Water Research 172 (2020) 115474 Contents lists available at ScienceDirect Water Research journal homepage: www.elsevier.com/locate/watres Evaluation of the integration of P recovery, polyhydroxyalkanoate production and short cut nitrogen removal in a mainstream wastewater treatment process Oriol Larriba, Eric Rovira-Cal, Zivko Juznic-Zonta, Albert Guisasola*, Juan Antonio Baeza gica i Ambiental. Escola d’Enginyeria. Universitat Auto GENOCOV. Departament d’Enginyeria Química, Biolo noma de Barcelona, 08193, Bellaterra, Barcelona, Spain a r t i c l e i n f o a b s t r a c t Article history: Wastewater treatment systems are nowadays evolving into systems where energy and resources are Received 1 August 2019 recovered from wastewater. This work presents the long term operation of a demo-scale pilot plant Received in revised form (7.8 m3) with a novel configuration named as mainstream SCEPPHAR (ShortCut Enhanced Phosphorus 3 January 2020 and polyhydroxyalkanoate (PHA) Recovery) and based on two sequencing batch reactors (R1-HET and Accepted 4 January 2020 Available online 7 January 2020 R2-AUT). This is the first report of an implementation at demo scale and under relevant operational conditions of the simultaneous integration of shortcut nitrification, P recovery and production of sludge with a higher PHA content than conventional activated sludge. An operating period under full nitrifi- Keywords: EBPR cation mode achieved successful removal efficiencies for total N, P and CODT (86 ± 12%, 93 ± 9% and nitrite 79 ± 6%). In the following period, nitrite shortcut (with undetectable activity of nitrite oxidising bacteria) Polyhydroxyalkanoate (PHA) was achieved by implementing automatic control of the aerobic phase length in R2-AUT using ammo- Struvite nium measurement and operating at a lower sludge retention time. Similar N, P and CODT removal ef- Recovery ficiencies to the full nitrification period were obtained. P-recovery from the anaerobic supernatant of R1- HET was achieved in a separate precipitator by increasing pH and dosing MgCl2, recovering an average value of 45% of the P in the influent as struvite precipitate, with a peak up to 63%. These values are much higher than the typical values of sidestream P-recovery (12%). Regarding PHA, a percentage in the biomass in the range 6.9e9.2% (gPHA$g1TSS) was obtained. © 2020 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/). 1. Introduction and, thus, for securing food supply. However, the main source of P, the phosphate rock, shows a faster rate of use relative to its for- The water sector has lately changed the paradigm from waste- mation and, this is why it can be considered as a non-renewable water treatment to resource recovery. Hence, wastewater treat- source which, depending on the assumptions taken, is envisaged ment plants (WWTPs) are nowadays known as water resource to be depleted in the next 50e300 years (Ahmed et al., 2015; Cieslik recovery facilities (WRRFs). The major aim of these facilities is to go and Konieczka, 2017; Cordell et al., 2011; Theregowda et al., 2019). one-step beyond wastewater treatment and is related to the re- It is estimated that 3 million tons of P are removed yearly via covery of energy (for example as biogas, biomethane, or hydrogen) wastewater treatment in the planet (which represents 15e20% of or water and material resources from wastewater (for example as the global P demand) and, therefore, implementing efficient P-re- NeP-based fertilisers, bioplastics or cellulose). In this frame, novel covery strategies would mitigate the dependence on the phosphate WRRFs configurations have been put forward recently. rocks (Mayer et al., 2016) Phosphorus arises as a perfect candidate in this paradigm shift Enhanced biological phosphorus removal (EBPR) is based on the since it is essential for our society in the production of fertilizers enrichment of the microbial community in polyphosphate accu- mulating organisms (PAO), which have a metabolism capable of accumulating P intracellularly as polyphosphate (Poly-P) when * Corresponding author. alternatively exposed to anaerobic/aerobic (or anoxic) conditions. E-mail addresses: Oriol.Larriba@uab.cat (O. Larriba), Eric.Rovira@uab.cat The fraction of PAO able to accumulate P under anoxic conditions (E. Rovira-Cal), Zivko.JuznicZonta@uab.cat (Z. Juznic-Zonta), albert.guisasola@uab. cat (A. Guisasola), JuanAntonio.Baeza@uab.cat (J.A. Baeza). with nitrate/nitrite as electron acceptor are called denitrifying PAO https://doi.org/10.1016/j.watres.2020.115474 0043-1354/© 2020 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/). 2 O. Larriba et al. / Water Research 172 (2020) 115474 Abbreviations PHV polyhydroxyvalerate PH2MV polyhydroxy-2-methylvalerate AOB ammonia oxidising bacteria Poly-P polyphosphate A2/O anaerobic/anoxic/oxic WWTP configuration PREL phosphorus release during the anaerobic phase for bAOB apparent decay rate of AOB PAO BCFS biologicalechemical phosphorus and nitrogen P/C P-release/C-consumption ratio under anaerobic removal process conditions for PAO BMP biomethane potential rbCOD readily biodegradable COD bNOB apparent decay rate of NOB R1-HET First SBR of the pilot plant, where mostly BOD biochemical oxygen demand heterotrophic processes occur CAPEX capital expenditures R2-AUT Second SBR of the pilot plant, where mostly COD chemical oxygen demand autotrophic processes occur CODCON COD consumption in the anaerobic phase for PAO R3-PRE Pilot plant reactor for struvite precipitation CODS soluble COD R4-INT Pilot plant vessel for supernatant exchange between CODT total COD R1-HET and R2-AUT DO dissolved oxygen SBR sequencing batch reactor DPAO denitrifying PAO SCEPPHAR Shortcut enhanced phosphorus and EBPR enhanced biological phosphorus removal polyhydroxyalkanoate recovery FNA free nitrous acid SRT sludge retention time GC gas chromatography TSS total suspended solids HRT hydraulic residence time UCT University of Cape Town WWTP configuration IRR internal rate of return V volume ISE ion-selective-electrode VER volume exchange ratio N/D nitrification/denitrification VFA volatile fatty acids NOB nitrite oxidising bacteria VSS volatile suspended solids OPEX operational expenditures WRRF water resource recovery facility PAO polyphosphate accumulating organisms WWTP wastewater treatment plant PHA polyhydroxyalkanoate mAOB apparent growth rate of AOB PHB polyhydroxybutyrate mNOB apparent growth rate of NOB (DPAO). Thus, the influent P is accumulated in the biomass purged volatile fatty acids (VFA) taken up. This PHA would be used as en- at the end of the aerobic stage, when the sludge has the maximum ergy and carbon source in a posterior phase under aerobic/anoxic amount of Poly-P accumulated. Purging biomass with high P con- conditions. Purging at the end of the anaerobic phase results in tent may have some practical drawbacks since undesired struvite biomass with the highest PHA content of the whole operation, precipitation that clogs pumps and pipes can occur when bio-P whereas the conventional aerobic purging results in a purge with sludge is subjected to anaerobic digestion. Decreasing P entrance low PHA content. Sludge rich in PHA shows two potential advan- to the anaerobic digester (for instance, implementing P-recovery) tages: i) it can be used as a precursor for bioplastics production should limit the extent of this uncontrolled precipitation. In fact, after its extraction and ii) if sent to the anaerobic digester, this struvite recovery has many benefits for the WRRFs, not only for its sludge would have a higher potential methane production due to commercial value but for the reduction of pipe blockage, sludge its higher PHA content (Chan et al., 2020; Huda et al., 2016; Wang production, ferric chloride dosage and operation and maintenance et al., 2016). costs of the plant (Lizarralde et al., 2019). However, the efficiency of Regarding N removal, the nitrite pathway approach, i.e. nitri- P recovery from the anaerobic digestate is only about 12% of the P tation (NHþ   4 / NO2 ) followed by denitritation (NO2 / N2), has entering in the influent (Remy and Jossa, 2015) and therefore al- been reported to have advantages when compared to conventional ternatives must be found to achieve a more significant percentage nitrification and denitrification processes (Türk and Mavinic, 1987): of P recovery. i) 25% lower oxygen consumption for N oxidation, thus reducing An option to increase the amount of P recovered and to avoid aeration costs and also lowering the carbon footprint of a WWTP, ii) undesired struvite precipitation is the implementation of main- up to 40% lower COD requirement for denitrification and iii) faster stream P-recovery based on EBPR activity. Some authors have denitrification rate (1.5e2 times), thus requiring smaller anoxic already shown that a significant P extraction from the anaerobic basin. The nitrite pathway has already been reported with low phase of EBPR systems can be maintained at long-term without any strength municipal wastewater (Guo et al., 2009; Ma et al., 2009; deleterious effect on PAO activity (Guisasola et al., 2019). Some Yang et al., 2007), and also integrated with EBPR (Marcelino et al., examples are the BCFS® process (Biologicalechemical phosphorus 2011). Moreover, nitrite has been also reported as a good electron and nitrogen removal) (Meijer et al., 2001; van Loosdrecht et al., acceptor for DPAO (Taya  et al., 2013). 1998) or other reports (Baeza et al., 2017; Kodera et al., 2013; Shi Nitritation, the first step of the nitrite pathway, is achieved by et al., 2012; Valverde-Pe rez et al., 2015). A common conclusion of creating conditions where nitrite oxidising bacteria (NOB) are these works is that up to 60% of the influent P could be recovered at eliminated from the system while ammonia oxidising bacteria the end of the anaerobic phase under well-controlled conditions. (AOB) are retained. Mathematically, equation (1) should be satisfied Anaerobic extraction of the supernatant can lead to another in order to eliminate NOB while retaining AOB: advantage in terms of resource recovery, since it opens the door to anaerobic purging of EBPR sludge. PAO accumulate poly- mNOB - bNOB < SRT1 < mAOB - bAOB (1) hydroxyalkanoate (PHA) under anaerobic conditions from the O. Larriba et al. / Water Research 172 (2020) 115474 3 where mNOB and mAOB are the apparent growth rates of NOB and settling, the supernatant of the reactor, which is rich in phosphates AOB, respectively, bNOB and bAOB the apparent decay rates of these and ammonia, is sent to R4-INT. Then, 150 L of R4-INT are sent to populations, and SRT the sludge retention time. Most of the stra- R3-PRE for P precipitation as struvite. The rest of content of R4-INT tegies enabling the nitrite pathway involve manipulating mNOB and is transferred to R2-AUT, where ammonia is oxidised to nitrate or mAOB such that equation (1) is satisfied. In most cases, it is inevitably nitrite (depending on the period). After settling in R2-AUT, the to reduce both mNOB and mAOB but incurring stronger reductions in supernatant is returned to R1-HET for an anoxic phase. During this mNOB. A common strategy reported in the literature to obtain NOB R1-HET phase, denitrifying PAO (DPAO) perform P-uptake using washout is to decrease the dissolved oxygen (DO) setpoint (Aslan nitrite/nitrate as electron acceptor. Finally, an aerobic phase is et al., 2009; Blackburne et al., 2008; Jiang et al., 2019; Jianlong required to complete P-uptake. After settling, the supernatant is and Ning, 2004; Pollice et al., 2002) since it is widely accepted discharged to the effluent and the cycle starts again. The complete that AOB have higher affinity with oxygen than NOB. The drawback sequence of the pilot-plant when operating with a cycle length of of operating at low DO value is that higher N2O emissions may 12 h is detailed in Table 3, while the 8 h cycle used during part of occur (Massara et al., 2018, 2017). Period I is reported in the supplementary information (Table S1). Hence, this work proposes a novel WRRF configuration named as mainstream SCEPPHAR (ShortCut Enhanced Phosphorus and 2.2. Pilot plant monitoring and control system PHA Recovery), which is one of the novel technologies involved in the SMART-Plant project (www.smart-plant.eu). The whole R1-HET, R2-AUT and R4-INT had on-line monitoring of level SMART-Plant project aims to prove the feasibility of novel waste- (Endress Hauser), which was essential for the automation of the water treatment technologies at demo pilot-scale towards a cir- interexchange steps between the reactors. R1-HET and R2-AUT cular economy scenario, not only to treat wastewater, but also to were monitored with DO (Hach Lange), temperature (Pt1000, recover resources. SCEPPHAR configuration redesigns the two- Axiomatic), pH and oxidation reduction potential (Hach Lange). R2- sludge system proposed by Marcelino et al. (2011) to include in AUT had also an on-line ion-selective-electrode (ISE) for ammonia/ the mainstream resource recovery (struvite and PHA-rich sludge). nitrate (Hach Lange) for the automation of the aeration phase The system is based on two sequential batch reactors (SBR): one control length. R3-PRE was monitored with a pH sensor (Hach mainly heterotrophic (R1-HET) performing EBPR under alternating Lange). Aeration flow of the reaction vessels was monitored with anaerobic/anoxic/aerobic conditions and another mainly autotro- gas rotameters (Iberfluid). R1-HET, R2-AUT and R3-PRE were stirred phic (R2-AUT) in charge of ammonia oxidation to nitrite (complete (Milton Roy Mixing) during the reaction phases. Centrifugal pumps nitritation). (Grundfos) were used for water interchange, and dosing pumps This work presents the critical assessment for long-term oper- (Seko) for sludge purge and dosage of Mg and acetic acid. All the ations of a demo-scale mainstream SCEPPHAR pilot plant treating mechanical equipment and monitoring systems were connected to real wastewater from a municipal WWTP, showing the resource an industrial computer (PPC-3170, Advantech) through a data recovery possibilities of this novel configuration. To the best of our acquisition system (PCI-1711 I/O card, Advantech). The software knowledge, this is the first report in the literature of an imple- Addcontrol developed by the research group was used for auto- mentation at demo scale and under relevant operational conditions mating all the operation, monitoring and control. DO was of the integration in a mainstream configuration of nitrite pathway, controlled by manipulating the aeration flow-rate through electric P-recovery and production of sludge with increased PHA content. control valves (Type 3241/3374, Samson Instruments) based on the DO measurement and a proportional-integral algorithm pro- 2. Materials and methods grammed in the control system. DO setpoint in the aerobic phases of R1-HET and R2-AUT was always maintained at 3 mg L1. 2.1. SCEPPHAR pilot plant configuration 2.3. Chemical analysis The demo-scale pilot plant was located in the municipal WWTP of Manresa (Barcelona, Spain). Fig. 1 shows the process diagram of All the samples for the determination of soluble compounds the pilot plant, which includes: i) R1-HET anaerobic/anoxic/aerobic were filtered with a 0.22 mm filter (Millipore) before the analysis. P SBR (V ¼ 2500L) for heterotrophic processes, ii) R2-AUT aerobic SBR was measured with a phosphate analyser (115 VAC PHOSPHAX sc, (V ¼ 2500L) for autotrophic nitrification, iii) R3-PRE reactor Hach-Lange) based on the Vanadomolybdate yellow method sam- (V ¼ 150L) for struvite precipitation, and iv) R4-INT vessel ples. Ammonium nitrogen was analysed with an ammonium ana- (V ¼ 2500L) for interchanging supernatants between R1-HET and lyser (AMTAXsc, Hach Lange), based on the potentiometric R2-AUT. The pilot plant operated with 70% of volume exchange determination of ammonia. Nitrite and nitrate were analysed with ratio (VER), i.e. 1750 L of wastewater were treated per cycle. Table 1 Ion Chromatography (DIONEX ICS-2000). summarizes the main operational parameters for the three periods Volatile suspended solids (VSS) and total suspended solids (TSS) reported: start-up, complete nitrification and nitrite pathway. Hy- in mixed liquor were analysed according to Standard Methods draulic residence time (HRT) is calculated considering both R1-HET (APHA, 1995). COD was analysed by using Lovibond kits (COD Vario and R2-AUT volume, and the independent SRT fixed for each Tube Test LR and COD Vario Tube Test MR) and the MD100 Lobivond reactor is also included. spectrophotometer. Soluble COD (CODS) was measured with sam- R1-HET and R2-AUT were inoculated with activated sludge from ples filtered through 0.22 mm filters, while total COD (CODT) was the Manresa WWTP (anoxic/aerobic configuration with N/D but no not filtered. The PHA content in the biomass was measured as EBPR) and R1-HET was bioaugmented with PAO-enriched sludge follows: 0.6 mL of formaldehyde was added to each sample to stop from an EBPR pilot plant (Rey-Martínez et al., 2019). The waste- the biological reactions and then the sample was lyophilised. The water used was the effluent of the primary settler of the Manresa PHA was extracted from lyophilised samples using hexane and WWTP with the average composition shown in Table 2. butanol according to the method of Werker et al. (2008). Subse- Each of the cycles follows the next operational sequence: the quently, the PHA was determined with a GC (Agilent Technologies process starts with the influent feeding to R1-HET. Then, an 7820A) as described in Montiel-Jarillo et al. (2017). The standards anaerobic phase takes place and PAO uptake organic matter, release used were 3-hydroxybutyric acid and 3-hydroxyvaleric acid P, and accumulate PHA. An anaerobic purge is performed and, after copolymer for polyhydroxybutyrate (PHB) and 4 O. Larriba et al. / Water Research 172 (2020) 115474 Fig. 1. Schematics of the SCEPPHAR configuration as implemented in the demosize pilot-plant. Table 1 Main operational parameters used for each period. Period Days Cycle length (h) HRT (d) SRT (d) SRT (d) Aerobic length control R1-HET R2-AUT I Start-up 8e12 1e1.5 15 15 N II Complete nitrification 0e225 12 1.5 10 12 N III Nitrite shortcut 291e439 12 1.5 10 5e7 Y Table 2 WTW), were connected to a multiparametric device (INOLAB 3, Average composition of the influent of the pilot plant (effluent of the primary settler WTW) which in turn was connected via RS232 to a computer with of the full-scale WWTP of Manresa). an own data acquisition software developed in Visual Basic. Compound Concentration Units In every experiment, 40 mL samples were extracted every PO3- 4 -P 4.2 ± 1.1 mgP$L1 30min for the analytical tests. The samples were processed with a NHþ 4 -N 40 ± 11 mgN$L1 centrifuge (Beckman Coulter, Avanti J-20 XP) and filtered through NO 2 -N 0.1 ± 0.1 mgN$L1 0.22 mm pore diameter filters (Millipore). NO 3 -N 1.3 ± 0.7 mgN$L1 CODT 228 ± 81 mgCOD$L1 CODS 136 ± 39 mgCOD$L1 2.4.2. PAO/DPAO activity batch experiments  Temperature 20 ± 6 C The PAO/DPAO activity tests were comprised of three sequential phases mimicking the cycle of R1-HET: an anaerobic phase (180 min), followed by an anoxic phase (60 min) and a final aerobic polyhydroxyvalerate (PHV) and 2-hydroxycaproic acid as standard phase (120 min). The first anaerobic phase was obtained by for polyhydroxy-2-methylvalerate (PH2MV). sparging N2 gas. At the beginning of the experiment, a pulse of nutrients was added so that the concentrations were: phosphorus 1 2.4. Batch experiments 10 mg PO3-4 -P L (KH2PO4), acetic acid 150 mg L1 and ammonium 20 mg NHþ 4 -N L 1 (NH4Cl). The anoxic conditions were obtained by 2.4.1. Sampling and monitoring adding at time 180 min a single dose of nitrite or nitrate so that the The biomass used in all the batch experiments was withdrawn concentrations were: nitrite 15 mg NO 2 -N L 1 (NaNO2) or nitrate  1 from the pilot plant at the end of a cycle. Then, biomass was washed 15 mg NO3 -N L (NaNO3). The nitrite/nitrate added was equivalent three times with tap water and was left overnight with magnetic to the ammonium that would have been oxidised during the stirring and aeration in order to consume the remaining COD and to nitrification phase in R2-AUT of the pilot plant. Finally, the aerobic achieve endogenous conditions at the beginning of the phase was accomplished by sparging air, maintaining a DO between experiments. 2 and 5 mg L1. The batch experiments were conducted in a 2 L vessel with on- line pH, DO and temperature monitoring as in (Guisasola et al., 2.4.3. Nitrification activity experiments 2007). The probes, pH (Sentix 81, WTW) and DO (Cellox 325, Nitrification activity tests comprised two aerobic phases of O. Larriba et al. / Water Research 172 (2020) 115474 5 Table 3 Typical configuration of the four reactors of the pilot plant with a 12 h cycle length. R1-HET (heterotrophic SBR) R2-AUT (autotrophic SBR) Time (min) Phase Time (min) Phase 0e37 Feeding from influent 37e3581 Aerobic 37e355 Anaerobic 3581e413 Settling 355e360 Purge 413e441 Extraction to R1-HET 360e385 Settling 441e465 Feeding from R4-INT 385e413 Extraction to R4-INT 465e470 Purge to R1-HET 413e441 Feeding from R2-AUT 470e372 Idle þ R3-PRE 441e541 Anoxic 541e661 Aerobic 661e691 Settling 691e720 Extraction to effluent R3-PRE (precipitation reactor) R4-INT (interchange vessel) Time (min) Phase Time (min) Phase 385e413 Settling 385e413 Feeding from R1-HET 413e425 Extraction to R1-HET 413e425 Idle 425e435 Feeding from R4-INT 425e435 Extraction to R3-INT 435e3852 Precipitation: Mg2þ addition 435e465 Extraction to R2-AUT 465e3852 Idle 1 Maximum value (the real value depends on the control of the aeration phase length). 2 Time of the following cycle. 120 min each, with a DO between 2 and 5 mg L1. At the beginning results can be found in the Supplementary Information. The major of the experiment, ammonium (NH4Cl) was added so that an initial design assumptions were: i) primary and secondary settlers are not concentration of 15 mg NHþ 4 -N L 1 was obtained. If required, a needed for the SCEPPHAR plant while they are considered for the second aerobic phase was planned with a pulse of nitrite (NaNO2). A2/O plant, ii) three independent SCEPPHAR process lines are needed to guarantee the continuous operation and to ease the 2.5. Calculations maintenance, and iii) just one warehouse spare equipment unit (i.e. blower and pumps) for all process lines is purchased as a backup. The observed production and consumption rates of the studied Incomes were derived from the net production of electricity (i.e. variables were calculated as follows: biogas) and wastewater tariff. The cash-flow relative to struvite was negligible compared to the other costs. PHA recovery incomes were DC excluded because of the low PHA content, but the increase in biogas Production or consumption rate ¼ (2) tf  to production due to the PHA concentration in the sludge was considered. External carbon addition was not accounted since the DC biochemical oxygen demand (BOD) content was considered high tf to Specific production or consumption rate ¼ (3) enough to promote both N and P removal. VSS Where DC is the concentration difference between the beginning 3. Results and the end of the period with linear behaviour in mg L1, to and tf are the initial and final time in minutes, and VSS concentration is in 3.1. Period I. Preliminary results and operational changes during the mg L1. start-up Furthermore, the ratio P-release/C-consumption (P/C) during the anaerobic phase was calculated as a PAO activity indicator: In a first start-up (Period I), the system was operated for more than three months with a configuration of 8 h per cycle (Table S1). , P Ammonia was fully oxidised since R2-AUT was enriched in nitri- REL molP M fying organisms. However, EBPR activity was not detected, indi- P Cð Þ ¼ CODP (4) molC CON cating the absence of PAO in R1-HET. The low COD present in the 32 influent (150 ± 40 mgCOD$L1) was a clear bottleneck for the implementation of EBPR. In addition, specific batch tests in R1-HET Where PREL is the phosphorus release during the anaerobic phase in demonstrated that COD remained almost constant during the 1 mg PO3- 4 -P L , MP is the P molecular weight (30.97 g/mol), CODCON anaerobic phase and that no VFA were detected, revealing that is the COD consumption in the anaerobic phase in mgO2 L1, and 32 besides the low COD content there was a lack of COD fermentation is the factor gO2/C mol for acetate. to VFA. To enhance fermentation and to increase the VFA avail- Total N removal was calculated as ammonium in the influent ability for PAO, the anaerobic phase was increased from 3 to 6 h and minus the sum of the nitrogen species measured in the effluent as the whole cycle length was extended from 8 to 12 h using the ammonium, nitrite and nitrate. configuration reported in Table 3. After several weeks with this extended cycle length, PAO activity remained very low. Finally, it 2.6. Economic assessment was decided to add an external carbon source (acetic acid, addi- tional concentration of 100 mg COD$L1) at the beginning of the The economic feasibility of the mainstream SCEPPHAR config- anaerobic phase. Fig. S1 compares an example of the anaerobic P uration was evaluated and compared to a traditional A2/O config- release obtained in two different cycles without and with VFA uration for a time span of 30 years. The detailed methodology and addition. 6 O. Larriba et al. / Water Research 172 (2020) 115474 Considering the low COD available in the wastewater used and (step 4), the supernatant of R1-HET (70% of the volume) was the results obtained during Period I, the SCEPPHAR pilot plant was transferred to R4-INT (step 5). In the following step 6, R1-HET operated from that moment onwards with a cycle configuration of received the supernatants of R2-AUT (containing the water of the 12 h and with an automated dosage of acetic acid (to increase the previous cycle after nitrification) and R3-PRE (150L). After this COD concentration by 100 mgCOD$L1 at the beginning of the water interchange, the concentration of ammonium (Fig. 3) at the anaerobic phase in R1-HET). After this operational change, R1-HET beginning of the anoxic phase decreased significantly down to was progressively enriched in PAO and EBPR activity increased, 8.5 mg NHþ 1 4 -N$L , as the content of R1-HET (30% remaining after promoting stable and good removal for P and N. settling) was diluted with the nitrified water from R2-AUT with ammonium concentration around 0 mg NHþ 1 4 -N$L . Phosphate also 1 3.2. Period II. Stable operation of the SCEPPHAR pilot plant with decreased to 27.1 mgPO3- 4 -P$L due to the added effect of three complete nitrification different processes: i) struvite precipitation in R3-PRE that reduced the concentration of phosphate in the supernatant of this reactor, ii) This section reports the results obtained during the long-term occurrence of some P-uptake in R2-AUT and iii) DPAO activity monitoring of Period II, when the pilot plant was operated under during the filling of R1-HET from R2-AUT, as nitrate content in this complete nitrification mode and had good PAO activity. Fig. 2 shows exchanged water can be used as electron acceptor for P-uptake. the ammonium, phosphate and CODS input/output profiles of the During the rest of the anoxic phase (step 7), the remaining ni- pilot plant once the PAO activity was enhanced and Table 4 shows trate could be used as electron acceptor for DPAO activity. In the the average solids concentrations. During the days 0e175, the cycle presented, initial nitrite/nitrate concentration was low, as it effluent P was consistently below the discharge limit, with an was almost depleted during the previous feeding phase, and hence 1 average of 0.2 ± 0.1 mgPO3- 4 -P$L . Ammonium oxidation was anoxic P-uptake was low. Regarding the other monitored variables, almost complete, with average effluent concentration of ammonium concentration did not change significantly, and a slight 1.9 ± 2.5 mg NHþ 1 4 -N$L . Nitrite, nitrate and TN average concen- COD consumption was observed. Finally, in the following aerobic tration in the effluent were 0.5 ± 1.2 mg NO 1 2 -N$L , 3.7 ± 0.9 mg phase (step 8), P-uptake was fast, reaching 0.06 mgPO3- 4 -P$L 1 at  1 1 NO3 -N$L and 5.2 ± 3.6 mgTN$L . Hence, the pilot plant was the end of the phase. Ammonium was fully oxidised and CODS was accomplishing its main objective, which was to meet legal only 16 mgCOD$L1 at the end of this phase. Finally, aeration and discharge limits for the Manresa WWTP (P < 1 mgP$L1, NHþ 4 -N < 4 stirring were turned off for the last settling (step 9) and discharge mgN$L1 and TN < 10 mgN$L1, (EEC Council, 1991)). The removal (step 10), where a very high quality effluent was obtained, fully efficiencies during Period II for total N, P and CODT were 86 ± 12%, accomplishing discharge limits. 93 ± 9% and 79 ± 6%. Ammonia oxidation started to decrease on day Regarding R2-AUT, it received the anaerobic supernatant of R1- 180, because of a significant decrease in temperature (Fig. S2) and HET through the buffer R4-INT (step 1 for R2-AUT). The aerobic because the aeration phase in R2-AUT was limited to 3 h. phase (step 2) was designed for ammonia oxidation to nitrate, and Concerning to CODS, Fig. 2C shows the low influent concentra- it accomplished its objective, reducing the ammonium concentra- tion and high variability (136 ± 39 CODS$L1), the values at the tion from 17.3 to 0 mg NHþ 4 -N$L 1 and increasing the concentration effluent of 31 ± 14 mg CODS$L1 and removal efficiencies higher  of nitrate up to 17.6 mg NO3 -N$L1 during its 2 h of duration than 75%. The lack of COD in this wastewater was alleviated by the (Fig. 3). The rest of the cycle time in R2-AUT was devoted to idle, addition of the 100 mgCOD$L1 of acetic acid. The need of this settling, discharge, filling and purge (steps 3, 4, 1). Fig. 3 also reveals addition was corroborated in some cycles (days 180e195) where a some P-removal in R2-AUT, probably due to the aerobic/extended- pump malfunctioning led to the absence of this extra VFA and idle configuration used in this reactor. The presence of PAO in this resulted in a lower P-removal. configuration is attributed to the non-aerated extended-idle phase Fig. 3 shows an example of a cycle operation, obtained at day acting as a post anaerobic zone (Wang et al., 2012). 139 of operation, when the system was under stable operation with Finally, R3-PRE receives 150 L of anaerobic supernatant from R4- complete nitrification. The cycle started with the feeding of 1750 L INT. This stream has a higher P content than the influent and a of wastewater (step 1 for R1-HET in Fig. 1). The influent wastewater similar high content of ammonium. Phosphate precipitation took 1 in this cycle contained 3.8 mgPO3- 4 -P$L and 31.3 mgNHþ 1 4 -N$L . place in this reactor when pH was increased up to 8.5 by air Considering the concentrations in the remaining mixed liquor in sparging for CO2 stripping and the addition of 19 mg Mg2þ$L1 as R1-HET before the feeding (30% of the reactor volume) and a concentrated solution of MgCl2. Typically, P concentration in R3- possible concentration similar to that of the end of this cycle PRE decreased around 65%e75% at the end of the phase. The su- 1 (around 0 mg PO3- 4 -P$L and 0 mgNHþ 1 4 -N$L ), the initial con- pernatant of this reactor was also sent to R1-HET during step 6 centrations in R1-HET after the feeding should have been around (Fig. 1). 1 2.7 mgPO3- 4 -P$L and 21.8 mg NHþ 1 4 -N$L . The first measurements (t ¼ 0.633 h) showed that the theoretical ammonium concentration 3.3. Period III. Achieving nitrite-shortcut in the SCEPPHAR (21.8 mgNHþ 1 4 -N$L ) agreed with the measured value (22.8 mg configuration þ 1 NH4 -N$L ), but did not agree for phosphate. The experimental 1 measurement was 13.2 instead of 2.7 mgPO3- 4 -P$L . This increase Before moving to shortcut nitrogen removal, the pilot plant was was related to the length of the feeding phase (38 min): PAO could stopped for maintenance and some valves and pumps were release P linked to the uptake of the VFA available in the replaced because of clogging issues. Hence, Period III starts at day wastewater. 275 after a new inoculation. The input/output profiles for phos- During the following anaerobic phase (step 2 for R1-HET) P phate and ammonium during this period are shown in Fig. 4, the concentration increased linked to VFA consumption by PAO, average solids concentration are reported in Table 4 and finally 1 reaching 34.8 mgPO3- 4 -P$L at the end of this phase (an increase of Fig. S3 shows the temperature profile in R1. The plant showed 9.2 times the initial wastewater concentration). On the other hand, complete nitrification on day 341 with good nutrient removal ammonium concentration remained almost constant during this performance. The system recovered a similar removal performance anaerobic phase, with 22.3 mg NHþ 4 -N$L 1 at the end of this phase. to that observed during Period II with an effluent below discharge Then, biomass was purged (step 3) to maintain the SRT around 10d limits. Fig. S4 shows a fully monitored cycle on day 362. The profiles and the stirring was turned off for biomass settling. After settling during this batch cycle were also similar to those obtained on day O. Larriba et al. / Water Research 172 (2020) 115474 7 Fig. 2. Period II. Inlet/outlet profiles for ammonium (A), phosphate (B) and soluble COD (C). Filled symbols: inlet, void symbols: outlet, dash line: discharge limit for TN and P. Table 4 oxygen surface transfer due to the stirring. Nevertheless, this Average solids concentrations in R1-HET and R2-AUT during all the operational problem was not observed in the conventional operation of the periods. plant. Operational period (days) Reactor VSS (gVSS$L1) TSS (gTSS$L1) Fig. 5B shows a batch experiment conducted on day 378, 15 days Period II R1-HET 1.44 ± 0.18 1.70 ± 0.22 after implementing the control strategy. During the first part of the R2-AUT 3.30 ± 0.80 4.09 ± 1.00 experiment, ammonium was oxidised at a rate of 5.7 mgNHþ 4- Period III R1-HET 2.03 ± 0.18 2.48 ± 0.12 N$L1$h1, nitrite was accumulated at 3.2 mgNO-2-N$L1$h1 and ± ± R2-AUT 2.15 0.12 2.80 0.16 nitrate was produced at 2.7 mgNO-3-N$L1$h1. These values indi- cated that AOB activity (oxidation of ammonium) duplicated the activity of NOB (production of nitrate), leading to accumulation of 139 (Fig. 3): only slightly lower P-release activity was observed nitrite. Moreover, in the second part of the experiment, after the during the anaerobic phase. addition of an excess of nitrite, NOB activity slightly decreased to Once the system was stable (day 363), two operational changes 2.2 mgNO-3-N$L1$d1 instead of increasing. This indicated that were implemented to decrease NOB activity in R2-AUT: i) real-time NOB were the limiting step and that the first actions towards control of the length of the aerobic phase using the on-line decreasing NOB activity were positive. ammonium measurement and ii) operating at lower SRT. The After a month of implementing the control strategy, NOB ac- proposed strategy does not require operating at a low DO value to tivity was undetectable and the nitrite-shortcut SCEPPHAR achieve nitritation and, hence, the DO setpoint was maintained at configuration was effectively obtained. Fig. 6 shows two examples 3 mg L1, avoiding a potential increase of N2O emissions reported at of pilot plant cycles obtained on days 410 and 439 of operation. lower DO (Massara et al., 2018, 2017). Ammonium in R2-AUT was completely oxidised to nitrite, without Fig. 5A shows an example of the ammonium and nitrate detecting nitrate and matching nitrite formation to ammonia monitored in some cycles of R2-AUT on day 365, two days after oxidation. The aeration phase length was shorter on day 439 since implementing the aerobic length control strategy. Ammonium was the controller adapts this length automatically. Fig. S5 shows an oxidised during the aerobic phase linked to the increase of nitrate example of a batch test with R2-AUT biomass on day 419, where a concentration, measured by the same on-line sensor. Aeration and high ammonium concentration was added at DO setpoints of 1, 2 stirring were automatically turned off when the ammonium con- and 3 mgO2$L1. Nitrate was not observed at any DO setpoint, centration decreased below 3 mgNHþ 1 4 -N$L . However, stirring was indicating the lack of NOB activity. not stopped in the cycles shown in Fig. 5A to have representative Returning to the experiments on Fig. 6, the nitrite-rich effluent values of the monitored variables. During the non-aerated phase, of R2-AUT entered the anoxic phase of R1-HET. Nitrite was not ammonium remained almost constant, probably related to the detected in the first sample of the anoxic phase because the feeding detection level of the on-line ISE analyser, while nitrate slightly time of 28 min was enough to reduce the available nitrite by DPAO, increased due to some nitrite oxidation, probably because of as already observed for nitrate in the previous cycles. No additional 8 O. Larriba et al. / Water Research 172 (2020) 115474 Fig. 3. Period II. Typical SCEPPHAR cycle with complete nitrification obtained at day 139 of operation for R1-HET and R2-AUT. Fig. 4. Period III. Inlet/outlet profiles for ammonium (A) and phosphate (B). Filled symbols: inlet, void symbols: outlet, dash line: discharge limit for TN and P. nitrite-based DPAO activity was observed during the rest of anoxic Tables 5 and 6 show the main parameters of R1-HET and R2-AUT phase due to the lack of nitrite. However, nitrite DPAO were very in the four cycles presented in this manuscript. The maximum active as demonstrated in the specific batch experiment shown in specific rates for conventional EBPR processes (i.e. COD uptake and Fig. 7, where a nitrite reduction rate of 0.072 mg NeNO-2$g1 P-release) were obtained for the case with full nitrification. How- VSS$min1 was observed. This high nitrite DPAO activity indicates ever, under nitrite-shortcut mode the P-removal efficiencies were that a fraction of P was uptaken in the anoxic feeding phase of R1- not affected in spite of the lower specific rates. The nature of the HET. The most significant result of the SCEPPHAR configuration in influent COD was very diverse: from complex COD to VFA. Thus, Period III was the successful P-removal using N-removal via nitrite. simultaneous fermentation and COD uptake could be occurring Moreover, some P-uptake was also observed in R2-AUT, similarly to under anaerobic conditions. The P/C values were quite high for Period II, and showing that the nitrite shortcut did not affect po- being a real wastewater, probably due to the acetic acid amend- tential PAO presence in R2-AUT. Regarding the overall removal ef- ment. Regarding the N-related processes, the specific ammonium ficiencies during Period III, they were similar to Period II: 82 ± 11% oxidation rate increased during time and it was higher when nitrite for total N, 94 ± 12% for P and 80 ± 7% for CODT. shortcut was applied, probably also due to the lower SRT used (5- O. Larriba et al. / Water Research 172 (2020) 115474 9 cycle of day 139. An average 45% of the P in the influent could be recovered as struvite following the next considerations: a typical precipitation of 70% of the initial P in R3-PRE, an influent of 1750L of 1 wastewater with 4 mgPO3- 4 -P$L and that 150L of anaerobic su- 1 pernatant with 30 mgPO3- 4 -P$L were sent to R3-PRE. When considering the best scenario (day 139), a supernatant of 35 1 1 mgPO3-4 -P$L , influent 3.8 mgPO4 -P$L 3- and 80% of precipitation, the recovery increased up to 63%. The biomass extracted at the end of the anaerobic phase has a higher PHA content than the typical purge obtained at the end of the aerobic phase. PHA analyses in a conventional cycle of the pilot plant showed percentages of PHA in the biomass (gPHA$g1TSS) of 3.1% and 6.9% at the start and end of the anaerobic phase. Batch experiments with excess of VFA and left overnight increased PHA content from 0.7% to 9.1%, which is the maximum value measured with this sludge. This value would be obtained by adding an excess of VFA in all or some of the cycles in the pilot plant. 3.5. Economic assessment This section reports the main results obtained for the economic evaluation, which is detailed in the Supplementary Information. The net present value for SCEPPHAR and A2/O were estimated at 58.2 and 63.4 MV, respectively. Considering a wastewater tariff of 0.73 V/m3, the internal rate of return (IRR) was very high for SCEPPHAR and A2/O: 21% and 29%, respectively. The slightly better outcome for the A2/O technology was mainly due to its lower capital expenditures (CAPEX) (15.9 MV) compared to SCEPPHAR (21.8 MV), due to the discontinuous nature of the SBR operation. The cost advantage of missing settler units in SCEPPHAR was off-set by the higher total tank volumes and mixing units. Cost of piping was also higher for SCEPPHAR because of its more complex liquid interchange system. Operational expenditures (OPEX) were slightly higher for SCEPPHAR, mainly due to the higher maintenance costs related to the higher CAPEX. The wastewater tariff required to obtain an Fig. 5. A) Example of on-line ammonium and nitrate measurements in R2-AUT during IRR ¼ 4% was 0.27 and 0.31 V/m3 for A2/O and SCEPPHAR, some cycles on day 365 of operation, two days after implementing the aeration length respectively. Hence, SCEPPHAR is outcompeted by a conventional control. The DO control loop was activated during the ON periods. B) Batch experiment A2/O if only the incomes from biogas production and struvite are for evaluating the decrease of NOB activity, performed with biomass obtained from R2- AUT on day 378 of operation. considered. However, the difference in terms of wastewater tariff aid for SCEPPHAR is only 15% higher than for A2/O. 4. Discussion 7d) that led to lower biomass concentration (Table 4). The ratio of nitrate with respect to the initial ammonium was close to one in the 4.1. COD requirements for the SCEPPHAR configuration complete nitrification cycles, while the ratio of nitrite versus initial ammonium was close to one in the two cycles reported with nitrite- The pilot plant during Period III showed good removal effi- shortcut. ciencies of P (94 ± 12%) and N (82 ± 11) with a wastewater enriched in VFA: CODT ¼ 338 ± 81 mgCOD$L1, CODS ¼ 236 ± 39 mgCOD$L1, ammonium 40 ± 11 mg NHþ 4 -N$L 1 and phosphate 4.2 ± 1.1 mg 1 3.4. Resource recovery PO3- 4 -P$L . Metcalf and Eddy (Tchobanoglous et al., 2013) recom- mend threshold ratios for the design of WWTP configurations The main purpose of the mainstream SCEPPHAR configuration aiming at simultaneous C, N and P removal (A2/O and UCT): a was to examine the feasibility of recovering resources from the readily biodegradable COD (rbCOD) content of 6.6 g wastewater besides obtaining a good effluent quality. Regarding the rbCOD$g1NO 1 3 -N and 10 g rbCOD$g P. These ratios, translated to recovery of P as struvite, the typical values obtained in R3-PRE the wastewater treated in this work, result in the need of 306 g showed P-precipitation to be in the range 65e90% depending on rbCOD$L1. This value is higher than the CODs available for the VFA- the specific cycle and precipitation conditions. Magnesium was enriched wastewater treated in this work (236 ± 39 mgCOD$L1). added in the range 1e1.5 molMg2þ$mol1P: a fixed amount of Hence, if the wastewater met the design recommendation of Mg2þ was dosed but the P concentration at the beginning of the Metcalf and Eddy, the SCEPPHAR configuration would be able to precipitation phase in R3-PRE was variable. P-recovery in R3-PRE treat it without the need of adding extra COD. was feasible due to the increased concentration of P obtained at If the COD concentration was too low, as in the case studied in the end of the anaerobic phase with respect to the influent. The this work, additional VFA would be required. Fermentation of average concentration of P in R3-PRE was 6.2 ± 2.4 times the excess sludge appears as a clear option to obtain extra VFA, influent concentration with a maximum increase of 9.2 times in the although the amount produced is highly dependent on the 10 O. Larriba et al. / Water Research 172 (2020) 115474 Fig. 6. Period III. Example of two SCEPPHAR pilot-plant cycles (days 410 and 439) with N-removal via nitrite. pretreatments used (Longo et al., 2015; Luo et al., 2019). For the R1- 4.2. Successful implementation of nitrite shortcut HET scenario (i.e. SRT around 10 d and with 2.7 gTSS$L1) the solids production would be around 675 gTSS$d1. To obtain an increase of The simultaneous implementation of real-time control of the VFA in the influent around 100 mgCOD$L1 for a daily influent of length of the aerobic phase and decrease of SRT led to a successful 3500 L d1, 350 gCOD$d1 would be required. Considering a ratio of operation with shortcut N-removal, that was maintained without 0.85 gTVS$g1TSS in the sludge, this amount of COD could be ob- observing any NOB adaptation during the operation of Period III. tained achieving a fermentation yield of 0.61 gCOD$g1TVS by us- The rationale of the first strategy is that if aeration is switched off ing one of the pretreatments reported in the literature (Luo et al., around complete ammonia oxidation, some residual nitrite should 2019). Additional COD could be also obtained from the fermenta- be present that could not be further oxidised by NOB due to the lack tion of primary sludge or other fermentable carbon sources such as of aeration, thereby limiting their growth (Fux et al., 2006; glycerol (Guerrero et al., 2015, 2012). Guisasola et al., 2010; Guo et al., 2009; Lemaire et al., 2008; O. Larriba et al. / Water Research 172 (2020) 115474 11 4.3. Resource recovery Most alternatives to recover P from WRRFs are based on the sludge generated (Cieslik and Konieczka, 2017; Egle et al., 2016; Mayer et al., 2016; Melia et al., 2017; Rittmann et al., 2011; Rotta et al., 2019). The most accepted strategy is, nowadays, its precipi- tation as struvite (NH4MgPO4$6H2O) from the supernatant of digested sludge (Jabr et al., 2019; Le Corre et al., 2009; Li et al., 2019; Rittmann et al., 2011). Struvite is a slow-release rate fertiliser with nitrogen and magnesium that has been reported to be a good op- tion for agricultural uses. Struvite precipitation in a tertiary step is not a viable option due to the low concentration of phosphate and ammonia. The most adequate stream for the struvite precipitation is the digestate of anaerobic digestion, due to its high ammonia and phosphate concentrations. In addition, when EBPR is integrated in a WRRF, P recovery as struvite from the digestate can be more effi- Fig. 7. Batch experiment with nitrite as electron acceptor under anoxic conditions, cient since the sludge entering anaerobic digestion contains up to performed on day 419 of operation with biomass obtained from R1-HET. 20 times more P as Poly-P that could be easily released in the digestate (Münch and Barr, 2001; Shu et al., 2006). However, typical values of P-recovery around 12% are obtained Marcelino et al., 2011). The reduced NOB growth would lead to a by struvite precipitation from the supernatant of anaerobic diges- slightly lower nitrite oxidation rate in the following cycle. As a tion (Remy and Jossa, 2015), in contrast to the much higher 45e63% result, nitrite accumulation when ammonia oxidation is completed of P-recovery reported in section 3.4, which is one of the main and the aeration phase is terminated would be slightly higher than achievements of the mainstream SCEPPHAR. in the previous cycle, further reducing the growth of NOB. Over The key operational factor in this configuration is obtaining a many cycles, the NOB population would decrease gradually, which high anaerobic P-release, as it determines the maximum resource will be accompanied by increased nitrite accumulation at the end of recovery that can be obtained: the amount of P that can be recov- the aeration phases. While NOB growth is repressed through this ered in R3-PRE and the percentage of PHA in the sludge. On the misalignment between oxygen and nitrite, AOB are not affected other hand, the precipitation reactor does not need to be designed because oxygen is provided until ammonium is depleted. Moreover, to treat all the influent wastewater, as less than 9% of the influent is it may also result in transitory high free nitrous acid (FNA) levels treated in this case (5.5% of the reactor volume). Another possibility when aeration is terminated and nitrite is accumulated in the bulk would be decreasing the number of cycles with precipitation: it liquid. While the reported FNA levels inhibitory to AOB and NOB could be plan a single recovery per day with some extra VFA in this varied considerably in different studies, it is evident that AOB are in specific cycle. This strategy has been proposed in the literature general much more tolerant to FNA than NOB (Jubany et al., 2009b; (Kodera et al., 2013; Lv et al., 2014), although in some cases it has Park and Bae, 2009). In summary, when aeration is suppressed, the provoked some instability in PAO activity (Acevedo et al., 2015). conditions are set so that it is likely that mNOB decreases more than Nevertheless, the long-term operation with an optimised volume of mAOB, thus contributing to the decrease of NOB activity. This strategy anaerobic supernatant extraction every cycle has been demon- can only be successful if it is combined with a proper SRT selected strated to be feasible and stable at lab-scale (Guisasola et al., 2019). to remove NOB faster than they grow as shown in equation (1) and With respect to PHA, the values of 6.9% at the end of the explained in previous works (Guisasola et al., 2010; Jubany et al., anaerobic phase or 9.1% after adding extra VFA obtained in this 2009a, 2009b; Lemaire et al., 2008; Marcelino et al., 2011). work are much lower than the 41% (gPHA$g1VSS) obtained by a Table 5 P release and COD consumption rates during R1-HET anaerobic phase. Cycle P-release P-release rate COD consumption rate P/C ratio mgP$L1 mgP$mg1VSS$min1 mgCOD$ mg1VSS$min1 molP/molC Day 139 31.0 0.129 0.212 0.35 Day 362 12.0 0.044 na na Day 410 27.0 0.096 0.055 0.68 Day 439 23.2 0.051 0.094 0.32 na: not available. Table 6 Ammonium consumption and nitrite and nitrate production rates in R2-AUT for the four cycles presented. Test NHþ 4 -N oxid. NO 2 -N prod. NO 3 -N prod. NO þ 3 -N/NH4 -N NO þ 2 -N/NH4 -N 1 1 1 1 1 1 gN$g VSS$min gN$g VSS$min gN$g VSS$min e Day 139 0.035 0.000 0.047 1.36 0.00 Day 362 0.029 0.000 0.027 0.94 0.00 Day 410 0.062 0.062 0.000 0.00 1.01 Day 439 0.087 0.122 0.000 0.00 1.40 12 O. Larriba et al. / Water Research 172 (2020) 115474 side-stream system fed with surplus full-scale activated sludge and specific biogas production. Struvite recovery is demonstrated, but acclimated with fest/famine cycles using fermented VFA liquors the incomes that could be obtained are nowadays negligible from industry or primary sludge sources (Werker et al., 2018). compared to the other costs. Current CAPEX estimation for the However, considering the amount of sludge produced per day and SCEPPHAR configuration is higher than for a typical A2/O configu- the COD load in the pilot plant, our 6.9% of PHA in the biomass ration without resource recovery, and hence, further development results in a yield around 5% of the inlet COD finally ending as PHA in and optimization is needed to become competitive from an eco- the produced sludge. This result seems low at a first glance, but it is nomic point of view. The 15% higher wastewater tariff required for much higher than the negligible yields that can be obtained in side- SCEPPHAR than for A2/O could justify its implementation if stream systems for producing PHA. resource recovery is considered strategic and incentives are The PHA content in the biomass could be even improved if R1- legislated. The current study did not consider incomes from PHA HET was operated at a lower SRT. This lower SRT would decrease because of its concentration below the recommended 40%. How- the total biomass and, thus, would increase the percentage of PHA ever, evidences from other projects, e.g. PHARIO (Bengsston et al., in the sludge if enough anaerobic VFA uptake is maintained. In a 2017; Werker et al., 2018), suggest that a PHA accumulation up to previous work (Chan et al., 2020), the PHA content of EBPR biomass 40% is feasible if short enrichment periods are planned at the increased from 8 to 18% (gPHA$g1VSS) when the SRT was expense of an additional reactor and external VFA dosage. This decreased from 15 to 5 days. This change can easily be imple- would open the possibility to obtain incomes from PHA recovery in mented in the mainstream SCEPPHAR configuration, since it is a SCEPPHAR that would potentially improve its economic feasibility two-sludge system that allows the independent optimization of (Bengsston et al., 2017). both SRT. To elucidate if the experimental PHA content in the biomass was 5. Conclusions reasonable, the amount of PHA produced with the extra VFA added was calculated considering the anaerobic stoichiometry for PAO The mainstream SCEPPHAR configuration was operated in a (Smolders et al., 1994) (1.33 molC-PHB$mol1C-Acet ¼ 0.95 demo-scale pilot plant and for a long-term with successful results gPHB$g1Acet). The inlet acetic acid (100 mgCOD$L1) to R1-HET in terms of C, N and P removal efficiencies under full nitrification with a typical biomass concentration of 2.7 gTSS$L1 would result and under nitrite shortcut mode. in an increase of 2.5% of the PHA percentage in the biomass. Therefore, the experimental PHA increase observed (3.8%) was  Regarding P recovery, the anaerobic supernatant had a con- reasonable, as some of the inlet COD could be fermented to VFA centration of P of 6e9 times that of the influent, allowing that during the anaerobic phase. In the hypothetic case that the biomass 45e63% of the P in the influent could be recovered as struvite in concentration was only 1.4 gTSS$L1, the increase of PHA would a separate precipitator. This value is much higher than the have been 4.8%, which supports the strategy of decreasing SRT for typical values around 12% reported for sidestream P-recovery. increasing the percentage of PHA. These results also show that an  Mainstream SCEPPHAR proposes a purge from the anaerobic important amount of VFA should be available or added in view of phase of the EBPR system rather the conventional purging at the obtaining biomass with a significant PHA percentage. end of the aerobic phase. This biomass contains 6.9e9.2% of Although the experimental PHA content obtained could be PHA, a higher percentage than typical activated sludge, but its improved by lowering SRT or with sporadic VFA dosages, its per- extraction does not seem economically feasible for bioplastic centage is still far from the requirements for an economical down- production with the current technologies. Nevertheless, purging stream recovery of the PHA polymer, which recommends a mini- biomass with high PHA content would increase the methane mum value of 40% (gPHA$g1VSS) (Werker et al., 2018). Never- production when diverting this sludge to anaerobic digestion theless, purging sludge with high PHA content can also be and, thus, it would enhance the energy recovery of the system. beneficial for an enhanced biogas production. Several works report  The economic feasibility study shows that the current SCEP- a significant increase of biomethane potential (BMP) of biomass PHAR technology would require a water tariff 15% higher than with higher PHA content (Guisasola et al., 2019; Huda et al., 2016; that of a conventional A2/O and hence it would be necessary to Wang et al., 2016). Chan et al. (2020) studied BMP versus the PHA legislate incentives to favour resource recovery to make it content in EBPR sludge at different operational conditions and economically viable. obtained a linear correlation BMP (mL CH4$g1VSS) ¼ 240 þ 15.3 * PHA (mmol C$g1VSS). They showed that the percentage of COD in the influent that could be recovered as methane could be increased Declaration of competing interest from 11% (SRT ¼ 15 d and purge extracted from the aerobic reactor) to 28% (SRT ¼ 5 d and purge extracted from the anaerobic reactor). The authors declare that they have no known competing Moreover, the main BMP increase was observed during the first financial interests or personal relationships that could have days of the test, due to the higher biodegradability of PHA. Thus, appeared to influence the work reported in this paper. digesting biomass with high PHA content would have a significant impact in anaerobic digesters working at relatively low HRT. Acknowledgements 4.4. Integrated assessment This work was funded by the SMART-Plant project (Scale-up of low-carbon footprint Material Recovery Techniques, EUH2020, This work reports the major benefits of the mainstream SCEP- grant agreement 690323). Oriol Larriba is grateful to the FI pre- PHAR configuration: i) EBPR can be combined with a high P re- doctoral scholarship from the Catalan Government (2019FI-B1- covery as struvite, ii) the sludge generated contains a higher 00051). The authors are members of the GENOCOV research group amount of PHA than conventional sludge, iii) it can remove nitro- (Grup de Recerca Consolidat de la Generalitat de Catalunya, 2017 gen via nitrite shortcut, and iv) high COD, N and P removal are SGR 1175, www.genocov.com). The authors want to acknowledge stably obtained if enough rbCOD is available in the wastewater. On the support on field by Anna Lupo  n, David Güell and Ricard Tom as the other hand, the PHA content in the sludge does not allow an from Aigües de Manresa. Special thanks to Borja Solís for his sup- economically feasible extraction, although this sludge increases the port during the plant operation. O. Larriba et al. / Water Research 172 (2020) 115474 13 Appendix A. Supplementary data conditions. Process Biochem. 39, 1223e1229. https://doi.org/10.1016/S0032- 9592(03)00249-8. Jubany, I., Baeza, J.A., Lafuente, F.J., Carrera, J., 2009a. Model-based study of nitrite Supplementary data to this article can be found online at accumulation with OUR control in two continuous nitrifying activated sludge https://doi.org/10.1016/j.watres.2020.115474. configurations. Water Sci. Technol. 60, 2685e2693. https://doi.org/10.2166/ wst.2009.694. Jubany, I., Lafuente, F.J., Baeza, J.A., Carrera, J., 2009b. Total and stable washout of References nitrite oxidizing bacteria from a nitrifying continuous activated sludge system using automatic control based on Oxygen Uptake Rate measurements. Water Acevedo, B., Camin ~ a, C., Corona, J.E., Borr as, L., Barat, R., 2015. The metabolic Res. 43, 2761e2772. https://doi.org/10.1016/j.watres.2009.03.022. versatility of PAOs as an opportunity to obtain a highly P-enriched stream for Kodera, H., Hatamoto, M., Abe, K., Kindaichi, T., Ozaki, N., Ohashi, A., 2013. Phos- further P-recovery. Chem. Eng. J. 270, 459e467. https://doi.org/10.1016/ phate recovery as concentrated solution from treated wastewater by a PAO- j.cej.2015.02.063. enriched biofilm reactor. Water Res. 47, 2025e2032. https://doi.org/10.1016/ Ahmed, M., Hasan, C.K., Rahman, H., Ali Hossain, M., Uddin, S.A., 2015. Prospects of j.watres.2013.01.027. using wastewater as a resource-nutrient recovery and energy generation. Am. J. Le Corre, K.S., Valsami-Jones, E., Hobbs, P., Parsons, S.A., 2009. Phosphorus recovery Environ. Sci. 11, 99e114. https://doi.org/10.3844/ajessp.2015.98.114. from wastewater by struvite crystallization: a review. Crit. Rev. Environ. Sci. APHA, 1995. Standard Methods for the Examination of Water and Wastewater. Technol. 39, 433e477. American Public Health Association, Washington, DC. USA. Lemaire, R., Marcelino, M., Yuan, Z., 2008. Achieving the nitrite pathway using Aslan, S., Miller, L., Dahab, M., 2009. Ammonium oxidation via nitrite accumulation aeration phase length control and step-feed in an SBR removing nutrients from under limited oxygen concentration in sequencing batch reactors. Bioresour. abattoir wastewater. Biotechnol. Bioeng. 100, 1228e1236. https://doi.org/ Technol. 100, 659e664. https://doi.org/10.1016/j.biortech.2008.07.033. 10.1002/bit.21844. Baeza, J.A., Guerrero, J., Guisasola, A., 2017. Optimising a novel SBR configuration for Li, B., Boiarkina, I., Yu, W., Huang, H.M., Munir, T., Wang, G.Q., Young, B.R., 2019. enhanced biological phosphorus removal and recovery (EBPR2). Desalin. Water Phosphorous recovery through struvite crystallization: challenges for future Treat. 68 https://doi.org/10.5004/dwt.2017.20468. design. Sci. Total Environ. 648, 1244e1256. https://doi.org/10.1016/ Bengsston, S., Werker, A., Visser, C., Korving, L., 2017. PHARIO Stepping stone to a j.scitotenv.2018.07.166. sustainable vaue chain for PHA bioplastic using municipal activated sludge. ndez-Are Lizarralde, I., Ferna valo, T., Manas, A., Ayesa, E., Grau, P., 2019. Model-based STOWA Report 2017-15. opti mization of phosphorus management strategies in Sur WWTP. Madrid. Blackburne, R., Yuan, Z., Keller, J., 2008. Partial nitrification to nitrite using low Water Res 153, 39e52. https://doi.org/10.1016/j.watres.2018.12.056. dissolved oxygen concentration as the main selection factor. Biodegradation 19, Longo, S., Katsou, E., Malamis, S., Frison, N., Renzi, D., Fatone, F., 2015. Recovery of 303e312. https://doi.org/10.1007/s10532-007-9136-4. volatile fatty acids from fermentation of sewage sludge in municipal waste- Chan, C., Guisasola, A., Baeza, J.A., 2020. Correlating the biochemical methane po- water treatment plants. Bioresour. Technol. 175, 436e444. https://doi.org/ tential of bio-P sludge with its polyhydroxyalkanoate content. J. Clean. Prod. 10.1016/j.biortech.2014.09.107. 242 https://doi.org/10.1016/j.jclepro.2019.118495, 118495. Luo, K., Pang, Y., Yang, Q., Wang, D., Li, X., Lei, M., Huang, Q., 2019. A critical review of Cieslik, B., Konieczka, P., 2017. A review of phosphorus recovery methods at various volatile fatty acids produced from waste activated sludge: enhanced strategies steps of wastewater treatment and sewage sludge management. The concept of and its applications. Environ. Sci. Pollut. Res. 26, 13984e13998. https://doi.org/ “no solid waste generation” and analytical methods. J. Clean. Prod. 142, 10.1007/s11356-019-04798-8. 1728e1740. https://doi.org/10.1016/j.jclepro.2016.11.116. Lv, J.-H., Yuan, L.-J., Chen, X., Liu, L., Luo, D.-C., 2014. Phosphorus metabolism and Cordell, D., Rosemarin, A., Schro € der, J.J., Smit, A.L., 2011. Towards global phosphorus population dynamics in a biological phosphate-removal system with simulta- security: a systems framework for phosphorus recovery and reuse options. neous anaerobic phosphate stripping. Chemosphere 117C, 715e721. https:// Chemosphere 84, 747e758. https://doi.org/10.1016/j.chemosphere.2011.02.032. doi.org/10.1016/j.chemosphere.2014.10.018. EEC Council, 1991. Council Directive of 21 May 1991 concerning urban waste water Ma, Y., Peng, Y., Wang, S., Yuan, Z., Wang, X., 2009. Achieving nitrogen removal via treatment (91/271/EEC). Off. J. Eur. Communities. nitrite in a pilot-scale continuous pre-denitrification plant. Water Res. 43, Egle, L., Rechberger, H., Krampe, J., Zessner, M., 2016. Phosphorus recovery from 563e572. https://doi.org/10.1016/j.watres.2008.08.025. municipal wastewater: an integrated comparative technological, environmental Marcelino, M., Wallaert, D., Guisasola, A., Baeza, J.A., 2011. A two-sludge system for and economic assessment of P recovery technologies. Sci. Total Environ. 571, simultaneous biological C, N and P removal via the nitrite pathway. Water Sci. 522e542. https://doi.org/10.1016/j.scitotenv.2016.07.019. Technol. 64, 1142e1147. https://doi.org/10.2166/wst.2011.398. Fux, C., Velten, S., Carozzi, V., Solley, D., Keller, J., 2006. Efficient and stable nitri- Massara, T.M., Malamis, S., Guisasola, A., Baeza, J.A., Noutsopoulos, C., Katsou, E., tation and denitritation of ammonium-rich sludge dewatering liquor using an 2017. A review on nitrous oxide (N2O) emissions during biological nutrient SBR with continuous loading. Water Res. 40, 2765e2775. https://doi.org/ removal from municipal wastewater and sludge reject water. Sci. Total Environ. 10.1016/j.watres.2006.05.003. 596e597. https://doi.org/10.1016/j.scitotenv.2017.03.191. Guerrero, J., Guisasola, A., Baeza, J.A., 2015. Controlled crude glycerol dosage to Massara, T.M., Solís, B., Guisasola, A., Katsou, E., Baeza, J.A., 2018. Development of an prevent EBPR failures in C/N/P removal WWTPS. Chem. Eng. J. 271, 114e127. ASM2d-N2O model to describe nitrous oxide emissions in municipal WWTPs https://doi.org/10.1016/j.cej.2015.02.062. under dynamic conditions. Chem. Eng. J. 335, 185e196. https://doi.org/10.1016/ Guerrero, J., Taya , C., Guisasola, A., Baeza, J.A., 2012. Glycerol as a sole carbon source j.cej.2017.10.119. for enhanced biological phosphorus removal. Water Res. 46, 2983e2991. Mayer, B.K., Baker, L.A., Boyer, T.H., Drechsel, P., Gifford, M., Hanjra, M.A., https://doi.org/10.1016/j.watres.2012.02.043. Parameswaran, P., Stoltzfus, J., Westerhoff, P., Rittmann, B.E., 2016. Total value of Guisasola, A., Chan, C., Larriba, O., Lippo, D., Sua rez-Ojeda, M.E., Baeza, J.A., 2019. phosphorus recovery. Environ. Sci. Technol. 50 https://doi.org/10.1021/ Long-term stability of an enhanced biological phosphorus removal system in a acs.est.6b01239. phosphorus recovery scenario. J. Clean. Prod. 214, 308e318. https://doi.org/ Meijer, S.C., van Loosdrecht, M.C.M., Heijnen, J.J., 2001. Metabolic modelling of full- 10.1016/j.jclepro.2018.12.220. scale biological nitrogen and phosphorus removing wwtp’s. Water Res. 35, Guisasola, A., Marcelino, M., Lemaire, R., Baeza, J.A., Yuan, Z., 2010. Modelling and 2711e2723. simulation revealing mechanisms likely responsible for achieving the nitrite Melia, P.M., Cundy, A.B., Sohi, S.P., Hooda, P.S., Busquets, R., 2017. Trends in the pathway through aeration control. Water Sci. Technol. 61, 1459. https://doi.org/ recovery of phosphorus in bioavailable forms from wastewater. Chemosphere 10.2166/wst.2010.017. 186, 381e395. https://doi.org/10.1016/j.chemosphere.2017.07.089. Guisasola, A., Vargas, M.D.M., Marcelino, M., Lafuente, F.J., Casas, C., Baeza, J.A., 2007. Montiel-Jarillo, G., Carrera, J., Su arez-Ojeda, M.E., 2017. Enrichment of a mixed On-line monitoring of the enhanced biological phosphorus removal process microbial culture for polyhydroxyalkanoates production: effect of pH and N and using respirometry and titrimetry. Biochem. Eng. J. 35, 371e379. https://doi.org/ P concentrations. Sci. Total Environ. 583, 300e307. https://doi.org/10.1016/ 10.1016/j.bej.2007.02.001. j.scitotenv.2017.01.069. Guo, J.H., Peng, Y.Z., Wang, S.Y., Zheng, Y.N., Huang, H.J., Ge, S.J., 2009. Effective and Münch, E.V., Barr, K., 2001. Controlled struvite crystallisation for removing phos- robust partial nitrification to nitrite by real-time aeration duration control in an phorus from anaerobic digester sidestreams. Water Res. 35, 151e159. https:// SBR treating domestic wastewater. Process Biochem. 44, 979e985. https:// doi.org/10.1016/S0043-1354(00)00236-0. doi.org/10.1016/j.procbio.2009.04.022. Park, S., Bae, W., 2009. Modeling kinetics of ammonium oxidation and nitrite Huda, S.M.S., Satoh, H., Mino, T., 2016. Anaerobic degradation of poly- oxidation under simultaneous inhibition by free ammonia and free nitrous acid. hydroxyalkanoate accumulated in activated sludge in the absence of anaerobic Process Biochem. 44, 631e640. https://doi.org/10.1016/j.procbio.2009.02.002. digested sludge. J. Water Environ. Technol. 14, 236e246. https://doi.org/ Pollice, A., Tandoi, V., Lestingi, C., 2002. Influence of aeration and sludge retention 10.2965/jwet.15-037. time on ammonium oxidation to nitrite and nitrate. Water Res. 36, 2541e2546. Jabr, G., Saidan, M., Al-Hmoud, N., 2019. Phosphorus recovery by struvite formation Remy, C., Jossa, P., 2015. Life Cycle Assessment of Selected Processes for P Recovery from Al Samra municipal wastewater treatment plant in Jordan. Desalin. Water from Sewage Sludge, Sludge Liquor, or Ash. Deliverable 9.2 of P-REX Project. Treat. 146, 315e325. https://doi.org/10.5004/dwt.2019.23608. Sustainable Sewage Sludge Management Fostering Phosphorus Recovery and Jiang, C., Xu, S., Wang, R., Feng, S., Zhou, S., Wu, Shimin, Zeng, X., Wu, Shanghua, Energy Efficiency. Bai, Z., Zhuang, G., Zhuang, X., 2019. Achieving efficient nitrogen removal from Rey-Martínez, N., Badia-Fabregat, M., Guisasola, A., Baeza, J.A., 2019. Glutamate as real sewage via nitrite pathway in a continuous nitrogen removal process by sole carbon source for enhanced biological phosphorus removal. Sci. Total En- combining free nitrous acid sludge treatment and DO control. Water Res. 161, viron. 657, 1398e1408. https://doi.org/10.1016/j.scitotenv.2018.12.064. 590e600. https://doi.org/10.1016/J.WATRES.2019.06.040. Rittmann, B.E., Mayer, B., Westerhoff, P., Edwards, M., 2011. Capturing the lost Jianlong, W., Ning, Y., 2004. Partial nitrification under limited dissolved oxygen phosphorus. Chemosphere 84, 846e853. https://doi.org/10.1016/ 14 O. Larriba et al. / Water Research 172 (2020) 115474 j.chemosphere.2011.02.001. growth medium for green microalgae cultivation. Water Res. 68, 821e830. Rotta, E.H., Bitencourt, C.S., Marder, L., Bernardes, A.M., 2019. Phosphorus recovery https://doi.org/10.1016/j.watres.2014.09.027. from low phosphate-containing solution by electrodialysis. J. Membr. Sci. 573, van Loosdrecht, M.C.M., Brandse, F., De Vries, A., 1998. Upgrading of waste water 293e300. https://doi.org/10.1016/j.memsci.2018.12.020. treatment processes for integrated nutrient removal-the BCFS® process. Water Shi, J., Lu, X., Yu, R., Zhu, W., 2012. Nutrient removal and phosphorus recovery Sci. Technol. 37, 209e217. performances of a novel anaerobic-anoxic/nitrifying/induced crystallization Wang, D.-B., Li, X.-M., Yang, Q., Zheng, W., Wu, Y., Zeng, T., Zeng, G., 2012. Improved process. Bioresour. Technol. 121, 183e189. https://doi.org/10.1016/ Biological Phosphorus Removal Performance Driven by the Aerobic/extended- j.biortech.2012.06.064. Idle Regime with Propionate as the Sole Carbon Source, Water Research. Shu, L., Schneider, P., Jegatheesan, V., Johnson, J., 2006. An economic evaluation of https://doi.org/10.1016/j.watres.2012.04.036. phosphorus recovery as struvite from digester supernatant. Bioresour. Technol. Wang, Q., Sun, J., Zhang, C., Xie, G.-J., Zhou, X., Qian, J., Yang, G., Zeng, G., Liu, Y., 97, 2211e2216. https://doi.org/10.1016/j.biortech.2005.11.005. Wang, D., 2016. Polyhydroxyalkanoates in waste activated sludge enhances Smolders, G.J.F., van der Meij, J., van Loosdrecht, M.C.M., Heijnen, J.J., 1994. Model of anaerobic methane production through improving biochemical methane po- the anaerobic metabolism of the biological phosphorus removal process: stoi- tential instead of hydrolysis rate. Sci. Rep. 6, 19713. https://doi.org/10.1038/ chiometry and pH influence. Biotechnol. Bioeng. 43, 461e470. https://doi.org/ srep19713. 10.1002/bit.260430605. Werker, A., Bengtsson, S., Korving, L., Hjort, M., Anterrieu, S., Alexandersson, T., Taya, C., Garlapati, V.K., Guisasola, A., Baeza, J.A., 2013. The selective role of nitrite in Johansson, P., Karlsson, A., Karabegovic, L., Magnusson, P., Morgan- the PAO/GAO competition. Chemosphere 93, 612e618. https://doi.org/10.1016/ Sagastume, F., Sijstermans, L., Tietema, M., Visser, C., Wypkema, E., van der j.chemosphere.2013.06.006. Kooij, Y., Deeke, A., Uijterlinde, C., 2018. Consistent production of high quality Tchobanoglous, G., Burton, F.L., Stensel, H.D., 2013. Wastewater Engineering: PHA using activated sludge harvested from full scale municipal wastewater Treatment and Reuse, Fourth. ed. Metcalf & Eddy, Inc., New York. McGraw-Hill. treatment e PHARIO. Water Sci. Technol. 78, 2256e2269. https://doi.org/ Theregowda, R.B., Gonza lez-Mejía, A.M., Ma, X., (Cissy), Garland, J., 2019. Nutrient 10.2166/wst.2018.502. recovery from municipal wastewater for sustainable food production systems: € m, F., 2008. Chlorinated-solvent-free gas Werker, A., Lind, P., Bengtsson, S., Nordstro an alternative to traditional fertilizers. Environ. Eng. Sci. 36, 833e842. https:// chromatographic analysis of biomass containing polyhydroxyalkanoates. Water doi.org/10.1089/ees.2019.0053. Res. 42, 2517e2526. https://doi.org/10.1016/j.watres.2008.02.011. Türk, O., Mavinic, D.S., 1987. Benefits of using selective inhibition to remove ni- Yang, Q., Peng, Y., Liu, X., Zeng, W., Mino, T., Satoh, H., 2007. Nitrogen removal via trogen from highly nitrogenous wastes. Environ. Technol. Lett 8, 419e426. nitrite from municipal wastewater at low temperatures using real-time control https://doi.org/10.1080/09593338709384500. to optimize nitrifying communities. Environ. Sci. Technol. 41, 8159e8164. Valverde-Pe rez, B., Ramin, E., Smets, B.F., Plo sz, B.G., 2015. EBP2R e an innovative https://doi.org/10.1021/es070850f. enhanced biological nutrient recovery activated sludge system to produce