Biomass and Bioenergy 149 (2021) 106077 Contents lists available at ScienceDirect Biomass and Bioenergy journal homepage: http://www.elsevier.com/locate/biombioe Comparing continuous and batch operation for high-rate treatment of urban wastewater Natalia Rey-Martínez, Aloia Barreiro-López, Albert Guisasola *, Juan A. Baeza GENOCOV, Department of Chemical, Biological and Environmental Engineering, Universitat Autònoma de Barcelona, Spain A R T I C L E I N F O A B S T R A C T Keywords: The water-energy nexus has changed the concept of wastewater treatment plants (WWTPs), which should move Anaerobic digestion from energy consumers into energy neutral or even energy positive facilities. The A/B process aims at achieving A-stage self-sufficient energy WWTPs: organic matter is removed in the first step (A-stage) and derived to biogas pro­ COD fractionation duction whereas autotrophic nitrogen removal is implemented in a second step (B-stage). This work compares Energy recovery High-rate two high-rate systems that can be used as A-stage in view of organic matter removal: a continuous high rate activated sludge (HRAS) reactor and a high-rate sequencing batch reactor (HRSBR). Both systems were operated with real urban wastewater at a short hydraulic retention time (2.5 h) and at short sludge retention time (SRT) of 1–2 d to minimize COD mineralization and to maximize organic matter diversion to methane production and, hence, energy recovery. The HRAS showed higher COD removal efficiencies and better energy recovery. On the other hand, the HRSBR was better to avoid undesired nitrification and provided lower COD mineralization for all the SRTs tested (ranging 20–48% for the HRSBR, and 41–58% for the HRAS). Then, the energy as methane recovered per unit of COD degraded was higher in the HRSBR. The HRSBR seems to be a good option, because the solids content in the effluent was similar for both systems and its COD removal efficiency can be further improved by optimizing the SBR cycle configuration. 1. Introduction COD seems a conservative approach [7,8]. Therefore, the energy content of a typical influent with concentration 400 mg COD⋅L− 1 would be The activated sludge (AS) process is still the most commonly used around 1.7 kWh⋅m− 3, which is 3–5 times higher than the energy needed technology for urban wastewater treatment since its first description in for AS treatment. This value is even higher with less loaded wastewaters: 1914 [1]. Although this technology is robust and provides a good Shizas and Bagley estimated that energy content of raw domestic sewage effluent quality, it presents high operational costs and does not profit could exceed the electricity requirements to operate a conventional AS from the potential energy of the organic compounds present in sewage plant by a factor of nine [9]. Hence, these rough estimations reveal that [2]. Moreover, nitrogen and phosphorus must be also removed to avoid WWTPs should not only aim at reducing their energy needs, but to eutrophication and oxygen depletion in water bodies. Conventional AS become energy self-sufficient. Nowadays, the chemical energy con­ systems for nitrogen removal through nitrification/denitrification tained in the influent can only potentially be recovered through the require high sludge retention times (SRTs), around 10–20 days, to anaerobic digestion of the excess sludge. Alsayed et al. [10] estimated a ensure nitrification [3]. The major costs in these WWTPs are related to range of potential values of energy recovery according to the efficiency the energy consumption and sludge treatment: aeration can use about of the primary settler. They estimated that about 15–35% of the influent 60% of the total energy consumption [4], while sludge treatment and COD could be converted to methane and subsequently to electricity in a disposal can account for up to 40% [5]. Combined Heat and Power (CHP) equipment. This means that only However, these intensive energy requirements of conventional AS 5–12% of the energy contained in the influent could be recovered as systems (0.3–0.6 kWh⋅m− 3 [6]) are in contrast to the fact that waste­ electricity, assuming an electricity efficiency of CHP technologies of water is a valuable energy resource. Calculating the energy content of 35% [11]. urban wastewater is not straightforward, but a value around 15 kJ g− 1 Urban WWTPs are currently energy-demanding facilities and thus * Corresponding author. E-mail addresses: Natalia.Rey@uab.cat (N. Rey-Martínez), aloiabarreiro@gmail.com (A. Barreiro-López), Albert.Guisasola@uab.cat (A. Guisasola), JuanAntonio. Baeza@uab.cat (J.A. Baeza). https://doi.org/10.1016/j.biombioe.2021.106077 Received 15 November 2020; Received in revised form 24 March 2021; Accepted 18 April 2021 Available online 28 April 2021 0961-9534/© 2021 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). N. Rey-Martínez et al. Biomass and Bioenergy 149 (2021) 106077 novel approaches are needed for achieving a self-sustainable or even Table 1 energy-positive WWTPs [4]. The four main bottlenecks to address are Summary of wastewater characteristics for primary settler influent (PSI) and [10]: i) the amount of COD in the effluent (4–10% of the influent COD), primary settler effluent (PSE). ii) the excessive COD mineralization in the aeration tank (30–50% of the Parameter Influent average entering COD), iii) the limited anaerobic degradability or methane yield PSI PSE of the produced sludge (around 30% of the influent COD may not be TSS (g⋅L− 1) 0.18 ± 0.17 0.08 ± 0.06 completely digested to biogas), because long sludge retention times VSS (g⋅L− 1) 0.16 ± 0.15 0.07 ± 0.05 (SRTs) lead to poor digestibility and conventional AS including nutrient Total COD (mg⋅L− 1) 496 ± 214 359 ± 128 removal typically operate at long SRT ≥ 10 d. Finally, iv) the loss of COD Particulate COD (mg⋅L− 1) 239 ± 203 124 ± 93 as heat when using biogas to produce electricity using CHP technologies Soluble COD (mg⋅L− 1) 185 ± 54 157 ± 53 (10–23% of the influent COD). Thus, the utilisation of short-SRT systems Colloidal COD (mg⋅L− 1) 72 ± 26 76 ± 27 − 1 N–NH+ 4 (mg⋅L ) 76 ± 13 65 ± 21 with minimal mineralization and lower aeration needs seems a preferred P-PO−4 3 (mg L− 1) 9±4 9±3 option to boost energy recovery when removing organic matter from pH 7.6 ± 0.2 7.6 ± 0.3 urban wastewaters. The two-stage A/B process is a configuration targeting this goal [12] that has recently received more attention [13,14]. In the first step the entire operation. (known as A-stage), the objective is to maximize the capture of carbon through mostly physicochemical processes (adsorption, flocculation and 2.2. High-rate activated sludge (HRAS) reactor coagulation) and to divert this sludge towards an anaerobic digestion system for maximising biogas production [15]. A-stage improvements The HRAS system consisted of a continuously stirred aerobic reactor for increasing COD removal through biosorption have been recently (19 L) and a settler (27 L). It was operated at room temperature proposed [16], showing the applicability of these A/B processes. An (18–22 ◦ C) with an average influent flowrate of 165 L d− 1 resulting in an A-stage system should operate at very short SRTs and short hydraulic HRT of 2.8 h. Daily sludge wastage was performed to maintain the SRT retention times (HRTs) to minimize COD mineralization. The effluent of between 1 and 2 days. The external recycle from the settler to the reactor the first step is treated in the second step (stage B) through autotrophic was 200 L d− 1 resulting in an external recycle ratio Rext of 1.2. The HRAS biological nitrogen removal (BNR): partial nitritation and anaerobic system was inoculated with sludge collected from the Rubí WWTP and ammonium oxidation (anammox) processes. Autotrophic BNR results in was operated for 175 days treating PSE wastewater. The operation significant economic and energetic cost savings when compared to the characteristics of each period are shown in Table 2. The HRAS reactor conventional nitrification/denitrification process. The performance of was managed with an on-line system based on an Advantech PCI-1711 I/ the B-stage depends on the nature of the potential COD inlet and on the O card and an industrial PC running the AddControl software developed efficiency in solids removal at the A-stage. Moreover, the A-stage acts as in our research group. DO was measured with a HACH-CRI6050 DO a shock absorber of unexpected toxic loads that could be very detri­ probe and controlled around 1 mg L− 1 by manipulating an on/off mental for the B-stage [17]. aeration valve. pH and temperature were monitored with a pH probe The A-stage is usually designed as a continuous high rate activated (HACH CRI5335) and a thermoresistance (Axiomatic Pt1000), sludge (HRAS) system under short SRT and HRT. For example, Jimenez respectively. et al. [18] studied the effect of SRT, HRT and dissolved oxygen (DO) on a HRAS pilot plant performance, while Ge et al. [19] evaluated the impact 2.3. High-rate sequencing batch reactor (HRSBR) of SRT on nutrient removal using a HRAS pilot plant. A review of new alternatives for carbon redirection to improve the energy balance in The HRSBR was a stirred aerobic system with a working volume of WWTPs [20], demonstrated that HRAS is a good technology to redirect 3.5 L and an exchange volume of 2.5 L. It was inoculated with sludge carbon in WWTPs. As an alternative to the continuous HRAS, a collected from the same WWTP and the HRT was fixed at 2.5 h. Daily sequencing batch reactor (SBR) operating at similar HRT and SRT as the sludge wastage was performed to maintain an SRT between 1 and 2 HRAS seems a priori also a good option because SBRs have some ad­ days. The reactor was monitored for oxygen, pH and temperature and vantages compared with continuous systems: i) higher tolerance facing controlled with a PLC. The SBR was operated at a controlled tempera­ toxic or inhibitory compounds [21], ii) higher operation flexibility [22], ture of 25 ± 1 ◦ C. pH was not controlled, however its value was main­ iii) they retain sludge more effectively than continuous reactors because tained in the range 7.5–8.6 during the entire operation. DO was the supernatant is discharged only after the sludge has been settled [23] controlled between 5 and 6 mg L− 1 by manipulating an on/off aeration and iv) they can have lower installation costs and less space requirement valve during the start-up and, afterwards, the set point was decreased to because the process takes place in only one tank and thus avoids the 3 mg L− 1 and maintained at this value during the rest of the operation. settler. The operation (253 days) was divided into different periods ac­ Hence, the main objective of this work was to compare the perfor­ cording to the operational conditions and the cycle configuration for mance in terms of COD removal and energy recovery of a continuous each period is reported in Table 3. The HRSBR was fed with PSI during HRAS and a high-rate SBR (HRSBR) acting as an A-stage under the same periods I and II and with PSE for the rest of the periods. SRT and HRT conditions and fed with real urban wastewater. On the basis of the obtained experimental results, the possible integration of 2.4. Anaerobic digestion batch tests both systems with a B-stage was discussed. The biochemical methane potential (BMP) of the sludge purged in 2. Materials and methods both systems was studied through anaerobic batch digestion tests. BMP 2.1. Characteristics of the real wastewater Table 2 Operational conditions for each period of the HRAS. Two different real wastewaters were used in this work: primary Period Days (d) Carbon source Purge flow (L⋅d− 1) SRT (d) settler influent (PSI) and primary settler effluent (PSE), both of them from a municipal WWTP (Rubí, Catalonia). Fresh wastewater was I 0–24 PSE 5 2.5 ± 0.2 collected once a week and stored in a refrigerated tank at 5 ◦ C. Table 1 II 26–87 PSE 3.5 2.1 ± 0.6 III 89–175 PSE 2.5 1.2 ± 0.6 summarizes the most important parameters of these wastewaters during 2 N. Rey-Martínez et al. Biomass and Bioenergy 149 (2021) 106077 Table 3 Operational conditions for each period of the HRSBR. Period Days (d) Carbon Purge flow (L⋅d− 1) SRT (d) Cycle length Reaction phase Settling phase Discharge phase source (min) (min) (min) (min) I 0–30 PSI modified to obtain a constant 1.8 ± 106 75 30 1 SRT 0.9 II 58–105 PSI 0.576 1.2 ± 116 85 30 1 0.5 IIIA 107–191 PSE 0.576 2.2 ± 116 85 30 1 1.0 IIIB 194–231 PSE 0.576 1.3 ± 116 85 30 1 0.8 IV 236–253 PSE 0.993 1.0 ± 116 85 30 1 0.3 assays were performed in 160 mL glass bottles (125 mL of working column. Helium was the carrier gas at 338 kPa, and the oven, injector volume) with rubber stoppers and screw caps according to Angelidaki and detector temperatures were 70, 150 and 180 ◦ C respectively. A total et al. [24]. The inoculum used in the tests was obtained from the sample volume of 100 μL was used for chromatography. anaerobic digester of the Rubí WWTP and was pre-incubated at 37 ◦ C in The observed yield (Yobs) under different SRT conditions was order to deplete the residual biodegradable organic matter. All tests calculated using the experimental biomass production (PX, equation were performed by triplicate in an incubator at T = 37 ◦ C and manually (2)). mixing of the bottles once a day. A substrate to inoculum (S/I) ratio of 1 PX = Qp ⋅Xp + Qe ⋅Xe − Qin ⋅Xin (2) (on VS basis) with 2 gVS⋅L− 1 of inoculum was used. Negative control assays with only inoculum and MilliQ water were conducted in parallel Where Qp, Qe and Qin are the flowrates of purge, effluent and influent to measure the background BMP from the inoculum and this was sub­ (L⋅d− 1) and Xp, Xe and Xin are the VSS concentrations in the purge, in the tracted from the test prior to parameter estimation. Additionally, a effluent and in the influent (gVSS⋅L− 1). positive control with cellulose as substrate was performed to confirm the The experimental observed yield (Yobs, gCOD⋅g− 1COD) was calcu­ activity of the inoculum. The liquid phase and the headspace of the lated with equation (3). bottles were flushed with nitrogen to provide anaerobic conditions. A pressure transducer was used to measure the pressure increase. The PX ⋅1.416 Yobs = (3) biogas was sampled regularly and its composition was determined by Qin ⋅(CODin − CODout ) gas chromatography. The moles of methane were calculated by the ideal gas law (Equation (1)): where 1.416 is a stoichiometric factor converting VSS into COD (gCOD⋅g− 1VSS), Qin is the influent flowrate (L⋅d− 1) and CODin and PT ⋅xCH4 ⋅Vheadspace CODout are the COD concentrations in the influent and effluent respec­ molesCH4 = (1) R⋅(T + 273) tively (mgCOD⋅L− 1). Two different energy recovery indices were used, the methane re­ where PT is the total pressure measured by the transducer (mmHg), xCH4 covery index (MRI) and the energy recovered index (ERI). MRI was is the methane molar fraction, Vheadspace is the headspace volume (mL), calculated as the COD content of the methane recovered (g CODCH4) per R is the ideal gas constant (62,320 mmHg mL⋅mol− 1⋅K− 1) and T is the gram of COD removed (g CODREM) using equation (4). MRI is function of temperature (◦ C). Then, the corresponding volume of methane can be the mineralization degree in the reactor and the digestibility of the calculated and converted to standard temperature and pressure condi­ produced solids. This value was calculated as: tions (0 ◦ C and 1 atm). [ ] [ ] [ ] gCODCH4 gVSS NLCH4 gCODCH4 MRI = =Yobs ⋅BMP ⋅ConvFact gCODREM gCODREM gVSS NLCH4 2.5. Specific analytical methods and calculations (4) Samples for phosphorus and ammonium, nitrate and nitrite were where ConvFact was calculated as function of the T and P as 2.86 g filtered through 0.22 μm filters (Millipore). Phosphorus concentration CODCH4/ NL CH4 was measured with a phosphate analyser (PHOSPHAXsc, Hach Lange) ERI was calculated as the energy that could be obtained in the pro­ based on the vanadomolybdate yellow method. Ammonium nitrogen duced methane (kJCH4) per energy in the influent (kJINF) using equation was analysed with an ammonium analyser (AMTAXsc, Hach Lange), (5). ERI includes the efficiency of the CHP, the energy content of the based on the potentiometric determination of ammonia. Analyses of influent wastewater and the percentage of COD removed. The theoret­ nitrite and nitrate samples were performed with ionic chromatography ical energy yield of methane is 13.9 kJ g− 1CH4 [8] and a CHP efficiency (DIONEX ICS-2000). of 35% was assumed [11]. The energy content of the COD in the influent Influent and effluent COD fractions were measured according to wastewater was assumed as 15 kJ g− 1 COD, according to literature [7,8]. Jimenez et al. [18]. Particulate COD (CODP) was the difference between [ ] total COD (CODT) and the COD filtered through a 1.5 μm filter (CODF). kJ CH4 prod [g CH4]⋅Energy CH4 g CH4 ⋅CHPEfficiency The soluble COD (CODS) was the flocculated and filtrated COD (CODff) kJCH4 ERI = = [ ] (5) as defined by Mamais et al. [25]. Colloidal COD fraction (CODC) was kJINF kJ Influent load [gCODINF ]⋅Energy of Influent g COD defined as the difference between CODP and CODS. These COD fractions INF were measured using COD kits (Lovibond Vario LR and Vario MR) and a spectrophotomer (Photometer system MD100 Lovibond). Total sus­ 3. Results and discussion pended solids (TSS) and volatile suspended solids (VSS) were analysed according to APHA [26]. 3.1. Continuous HRAS system performance Biogas composition (CH4 and CO2) was analysed with a Hewlett Packard gas chromatograph (HP 5890) equipped with a thermal con­ The continuous HRAS system was operated for 175 days under two ductivity detector (TCD) and a Supelco Porapack Q (250 ◦ C) 3 m x 1/8” 3 N. Rey-Martínez et al. Biomass and Bioenergy 149 (2021) 106077 different SRTs and real PSE as feed. During the start-up (period I), the HRSBR operation. The effluent CODT was below 200 mgCOD⋅L− 1 during average CODS removal efficiency was 66 ± 6%. Fig. 1 shows the average periods I, II and IIIA, and it increased around 300 mgCOD⋅L− 1 for pe­ values of influent/effluent CODT and the sludge volume index (SVI) riods IIIB and IV. High effluent CODT is linked to the poor settleability of measured for periods II and III. The CODT in the effluent was not affected the sludge (SVI reached 300–500 mL⋅g− 1VSS). The fractionation of the by the reduction of the SRT from 2 d (period II) to 1 d (period III). In fact, effluent COD (Fig. 5) shows that during periods IIIB and IV the effluent the CODT removal efficiencies were 70 ± 16% and 65 ± 15% respec­ CODP increased. This fraction is strongly related to the sludge settling tively. The slight increase in the effluent CODT concentration from characteristics [28]. period II to period III was probably due to the 18% increase in the influent CODT (Fig. 1). 3.3. Comparison of effluent quality and carbon removal in both systems The settleability of the sludge in the HRAS reactor was never good as can be deduced from the high SVI values (Fig. 1), being 568 ± 329 mL g The effluent quality of both systems was evaluated in terms of solids VSS− 1 the average value during the entire operation. The settleability concentration and effluent COD fractions (Table 4 and Fig. 6 respec­ problems led to an extremely low solids concentration (Fig. 2): the VSS tively). The average effluent TSS in each reactor were similar but the reached values around 0.13 g L− 1 during period III. In fact, from day average SVI for the HRAS sludge (568 ± 329 mL g− 1 VSS) was much 140–168 of operation, the purge had to be turned off because of this low higher than the observed for the HRSBR sludge (206 ± 192 mL g− 1 VSS), VSS concentration. Settleability issues are very relevant when imple­ indicating that the former had worse settleability properties (Table 4). menting novel technologies at low SRT and, in this case, a settler with Fig. 6 shows the average values of the different COD fractions as higher surface than the conventional ones would be needed after the function of the SRT in both experimental systems. In the case of the continuous HRAS reactor. In this sense, Rahman et al. [27] proposed a HRAS reactor, the effluent CODT, CODF and CODP concentrations settler three times bigger than the reactor when operating A-stage and increased (23.5, 11.7 and 31.2% respectively) when SRT decreased from high-rate contact-stabilisation systems. Jimenez et al. [18] did not 2.1 d to 1.2 d. CODC had similar concentrations between 8 and 15 mg report settleability problems working with a biological reactor of 260 L COD⋅L− 1 in all cases and CODS was below 50 mg COD⋅L− 1 for both SRT. and a sedimentation tank of 280 L. Similar results of the effect of SRT on the effluent COD fractions was also observed by Rahman et al. [29] when comparing continuous systems 3.2. HRSBR performance treating raw wastewater. Regarding the HRSBR operation, effluent CODT, CODF and CODP The HRSBR was operated for 253 days under different conditions increased when SRT decreased, except for SRT = 1.2 d (period II). This (Table 3). Period I corresponds to the start-up of the reactor. The fact could be explained by the attached biomass on the walls of the objective of this period was operating with a constant SRT of 2 days, but reactor, which resulted in an effective SRT higher than that calculated this required a daily modification of the purge flow, which could lead to considering only planktonic biomass. The rest of the COD fractions were an unstable operation. Thus, it was decided to maintain a constant purge less affected by the SRT: CODS was in the range of 59.4–80.8 mgCOD⋅L− 1 flow in the subsequent periods II and III (0.576 L d− 1 and 0.993 L d− 1 in all cases whereas CODC had a minor role due to the low influent respectively) and, thus, the SRT fluctuated in a range of 1.2–2.2 days. concentrations (14.5–21.2% of influent CODT) (Table 1). Period III was divided into IIIA and IIIB because there was an incident COD removal efficiencies for both systems are shown in Table 5. The and the reactor had to be inoculated again with biomass from the pre­ HRAS system showed lower COD removal efficiency (from 70 to 65% vious purges, which had been stored at 4 ◦ C. Finally, the purge flow was CODT removal) when decreasing the SRT from 2.1 to 1.2 d. Higher maintained at 0.993 L d− 1 in period IV and the system operated at a SRT variations were observed for the HRSBR. If the results of periods I and II of 1.0 ± 0.3 days. The evolution of the solids in the HRSBR is presented are discarded due to the problems of biomass attached to the reactor in Fig. 3. It is worth mentioning that biomass attachment to the walls of walls, a clear decreasing trend of COD removal is observed when SRT the reactor was observed in periods I and II. Therefore, the decrease in decreases. For example, the CODT removal efficiency decreased from 54 the solids concentration during period II might be unreliable. The walls to 15% when the SRT decreases from 2.2 to 1 d. Thus, the lower the SRT, of the reactor were scratched once a week from period IIIA onwards in the higher the effluent CODT. Understanding the behaviour of the order to detach this biomass. Attached biomass can distort the total removal efficiencies of each individual COD fraction is not straightfor­ solids concentration in the reactor and consequently the real value of the ward since many non-controlled interactions happen (from hydrolysis to SRT, being higher than desired. The solids concentration increased aggregation). Moreover, the values of these fractions in the reactor are during period IIIA because of the detachment of the biomass grown on dynamic and, thus, some positive or negative accumulation of these the walls of the reactor, however TSS and VSS diminished again during fractions may occur. Removal efficiencies lower than 50% were ach­ periods IIIB and IV when the SRT was decreased from 2.2 ± 1.0 d (period ieved for CODP during the entire operation. In fact, negative CODP IIIA) to 1.3 ± 0.8 and 1.0 ± 0.3 d respectively. removal efficiencies were obtained when SVI increased and the solids Fig. 4 shows the evolution of the CODT and the average SVI in the content in the reactor was reduced (periods IIIB and IV corresponding to Fig. 1. HRAS operation. Average influent and effluent CODT and SVI in periods II and III. 4 N. Rey-Martínez et al. Biomass and Bioenergy 149 (2021) 106077 Fig. 2. HRAS operation: suspended solids in the reactor. (●) TSS and (○) VSS. Fig. 3. HRSBR operation: suspended solids in the reactor. (●) TSS and (○) VSS. Fig. 4. HRSBR operation: CODT in the influent and in the effluent and SVI. the lower SRT tested of 1 d). That led to a high effluent CODP value, The presence of either solids or organic matter in the influent of a B- which was transitorily higher than the influent CODP. A decrease in the stage can decrease its performance [17]. Thus, the effluent of the CODC removal efficiency during the same periods (Table 5) was also high-rate system should not contain a significant organic load to the observed. Therefore, the efficiency removal of CODP and CODC B-stage process. Otherwise, this organic matter would promote the decreased in parallel to the decrease in the SRT but with limited impact growth of heterotrophic bacterial population, which would compete on the CODS. Faust et al. [30] reported that a lower amount of extra­ with the autotrophic bacteria for the electron acceptor (i.e. mostly ni­ cellular polymeric substances was detected at lower SRTs. Thus, bio­ trite) resulting in a limitation of the autotrophic nitrogen removal [32, flocculation, the mechanism responsible for removing particulate and 33] and an increase of the aeration requirements. Considering this COD colloidal COD from wastewater [31], was unfavourable at low SRT removal criterion, the effluent of the HRAS reactor was more adequate values. Jimenez et al. [18] obtained a similar trend, the efficiencies for than that from the HRSBR, as the CODT removal efficiencies were sys­ CODP and CODC removal decreased from 70% to around 40% and 60% tematically higher in the former case. Moreover, a higher impact of SRT respectively when the SRT decreased from 2 to 1 day. on COD removal efficiencies was observed for the HRSBR (Table 5). The 5 N. Rey-Martínez et al. Biomass and Bioenergy 149 (2021) 106077 Fig. 5. HRSBR operation: Effluent COD fractionation: Total (●), particulate (○), colloidal (●) and soluble (Δ). (Fig. 7C), where values around 41–58% were observed. These values are Table 4 in the range obtained by Jimenez et al. [18] and Akanyeti et al. [2] at Total suspended solids concentration in the reactor and in the effluent for both SRT of 1 d. Moreover, the fraction of the inlet COD directed to the purge systems. decreased from 34% to 30% when the SRT decrease from 2 to 1 d. System TSS reactor (g⋅L− 1) TSS effluent (g⋅L− 1) SVI (mL⋅gVSS− 1) Previous works [2,18] reported carbon redirection improvement when HRAS lowering the SRT. Our observation (i.e. similar COD distribution inde­ I (start-up) 2.5 ± 0.2 0.059 ± 0.039 171 pendently of the SRT) could be attributed to the SRT range tested, al­ II 2.1 ± 0.6 0.060 ± 0.068 553 ± 198 ways higher than 1 day. In fact, Jimenez et al. [18] and Rahman et al. III 1.2 ± 0.6 0.075 ± 0.069 673 ± 448 [29] did not observe differences between the COD distribution for 1–2 HRSBR d SRT, and a clear difference was only observed at SRT below 0.5 d. Akanyeti et al. [2] also worked with SRT below 1 d (0.25, 0.5 and 1 d), I 1.8 ± 0.9 0.097 ± 0.048 95 ± 3 II 1.2 ± 0.5 0.134 ± 0.058 71 ± 8 obtaining a mineralization of 27%, 41% and 54%, respectively. IIIA 2.2 ± 1.0 0.070 ± 0.040 – Regarding the HRSBR, a decrease in the SRT implied an increase of IIIB 1.3 ± 0.8 0.081 ± 0.038 282 the COD redirected to the purge, up to 62% when the SRT was 1 d. IV 1.0 ± 0.3 0.097 ± 0.040 510 Consequently, the COD fraction in the effluent decreased considerably: the values obtained in this hypothetical scenario were in the range of periods with the same SRT for both systems, as for example when SRT 16–19%, whereas the values reached in the real mass balance were in was around 2 days (period II of the HRAS and period IIIA of the HRSBR), the range 42–74%. The COD redirected to purge obtained in this work at had similar CODS removal efficiency (60 vs. 57%). However, higher an SRT of 1 d (62%) was higher than the values observed in other studies CODP removal efficiency was obtained for the HRAS (70 vs. 25%), with continuous systems operating at even lower SRTs. For example, showing the higher capacity of separation of particulate compounds in Rahman et al. [27] observed percentages of COD in purge around 43% the continuous HRAS. only at SRT of 0.3 d. The extent of mineralization was lower in the HRSBR (20–48%) than in the continuous HRAS system (41–58%), with the highest difference for SRT = 1d: 20% for HRSBR and 58% for HRAS. 3.4. CODT mass balances Then, mineralization in the HRSBR was lower than that obtained by Jimenez et al. (2015) for the same SRT (60%). However, when the au­ CODT mass balances were performed at different SRTs for both sys­ thors worked at shorter SRTs (0.3–0.1 d) the mineralization decreased tems using the steady-state data (Fig. 7A and B). In addition, taking into until 20%, which is the value obtained for the HRSBR at SRT = 1d. As a account the floc formation limitations usually found in high-rate systems conclusion, the HRSBR allowed a better control of the extent of miner­ [34], which lead to high biomass concentration in the effluent, the COD alization process, which means that there was more COD in the purge mass balance distribution in a perfect settling scenario (no solids in the and available for redirection to anaerobic digestion. effluent) was also evaluated. Fig. 7C and D shows this scenario consid­ ering that all of the effluent biomass could be redirected to the anaerobic digestion. Mineralization was estimated as in Akanyeti et al. [2], i.e. the 3.5. Observed biomass growth yield (YOBS) COD fraction required to close the COD mass-balance. Fig. 7A and B shows the COD distribution for the continuous HRAS The values of YOBS calculated for each SRT are presented in Table 6. and for the HRSBR. The data of the mineralization of the COD and the YOBS seems to be higher at high SRT in the HRAS. The start-up (period I) COD distribution was very dependent on the effluent solids. These var­ was the period with the highest SRT value (2.5 d), and with highest YOBS iable concentrations of effluent solids can be attributed to a settling issue (1.05 ± 0.38 gVSS⋅g− 1COD). Values higher than one are not possible rather than a problem with the reactor configuration and, hence, the when considering only biological processes. However, higher YOBS discussion of Fig. 7C and D becomes more meaningful. values make sense if other non-biological processes such as adsorption The COD distribution obtained in the perfect settling scenario occur. There may be a fraction of COD, which is not degraded, that is (Fig. 7C and D) was estimated assuming that: i) effluent CODS was the adsorbed in the solids of the effluent, leading to the overestimation of only effluent COD fraction and ii) the experimental solids in the effluent VSS. Ge et al. [19] also observed the same behaviour operating a were considered as part of the purge and the COD content in these solids lab-scale HRAS system with YOBS values between 10 and 13 was estimated using the same procedure as in the purge and with the gVSS⋅g− 1COD for SRT of 2–3 d and 3–6 gVSS⋅g− 1COD for SRT of 0.5–1 stoichiometric factor 1.416 gCOD⋅g− 1VSS. d (Table 6). In contrast, in the HRSBR and at higher SRTs (2.2 d), the The extent of mineralization in the HRAS system was important YOBS (0.58 ± 0.34 gVSS⋅g− 1COD) was lower compared than at short SRT 6 N. Rey-Martínez et al. Biomass and Bioenergy 149 (2021) 106077 Fig. 6. Average COD fractionation in the effluent of A) HRAS and B) HRSBR. 3.6. Nutrient removal Table 5 COD removal efficiencies (%) for both high rate systems. Fig. 8 shows the nitrogen and phosphorus concentrations in the SRT (d) Period CODT CODF CODP CODS CODC influent and effluent of the continuous HRAS system and the HRSBR HRAS Start I – – – – – during the entire operation. Some nitrification was observed in the case up of continuous HRAS reactor (Fig. 8A): ammonia removal efficiencies of 2.1 ± II 70 ± 64 ± 5 70 ± 60 ± 7 77 ± 7 53 ± 11% and 47 ± 17% were obtained during periods I and II 0.6 16 32 1.2 ± III 65 ± 63 ± 66 ± 55 ± 71 ± respectively. However, from day 115 until day 145, ammonia removal 0.6 15 18 23 17 22 efficiency decreased down to 10 ± 5%. This ammonia removal could be attributed to assimilation for microbial growth rather than to nitrifica­ 1.8 ± I 60 ± 68 ± 5 38 ± 58 ± 72 ± tion. The amount of nitrogen incorporated into the biomass during the 0.9 10 36 27 17 cellular synthesis (Table 6) was estimated with mass balances, consid­ HRSBR 1.2 II 66 72 ± 8 44 ± 73 ± 7 67 ering the COD removal values, the sludge yield and the typical stoi­ ± ± ± 0.5 13 39 27 2.2 ± IIIA 54 ± 62 ± 9 25 ± 57 ± 57 ± chiometric factor of 0.086 g N⋅g− 1COD reported by Henze et al. [35] for 1.0 17 99 12 20 modelling studies. It can be inferred that 31% and 23% of the influent 1.3 ± IIIB 27 ± 50 ± 4 – 51 ± 42 ± nitrogen was removed due to heterotrophic growth during periods I and 0.8 19 10 15 II. The assimilative nitrogen uptake was higher in period I; however, 1.0 ± IV 15 ± 43 ± 5 – 47 ± 33 ± 0.3 27 10 18 ammonia assimilation in this case could be overestimated since it was considered that all the solids in the effluent were biomass and, as dis­ cussed above, there was an unaccounted fraction of adsorbed COD. The conditions (1 d) with YOBS values of 1.35 ± 0.54 gVSS⋅g− 1COD. This occurrence of nitrification in periods I and II at SRT values of 2.5 days effect of SRT on the YOBS was similar to the observed by Jimenez et al. and T = 20 ◦ C was unexpected. However, the proliferation of some ni­ [18] and Rahman et al. [29] (Table 6). It is worth mentioning that, in the trifiers as a biofilm grown on the reactor walls could explain the case of the HRSBR, the values of YOBS for periods I and II were not detection of some nitrification activity. On the other hand, nitrification considered due to the observation of biomass attachment to the walls of was not detected during period III (SRT = 1 d) onwards due to the pe­ the reactor. riodic scratching of the walls. Then, the percentage of nitrogen 7 N. Rey-Martínez et al. Biomass and Bioenergy 149 (2021) 106077 Fig. 7. Ouput COD distribution as a fraction of influent COD at different SRT conditions for A) HRAS and B) HRSBR. C) and D) correspond to HRAS and HRSBR in the perfect settling scenario. Effluent ( ), purge ( ) and mineralization ( ). Table 6 Observed yield as a function of SRT. System Period SRT (d) Yobs (gCOD⋅g− 1COD) Yobs (gVSS⋅g− 1COD) Ngrowth N-removal (%) (%) Ninfluent HRAS III 1.2 ± 0.6 0.53 ± 0.35 0.37 ± 0.25 22 ± 19 11 ± 4 (This study) II 2.1 ± 0.6 0.92 ± 0.80 0.65 ± 0.56 23 ± 16 19 ± 7 I 2.5 ± 0.2 1.48 ± 0.54 1.05 ± 0.38 31 ± 4 32 ± 9 HRSBR IV 1.0 ± 0.3 1.91 ± 0.77 1.35 ± 0.54 24 ±2 6±2 (This study) II 1.2 ± 0.5 20 ±6 20 ± 10 IIIB 1.3 ± 0.8 0.83 ± 0.43 0.59 ± 0.31 13 ±7 4±2 I 1.8 ± 0.9 21 ±9 22 ± 6 IIIA 2.2 ± 1.0 0.81 ± 0.49 0.58 ± 0.34 18 ± 10 6±3 HRAS 0.5 0.47 ± 0.11 [18] 1.0 0.38 ± 0.08 2.0 0.33 ± 0.10 CSTR 0.2 0.54 ± 0.06 [29] 0.8 0.47 ± 0.08 2.2 0.45 ± 0.13 HRAS 0.5–1 3–6 [19] 2–3 10–13 attributed to growth was higher than the nitrogen removal efficiency scale systems. (Table 6). Regarding phosphorus, the slight phosphate consumption observed In the case of the HRSBR, the average nitrogen removal was 21 ± 8% could be due to assimilation during microbial growth for both systems. for periods I and II, however in the subsequent periods this value P-removal efficiencies of 9.5–12% were obtained in HRAS whereas decreased down to 7 ± 1% (Fig. 8B). In this case, 18% of the influent higher values between 7 and 28% were obtained in the HRSBR (Fig. 8C nitrogen was removed via biomass assimilation or adsorption for the and D respectively). The oscillations in the influent concentration were higher SRT tested (2 d). A 24% of the nitrogen influent was removed by due to variations of the real wastewater. heterotrophs growth when SRT was 1 d (Table 6). Nitrification is neither expected nor desired in a high rate system, as only COD should be removed during this step. It was only observed in 3.7. Methane yield and hydrolysis constants the HRAS when the growth of a biofilm in the reactor walls was not prevented. The significance of the reactor walls is an inherent problem Anaerobic digestion assays of the sludge generated in both systems only for lab-scale reactors and it is not expected to be important at full- and in periods with different SRT were performed (Fig. 9 and Table 7). The highest BMP (307.8 ± 6.7 L CH4⋅kg− 1VS-added) was obtained for 8 N. Rey-Martínez et al. Biomass and Bioenergy 149 (2021) 106077 Fig. 8. Ammonium concentration in the influent (●) and in the effluent (○) A) for HRAS and B) for HRSBR. Nitrite ( ) and nitrate (Δ) concentrations in the effluent for HRAS. Phosphate concentration in the influent (●) and in the effluent (○) C) for HRAS and D) for HRSBR. hypothesis of anaerobic degradability of the sludge increase when decreasing the SRT [36]. Finally, the sludge generated in the HRAS at SRT = 1.2d had a BMP of 275.6 ± 7.4 L CH4⋅kg− 1VS-added (Table 7). Therefore, no quantitative differences were observed in terms of BMP between the two high rate systems. The BMP values obtained in this study were lower than those obtained by Ge et al. [19] (335.5 ± 8.7 and 357 ± 12.8 L CH4⋅kg− 1VS-added) for the same SRT interval (2-1 d). 3.8. Energy balance MRI and ERI were the two energy-based indices used to estimate the potential upgrade in energy recovery in our system. MRI provides the efficiency of the transformation of COD in the influent to methane, but without considering the COD removal efficiency, while ERI represents the fraction of energy that could be recovered from the influent. Considering, as detailed in the introduction of this work, that the energy content of wastewater is usually around 3–5 times higher than that 1 needed for its treatment, an ERI value in the range 0.2–0.3 kJCH4⋅kJ−INF would provide enough energy for WWTP autarky. Moreover, due to the assumed CHP conversion efficiency, the maximum theoretical ERI 1 would be limited to a maximum value of 0.35 kJCH4⋅kJ−INF . Fig. 10 compares the indices obtained in this work (period III of HRAS and periods IIIA and IIIB of the HRSBR) to those in Meerburg et al. [15] (marked with an asterisk) for different experimental systems: a Fig. 9. Cumulative BMP from mesophilic anaerobic digestion batch tests for conventional AS, two HRSBRs, a contact stabilisation (CS) system and different SRTs sludge. HRAS: 1.2 d ( ). HRSBR: 2 d (○), 0.8 d (●). two high-rate contact stabilisation (HRCS) systems. The idea beneath CS is to enhance sorption and storage process when removing organic matter. In short, in a CS system, part of the sludge of the external recycle Table 7 is aerated for its stabilisation. Then, this stabilised sludge is mixed for Ultimate methane production for the different sludge studied in this work. the influent under anoxic conditions and very short HRT to boost sub­ System SRT (d) Ultimate methane production strate sorption without high mineralization. Subjecting the sludge to (NL CH4⋅kg− 1VS-added) feast and famine phases increases the selective pressure to favour stor­ HRAS 1.2 275.6 ± 7.4 age and sorption processes. As can be observed in Fig. 10, the high-rate systems (i.e. those HRSBR 0.8 307.8 ± 6.7 operating at low SRTs) have at least three times higher MRI and ERI than 2.0 187.2 ± 4.2 those operating at higher SRTs. This agrees with the hypothesis of the high-rate systems that are designed to boost energy recovery from wastewaters. HRAS and HRSBR systems are in the ERI range of the lowest SRT (0.8 d) and the lowest BMP (187.2 ± 4.2 L CH4⋅kg− 1VS- 1 0.18–0.28 kJCH4⋅kJ−INF and hence seem able to provide WTTP autarky. Its added) was obtained for the sludge generated at SRT of 2 d (both from values are only outperformed by the operation of some HRCS systems, the HRSBR). This observation is in agreement with the reported but this technology also shows an important variability with lower 9 N. Rey-Martínez et al. Biomass and Bioenergy 149 (2021) 106077 Fig. 10. Energy recovery indices obtained in this work compared to those obtained in Meerburg et al. [15] (indicated with an asterisk) for different configurations. HRAS corresponds to data from period III of the HRAS, and HRSBR-1 and HRSBR-2 to data from periods IIIA and IIIB of the HRSBR. performance as observed for HRCS-1*. The differences observed be­ • COD mineralization was calculated for the perfect settling scenario, tween both HRCSs lay on the contact and stabilisation times. According obtaining values around 20–48% for the HRSBR, and 41–58% for the to the authors, these parameters should be further optimized together HRAS. However, the lower COD removal in the HRSBR led to lower with SRT and sludge separation. overall energy recovery for this reactor (ERI = 0.283 and 0.229, for When comparing continuous (HRAS) and SBR systems at similar HRAS and HRSBR, respectively). SRT, the results show that the latter have higher MRI and thus, a higher • Nitrification was more easily repressed in the HRSBR. This is an energy recovery is obtained per gram of COD removed. However, the important factor to guarantee a suitable ammonia influent concen­ ERI is lower. This is correlated to the abovementioned observation of a tration for the subsequent B-stage. higher percentage of COD removal in the HRAS. In fact, the ERI for the • Similar BMP values were obtained for sludge in both systems under HRAS (0.283) is not that different from those obtained in the HRCS similar SRT conditions, with values in the range of 187–308 NL (0.15–0.35), which show much higher MRI. Thus, HRSBR systems CH4⋅kg− 1VS. should focus on increasing the amount of COD removed in order to be • Similar solids content in the effluent was observed for both systems. more competitive in this aspect. Obtaining low COD content and solids presence in the effluent is an essential feature of an A-stage system, since they affect the perfor­ 3.9. Overall comparison mance of the B-stage. Biomass separation should be improved in high-rate systems to allow maximum redirection of COD to anaerobic A dilemma arises between the two high-rate systems tested in this digestion. work. On the one hand, better results were obtained with the HRAS in • Values of YOBS higher than 1 were obtained for both systems, indi­ terms of COD removal, ERI and solids concentration in the reactor. On cating that other non-biological processes (such as adsorption) the other hand, the HRSBR was better to avoid nitrification, provided occurred. Consequently, a fraction of COD was adsorbed in the solids better settleability and the extent of mineralization was lower, leading to of the effluent, leading to biomass concentration overestimation. a better MRI. However, the ERI was lower, but the COD removal effi­ ciency and thus, the ERI, could be improved by increasing the oxygen Acknowledgements transfer with better mixing conditions or by lengthening the aerobic phase. Therefore, the HRSBR seems to be a promising option, although This work was supported by the Spanish Ministerio de Economía y an optimization of the cycle configuration is recommended to demon­ Competitividad (CTQ2017-82404-R) with funds from the Fondo strate the feasibility and reliability of the HRSBR as an A-stage in an A/B Europeo de Desarrollo Regional (FEDER). Natalia Rey is grateful for the configuration. PIF PhD grant funded by the Universitat Autònoma de Barcelona. The authors are members of the GENOCOV research group (Grup de Recerca 4. Conclusions Consolidat de la Generalitat de Catalunya, 2017SGR 1175, www.ge nocov.com). A continuous HRAS was compared to a HRSBR under different operational conditions and the following conclusions were drawn: References • The continuous HRAS system provided higher COD removal effi­ [1] M.C.M. van Loosdrecht, D. Brdjanovic, Anticipating the next century of wastewater treatment, Science (80-.) 344 (2014) 1452–1453, https://doi.org/10.1126/ ciencies for all fractions, but the operation led to poorer sludge science.1255183. settleability. 10 N. Rey-Martínez et al. Biomass and Bioenergy 149 (2021) 106077 [2] I. Akanyeti, H. Temmink, M. Remy, A. Zwijnenburg, Feasibility of bioflocculation wastewater, Bioresour. Technol. 218 (2016) 146–152, https://doi.org/10.1016/j. in a high-loaded membrane bioreactor for improved energy recovery from sewage, biortech.2016.06.055. Water Sci. Technol. 61 (2010) 1433–1439, https://doi.org/10.2166/wst.2010.032. [22] M.C. Tomei, J. Soria Pascual, D. Mosca Angelucci, Analysing performance of real [3] B. Kartal, J.G. Kuenen, M.C.M. Van Loosdrecht, Sewage treatment with anammox, textile wastewater bio-decolourization under different reaction environments, Science 328 (2010) 702–703, https://doi.org/10.1126/science.1185941, 80. J. Clean. Prod. 129 (2016) 468–477, https://doi.org/10.1016/j. [4] Y. Gu, Y. Li, X. Li, P. Luo, H. Wang, Z.P. Robinson, X. Wang, J. Wu, F. Li, The jclepro.2016.04.028. feasibility and challenges of energy self-sufficient wastewater treatment plants, [23] M. Sekine, S. Akizuki, M. Kishi, T. Toda, Stable nitrification under sulfide supply in Appl. Energy 204 (2017) 1463–1475, https://doi.org/10.1016/j. a sequencing batch reactor with a long fill period, J. Water Process Eng. 25 (2018) apenergy.2017.02.069. 190–194, https://doi.org/10.1016/j.jwpe.2018.05.012. [5] W. Verstraete, S.E. Vlaeminck, ZeroWasteWater: short-cycling of wastewater [24] I. Angelidaki, M. Alves, D. Bolzonella, L. Borzacconi, J.L. Campos, A.J. Guwy, resources for sustainable cities of the future, Int. J. Sustain. Dev. World Ecol. 18 S. Kalyuzhnyi, P. Jenicek, J.B. van Lier, Defining the biomethane potential (BMP) (2011) 253–264, https://doi.org/10.1080/13504509.2011.570804. of solid organic wastes and energy crops: a proposed protocol for batch assays, [6] J. Wan, J. Gu, Q. Zhao, Y. Liu, COD capture: a feasible option towards energy self- Water Sci. Technol. 59 (2009) 927–934, https://doi.org/10.2166/wst.2009.040. sufficient domestic wastewater treatment, Sci. Rep. 6 (2016) 1–9, https://doi.org/ [25] D. Mamais, D. Jenkins, P. Prrr, A rapid physical-chemical method for the 10.1038/srep25054. determination of readily biodegradable soluble COD in municipal wastewater, [7] B. Korth, T. Maskow, S. Günther, F. Harnisch, Estimating the energy content of Water Res. 27 (1993) 195–197, https://doi.org/10.1016/0043-1354(93)90211-Y. wastewater using combustion calorimetry and different drying processes, Front. [26] APHA, Standard Methods for the Examination of Water and Wastewater, 1995, Energy Res 5 (2017) 23. https://www.frontiersin.org/article/10.3389/fenrg.2 19th Editi. 017.00023. [27] A. Rahman, H. De Clippeleir, W. Thomas, J.A. Jimenez, B. Wett, A. Al-Omari, [8] E.S. Heidrich, T.P. Curtis, J. Dolfing, Determination of the Internal Chemical S. Murthy, R. Riffat, C. Bott, A-stage and high-rate contact-stabilization Energy of Wastewater 45 (2011) 827–832, https://doi.org/10.1021/es103058w. performance comparison for carbon and nutrient redirection from high-strength [9] I. Shizas, D.M. Bagley, M. Asce, Experimental Determination of Energy Content of municipal wastewater, Chem. Eng. J. 357 (2019) 737–749, https://doi.org/ Unknown Organics in Municipal Wastewater Streams, vol. 130, 2005. 10.1016/j.cej.2018.09.206. [10] A. AlSayed, M. Soliman, A. Eldyasti, Anaerobic-based water resources recovery [28] W. Rössle, W. Pretorius, A review of characterisation requirements for in-line facilities: a review, Energies 13 (2020), https://doi.org/10.3390/en13143662. prefermenters: paper 1: wastewater characterisation, WaterSA 27 (2001) 405–412, [11] P.L. McCarty, J. Bae, J. Kim, Domestic wastewater treatment as a net energy https://doi.org/10.4314/wsa.v27i3.4985. producer–can this be achieved? Environ. Sci. Technol. 45 (2011) 7100–7106, [29] A. Rahman, F.A. Meerburg, S. Ravadagundhi, B. Wett, J. Jimenez, C. Bott, A. Al- https://doi.org/10.1021/es2014264. Omari, R. Riffat, S. Murthy, H. De Clippeleir, Bioflocculation management through [12] A.I. Versprille, B. Zuurveen, T. Stein, The A–B process: a novel two stage high-rate contact-stabilization: a promising technology to recover organic carbon wastewater treatment system, Water Sci. Technol. 17 (1985) 235–246. from low-strength wastewater, Water Res. 104 (2016) 485–496, https://doi.org/ [13] Y.J. Liu, J. Gu, Y. Liu, Energy self-sufficient biological municipal wastewater 10.1016/j.watres.2016.08.047. reclamation: present status, challenges and solutions forward, Bioresour. Technol. [30] L. Faust, H. Temmink, A. Zwijnenburg, A.J.B. Kemperman, H.H.M. Rijnaarts, High 269 (2018) 513–519, https://doi.org/10.1016/j.biortech.2018.08.104. loaded MBRs for organic matter recovery fromsewage: effect of solids retention [14] G. Xu, H. Wang, J. Gu, N. Shen, Z. Qiu, Y. Zhou, Y. Liu, A novel A-B process for time on bioflocculation and on the role of extracellular polymers, Water Res. 56 enhanced biological nutrient removal in municipal wastewater reclamation, (2014) 258–266, https://doi.org/10.1016/j.watres.2014.03.006. Chemosphere 189 (2017) 39–45, https://doi.org/10.1016/j. [31] A. Rahman, T. Wadhawan, E. Khan, R. Riffat, I. Takács, H. De Clippeleir, B. Wett, J. chemosphere.2017.09.049. A. Jimenez, A. Al-Omari, S. Murthy, Characterizing and quantifying flocculated [15] F.A. Meerburg, N. Boon, T. Van Winckel, J.A.R. Vercamer, I. Nopens, S. and adsorbed chemical oxygen demand fractions in high-rate processes, Glob. E. Vlaeminck, Toward energy-neutral wastewater treatment: a high-rate contact Challenges Sustain. Wastewater Treat. Resour. Recover. IWA Spec. Conf. (2014), stabilization process to maximally recover sewage organics, Bioresour. Technol. https://doi.org/10.1103/PhysRevB.91.155112. 179 (2015) 373–381, https://doi.org/10.1016/j.biortech.2014.12.018. [32] M.W. Miller, J. De Armond, M. Elliot, M. Kinyua, D. Kinnear, B. Wett, S. Murthy, C. [16] B. Wett, P. Aichinger, M. Hell, M. Andersen, L. Wellym, Y. Fukuzaki, Y.S. Cao, B. Bott, Settling and dewatering characteristics of an A-stage activated sludge G. Tao, J. Jimenez, I. Takacs, C. Bott, S. Murthy, Operational and structural A-stage process proceeded by shortcut biological nitrogen removal, Int. J. Water improvements for high-rate carbon removal, Water Environ. Res. (2020) 1354, Wastewater Treat. 2 (2016), https://doi.org/10.16966/2381-5299.133. https://doi.org/10.1002/wer.1354, wer. [33] M.S.I. Mozumder, C. Picioreanu, M.C.M. Van Loosdrecht, E.I.P. Volcke, Effect of [17] J. Smitshuijzen, J. Pérez, O. Duin, M.C.M. van Loosdrecht, A simple model to heterotrophic growth on autotrophic nitrogen removal in a granular sludge describe the performance of highly-loaded aerobic COD removal reactors, reactor, Environ. Technol. 35 (2014) 1027–1037, https://doi.org/10.1080/ Biochem. Eng. J. 112 (2016) 94–102, https://doi.org/10.1016/j.bej.2016.04.004. 09593330.2013.859711. [18] J. Jimenez, M. Miller, C. Bott, S. Murthy, H. De Clippeleir, B. Wett, High-rate [34] T. Van Winckel, X. Liu, S.E. Vlaeminck, I. Takács, A. Al-Omari, B. Sturm, B. activated sludge system for carbon management - evaluation of crucial process V. Kjellerup, S.N. Murthy, H. De Clippeleir, Overcoming floc formation limitations mechanisms and design parameters, Water Res. 87 (2015) 476–482, https://doi. in high-rate activated sludge systems, Chemosphere 215 (2019) 342–352, https:// org/10.1016/j.watres.2015.07.032. doi.org/10.1016/j.chemosphere.2018.09.169. [19] H. Ge, D.J. Batstone, M. Mouiche, S. Hu, J. Keller, Nutrient removal and energy [35] M. Henze, C.P.L. Grady, W. Gujer, G.v.R. Marais, T. Matsuo, Activated sludge recovery from high-rate activated sludge processes – impact of sludge age, model No . 1, IAWPRC Sci. c Tech. Reports. 1223 (1987) 19–21, https://doi.org/ Bioresour. Technol. 245 (2017) 1155–1161, https://doi.org/10.1016/j. 10.1016/S0273-1223(98)00785-9. biortech.2017.08.115. [36] H. Ge, D.J. Batstone, J. Keller, Operating aerobic wastewater treatment at very [20] I. Sancho, S. Lopez-Palau, N. Arespacochaga, J.L. Cortina, New concepts on carbon short sludge ages enables treatment and energy recovery through anaerobic sludge redirection in wastewater treatment plants: a review, Sci. Total Environ. 647 digestion, Water Res. 47 (2013) 6546–6557, https://doi.org/10.1016/j. (2019) 1373–1384, https://doi.org/10.1016/j.scitotenv.2018.08.070. watres.2013.08.017. [21] Y. Jiang, L. Wei, H. Zhang, K. Yang, H. Wang, Removal performance and microbial communities in a sequencing batch reactor treating hypersaline phenol-laden 11