attributed to the acclimated biomass and gradual increase in MLSS (Fig. 2) due to the availability of abundant nutrients. A slight increase in COD and PO₄³⁻-P removals was also attained. The efficiency of the SBR in terms of nutrient removal was higher at a cycle time of 15 h compared to 9 h (Table 3).

In phase II, two feed cycles per day was maintained as such, while the average influent COD was increased from 470 to 640 mg/L. The COD/N ratio increased from 9 in previous phases to 12 in this phase. The aeration was decreased to enhance denitrification and PO₄³⁻-P removal and the corresponding DO levels were in the range of 0.98 mg/L and 1.87 mg/L. Slight increase in removals of NH₄⁺-N and PO₄³⁻-P was observed compared to previous phases. The increased organic loading rate did not affect the nitrification efficiency of the system. An increase in denitrification was also observed giving a TN removal of 64% which could be due to availability of COD and utilization of DO due to higher substrate concentration. TN removal efficiency increased at high COD/N ratio as a drastic decrease is seen in the effluent NO₂⁻-N and NO₃⁻-N concentrations. Similar results were obtained by Zaman et al. [32], where at a COD/N ratio of 7, 69% of TN was removed, while at COD/N ratio of 10 it increased to 86%. The COD/N ratio and DO levels show a direct effect on TN and PO₄³⁻-P removal. At a DO of 1 mg/L and COD/N ratio of 4, only 65% of TN and 37% of PO₄³⁻-P removals were achieved by Wang et al. [33], whereas the removal efficiencies increased at high COD/N ratio. The COD removal percentage was slightly less in this phase which could be attributed to increased influent concentrations. The increase in PO₄³⁻-P removal efficiency could be due to the decrease in DO levels and also less effluent NO₂⁻-N and NO₃⁻-N concentrations. Denitrifiers dominate the PAOs and dPAOs in utilizing COD in the presence of high NO₂⁻-N and NO₃⁻-N levels. Also, high DO levels are not favorable for the enrichment of dPAOs as the genes responsible for nitrite reductase cannot be expressed at high DO. In such studies with high DO, phosphorus removal by dPAOs was not emphasized [34]. Several studies related to aerobic granules and simultaneous nutrient removal were carried out at DO levels of 2 mg/L to 5 mg/L and COD/N ratio of 7 to 11 [32]. By the end of this phase, few small granules were formed in the SBR which would lead to development of denitrifiers and PAOs in the granule interior thereby leading to effective TN and PO₄³⁻-P removal. However, complete granulation and mature granules were not obtained even in phase II of operation which could be due to overgrowth of filamentous biomass on the outer layers of flocs at high influent COD concentrations and low DO.

In phase III the reactor was fed only once a day with a cycle time of 24 h. In this phase complete granulation was seen after day 236 with stable granules in the size of 0.9 ± 0.3 mm. Less DO levels would decrease the diffusion depth of oxygen in the granules leading to small aerobic zones and large anoxic zones which impact nitrification and denitrification [35]. Phase III was operated for a very long time and no decrease in the removal efficiencies was noticed. After stable granule formation, the MLSS in the SBR stabilized. Effective average removals of COD (90%), NH₄⁺-N (91%), TN (87%), and PO₄³⁻-P (83%) were achieved in this phase. Generally, removal of PO₄³⁻-P is efficient at low sludge age. However, in this reactor, though the sludge age was high, no effect on PO₄³⁻-P removal was noticed which could be due to the formation of precipitates of phosphorous in the inside of granules. The MLVSS was less during the initial few days of phase III (0.23 g/L). However, by the end of phase III the MLVSS increased to only 0.52 g/L though the SVI decreased which confirms that a part of PO₄³⁻-P is removed by precipitation. The presence of Ca²⁺ and Mg²⁺ in the mineral media helps in the precipitation and this is enhanced by the presence of PAOs. Formation of precipitates is also beneficial to the system as the granules get densified and settle easily eliminating the need for clarifiers [36].

The reactor was inoculated with activated sludge flocs with an SVI₃₀ of 380 mL/g and an MLSS of 0.26 g/L. As the MLSS was very less, a settling time of 30 min was maintained to avoid the washout of biomass. However, before inoculating, the seed sludge was washed thrice with mineral media with a settling time of 10 min to allow the washout of low settling biomass. The physical observation of the biomass by the end of start-up phase did not show the formation of granules and bigger flocs were still dominant in the reactor. A slight increase in the SVI₃₀ to 440 mL/g was observed by the end of start-up phase which could be due to the sludge lysis during acclimation period. During start-up and phase I of reactor operation the DO levels in the reactor were ranging between 3.21 mg/L and 4.26 mg/L. By the end of phase I of operation, the size of activated sludge flocs decreased and smaller flocs with good settling abilities were formed. SVI₃₀ also decreased greatly from the inoculation to day 53. In phase II, the aeration was decreased and the DO levels in the system were in the range of 0.98 mg/L to 1.87 mg/L. As the influent substrate concentration was increased, during the initial few days of phase II slight filamentous growth was observed on the exterior of the flocs. Slight increase in the SVI₃₀ was also noticed compared to phase I of operation. At incomplete organic substrate uptake, aerobic degradation of organic matter occurs at the surface of flocs or granules leading to the development of fluffy granules and similar results were reported [37].

In phase III of operation, the cycle time was again increased to 24 h. After stable operation from day 88 without varying the operating parameters, from day 126 onwards formation of granules was noted. By day 236 mature granules of size 0.9 ± 0.3 mm were developed without any filamentous growth. The size was measured as reported by Desireddy and Sabumon [21]. To find the size of granules homogenous sample of 500 mL was collected and passed through a series of sieves with various mesh sizes (1.5 mm, 1.2 mm, 1 mm, 800 μm, and 500 μm). The corresponding MLSS and MLVSS in the reactor were 2.15 g/L and 0.52 g/L (24%), respectively. The MLSS and MLVSS on the first day of phase III was 1.76 g/L and 0.23 g/L (13%), respectively. The percentage of inorganic content increased while the SVI₃₀ and SVI₁₅ decreased, which indicates the formation of phosphorous precipitates in the core of granules after long-term operation. Decreased DO levels would have helped in the generation of layers in the granule with nitrifiers, denitrifiers, and PAOs on the outer layers and dPAOs in the interiors of granules. The reactor was operated for 240 days in phase III. By the end of reactor operation (day 328) also the granules were not disintegrated even at a very high SRT of 328 days: considering negligible biomass in the decanted effluent in each cycle. Slight fluctuations in the granules size ranging between 0.6 mm and 1.2 mm was observed in phase III. However, the granules remained intact and stable which could be due to their smaller size. The conical geometric configuration and induced shear force by low rate of aeration would have helped in maintaining the granule size. If the granules were bigger, they easily get breakdown into smaller granules and again regrow. Sometimes due to endogenous respiration, the granules get destabilized leading to smaller fractions. By the end of phase III, the SVI₃₀ decreased to 65 mL/g while SVI₅ was 67 mL/g giving an SVI₅/SVI₃₀ ratio of 1.03 indicating good settling nature of the developed granules.

There are fluctuations in the EPS during the study period (Fig. 2). EPS during start-up was high which could be due to the sludge lysis and enhanced aerobic digestion at high DO on prolonged starvation period. From phase III of operation gradual increase in the EPS was noticed which helps in the formation of granules. Also, it is observed that there were not many fluctuations in EPS after 200 days of operation till the end of the study, which could have helped in maintaining stable granules in the reactor.

A measuring cylinder of 1 L capacity was used as a control reactor operated at the same operational conditions as shown in Table 2. The average percentage removals of COD, NH₄⁺-N, TN, and PO₄³⁻-P in various phases of reactor operation are given in Table S2, and the profile of SBR is shown in Fig. 3 and Fig. 4. Up to phase I of operation, the nutrient removal efficiencies in control reactor were almost similar to that of conical SBR while no improvement in COD, TN, and PO₄³⁻-P removals occurred. The SVI₃₀ and SVI₅ are not comparable with that of conical SBR. This could be due to the non-uniform aeration in the measuring cylinder which led to settling of some amount of biomass at the bottom corners of the control SBR. There was no proper shear force and velocity gradient to help in the enhanced formation of aerobic granules. Hence effective granulation was not seen in the control reactor. However, the nitrification efficiency throughout the study was similar to that of conical SBR, while the TN removal efficiency is not satisfactory. The ineffective COD, TN, and PO₄³⁻-P removals could be due to the existence of non-granular biomass in the reactor. The MLSS in the conical SBR on the last day of reactor operation was 2.31 g/L, while it was 1.96 g/L in the control reactor. As there were no proper granule formation and the settling was not effective, there would be a higher chance for the escape of biomass from the reactor during decant phase. Also, the MLSS measured occasionally (results not shown) confirms that the effluents from conical SBR are clear with negligible solids compared to the control reactor. The EPS on day 328 in conical SBR and control SBR were 124 mg/g/VSS and 98 mg/g/VSS, respectively. The microscopic images (4 times magnification using phase contrast microscope; Zeiss-Primo star) of biomass drawn on day 150 from the conical SBR compared with that of control SBR is shown in the supplementary section (Fig. S1). From the images, it is evident that conical SBR is able to hold denser and granular biomass compared to the control reactor. Fig. S2 (supplementary) shows the schematic of SBRs with the highlights of the results obtained. Based on these results, it can be confirmed that the SBR with conical configuration is beneficial in terms of granule formation and simultaneous removal of COD, NH₄⁺-N, and PO₄³⁻-P cost-effectively without needing any costly instrumentation control. Also, the results obtained on the long-term operation are helpful in further designing and piloting conical configurations of SBR for cultivating aerobic granules in low DO concentrations for simultaneous removal of COD, nitrogen, and phosphorous.

Kinetics data of COD, NH₄⁺-N, NO₂⁻-N, NO₃⁻-N, pH, and PO₄³⁻-P are plotted in Fig. 5. Results show a gradual decrease in NH₄⁺-N up to 8 h with slight accumulation of NO₂⁻-N and NO₃⁻ N whereas slight increase in NH₄⁺-N and high accumulation of NO₂⁻-N and NO₃⁻-N levels were observed at 24 h. Similar trends were observed in the removals of COD. This indicates the lysis of biomass after depletion of nutrients. Immediately after adding the feed, homogenous sample was collected and measured for COD. While the feed contained 640 mg/L of COD, it swiftly decreased to 313 mg/L at 0 h. The high concentration of COD leads to penetration of COD into the interior of granules swiftly and ultimately could be stored as PHB by heterotrophic microbes inside the granules. At 0 h slight increase in PO₄³⁻-P (10.2 mg/L) was also noticed which could have been released due to assimilation of carbon during the feeding stage. However, the phosphorous released to COD used ratio (0.004) is negligible compared to the typical ratios (0.3) of anaerobic periods as reported in the literature [38]. Up to 8 h, no much accumulation of NO₂⁻-N and NO₃⁻-N was seen even though effective nitrification occurred. Also, a gradual decrease in COD and PO₄³⁻-P removals occurred which implies the occurrence of nitrification, denitrification, and denitrifying P removal simultaneously. After majority of the PO₄³⁻-P is removed by 8 h, at 24 h increase in accumulation of NO₂⁻-N and NO₃⁻-N was seen which signifies the presence of dPAOs in the reactor. The DO was in the range of 0.9 mg/L to 1.5 mg/L during the kinetic studies. The presence of heterotrophic nitrification and aerobic denitrifcation (HNAD) species is also possible in the SBR under these operating conditions, which have unique ability to denitrify in both aerobic and anoxic environments [39]. The reactor might be dominant in nitrifiers compared to denitrifiers due to the prevalence of continuous aerobic conditions. However, the gradual decrease in COD and PO₄³⁻-P removal shows that the granules contained mixed microbial consortium that involves in SNDPR. Slight decrease in pH was observed at 4 h which could be due to aerobic nitrification, however, the pH again increased at 8 h and 24 h indicating simultaneous denitrification. The pH was in the range of 6.6 to 7.2 during kinetic studies. Filipe et al. [40] reported that at a pH higher than 7.25, denitrifying glycogen accumulating organisms (dGAOs) dominate dPAOs. Hence in this system dPAOs would be the dominant organisms that are favorable, as dPAOs are known to reduce NO₂⁻-N and NO₃⁻-N more effectively than dGAOs. Also, literature shows that dPAOs prevail effectively at lower DO concentrations compared to dGAOs due to the high oxygen affinity of PAOs. Carvalheira et al. [41] reported that phosphorous uptake rate of PAOs decreased by 20% while that of GAOs decreased by 77% when DO was decreased from 8 mg/L to 0.6 mg/L. Hence the low DO and pH values could have led to the favoritism of PAOs or dPAOs in this system. However, the composition of mixed culture was not done in this study by advanced microbiological identification techniques, due to lack of facility.

NH₄⁺-N and PO₄³⁻-P removals followed zero-order kinetics with a rate constant of k = 6.22 M/h (R² = 0.98), and k = 1.03 M/h (R² = 0.98), respectively. The kinetics of COD removal was fitted well with first-order kinetic model with k = 0.28/h and R² = 0.98. Though the pH was not controlled in the reactor, it was in the neutral range all through the study and hence there was no need for an external supply of alkalinity. The obtained results indicate the possibility of SNDPR in a single unit. Also, the kinetics study reveals that for effective and optimal treatment, the cycle time required is 8 h. Therefore, it is possible to operate conical SBR in 8 h cycle time yielding effective SNDPR.