Chemical Engineering Journal 468 (2023) 143700 Contents lists available at ScienceDirect Chemical Engineering Journal journal homepage: www.elsevier.com/locate/cej Benefits and drawbacks of integrating a side-stream sludge fermenter into an A2O system under limited COD conditions Congcong Zhang a, Albert Guisasola a, *, Adrian Oehmen b, Juan Antonio Baeza a a GENOCOV, Departament d’Enginyeria Química, Biològica i Ambiental, Escola d’Enginyeria, Universitat Autònoma de Barcelona, 08193 Bellaterra (Barcelona), Spain b School of Chemical Engineering, The University of Queensland, St Lucia, Queensland 4072, Australia A R T I C L E I N F O A B S T R A C T Keywords: The implementation of a side-stream sludge fermenter (SSSF) has been identified as a possible solution to Anaerobic/anoxic/aerobic (A2O) improve the performance of an anaerobic/anoxic/aerobic (A2O) configuration when treating low COD waste­ COD limitation water. This study systematically evaluated the effects of incorporating a SSSF into an A2O configuration (side- Enhanced biological phosphorus removal stream enhanced biological phosphorus removal, S2EBPR) for P/N/COD removal under a limited influent COD (EBPR) Side-stream enhanced biological phosphorus (CODINF) condition. The performance of the S2EBPR (with the SSSF receiving 6% of the recycled activated sludge removal (S2EBPR) and operating with a hydraulic retention time (HRT) of 2.4 d) and A2O were compared under the same limited Side-stream sludge fermenter (SSSF) CODINF (350 mg/L) condition. S2EBPR improved the amount of P removed (26.6%) under a low influent COD/P of only 26.3 compared with A2O of 32.6, and enhanced denitrification (11%) without compromising full ammonium and COD removal. However, the PLOAD to the plant increased due to the P-release in SSSF, resulting in higher effluent P concentration. The methane and energy recovery indexes were around 45% lower than those of A2O. Sequencing analysis revealed a high abundance of PAO in accordance to its higher P removal. This study represents a comprehensive evaluation of the S2EBPR configuration and provides an assessment of its suitability. 1. Introduction COD requirements [10]. The influent of the SSSF can be a fraction of the anaerobic mixed liquor or the return activated sludge (RAS) (4%-30%) The increasingly serious eutrophication problem led by over- [7,24,31,49]. There are more than 80 full-scale applications of S2EBPR discharging of phosphorus (P) and nitrogen (N) drives the research for facilities worldwide [12,44,46,48], and most of them are implemented an efficient biological nutrient removal from wastewaters. Among the in Europe (60) and United States (12) [31]. Contrasting results exist in different wastewater treatment plants (WWTPs) configurations, the the literature regarding the effectiveness of S2EBPR vs conventional anaerobic/anoxic/aerobic (A2O) configuration is the most common for EBPR processes. Some studies have shown that the S2EBPR configura­ simultaneous biological nutrient and organic matter removal [42]. P is tion has improved the P removal performance and stability compared to removed by promoting the proliferation of polyphosphate-accumulating traditional EBPR configurations [22,31,49]. However, it has also been organisms (PAOs) with alternative anaerobic and aerobic/anoxic con­ reported that S2EBPR showed high fluctuations of effluent P (0.6 ± 1.0 ditions in the so-called enhanced biological phosphorus removal (EBPR) mgP/L) [46,49]. It has yet to be established how the S2EBPR configu­ process [11]. ration impacts WWTPs from a holistic perspective, especially when the Low influent COD concentration and nitrate intrusion through the influent COD is limited. external recycle are common causes of A2O failure [6,17,28]. Additional PAOs in S2EBPR systems can be promoted independently of the COD dosage can improve EBPR performance under these conditions at nature of the influent carbon source. For example, Vollertsen et al. [48] expense of higher cost and higher carbon footprint [39]. Besides that, reported P concentrations about 40 mg/L and considerable COD in the the integration of a side-stream sludge fermenter (SSSF) into a conven­ SSSF of two S2EBPR-WWTPs, which made the plant less dependent on tional EBPR process (also known as S2EBPR) has been suggested as a the input wastewater quality compared with a traditional EBPR-WWTP. potential strategy to overcome these issues since the volatile fatty acids The reasons are: (VFAs) production from sludge fermentation could cover these extra * Corresponding author. E-mail addresses: congcong.zhang@uab.cat (C. Zhang), albert.guisasola@uab.cat (A. Guisasola), a.oehmen@uq.edu.au (A. Oehmen), JuanAntonio.Baeza@uab.cat (J.A. Baeza). https://doi.org/10.1016/j.cej.2023.143700 Received 21 March 2023; Received in revised form 12 May 2023; Accepted 21 May 2023 Available online 24 May 2023 1385-8947/© 2023 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). C. Zhang et al. Chemical Engineering Journal 468 (2023) 143700 i) the biomass fermentation products are mostly VFA, a preferred temperature (22 ± 2 ℃). electron donor for PAOs [24,31], The composition of the concentrated solution is shown in Table S1, ii) the extended anaerobic phase has been hypothesised to provide a which contained 1.4 g P-PO3− 4 /d and 5.6 gN-NH4 /d. The initial organic + competitive advantage for PAOs with respect to glycogen accu­ matter concentration was 585 mgCOD/L during the start-up period, then mulating organisms (GAOs, i.e. PAOs competitors) or other het­ decreased to 500 mgCOD/L and progressively decreased in steps of 50 erotrophic organisms [7,6,49]. Wang et al. [49] showed a higher mgCOD/L for each operational period, in order to determine the COD- PAO activity in S2EBPR (higher PAOs and lower GAOs abun­ limited A2O operation (Table 1). dance). Nevertheless, there were no high differences between the The S2EBPR configuration was implemented by installing a SSSF (20 abundance of putative PAOs, such as Ca. Accumulibacter and L) reactor treating part of the external recycle from the settler. The SSSF Tetrasphaera, and reactor was mixed with a magnetic stirrer at 200 rpm and was kept iii) besides VFAs, the SSSF can contain extra readily biodegradable under anaerobic conditions to favour fermentation processes. The COD (rbCOD), which could be further fermented to VFAs in the flowrate to the SSSF was set to 6% of the RAS flowrate (8.4 L/d), which anaerobic reactor by fermenting-PAO such as Tetrasphaera led to an HRT of 2.4 d in the SSSF. The percentage was selected based on [7,15,30]. preliminary experiments and a literature review showing typical values of RAS to SSSF in the range 4%-30%. The effluent of the SSSF was fed to Coats et al., [10] showed that the integration of an SSSF could pro­ R1 most of the time, although it was connected to R2 or R3 for some mote GAO formation which could hinder EBPR activity. They studied shorter periods to investigate its effect. The relative configurations and this integration in an A/O EBPR sequencing batch reactor (i.e. without diagrams are shown in Figure S1 and S2. HRT in the A2O was about 23 h biological nitrogen removal). This contradiction on the results needs considering only the reactors and 31 h when considering the settler. In further investigation in plants aiming at simultaneous C/N/P removal. the case of S2EBPR, HRT was about 26 h considering only the reactors S2EBPR systems seem a promising option in certain scenarios, and 34 h with the settler. particularly under limited COD conditions. However, the effectiveness SRT was calculated with equation (1) for the A2O plant and equation of S2EBPR vs traditional EBPR has not before been systematically (2) for the S2EBPR: evaluated at limited influent COD/P ratios, nor how its operation im­ VANA ⋅XANA + VANOX ⋅XANOX + VAER ⋅XAER pacts adjoining unit processes such as anaerobic digesters. SRT = (1) QPUR ⋅XAER + QEFF ⋅XEFF In this work, the EBPR performance when a SSSF reactor was inte­ grated into an A2O system under a low influent COD (CODINF) scenario VANA ⋅XANA + VANOX ⋅XANOX + VAER ⋅XAER was comprehensively evaluated in a pilot-scale plant and on a long-term SRT = (2) QPUR ⋅XAER + QEFF ⋅XEFF + QSSSF ⋅ΔXSSSF basis. A comprehensive understanding of the benefits and drawbacks of these systems is essential in view of identifying the scenarios where its where VANA, VANOX and VAER (L) are the volume of the anaerobic, anoxic industrial adoption would be positive. and aerobic reactors, XANA, XANOX and XAER (g/L) the biomass concen­ Thus, the main objectives were: (1) to explore the CODINF limits for tration in these reactors, QPUR, QEFF and QSSSF mean the flow rate (L/d) the A2O vs S2EBPR configurations, (2) to study the impact on the plant of waste activated sludge (WAS), effluent and SSSF. XEFF is the biomass performance of integrating a SSSF reactor in different zones of an A2O concentration in the effluent. ΔXSSSF was the decay of biomass in the plant, (3) to assess how S2EBPR affects downstream biogas production, SSSF reactor, which was calculated as the input biomass concentration if employed instead of an A2O process, (4) to investigate the difference minus the biomass concentration of SSSF, where the input biomass of microbial communities in the A2O and S2EBPR processes under concentration is the theoretical concentration from the external recycle different operational conditions and COD/P ratios. (XSETTLER) determined with equation (3): 2. Materials and methods Table 1 Operational parameters for each period: COD concentration in the influent, SSSF 2.1. Equipment and operation parameters connection, waste activated sludge (WAS) flow, and DO setpoint. The initial pilot-scale A2O configuration consisted of three contin­ Period Day operation CODINF SSSF connection WAS flow DO (mg/L) (L/d) (mg/ uous stirred tank reactors for simultaneous C/N/P removal, with L) anaerobic reactor (R1, 28 L), anoxic reactor (R2, 28 L), aerobic reactor (R3, 90L) and settler (50 L). The feeding solution introduced to the Ia 1 — — 0 to10 2 Ib 18 585 — 10 2 anaerobic reactor consisted of tap water (144 L/d) and a concentrated Ic 32 500 — 10 2 solution (7 L/d). The internal recycle (IR) from R3 to R2 (450 L/d) was Id 42 450 — 10 2 used to keep the anoxic condition in R2. Waste sludge was discharged Ie 50 400 — 10 2 from R3 automatically with a flowrate selected to maintain the desired If 57 350 — 10 2 Ig 70 300 10 2 sludge retention time (SRT). The settler produced the effluent stream — Ih 78 500 — 10 2 and an enriched biomass stream (external recycle, ER) which was Ii 84 400 — 10 2 recycled to R1 in a flow of 140 L/d. The percentage of ER with respect to Ij 92 450 — 10 2 the influent was in the recommended range of 0.5–1[42]. II a 107 450 anaerobic 10 2 The three reactors were monitored on-line with DO (HACH reactor II b 119 350 anaerobic 1 2 CRI6050), pH (HACH CRI5335) and temperature probes (Axiomatic reactor Pt1000) connected to multimeters (HACH CRI-MM44). On-line data was II c 133–175 350 anaerobic 5 2 acquired with a data acquisition card (Advantech PCI-1711), which was reactor connected to a PC with the AddControl software [5] developed in the III a 206 350 anaerobic 7 2 reactor research group for process monitoring and control. DO in R3 was III b 217 350 disconnect SSSF 7 2 controlled with a proportional-integral algorithm manipulating the III c 232 350 anoxic reactor 7 2 aeration flow rate with a mass flow controller (MFC F-201CV, Bronk­ III d 246 350 aerobic reactor 7 2 horst) using a DO setpoint of 2 or 3 mg/L in different periods. pH in R3 III e 255 350 aerobic reactor 7 3 was controlled with an on–off controller dosing a sodium carbonate III f 268–283 350 anaerobic 7 3 reactor solution to adjust the pH about 7.5. The system was operated at room 2 C. Zhang et al. Chemical Engineering Journal 468 (2023) 143700 (QEFF + QER + QSSSF )⋅XAER − QEFF ⋅XEFF in Zhang et al. [51]. XSETTLER = (3) QSSSF + QER 2.3.3. Microbiological analyses where QER means the flow rate of external recycle. Sludge samples were collected from the aerobic reactor on day 36 SRT of the A2O and S2EBPR was controlled around 13 ± 3 days by (Period I c), 56 (I e), 69 (I f), 77 (I g), 105 (I j), and biomass from the manipulating the WAS flowrate from the aerobic reactor. The actual SRT aerobic reactor and SSSF on day175 (II c), 245 (period IIIc), 266 (period for each period is shown in Table S5. IIIe) and 283 (period IIIf) when the system reached stable operation. The Four different operational periods were run. Period I (days 1–106) bacterial population was identified by the Illumina amplicon sequencing aimed at assessing the lower limit of the CODINF under an A2O config­ of the 16S rRNA gene. The detailed process was as follows: the samples uration. Period II (days 107–205) explored the S2EBPR performance from the system were washed by PBS for three times and centrifugated under different operational conditions. Finally, period III (day 206–283) for further DNA extractions. Soil DNA isolation plus kit (Norgen Biotek focused on studying the effect of connecting the SSSF effluent to CORP, Ontario., Canada) were used for Genomic DNA extraction pro­ different reactors of the S2EBPR. The micronutrients composition was cess. Further, the obtained extracted DNA was detected and quantified adapted from Smolders et al. [40]. The biomass for inoculation was by a DNA NanoDrop 1000 Spectrophotometer (Waltham, MA, USA), and obtained from the municipal WWTP of Baix Llobregat (Barcelona, the purified DNA was analysed in an Illumina MiSeq platform service Spain). center in the Autonomous University of Barcelona (Barcelona, Spain). Universal primer pair 515F (GTGCCAGCMGCCGCGGTAA) and 806R 2.2. Chemical and biochemical analyses (GGACTACHVGGGTWTCTAAT) were applied to amplify the V3–V4 re­ gions of the small subunit (SSU) rRNA prokaryote gene (16S) [43]. The Samples for phosphate, COD, ammonium, nitrate and nitrite were database used for the classification of organisms was based on the withdrawn from R1, R2, R3 and SSSF almost daily and filtered with 0.22 Greengenes database. The sequence reads were processed through μm filters (Millipore). The concentration of phosphate was analysed by a Usearch software. Operational taxonomic units (OTUs) were generated phosphate analyser (115 VAC PHOSPHAX sc, Hach-Lange) based on the with the open reference methodology. Raw sequencing data of Illumina Vanadomolybdate yellow method [23]. The concentration of ammo­ amplicon sequencing of the 16S rRNA gene of sludge samples can be nium was measured with an ammonium analyser (AMTAXsc, Hach found in NCBI Sequence Read Archive with accession numbers of Lange) based on a gas selective electrode potentiometric determination SUB13291801. of ammonia [25]. Nitrite and nitrate were determined by Ion Chroma­ tography (DIONEXICS-2000). COD was measured by kits (HACH LCK 3. Results 314 and LCK 714) and a spectrophotometer. Sludge samples were withdrawn from R1, R2, R3, SSSF and effluent, 3.1. Exploring the effect of influent COD limitation on the A2O and evaluated by mixed liquor volatile suspended solids (VSS) and total performance suspended solids (TSS) according to Standard Methods [4]. The corre­ sponding sludge volume index (SVI) was calculated as the observed Figs. 1 and 2 exhibit the removal performance of the different con­ volume (mL) of sludge from the aerobic reactor after settling for 30 min figurations on a load basis. Figure S3-S4 show the P fate in a concen­ divided by the TSS (g/L) measured on the same day. tration basis. Table S2 to Table S5 show the performance of the system All the performance indicators are described in the Supplementary regarding the different species (P, N, COD and solids, respectively) Information section SI-2. during the operational process. After the system start-up, it took 18 days to reach a pseudo steady 2.3. Batch tests state with full P, ammonium and COD removal under A2O configuration (Period Ia). Period Ib (day 18–31) had the highest influent COD con­ 2.3.1. PAOs batch activity tests centration (585 mg/L) and PANA and PAER reached 28.7 and 0.6 mg/L, Four batch tests were conducted to estimate PAO activity. The respectively, with a PREMOVAL_ABSOLUTE about 1.33 g/d. In terms of N, biomass was obtained from the A2O aerobic reactor when the system there was neither ammonium, nor nitrite detected in the effluent. NO−3 -N was at stable operation in period I c: (day 36), I e (day 57), I f (day 70), was about 7.0 mg/L, indicating that all entering ammonium was and I g (day 78). Batch tests were carried out in a system equipped with a oxidized into nitrate. Total NREMOVAL_ABSOLUTE was about 4.79 g/ magnetically stirred vessel of 2 L, DO probe (Cellox 325, WTW) and pH d (Fig. 1b). Regarding organic carbon, total COD REMOVAL_ABSOLUTE (76.1 probe (Sentix 81, WTW). The anaerobic and anoxic conditions were g/d COD) was reached (Fig. 1c). VSS concentrations in the reactor and in maintained 2 h by nitrogen gas sparging, followed by 2 h of aerobic the effluent were around 1.40 and 0.034 g/L Fig. 2, which indicated a condition with a mass flowmeter (MFC F-201CV, Bronkhorst). The period of stability in terms of biomass concentration. In short, the A2O carbon source for the anaerobic phase was the same composition as the plant with an excess of CODINF could operate successfully. feed for the A2O pilot plant to reach a concentration about 200 mg/L. CODINF was decreased down to 500 mg/L in Period Ic (day 32–41) Nitrate was dosed at the end of the anaerobic phase to reach 10 mg N/L. for 10 days. All the performance indices were still positive and CODINF The temperature and pH were controlled about 25 ◦ C and 7.5 ± 0.3 was successively decreased to 350 mg/L from Period Id to If (day 42–69) throughout the process. Samples for phosphate, nitrogen species and and, in all the cases, the high removal percentages of P, N and COD were COD were taken every 30 min and filtered with 0.22 μm filters maintained. The VSS concentration in the system showed an expected (Millipore). decreasing trend with the decrease of CODINF (Fig. 2) (except for the Ib to Ic when undesired tap water flowrate fluctuations occurred). 2.3.2. Biochemical methane potential (BMP) batch tests in S2EBPR system The CODINF was reduced to 300 mg/L on day 70 (Period Ig) and PANA Three sets of BMP experiments as in [3] were conducted to investi­ declined from 26.1 to 17.6 mg/L due to the COD limitations. An gate methane production from different sludge samples (anaerobic, increased concentration of PAER (from 0.6 to 2.2 mg/L) was observed. aerobic and SSSF) under stable S2EBPR system performance. The inoc­ Thus, the limit of CODINF needed to maintain successful EBPR perfor­ ulum sludge was from the anaerobic digester of an urban WWTP mance in the A2O system was 350 mg/L. When CODINF was increased to (Manresa, Barcelona) and degassed at 37 ◦ C for at least 3 days before 500 mg/L aiming at EBPR recovery (Period Ih), the system showed use. The anaerobic digestion tests were conducted in 160 mL serum bulking issues. Dispersed growth of biomass was detected probably due bottles with 125 mL of effective volume and 35 mL headspace for biogas to the sudden and large increase of CODINF. Floc-forming species may production. All the tests lasted for 42 days. Further details are provided grow in a non-settleable form when exposed to a sudden high organic 3 C. Zhang et al. Chemical Engineering Journal 468 (2023) 143700 Fig. 1. The fate of P (a), N (b) and COD (c) and the removal performance of A2O (period I) and S2EBPR (period II) configurations. loading. The best response to this problem is a reduction ratio in the F/M lower CODINF and lower WAS flow. The system recovered subsequently of the system [14,34]. Thus, the CODINF was decreased to 400 mg/L on without affecting the P performance. Then, the WAS was increased again day 84 and increased to 450 mg/L on day 92 (Periods Ii and Ij) to lessen in day 133 (Period IIc) to reach the targeted SRT (11.7 ± 0.1 d) and the the growth of dispersed biomass. PANA and PAER recovered to the pre­ same CODINF of 350 mg/L as Period If to allow a thorough comparison of vious condition and the bulking problem improved. The amount of S2EBPR vs A2O (sections 3.3 and 3.4 and the Discussion section are PREMOVAL_ABSOLUTE and P in the effluent were similar to those of Period based on Period If and Period IIc). The SSSF performance for the Id, at 1.38 g/d and 0.02 g/d, respectively. different periods is shown in Tables S2 to S5 (the concentrations of P, N, Four batch tests were performed to assess PAO activity (Table S6). COD and solids in the SSSF were indicated as PSSSF, NH+ 4 -NSSSF, CODSSSF, High anaerobic release and aerobic uptake rates of P were obtained with VSSSSSF and TSSSSSF) and Figure S3. the system operation at 500 mg/L CODINF (0.37 and 0.18 mgP/gVSS min Significant P release was observed in the SSSF due to VFA production respectively). The lowest aerobic and anoxic P uptake rates were and its simultaneous in-situ consumption by PAO, as reported in previous observed for the biomass withdrawn under the lowest CODINF condition works [31,46,48]. The SSSF showed an absolute P release of about 88.7 (300 mg/L). Complete P removal was observed under A2O configuration mg/L and a concentration of COD around 40 mg/L. Thus, as a result of except for the period with limited CODINF (300 mg/L). SSSF integration, the system received higher PLOAD compared to A2O in period If (2.07 vs 1.40 g/d), and showed higher PANA (37.5 vs 26.1 mg/ L), and PREMOVAL_ABSOLUTE (1.76 vs 1.39 g/d). Therefore, considering the 3.2. Performance of the S2EBPR configuration amount of P removed, EBPR performance with SSSF integration improved by around 27%. However, the high increase in PLOAD led to an On day 107, the SSSF was installed and fed with a biomass-enriched undesired and unstable effluent quality, with PAER about 2.0 ± 1.6 mg/ stream from the settler (8.4 L/d, i.e. 6% of the RAS flowrate). The SSSF L. effluent was connected to the anaerobic reactor (Period IIa). The initial In terms of N, the SSSF effluent provided around 11.0 mg NH+ 4 -N/L integration of the SSSF with a high daily WAS volume (10 L) and 450 (without nitrate and nitrite) due to the fermentation/hydrolysis of the mg/L of CODINF exhibited bulking issues (VSS in the reactor decreasing biomass and cell lysis or decay in SSSF. Thus, the S2EBPR showed almost from 0.98 to 0.75 g/L and in the effluent increasing from 0.023 to 0.095 no change in the NLOAD (5.69 vs 5.60 g/d, i.e. + 1.6%) and a slightly g/L). Period IIb (day 119–132) was designed as a VSS upturn period with 4 C. Zhang et al. Chemical Engineering Journal 468 (2023) 143700 Fig. 2. Evolution of solids during the A2O (Period I) and S2EBPR (Period II) operation. (a) Solids concentration in the reactor and the SSSF, (b) Solids concentration in the effluent, and (c) Ratio of volatile suspended solids (VSS) to total suspended solids (TSS). higher value of NEFF (1.91 vs 1.64 g/d) without any observed influence Period III (day 206–283) aimed at comparing the different combi­ on nitrification. Mass balances showed that the percentage of influent N nation of locations for the effluent of the SSSF (Figures S5 and S6): incorporated into the biomass (NBIOMASS) decreased from 33% in A2O to anaerobic, anoxic and aerobic reactors. More detailed information can 19% in S2EBPR, and it was reasonable due to the lower WAS discharged. be found in the Supplementary Information. According to the results Accordingly, the percentage of NDENITRIFIED improved from 35% (A2O) obtained, the optimum integration position of SSSF to A2O for EBPR to 46% (S2EBPR), which showed that the integration of SSSF improved performance under low CODINF conditions is the anaerobic reactor. The the denitrification extent. worst EBPR performance was obtained with the connection of the SSSF Regarding the COD, the SSSF effluent contained around 40 mg/L to the anoxic reactor, but the reason for this needs further study. COD, which only represented a 0.5% increase of the CODLOAD in S2EBPR Connection to the aerobic reactor could enhance P-removal activity, but (from 53.3 to 53.6 g/d). Full COD removal efficiency indicated that the at the expense of a detrimental effect on nitrification. SSSF integration did not affect the COD removal since most of the COD produced was used in-situ in the SSSF. The COD mass balance showed a 3.3. Energy recovery based on BMP in S2EBPR higher percentage of CODMINERALIZED in S2EBPR (77%) vs A2O (60%), which agreed with part of the theoretical WAS being degraded in the Recovering part of the chemical energy contained in the wastewater SSSF. The solids in the reactors in both scenarios (S2EBPR and A2O) is a hot trend in current water resource recovery facilities. The inte­ were around 1 g/L. However, the solids in the SSSF were always higher gration of the SSSF into the A2O process decreased the amount of WAS (around 2 g/L) due to the use of the concentrated biomass stream from (due to the hydrolysis and fermentation of solids in the SSSF) and, thus, the settler. The ratio of VSS/TSS in the SSSF tended to be higher than biogas production would be minimized. The sludge coming from an that of A2O (Fig. 2c), which indicated that the biomass in SSSF was EBPR system has different BMP depending on the WAS location because releasing P and that the polyphosphate (PolyP) levels were much lower. of the different polyhydroxyalkanoate (PHA) content [8,20]. Fig. 3 and In addition, the sludge production decreased from 15.1 to 8.5 g/ Table 2 compare the BMP of different sludge samples from the S2EBPR d because the WAS in S2EBPR was half of that of A2O to maintain a system: sludge from anaerobic, aerobic reactors and SSSF in Period II c. similar SRT (see below in the discussion section). The highest BMP was obtained from the anaerobic sludge with about 5 C. Zhang et al. Chemical Engineering Journal 468 (2023) 143700 when CODINF was 350 mg/L in the aerobic reactor and SSSF respec­ tively). Dechloromonas, which can use oxygen or NO−X -N as electron ac­ ceptors, have been reported to exist frequently in full-scale WWTPs and have been considered as functional PAOs [32]. The relative abundance of Dechloromonas in the A2O system was in the range of 1.22–5.15%, and in the S2EBPR system was found to be 5.38% in the aerobic reactor and 5.78% in the SSSF. Rhodobacter [19] and Thauera [52] have been implicated as potential PAOs in EBPR systems. Interestingly, these organisms were found to be of higher abundance during the A2O operation rather than S2EBPR. The most common PAO, Ca. Accumulibacter, was not detected in the system. However, Thiothrix was much higher in abundance in the S2EBPR pro­ cess (more than 6.5%) as compared to A2O (from 0.01% to 2.7%). Thi­ othrix has been recognized as candidate PAO in a broad range of reports [27,33,37] and grew in a low COD condition [37]. This correlated well with the results of this study, since the ratio of C/P reached the lowest level under the operation of the S2EBPR system. Thiothrix was also considered as a typical filamentous bacteria which could lead to settling Fig. 3. Anaerobic biochemical methane potential tests for biomass samples problems of activated sludge [45]. Notably, the percentage of Thiothrix obtained from anaerobic, aerobic and SSSF reactors under S2EBPR operation in began to increase in the A2O process when the CODINF was 450 mg/L period IIc. and sludge bulking was observed at that time (SVI around 571 mL/g). However, In S2EBPR there was no sludge bulking, with the SVI about 200 mL/g, which indicated that the presence of Thiothrix in S2EBPR Table 2 system doesn’t seem to affect the stability of the sludge despite of their Biochemical methane potential (BMP) tests with biomass from anaerobic, aer­ filamentous cell morphology. obic reactor and SSSF. In addition, fermenters Dysgonomonas and Propionispora were re­ Sludge sample a BMP b Initial methane production rate c ported to ferment organics to VFA [53], which accounted for significant (mL CH4/gVSS) (mL CH4⋅/(gVSS⋅d)) abundances with 1.5%-3.4% and 0.8%-2.2%, respectively, during the Anaerobic reactor 250 ± 10 27.3 ± 0.8 whole operation. In terms of Propionivibrio and Defluviicoccus, which are Aerobic reactor 221 ± 3 24.7 ± 0.6 groups of known GAO [1,35], they were less abundant in the A2O system SSSF 215 ± 11 22.3 ± 1.7 with less than 0.3% and 0.8% observed during period I, respectively, a The sludge samples were taken from anaerobic, aerobic and SSSF reactors in which corresponded with successful EBPR performance. However, the Period IIc during stable state. presence of GAO is not a necessary indicator of the EBPR deterioration if b The final cumulative CH4 yield obtained from different sludge samples in 42 PAOs are favoured kinetically [24,30]. It is interesting that in the days. S2EBPR system, both organisms showed higher proportions. Especially, c The rates for the first 10 days of the BMP test. Propionivibrio accounted for 1.5% in the aerobic reactor and 5.5% in the SSSF. These results were contrary to that of Wang et al., [49], since they 250 mL CH4/gVSS, followed by aerobic and SSSF, with 221 and 215 mL found Propionivibrio showed an inferior population in conventional CH4/gVSS, respectively. Anaerobic biomass showed a 16% higher BMP EBPR than S2EBPR. Nevertheless, in our S2EBPR system, PAOs still held than aerobic sludge. Unexpectedly, the BMP of SSSF was closer to that in a competitive advantage over GAOs, both in terms of total population the aerobic reactor rather than being similar to the anaerobic BMP. It abundance (Table 4) Table S8 and in terms of P removal performance. could be speculated that the internal levels of PHA in the biomass from the SSSF were lower to those from the anaerobic reactor, however PHA 4. Discussion was not measured. The initial methane production rates in the anaerobic sludge were the highest: 27.3 mL CH4/(gVSS⋅d), whereas 24.7 and 22.3 4.1. Integrating SSSF in an A2O process mL CH4/(gVSS⋅d) were observed for aerobic and SSSF sludge, respectively. The main objective of integrating a SSSF into an A2O system is to increase the stability of biological P removal under potentially detri­ 3.4. The microbiological community in A2O and S2EBPR mental situations (i.e. low influent rbCOD or excessive nitrate entering the anaerobic reactor). The SSSF should provide an extra source of VFAs The variations and relative abundances of the bacteria selected in that would come from the degradation of part of the WAS. When a SSSF A2O (period I) and S2EBPR (period II and III) were analyzed by 16S is integrated into a non-EBPR system, the SSSF effluent contains a high rRNA gene sequencing. Fig. 4 compares the microbiological community amount of VFA that could be used, for example, to enhance denitrifi­ observed at the genus level in the operation of periods I and II (identity cation. In this work, the effluent of the SSSF was not VFA-rich but it derived at an OTU threshold of 96.5% similarity). The number of OTUs contained a significant amount of P [21,48,49]. PAO had consumed the was similar in each operational period: about 1414 in the A2O system in VFA produced in-situ in the SSSF linked to P release, with an average of P period I (Table S7) and in periods II and III from S2EBPR, the OTUs were concentration about 88.7 mg/L in SSSF. Thus, the SSSF integration 1430 in the reactor and 1480 in the SSSF. increased the average PLOAD (47.9%) and concentration of PANA (43.7%) Among the observed OTUs, the clusters Desulfovibrio, Anaerosinus, in the S2EBPR system (2.07 g/d and 37.5 mgP/L) when compared to Insolitispirillum and Dechloromonas were found to be dominant along the A2O (1.40 g/d and 26.1 mgP/L) under the same CODINF conditions. The whole operational process. Thereinto, Desulfovibrio was considered as increase of PLOAD was partially mitigated by the better P performance of heterotrophic denitrifying bacteria [9], and it has been observed in a the S2EBPR configuration: the P removal capacity in S2EBPR increased system with the simultaneous removal of N and P [41]. Desulfovibrio was by 26.6 % (1.76 vs 1.39 g P/d). Wang et al. [49] reported a 24.5 % generally the most abundant organism during the whole study, no higher P release (132 kg P/d vs 106 kg P/d) and a 80% higher P removal matter if the system was A2O (4%-10% when the CODINF was 500, 400, efficiency when comparing S2EBPR with A2O. Onnis-Hayden et al. [31] 350, 300 and 450 mg/L respectively), or S2EBPR (about 6.3% and 4.2% also reported higher P removal performance with S2EBPR vs A2O (90% 6 C. Zhang et al. Chemical Engineering Journal 468 (2023) 143700 Fig. 4. Microbial communities at the genus level observed during the different operational periods: 500, 400, 350, 300, 450 mg/L of CODinf in the A2O system (period I); 350 + and 350 + S are samples from the aerobic reactor and SSSF of the S2EBPR system (period IIc) obtained operating with CODinf of 350 mg/L. The order shown is based on the total abundance of that genus calculated by adding up its abundance across all analyzed samples. vs 82%). from 3.96 to 3.79 g/d with the N removal efficiency decreasing from This extra PLOAD can challenge the plant performance if the ratio of 69% to 63% due to the lower amount of WAS from the S2EBPR (5L/d) RAS diverted to the SSSF is too high. In our case, PAER was about 2.0 ± compared with A2O (10L/d) to keep the same SRT (see equations (1) and 1.6 mgP/L since not all the entering P could be removed despite the (2). observed increased in P removal performance. Vale et al. [46] showed Equation (1) shows the typical SRT calculation, while equation (2) 2.8 mg/L of P in the effluent in a full-scale S2EBPR with 6% of RAS to accounts for the amount of biomass degraded in the SSSF. It is important SSSF (without reporting the specific extra PLOAD). A potential solution to include this biomass degraded in the SSSF, as it has a comparable would be to integrate P removal/recovery strategies at this point to impact to an additional purge. This means that less WAS is required in avoid a drastic increase of the P entering the plant. S2EBPR compared to A2O to maintain the SRT. The SRT concept is only The produced VFA in the SSSF was up-taken in-situ by PAO and related to the fate of the solids: part of the COD as biomass was degraded others leading to a low COD SSSF effluent (around 40 mg/L). Vollertsen in the SSSF, however, P and N were not degraded but released to the et al. [48] estimated that about half of the rbCOD was stored by PAO in medium because of the biomass fermentation. As mentioned above, the the SSSF with HRT about 30 to 35 h, which left half of the COD in the SSSF integration decreased the amount of WAS and increased the SSSF effluent (50–90 mg rbCOD/L). Andreasen et al. [2] and Wang et al. nutrient load to the plant, and these factors pose complex interactions to [49] reported that an increase of 14% and 16% of influent COD could be the system and may compromise the effluent quality. due to the implementation of a SSSF in full-scale S2EBPR. The extent of the CODLOAD increase depends on both the HRT of SSSF and the pro­ portion of PAO in the sludge. Too low HRTs (less than 1 day) would lead 4.2. Influent C/P ratio and the COD requirement to low VFA production and combined with a high amount of PAO would lead to a low COD increase since all the potential COD released would be In our case, different influent COD/P ratios were obtained by the directly used in the SSSF. integration of SSSF despite the plant influent was not changed. In addition, the biomass fermentation also led to ammonium release Considering the increase of PLOAD from the SSSF and the low increase of and about 11 mgN/L was detected in the SSSF effluent, which resulted in COD, the COD/P (g/g) was only 26.3 in the S2EBPR, being 31% lower a slight increase of the total NLOAD to the system of 1.6%. The VFA than that of the A2O system under the same CODINF (COD/P ratio about released by biomass fermentation was not only used by PAO, but also 38). This COD/P ratio was even lower than that of the A2O system under acted as an electron donor for denitrification, since the nitrate in the ER the CODINF = 300 mg/L condition (period I g; COD/P ratio about 32.6), recycled to the SSSF was fully denitrified. The concept is similar to that which resulted in EBPR deterioration. Thus, the S2EBPR system showed of the so-called Johannesburg WWTP configuration that aims at avoid­ a higher PREMOVAL_ABSOLUTE at a much lower COD/P ratio (26.3), which ing nitrate entering to the anaerobic reactor by providing an external indicated the promising advantage on P removal under a low COD/P carbon source as electron donor. Thus, the percentage of NDENITRIFIED ratio scenario of S2EBPR vs A2O. increased from 35% to 46%. However, the NREMOVAL_ABSOLUTE decreased The influent COD/P ratio plays an important role in EBPR perfor­ mance [17,38]. However, the carbon biodegradability/fractionation is 7 C. Zhang et al. Chemical Engineering Journal 468 (2023) 143700 as important as the amount. PAO need VFA-like organic matter for their S2EBPR system were around 45 % lower than those from the A2O. anaerobic metabolism. Calculating the minimum COD required for a successful N and P removal is not a straightforward issue and depends 4.4. Key functionally populations- relevant PAO and GAO not only on the N and P loads but also on the COD biodegradability. Metcalf and Eddy [42] suggest for simultaneous C, P and N removal in an Reactor operation with a S2EBPR process showed a higher percent­ A2O configuration, the following values of readily biodegradable COD age of known candidate PAOs than the A2O system (Table 4), which was (rbCOD): 10 g rbCOD/gP and 6.6 g rbCOD/gNO−3 -N. Then, 334 mg/L in accordance with the higher P removal capability of the S2EBPR. rbCOD would be required to remove the influent concentrations of 9 Specifically, under the same CODINF condition of 350 mg/L, more PAO mgP/L and 37 mgN/L. This theoretical value agrees with the required clades (e.g. Thiothrix and Dechloromonas) could be enriched in S2EBPR. CODINF experimentally observed during the A2O evaluation (350 mg/L). GAO seemed to hold higher percentages in S2EBPR (2.34% in the aer­ In the case of the S2EBPR, PLOAD and NLOAD increased to 2.07 gP/d and obic reactor and 6.55% in the SSSF) than in A2O (1.02%), in contrast to 5.69 gN/d with the input from the SSSF, which would correspond to an previous findings that S2EBPR showed lower GAO abundance than influent concentration of 13.3 mgP/L and 37.6 mgN/L. Considering the conventional EBPR [31,49]. Different from other investigations effluent P concentration during period IIc (2.0 mgP/L), 11.3 mgP/L were [29,31,49], Tetrasphaera was not observed in this study, despite of its removed in the S2EBPR system with the same CODINF = 350 mg/L reported low decay rate and fermenting ability. Dold and Conidi [13] instead of only 9 mgP/L of A2O. This would imply a decrease in the COD also pointed out some potential conflicts for Tetrasphaera, such as the requirements down to 9 g rbCOD/g P, i.e. 10% lower than for the A2O conflict in its abundance due to different quantification procedures, the configuration. Since the generated organic matter is mostly used by PAO implied importance in RAS fermentation processes and its importance in in the anaerobic reactor and not for denitrification, it is reasonable to its ability to ferment in EBPR systems. Apart from that, it should be assume organic matter requirements of 9 g rbCOD/g P and 6.6 g rbCOD/ noted that in S2EBPR the microbial communities showed a relatively g NO−3 -N for simultaneous C, P and N removal in a S2EBPR system. lower percentage of PAO and higher percentage of GAO in the SSSF compared with the aerobic reactor. SSSF was speculated to induce more 4.3. Energy recovery indices decay of GAO and other ordinary heterotrophic organisms due to the extended anaerobic condition, thus giving PAO a competitive advantage Table 3 shows the energy recovery indices of A2O and S2EBPR at the [13,18,49], which was not consistent with the observed community same CODINF conditions. One should previously think whether low dynamics of this study. It suggests the existence of some unknown PAO sludge production is beneficial or detrimental for the plant performance. in S2EBPR. For example, Ca. Accumulibacter PAO were not detected in When considering energy recovery, the integration of the SSSF should this study, but it should be considered that its quantification by 16S decrease the potential energy recovery of the plant since not all the WAS rRNA techniques has usually led to lower values than other techniques is derived to biogas production but part of it is degraded in the SSSF. as fluorescent in-situ hybridization [36,47]. As a result, future work However, lower sludge production would as well lead to lower sludge should also focus on identifying other potential groups of PAOs that may handling costs. Operational costs of a WWTP are typically about 40–50% exist in S2EBPR processes. In any case, the S2EBPR sustained the due to sludge production, with only 10% related to energy [16,50], so coexistence of an abundant PAO community that was appropriate for less sludge produced with less energy produced could still be an eco­ EBPR, despite the higher P effluent observed. nomic win from an operational costs standpoint. Besides that, less WAS would also mean a smaller digester too (and, thus, lower capital costs 4.5. Practical implications and operation costs). The WAS in the S2EBPR was lower than in the A2O for the same This comprehensive work systematically evaluated the effect of targeted SRT. We can assume similar BMP values from these two systems introducing a SSSF in an A2O plant (S2EBPR configuration) under a wide since the BMP value is mostly dependent on the internal PHA content range of influent COD scenarios. The S2EBPR system showed improve­ and, thus, on the SRT [8,51]. We adopted 221 mL CH4/gVSS from aer­ ment of EBPR and denitrification even at lower COD/P ratios than that obic sludge under the condition of S2EBPR for the calculation process being minimum threshold in an A2O configuration. Besides the con­ (see section 3.3). Thus, both energy recovery indices for the S2EBPR ventional performance, we assessed the possibility of improving energy system were around 45 % lower than those from the A2O for the period recovery and the connection of a SSSF to different locations of the A2O. with the same SRT, since half volume of wasted sludge was discharged in Implementing a SSSF into a conventional A2O system can have the S2EBPR with similar amount of CODLOAD. MRI and ERI are about 0.253 following implications: gCODCH4/gCODREM and 0.166 kJCH4/kJINF for the A2O case and 0.139 gCODCH4/gCODREM and 0.093 kJCH4/kJINF for the S2EBPR. MRI is also 1) The connection of a SSSF to the anaerobic reactor can allow higher an indicator for mineralization degree, which was in accordance with SRT (i.e. lower WAS) for fermentation processes and thus lead to the COD mineralization degree in these two systems (60% in A2O and lower COD/P requirements for EBPR. The connection of a SSSF to the 77% in S2EBPR), and higher mineralization results in higher cost in anaerobic reactor of A2O showed the optimum EBPR performance S2EBPR. Literature reports show that the conversion efficiency from compared with other locations (i.e. to the anoxic or aerobic reactor). influent COD to methane is in the range of 15–35% and this energy could If an energy recovery process is implemented (i.e. anaerobic diges­ be recovered as a form of electricity by combined heat and power tion), the energy recovery indices would decrease due to the lower technologies. Assuming 35% of the transformation efficiency from volume of WAS. Despite the energy recovery decline, it would still be methane to electricity, only 5% (S2EBPR) or 12% (A2O) of the CODINF advantageous from the economic point of view due to the savings in could be recovered in the form of electricity [26]. terms of the sludge disposal and the need of a smaller digester. As a consequence, the S2EBPR would still be an economic win with 2) Combining the S2EBPR with P recovery provides a novel opportunity half sludge produced though both energy recovery indices for the for P recovery by chemical precipitation due to the higher P Table 3 Comparison of energy recovery indices of A2O and S2EBPR under the same CODINF conditions. Systems Period VSS (g/L) YOBS (gCODx/gCODs) Influent load (gCOD/d) CH4 production (gCOD/d) MRI (gCODCH4)/(gCODREM) ERI (kJCH4)/(kJINF) A2O If 1.13 ± 0.13 0.40 ± 0.04 53.3 9.544 0.253 0.166 S2EBPR IIc 0.95 ± 0.12 0.22 ± 0.03 53.6 5.373 0.139 0.093 8 C. Zhang et al. Chemical Engineering Journal 468 (2023) 143700 Table 4 The read abundance of relative PAOs and GAOs at the genus level observed during different operational periods: 500, 400, 350, 300, 450 mg/L of CODINF in the A2O system (period I); 350 + and 350 + S are samples from the aerobic reactor and SSSF of the S2EBPR system (period IIc) obtained operating with CODINF of 350 mg/L. CODINF(mg/L) 500 400 350 300 450 350+ 350 + S Genus % % % % % % % Thiothrix 0.01 ± 0.00 0.28 ± 0.31 0.66 ± 0.04 0.14 ± 0.02 2.67 ± 1.31 6.47 ± 1.40 6.89 ± 0.39 Dechloromonas 1.22 ± 1.20 3.24 ± 3.18 2.70 ± 0.01 5.14 ± 1.54 3.53 ± 0.70 5.38 ± 1.35 5.78 ± 1.47 Desulfuromonas 2.04 ± 0.57 3.70 ± 3.72 0.75 ± 0.02 1.02 ± 0.06 0.80 ± 0.17 1.27 ± 0.31 0.83 ± 0.00 Desulfovibrio 9.54 ± 2.80 7.05 ± 2.53 4.08 ± 0.36 5.56 ± 0.22 4.33 ± 1.50 6.31 ± 1.23 4.20 ± 0.64 Rhodobacter 2.36 ± 1.67 0.81 ± 0.90 1.90 ± 0.08 2.76 ± 0.34 1.55 ± 0.56 0.19 ± 0.06 0.24 ± 0.03 Desulfobulbus 1.95 ± 0.82 1.16 ± 0.29 0.76 ± 0.09 0.99 ± 0.07 0.89 ± 0.33 1.18 ± 0.26 0.76 ± 0.13 Thauera 0.66 ± 0.43 1.25 ± 1.11 2.53 ± 0.10 2.31 ± 0.87 0.42 ± 0.41 0.07 ± 0.01 0.09 ± 0.02 Desulfomicrobium 1.07 ± 0.25 1.44 ± 1.16 0.43 ± 0.04 0.58 ± 0.13 0.47 ± 0.15 0.66 ± 0.09 0.55 ± 0.00 Total PAO 18.84 ± 7.84 18.91 ± 13.20 13.81 ± 0.76 18.50 ± 3.25 14.66 ± 5.13 21.54 ± 4.72 19.35 ± 2.67 Propionivibrio 0.14 ± 0.08 0.30 ± 0.06 0.22 ± 0.01 0.23 ± 0.20 0.11 ± 0.04 1.52 ± 0.38 5.52 ± 0.67 Defluviicoccus 0.67 ± 0.60 0.81 ± 0.92 0.80 ± 0.15 0.73 ± 0.28 0.70 ± 0.31 0.82 ± 0.14 1.03 ± 0.01 Total GAO 0.81 ± 0.69 1.11 ± 0.98 1.02 ± 0.16 0.96 ± 0.48 0.81 ± 0.35 2.34 ± 0.52 6.55 ± 0.69 concentrations attained: on one hand, it would allow removing part Data availability of the P accumulated in the system due to the lower WAS volume required to maintain biomass concentration in this configuration, on Data will be made available on request. the other hand, it would decrease the potential of additional P load from SSSF to threaten the quality of mainstream effluent. Acknowledgements 3) In this work, we have not considered another potential P input to the plant: the reject wastewater or the effluent from the waste sludge This work was supported by Grant PID2020-119018RB-I00 funded treatment. We only considered the P-load increase due to part of the by MCIN/AEI/10.13039/501100011033 and by European Union Next­ RAS being diverted to the SSSF. In real plants, the whole reject water GenerationEU/PRTR. Congcong Zhang would like to thank the financial may be recycled to the plant. This effluent would increase the P-load support from China Scholarship Council. The authors from UAB are to the plant without providing any extra COD and, thus, would even members of the GENOCOV research group (Grup de Recerca Consolidat make more evident the need of novel configurations such as the de la Generalitat de Catalunya, 2021 SGR 515, www.genocov.com). SSSF. Appendix A. Supplementary data 5. Conclusions Supplementary data to this article can be found online at https://doi. The S2EBPR configuration with the connection of the SSSF to the org/10.1016/j.cej.2023.143700. anaerobic reactor was studied by introducing 6% of the RAS with HRT = 2.4 d and CODINF of 350 mg/L (minimum value for successful P/N removal in the A2O configuration). References The main benefits with S2EBPR compared with A2O were: [1] M. Albertsen, S.J. McIlroy, M. Stokholm-Bjerregaard, S.M. Karst, P.H. Nielsen, “Candidatus Propionivibrio aalborgensis”: A Novel Glycogen Accumulating 1. Higher P and N removal capacity (26.6% and 11%) without Organism Abundant in Full-Scale Enhanced Biological Phosphorus Removal Plants, compromising full COD and ammonium oxidation. Front. Microbiol. 7 (2016) 1–17, https://doi.org/10.3389/fmicb.2016.01033. [2] K. Andreasen, G. Petersen, H. Thomsen, R. Strube, Reduction of nutrient emission 2. EBPR could be sustained with a low influent COD/P of only 26.3, by sludge hydrolysis, Water Sci. Technol. 35 (1997) 79–85, https://doi.org/ while EBPR deterioration in A2O was observed with a ratio about 10.1016/S0273-1223(97)00215-1. 32.6. Organic matter needs decreased from 10 g rbCOD/gP to 9 g [3] I. Angelidaki, M. Alves, D. Bolzonella, L. Borzacconi, J.L. Campos, A.J. Guwy, S. Kalyuzhnyi, P. Jenicek, J.B. van Lier, Defining the biomethane potential (BMP) rbCOD/gP. of solid organic wastes and energy crops: a proposed protocol for batch assays, 3. A higher abundance of functional PAO were observed in S2EBPR, Water Sci. Technol. 59 (2009) 927–934, https://doi.org/10.2166/wst.2009.040. which gave PAO the advantage for EBPR performance. [4] APHA. Stand. methods exam. water wastewater, 20th ed., 1995. [5] J.A. Baeza, Advanced Direct Digital Control (AddControl): lessons learned from 20 years of adding control to lab and pilot scale treatment systems, in: in: 13th IWA However, some underlying drawbacks were: Conference on Instrumentation, Control and Automation. ICA 2022, Tsinghua University, Beijing (China), 2022, pp. 13–15. [6] J.L. Barnard, K. Abraham, Key features of successful BNR operation, Water Sci. 1. Lower effluent quality (with 2.0 ± 1.6 mgP/L) was observed due to Technol. 53 (2006) 1–9, https://doi.org/10.2166/wst.2006.400. the additional PLOAD, accompanied with a relatively insignificant [7] J.L. Barnard, P. Dunlap, M. Steichen, Rethinking the Mechanisms of Biological increase of residual COD (0.5%) and N (1.6%). Phosphorus Removal, Water Environ. Res. 89 (2017) 2043–2054, https://doi.org/ 10.2175/106143017X15051465919010. 2. The energy recovery indices were around 45% lower than those of [8] C. Chan, A. Guisasola, J.A. Baeza, Correlating the biochemical methane potential of A2O (with the same SRT), and less of the input COD could be bio-P sludge with its polyhydroxyalkanoate content, J. Clean. Prod. 242 (2020), recovered as electricity in S2EBPR (5%) compared with A2O (12%), 118495, https://doi.org/10.1016/j.jclepro.2019.118495. which may not a real drawback from an economic standpoint due to [9] L. Chu, J. Wang, Denitrification performance and biofilm characteristics using biodegradable polymers PCL as carriers and carbon source, Chemosphere 91 the sludge reduction. (2013) 1310–1316, https://doi.org/10.1016/j.chemosphere.2013.02.064. [10] E.R. Coats, K. Eyre, C. Bryant, T. Woodland, C.K. Brinkman, Assessing the Effects of Declaration of Competing Interest RAS Fermentation on EBPR Performance and Associated Microbial Ecology, Water Environ. Res. 90 (2018) 659–671, https://doi.org/10.2175/ 106143017X15131012153130. The authors declare that they have no known competing financial [11] Comeau, Y., Oldham, W.K., Hall, K.J., 1987. Dynamics of Carbon Reserves in interests or personal relationships that could have appeared to influence Biological Dephosphatation of Wastewater, Biological Phosphate Removal from Wastewaters. International Association on Water Pollution Research and Control. the work reported in this paper. https://doi.org/10.1016/b978-0-08-035592-4.50010-9. [12] Copp, J.B., Belk, I., Vale, P., 2012. Operational Control of a RAS Fermentation Process for Enhancing Biological Phosphorus Removal. 9 C. Zhang et al. Chemical Engineering Journal 468 (2023) 143700 [13] P. Dold, D. Conidi, Achieving enhanced biological P removal: Have we forgotten [35] S. Roy, Q. Guanglei, R. Zuniga-Montanez, R.B. Williams, S. Wuertz, Recent how to design a bioP plant? WEFTEC 2019–92nd Annu, Water Environ. Fed. Tech. advances in understanding the ecophysiology of enhanced biological phosphorus Exhib. Conf. (2019) 1452–1466. removal, Curr. Opin. Biotechnol. 67 (2021) 166–174, https://doi.org/10.1016/j. [14] W. Eckenfelder, J. Musterman, Activated sludge: treatment of industrial copbio.2021.01.011. wastewater, CRC Press, 1998. [36] F.J. Rubio-Rincón, D.G. Weissbrodt, C.M. Lopez-Vazquez, L. Welles, B. Abbas, [15] E.Y. Fernando, S.J. McIlroy, M. Nierychlo, F.A. Herbst, F. Petriglieri, M.C. Schmid, M. Albertsen, P.H. Nielsen, M.C.M. van Loosdrecht, D. Brdjanovic, “Candidatus M. Wagner, J.L. Nielsen, P.H. Nielsen, Resolving the individual contribution of key Accumulibacter delftensis”: A clade IC novel polyphosphate-accumulating microbial populations to enhanced biological phosphorus removal with organism without denitrifying activity on nitrate, Water Res. 161 (2019) 136–151, Raman–FISH, ISME J. 13 (2019), https://doi.org/10.1038/s41396-019-0399-7. https://doi.org/10.1016/j.watres.2019.03.053. [16] R. Ferrentino, M. Langone, F. Merzari, L. Tramonte, G. Andreottola, A review of [37] F.J. Rubio-Rincón, L. Welles, C.M. Lopez-Vazquez, M. Nierychlo, B. Abbas, anaerobic side-stream reactor for excess sludge reduction: Configurations, M. Geleijnse, P.H. Nielsen, M.C.M. van Loosdrecht, D. Brdjanovic, Long-term mechanisms, and efficiency, Crit. Rev. Environ. Sci. Technol. 46 (2016) 382–405, effects of sulphide on the enhanced biological removal of phosphorus: The https://doi.org/10.1080/10643389.2015.1096879. symbiotic role of Thiothrix caldifontis, Water Res. 116 (2017) 53–64, https://doi. [17] A.Z. Gu, A. Saunders, J.B. Neethling, H.D. Stensel, L.L. Blackall, Functionally org/10.1016/j.watres.2017.03.017. Relevant Microorganisms to Enhanced Biological Phosphorus Removal [38] F. Sabba, M. Farmer, Z. Jia, F. Di Capua, P. Dunlap, J. Barnard, C.D. Qin, J. Performance at Full-Scale Wastewater Treatment Plants in the United States, Water A. Kozak, G. Wells, L. Downing, Impact of operational strategies on a sidestream Environ. Res. 80 (2008) 688–698, https://doi.org/10.2175/106143008x276741. enhanced biological phosphorus removal (S2EBPR) reactor in a carbon limited [18] D. Herbst, N. Wimmer, The Proteome of Tetrasphaera elongata is adapted to wastewater plant, Sci. Total Environ. 857 (2023), 159280, https://doi.org/ Changing Conditions in Wastewater Treatment Plants, Proteomes 7 (2019) 16, 10.1016/j.scitotenv.2022.159280. https://doi.org/10.3390/proteomes7020016. [39] N. Shen, Y. Zhou, Enhanced biological phosphorus removal with different carbon [19] A. Hiraishi, A. Yanase, H. Kitamura, Polyphosphate accumulation by Rhodobacter sources, Appl. Microbiol. Biotechnol. 100 (2016) 4735–4745, https://doi.org/ sphaeroides grown under different environmental conditions with special emphasis 10.1007/s00253-016-7518-4. on the effect of external phosphate concentrations, Bull. Japanese Soc. Microb. [40] G.J.F. Smolders, J. van der Meij, M.C.M. van Loosdrecht, J.J. Heijnen, Model of the Ecol. 6 (1991) 25–32, https://doi.org/10.1264/microbes1986.6.25. anaerobic metabolism of the biological phosphorus removal process: Stoichiometry [20] S.M.S. Huda, H. Satoh, T. Mino, Anaerobic degradation of polyhydroxyalkanoate and pH influence, Biotechnol. Bioeng. 43 (1994) 461–470, https://doi.org/ accumulated in activated sludge in the absence of anaerobic digested sludge, 10.1002/bit.260430605. J. Water Environ. Technol. 14 (2016) 236–246, https://doi.org/10.2965/jwet.15- [41] H. Sun, Q. Zhou, L. Zhao, W. Wu, Enhanced simultaneous removal of nitrate and 037. phosphate using novel solid carbon source/zero-valent iron composite, J. Clean. [21] K. Jönsson, J. la C. Jansen, Hydrolysis of return sludge for production of easily Prod. 289 (2021), https://doi.org/10.1016/j.jclepro.2020.125757. biodegradable carbon: effect of pre-treatment, sludge age and temperature, Water [42] G. Tchobanoglous, F.L. Burton, H.D. Stensel, Wastewater Engineering: Treatment Sci. Technol. 53 (2006) 47–54, https://doi.org/10.2166/wst.2006.405. and Reuse, Metcalf & Eddy Inc, Fourth. ed., McGraw-Hill, New York, 2013. [22] A.B. Lanham, A. Oehmen, A.M. Saunders, G. Carvalho, P.H. Nielsen, M.a M. Reis, [43] H. Tian, X. Xu, J. Qu, H. Li, Y. Hu, L. Huang, W. He, B. Li, Biodegradation of Metabolic versatility in full-scale wastewater treatment plants performing phenolic compounds in high saline wastewater by biofilms adhering on aerated enhanced biological phosphorus removal, Water Res. 47 (2013) 7032–7041, membranes, J. Hazard. Mater. 392 (2020), 122463, https://doi.org/10.1016/j. https://doi.org/10.1016/j.watres.2013.08.042. jhazmat.2020.122463. [23] APHA. Method 4500-P C. Stand. methods exam. water wastewater, 20th ed., 1995. [44] N.B. Tooker, G. Li, C. Bott, P. Dombrowski, P. Schauer, A. Menniti, A. Shaw, J. [24] Li, G., Tooker, N.B., Wang, D., Srinivasan, V., Barnard, J.L., Russell, A., Takacs, I., L. Barnard, B. Stinson, G. Stevens, P. Dunlap, I. Takacs, H. Phillips, H. Analla, Bott, C., Dobrowski, P., Onnis-Hayden, A., Gu, A.Z., 2020. Modeling side-stream A. Russell, A. Ellsworth, J. McQuarrie, K. Carson, A. Onnis-Hayden, A.Z. Gu, enhanced biological phosphorus removal (S2EBPR) system using agent-based Rethinking and Reforming Enhanced Biological Phosphorus Removal (EBPR) model with adaptive maintenance, decay and TCA metabolism. bioRxiv. https:// strategy - Concepts and mechanisms of side-stream EBPR, in: in: Water doi.org/10.1101/2020.11.18.387589. Environment Federation Technical Exhibition and Conference 2017, WEFTEC [25] APHA. Method 4500-NH3 D. Stand. methods exam. water wastewater, 20th ed., 2017, Water Environment Federation, 2017, pp. 4387–4404. https://doi. 1995. org/10.2175/193864717822153076. [26] P.L. McCarty, J. Bae, J. Kim, Domestic Wastewater Treatment as a Net Energy [45] E. Vaiopoulou, P. Melidis, A. Aivasidis, Growth of filamentous bacteria in an Producer-Can This be Achieved? Environ. Sci. Technol. 45 (2011) 7100–7106, enhanced biological phosphorus removal system, Desalination 213 (2007) https://doi.org/10.1021/es2014264. 288–296, https://doi.org/10.1016/j.desal.2006.02.101. [27] Q. Meng, W. Zeng, B. Wang, Z. Fan, Y. Peng, New insights in the competition of [46] P. Vale, J. Barnard, D. Thomas, P. Dold, RAS Fermentation to Enhance Biological polyphosphate-accumulating organisms and glycogen-accumulating organisms Phosphorus Removal, Proc. Water Environ. Fed. 2008 (2008) 1775–1789, https:// under glycogen accumulating metabolism with trace Poly-P using flow cytometry, doi.org/10.2175/193864708788733981. Chem. Eng. J. 385 (2020), 123915, https://doi.org/10.1016/j.cej.2019.123915. [47] B. Valverde-Pérez, D.S. Wágner, B. Lórant, A. Gülay, B.F. Smets, B.G. Plósz, Short- [28] J.B. Neethling, Factors Influencing the Reliability of Enhanced Biological sludge age EBPR process - microbial and biochemical process characterisation Phosphorus Removal, Water Intell. Online 5 (2006), https://doi.org/10.2166/ during reactor start-up and operation, Water Res. 104 (2016) 320–329, https://doi. 9781780404479. org/10.1016/j.watres.2016.08.026. [29] H.T.T. Nguyen, V.Q. Le, A.A. Hansen, J.L. Nielsen, P.H. Nielsen, High diversity and [48] J. Vollertsen, G. Petersen, V.R. Borregaard, Hydrolysis and fermentation of abundance of putative polyphosphate-accumulating Tetrasphaera-related bacteria activated sludge to enhance biological phosphorus removal, Water Sci. Technol. 53 in activated sludge systems, FEMS Microbiol. Ecol. 76 (2011) 256–267, https:// (2006) 55–64, https://doi.org/10.2166/wst.2006.406. doi.org/10.1111/j.1574-6941.2011.01049.x. [49] D. Wang, N.B. Tooker, V. Srinivasan, G. Li, L.A. Fernandez, P. Schauer, A. Menniti, [30] P.H. Nielsen, S.J. McIlroy, M. Albertsen, M. Nierychlo, Re-evaluating the C. Maher, C.B. Bott, P. Dombrowski, J.L. Barnard, A. Onnis-Hayden, A.Z. Gu, Side- microbiology of the enhanced biological phosphorus removal process, Curr. Opin. stream enhanced biological phosphorus removal (S2EBPR) process improves Biotechnol. 57 (2019) 111–118, https://doi.org/10.1016/j.copbio.2019.03.008. system performance - A full-scale comparative study, Water Res. 167 (2019), [31] A. Onnis-Hayden, V. Srinivasan, N.B. Tooker, G. Li, D. Wang, J.L. Barnard, C. Bott, 115109, https://doi.org/10.1016/j.watres.2019.115109. P. Dombrowski, P. Schauer, A. Menniti, A. Shaw, B. Stinson, G. Stevens, P. Dunlap, [50] S.-S. Yang, W.-Q. Guo, X.-J. Zhou, Z.-H. Meng, B. Liu, N.-Q. Ren, Optimization of I. Takács, J. McQuarrie, H. Phillips, A. Lambrecht, H. Analla, A. Russell, A.Z. Gu, operating parameters for sludge process reduction under alternating aerobic/ Survey of full-scale sidestream enhanced biological phosphorus removal (S2EBPR) oxygen-limited conditions by response surface methodology, Bioresour. Technol. systems and comparison with conventional EBPRs in North America: Process 102 (2011) 9843–9851, https://doi.org/10.1016/j.biortech.2011.07.079. stability, kinetics, and microbial populations, Water Environ. Res. 92 (2020) [51] C. Zhang, A. Guisasola, J.A. Baeza, Achieving simultaneous biological COD and 403–417, https://doi.org/10.1002/wer.1198. phosphorus removal in a continuous anaerobic/aerobic A-stage system, Water Res. [32] F. Petriglieri, C. Singleton, M. Peces, J.F. Petersen, M. Nierychlo, P.H. Nielsen, 190 (2021), 116703, https://doi.org/10.1016/j.watres.2020.116703. “Candidatus Dechloromonas phosphoritropha” and “Ca. D. phosphorivorans”, [52] Y. Zhang, Z. Hua, H. Lu, A. Oehmen, J. Guo, Elucidating functional microorganisms novel polyphosphate accumulating organisms abundant in wastewater treatment and metabolic mechanisms in a novel engineered ecosystem integrating C, N, P and systems, ISME J. 15 (2021) 3605–3614, https://doi.org/10.1038/s41396-021- S biotransformation by metagenomics, Water Res. 148 (2019) 219–230, https:// 01029-2. doi.org/10.1016/j.watres.2018.10.061. [33] N. Rey-Martínez, M. Badia-Fabregat, A. Guisasola, J.A. Baeza, Glutamate as sole [53] L. Zhao, H. Chen, Z. Yuan, J. Guo, Interactions of functional microorganisms and carbon source for enhanced biological phosphorus removal, Sci. Total Environ. 657 their contributions to methane bioconversion to short-chain fatty acids, Water Res. (2019) 1398–1408, https://doi.org/10.1016/j.scitotenv.2018.12.064. 199 (2021), 117184, https://doi.org/10.1016/j.watres.2021.117184. [34] Richard, M., 2003. Michael Richard, Ph.D. Sear-Brown Fort Collins, CO. 20th Annu. USEPA Natl. Oper. Trainers Conf. 1–21. 10