Science of the Total Environment 849 (2022) 157626 Contents lists available at ScienceDirect Science of the Total Environment journal homepage: www.elsevier.com/locate/scitotenv Exploring GHG emissions in the mainstream SCEPPHAR configuration during wastewater resource recovery ⁎ Borja Solís a, Albert Guisasola a, , Maite Pijuan b,c, Juan Antonio Baeza a a GENOCOV, Departament d'Enginyeria Química, Biològica i Ambiental, Escola d'Enginyeria, Universitat Autònoma de Barcelona, 08193 Bellaterra, Barcelona, Spain b Catalan Institute for Water Research (ICRA), Emili Grahit 101, 17003 Girona, Spain c University of Girona, Girona, Spain H I G H L I G H T S G R A P H I C A L A B S T R A C T • Long-term assessment of GHG emissions from the mainstream SCEPPHAR configu- ration • Low GHG emissions even with high nitrite accumulation. • Low DO set-points led to lower emissions by hindering N2O transfer to the gas phase. • A peak of N2O emission was found attrib- uted to the transient conditions of AOB. • The intermittent aeration strategy miti- gated N2O emissions by 40 %. A R T I C L E I N F O A B S T R A C T Editor: Paola Verlicchi The wastewater sector paradigm is shifting from wastewater treatment to resource recovery. In addition, concerns re- garding sustainability during the operation have increased. In this sense, many water utilities have become aware of Keywords: the potential GHG emissions during the operation of wastewater treatment. This study assesses the nitrous oxide Enhanced biological phosphorus removal (EBPR) and methane emissions during the long-term operation of a novel wastewater resource recovery facility (WRRF) con- GHG mitigation figuration: the mainstream SCEPPHAR. The long-term N2O and CH4 emission factors calculated were in the low range Greenhouse gases (GHG) Pilot plant of the literature, 1 % and 0.1 %, respectively, even with high nitrite accumulation in the case of N2O. The dynamics and Water resource recovery facility (WRRF) possible sources of production of these emissions are discussed. Finally, different aeration strategies were implemented to study the impact on the N2O emissions in the nitrifying reactor. Results showed that operating the pilot-plant under different dissolved oxygen concentrations (between 1 and 3 g O2 m−3) did not have an effect on the N2O emission fac- tor. Intermittent aeration was the aeration strategy that most mitigated the N2O emissions in the nitrifying reactor, obtaining a reduction of 40 % compared to the normal operation of the pilot plant. 1. Introduction towards a resource recovery scenario. Therefore, wastewater treatment plants (WWTPs) are being transformed into water resource recovery facili- Over the past years, the scarcity of natural resources and the concern for ties (WRRFs). The objective of these facilities is not only to achieve a good sustainability, with particular emphasis on climate change due to green- effluent quality under the constraints of technical feasibility and cost but house gas (GHG) emissions, have shifted the wastewater sector paradigm also to recover resources (bioplastics, cellulose or nitrogen (N), and phos- phorus (P) into fertilizers such as struvite), water or energy (in energy ⁎ Corresponding author. carriers such as methane (CH4) or hydrogen) in a sustainable way. Novel E-mail addresses: Borja.Solis@uab.cat (B. Solís), Albert.Guisasola@uab.cat (A. Guisasola), WRRFs configurations as well as innovative operational and control strate- mpijuan@icra.cat (M. Pijuan), JuanAntonio.Baeza@uab.cat (J.A. Baeza). gies have arisen recently to promote this new paradigm shift. http://dx.doi.org/10.1016/j.scitotenv.2022.157626 Received 20 May 2022; Received in revised form 19 July 2022; Accepted 21 July 2022 Available online 25 July 2022 0048-9697/© 2022 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). B. Solís et al. Science of the Total Environment 849 (2022) 157626 Regarding GHG emissions, nitrous oxide (N2O) is produced and emitted Abbreviations during biological N removal on WRRFs through nitrification and denitrifi- cation (Kampschreur et al., 2009; Massara et al., 2017). N2O is a potent AOB Ammonia oxidizing bacteria GHG with a global warming potential 265 times higher than that of carbon AOR Ammonia oxidation rate dioxide (CO2) (IPCC, 2014), therefore, the carbon (C) footprint of WRRF is C Carbon highly sensitive to N2O emissions. N2O can be produced during biological CH4 Methane oxidation of ammonium to nitrite via two pathways: i) the incomplete CH4-EF Methane emission factor NH2OH oxidation pathway, and ii) the nitrifier denitrification pathway CO2 Carbon dioxide (Kampschreur et al., 2009;Wunderlin et al., 2012; Chen et al., 2020). The COD Chemical oxygen demand parameters contributing the most to the activation of these pathways are: CODS Soluble chemical oxygen demand low dissolved oxygen (DO) levels, NO− 2 accumulation, ammonia oxidation CODT Total chemical oxygen demand rate (AOR) and transient conditions from low to high activity (Law et al., DO Dissolved oxygen 2012; Massara et al., 2017; Vasilaki et al., 2019). During denitrification, DOSP Dissolved oxygen set point N2O is produced as an intermediate of the four step reduction reactions DPAO Denitrifying polyphosphate accumulating organisms from NO− − 3 to NO2 , nitric oxide (NO), N2O and finally to nitrogen gas EBPR Enhanced biological phosphorus removal (N2). In most wastewater treatment systems, the N2O reducing capacity GHG Greenhouse gas exceeds the capacity of producing N2O, and therefore denitrification is Mg2+ Magnesium ion often considered as an N2O sink (Conthe et al., 2019). However, under MgCl2 Magnesium chloride certain conditions, N2O can also accumulate during denitrification. The N Nitrogen parameter that most affects N2O accumulation during denitrification N2 Nitrogen gas is the low chemical oxygen demand (COD) to N ratio (COD/N ratio) (Law N2O Nitrous oxide et al., 2012). N2O-EF Nitrous oxide emission factor CH4 is also a GHG that is emitted during wastewater treatment. CH4 is NH2OH Hydroxylamine the second most important GHG after CO2 and has a global warming poten- NH+4 Ammonium tial 21 times higher than that of CO2 (IPCC, 2014). Previous studies have NO Nitric oxide demonstrated that CH4 emitted during wastewater treatment can be pres- NO−2 Nitrite ent in the influent of the WRRF, produced under the anaerobic environ- NO−3 Nitrate ments in the sewer network (Guisasola et al., 2008; Gutierrez et al., 2014; NOB Nitrite oxidizing bacteria Mannina et al., 2018) or present in the reject water recirculated from the ORP Oxidation-reduction potential anaerobic digester (Ribera-Guardia et al., 2019; Rodriguez-Caballero P Phosphorus et al., 2014). PAO Polyphosphate accumulating organisms The novel mainstream SCEPPHAR (Short Cut Enhanced Phosphorus PHA Poly-hydroxyalkanoate and PHA recovery) configuration of WRRF has demonstrated at demo PO3− 4 Phosphate scale and under real influent conditions the feasibility of implementing R1-HET Heterotrophic reactor resource recovery in the mainstream line (struvite and PHA-rich sludge) R2-AUT Autotrophic reactor (Larriba et al., 2020). The mainstream SCEPPHAR is one of the novel R3-PRE Precipitation reactor technologies involved in the SMART-Plant project (www.smart-plant. R4-INT Interchange reactor eu). The whole project aimed to prove the feasibility of novel wastewa- SBR Sequencing batch reactor ter treatment technologies at pilot-scale towards a circular economy SCEPPHAR Short-cut enhanced phosphorus and PHA recovery scenario. The mainstream SCEPPHAR pilot plant contains three se- configuration quencing batch reactors (SBR), the first mainly heterotrophic (R1- SP Set point HET), designed to promote EBPR, the second reactor mainly autotro- SRT Sludge retention time phic (R2-AUT) and a third one that acts as an interchange vessel. During SI Supplementary information the first long term operation of the mainstream SCEPPHAR, the pilot TN Total nitrogen plant achieved successful removal efficiencies for C, P and N under TSS Total suspended solids nitrite shortcut N-removal, i.e. the N-removal via nitritation (NH+ 4 to V Volume NO− − 2 ) and denitritation (NO 2 to N 2 ). The advantages of the nitrite VFA Volatile fatty acid pathway approach are the lower oxygen requirements for N oxidation, VSS Volatile suspended solids the lower COD requirements for denitrification processes and faster WRRF Water resource recovery facility denitrification rate (Mavinic and Turk, 1987). However, nitrite accumu- WWTP Wastewater treatment plant lation can have a negative effect on the N 2 O emissions (Law et al., 2012). Regarding nutrient recovery results, up to 45–63 % of the influent P load could be recovered as struvite in a separate reactor in the mainstream line (Larriba et al., 2020), which is higher than the 12 % of influent P reported within side-stream P-recovery (Remy and Regarding resource recovery, P arises as a perfect candidate for its Jossa, 2015). recovery from wastewater. P is essential for our society in the production Therefore, the aim of this work was to monitor the long-term operation of fertilizers. However, the main source of P is the phosphate rock and it of the mainstream SCEPPHAR pilot plant, under real conditions and under is estimated to be depleted in the next 50–300 years (Cieślik and nitrite shortcut N-removal in view of assessing the overall GHG emissions Konieczka, 2017; Cordell et al., 2011). During wastewater treatment, it is and the dynamics of these emissions through different pilot plant cycles. estimated that 3 million tons of P are removed yearly, showing that the im- Different aeration strategies were implemented to critically assess its effect plementation of P-recovery strategies would mitigate the current depen- on the N2O liquid concentration and emissions. Different parameters re- dency on phosphate rocks (Mayer et al., 2016). Biological P removal ported to have an effect on the N2O production and emission were explored during wastewater treatment is based on promoting the growth of and discussed. Finally, the measured GHG emissions were compared to polyphosphate accumulating organisms (PAO) during the enhanced biolog- other WRRF configurations and the sources that trigger these emissions ical phosphorus removal (EBPR) process. were discussed. 2 B. Solís et al. Science of the Total Environment 849 (2022) 157626 Table 1 operated with 70 % of volume exchange ratio of R1-HET, resulting in 1.75 m3 Average composition and temperature of the influent of the pilot plant. of wastewater treated per cycle. The pilot plant was operated in cycles of Variable Value Units 8- and 12-hour duration, being able to treat 5.2 and 3.5 m3 of wastewa- PO3− 4.1 ± 1.3 g P m−3 ter per day, respectively. 4 NH+4 39.4 ± 10.5 g N m−3 Each cycle of the mainstream SCEPPHAR configuration operated in the NO−2 0.2 ± 0.4 g N m−3 following sequence (Fig. 1): the cycle started with the filling of R1-HET, − NO3 0.2 ± 0.4 g N m−3 then an anaerobic phase for promoting COD fermentation, VFA uptake, Total COD 300 ± 128 g COD m−3 PO3−4 release and PHA accumulation. After an anaerobic purge to obtain Soluble COD 179 ± 62 g COD m−3 Temperature 15.6 ± 4.2 °C a PHA-rich sludge and the settling of the reactor, the supernatant of R1- HET, rich in PO3− 4 and NH+ 4 , was transferred to the R4-INT vessel. In R4- 3 INT, 0.15 m of the supernatant were sent to the R3-PRE reactor to precip- 2. Materials and methods itate struvite. The rest of the volume was transferred to the R2-AUT reactor and, once R2-AUT was filled, an aerobic phase with controlled DO took 2.1. Pilot plant configuration and influent place to promote autotrophic nitritation. After the settling period, the su- pernatant of R2-AUT, rich in PO3− 4 and NO− 2 was transferred back to R1- The SCEPPHAR pilot plant was located in the municipal WWTP of HET. The next phase of R1-HET was an anoxic phase in which DPAO took Manresa (Barcelona, Spain). The pilot plant was fed with the primary settler up P while reducing NO− 2 . The last phase of R1-HET was aerobic, when effluent of the Manresa WWTP. The average composition of the influent PAO captured all the remaining P. After settling, the R1-HET supernatant wastewater is shown in Table 1. The pilot plant consisted of two SBRs was discharged to the effluent and the cycle started again. The detailed se- (R1-HET and R2-AUT), a precipitation reactor (R3-PRE) and an interchange quence of the cycles and the duration of each step for the 8- and 12-hour vessel (R4-INT). The process diagram of the pilot plant is shown in Fig. 1. configuration are shown in Table S1 of the supplementary information The R1-HET reactor was an anaerobic/anoxic/aerobic SBR (V = 2.5 m3) (SI) section. designed for heterotrophic processes (EBPR, carbon removal and denitrifi- cation). The R2-AUT reactor was an aerobic SBR (V = 2.5 m3) designed for 2.2. Pilot plant monitoring and control architecture autotrophic nitrification. R3-PRE was a precipitation reactor (V = 0.15 m3) designed for struvite precipitation with controlled magnesium dosage. R4- The pilot plant was highly equipped with online monitoring sensors and INT was the interchange vessel (V = 2.5 m3) designed for the exchange of automatic control loops. R1-HET, R2-AUT and R4-INT had on-line monitor- supernatants among R1-HET, R2-AUT and R3-PRE reactors. The pilot plant ing of reactor level (Micropilot FMR20, Endress Hauser), which was key for Influent Effluent Filling 1 Discharge 10 2 Anaerobic phase 3 PHA-rich sludge 9 4 Purge Settling Settling R1-HET Filling SBR cycle 1 Aerobic phase 8 5 Discharge R2-AUT R3-PRE discharge discharge 3 R4-INT 2 7 6 Anoxic phase Filling Discharge 5 Precipitation R3-PRE reactor Filling Struvite Settling Magnesium 4 1 R2-AUT SBR cycle 2 3 Purge to R1-HET Aerobic phase Fig. 1. SCEPPHAR pilot plant configuration and cycle operation (adapted from Larriba et al., 2020). 3 B. Solís et al. Science of the Total Environment 849 (2022) 157626 the automation of the interexchange steps between the reactors. The super- where Qgas is the aeration flowrate, given by the data acquisition system of the natants were transferred through centrifugal pumps (JP6 B-A-CVBP, pilot plant. Finally, the total N2O emitted per cycle is calculated with Eq. (3): Grundfos). R1-HET, R2-AUT and R3-PRE were stirred (HR4A-020/100, Milton Roy Mixing) during the reaction phases. R1-HET and R2-AUT    were equipped with probes for temperature (Pt1000, Axiomatic), DO N2 O emitted ½g ¼ ∑ni¼1 N2 O emission rate g d−1  Δt ½d i ð3Þ (LDO sc, Hach), pH (PC1R1N, Hach) and oxidation-reduction potential (ORP) (RC1R5N, Hach). In addition, R2-AUT was monitored with an online where Δt is the time interval for off-gas N2O recording (1 min) and n is the ion-selective electrode for ammonia (AN-ISE sc, Hach). Dosing pumps were total number of data points recorded in the cycle. CH4 emission was analo- used for sludge purge (PS2, Seko) and magnesium addition (Tekna EVO, gously calculated using Eq. (1) for CH4 concentration (in ppmv), using the Seko). DO in R1-HET and R2-AUT was controlled by manipulating the aer- CH4 molecular weight (16 g mol−1). ation flowrate through electric control valves (Type 3241/3374, Samson Finally, the N2O emission factor (N2O-EF) of each cycle was calculated Instruments), based on the DO measurement and a proportional-integral as the percentage of N2O-N emitted during the cycle (in R2-AUT or R1- (PI) algorithm implemented in the control system. The aeration flow to HET) of the total influent NH+ 4 -N load (Eq. (4)). Similarly, The CH4 emis- each reactor was monitored with gas rotameters (Iberfluid). The aerobic sion factor (CH4-EF) was calculated as the percentage of influent COD phase length in R2-AUT was controlled via the online ammonium sensor, emitted as CH4 (Eq. (5)). i.e. the control system deactivated the DO PI controller and stopped the aer- ation when the ammonium was depleted (Larriba et al., 2020). N2O and N2 O  N emitted N2 O − EF ½% ¼  100 ð4Þ CH4 gas emissions during the aerobic phases were continuously analysed NHþ 4 − Ninfl  VOLinfl via an online infra-red gas analyser (VA 3000, Horiba). The typical location of the gas measurement analyser was R2-AUT. However, in some cycles, the CH4 emitted CH4 −EF ½% ¼  100 ð5Þ analyser monitored R1-HET. The dissolved N2O in the liquid phase was CODinfl  VOLinfl monitored in R2-AUT with an on-line dissolved N2O microsensor (N2O-R, Unisense A/S). All the sensors and the mechanical equipment were con- where NH+ 4 -Ninfl and CODinfl are the ammoniacal nitrogen and COD in- nected to a computer (PPC-3170, Advantech) through a data acquisition fluent concentrations, respectively, and VOLinfl is the influent volume system (PCI-1711U, PCLD-8710, PCLD-885 I/O, Advantech). The software (1.75 m3 ). In the cycles where the emissions were measured in both AddControl developed by the research group was used for automating all R1-HET and R2-AUT, the total emission factor was calculated as the the operation, monitoring and control. sum of the emission factor in each reactor. 3. Results 2.3. Chemical analysis 3.1. Long term operation of the pilot plant − − Chemical analysis of the influent and effluent COD, NH+ 4 , NO2 , NO3 and PO3−4 concentrations was periodically performed. In addition, one The SCEPPHAR pilot plant was operated for two years. The operation of cycle per week was thoroughly monitored. Soluble components were fil- the plant was divided into three periods (Larriba et al., 2020): tered with a 0.22 μm filter (Millipore). COD was analysed using Lovibond - Period I (first three months) corresponded to the start-up of the pilot kits (COD Vario Tube Test LR and COD Vario Tube Test MR) and the plant. During this period, the operation was adapted to achieve com- MD100 spectrophotometer (Lovibond). Soluble COD (CODS) was measured plete nitrification and EBPR. after the sample filtration while total COD (CODT) was not filtered. NH+ 4 - Period II (from day 0 to 225) corresponded to the operation under com- was measured with an ammonium analyser (AMTAXsc, Hach Lange). Phos- plete nitrification and high PAO activity. The pilot plant met the effluent phate was measured with an analyser based on the Vanadomolybdate yel- legal discharge limits. low method (115 VAC PHOSPHAX sc, Hach-Lange). NO− − 2 and NO3 were - Period III (from day 275 to 700) corresponded to the N removal via analysed with Ion Chromatography (DIONEX ICS-2000). Volatile nitrite while maintaining good PAO activity and meeting the legal suspended solids (VSS) and total suspended solids (TSS) were analysed fol- discharge limits. The plant operation was optimized by decreasing of lowing Standard Methods (APHA, 1995). the cycle length from 12 to 8 h or by implementing different aeration strategies. 2.4. GHG emissions and emission factor calculations The results showed in this study belong to the last 300 days of operation The N2O gas concentration (in mg m−3) in the off-gas was calculated of period III in which the pilot plant achieved N removal via nitrite and with Eq. (1): GHG emissions were monitored in selected cycles of the pilot plant.   3.2. Process performance of the pilot plant  −3  CN−N2 O ½ppmv  P½1 atm  MWN−N2 O 28 g mol−1 CN−N2 O mg m ¼   ð1Þ R 0:082 atm L mol−1 K−1  T½K Fig. 2 shows the influent and effluent profiles for NH+ 3− 4 , PO4 and COD − − and the effluent profiles for NO2 and NO3 of the pilot plant during the last 300 days of operation. The influent NO− − 2 and NO3 values are shown in −3 CN−N2O [ppmv] and T[K] are the N2O gas concentration, provided by Table 1. NH+ + 4 effluent concentration was 6.0 ± 7.3 g NH4 -N m . EBPR the Horiba analyser, and temperature. MWN−N2O is the N2O molecular performance was good during most of the time with an average effluent P −3 weight and R is the gas constant. The emission flowrate for N2O was calcu- of 0.72 ± 0.99 g PO3− 4 -P m . The effluent COD average concentration lated with Eq. (2): −3 was 70 ± 68 g COD m . Effluent NO− − 2 , NO3 and total nitrogen (TN) av- erage concentrations were 3.2 ± 2.0 g NO2 -N m−3, 0.6 ± 0.7 g NO− − 3 -N m−3 and 11.5 ± 6.8 g N m−3, respectively. Therefore, the pilot plant       was meeting the required discharge legal limits most of the time, except N2 O emission rate g d−1 ¼ CN−N2 O mg m−3  Qgas m3 d−1 for the ammonium (NH+ 4 < 4 g NH4 -N m + −3 , P < 1 g PO3− 4 -P m −3 , COD < 1g −3 −3 125 g COD m and TN < 10 g N m (EEC Council, 1991)). Furthermore,  ð2Þ 1000 mg good removal efficiencies were obtained for TN, P and COD: 67 ± 23 %, 4 B. Solís et al. Science of the Total Environment 849 (2022) 157626 12-hours configuration 8-hours configuration To implement this strategy in the plant, a stepwise increase of R2-AUT A purge was applied to decrease the SRT until detecting high nitrite and [g N m-3] low nitrate concentration at the end of the aeration phase. Both the control NH4+ of the aeration phase length and the fixed SRT led to a gradually decrease of the NOB population and a subsequent increase on NO− 2 accumulation at the end of aerobic phases (Larriba et al., 2020). Fig. 3 shows an example of the SCEPPHAR cycle operation (day 410), when the pilot plant was operated under a 12-hour cycle configuration. Fig. 4 shows all the monitored variables of the pilot plant for the same B −3 cycle. The influent composition for this cycle was 5.8 g PO3− 4 -P m and [g P m -3] + −3 PO 43- 41.0 g NH4 -N m . The cycle started with the feeding of R1-HET (step 1 of R1-HET in Fig. 1 and Table S1 of SI). The first measurement just after the feeding ended (time = 0.63 h, Fig. 4A) was 14.8 g PO3− 4 -P, showing that during the feeding some P was released by PAO. During the second step of R1-HET, the anaerobic phase, the PO3− 4 concentration increased up to 31.8 g P m−3 (Fig. 3A) due to the P-release linked to VFA consump- C tion by PAO. The ORP probe (Fig. 4D) shows that anaerobic conditions [g O2 m -3] were reached (ORP value below zero) after the feeding of the reactor was COD completed. The ORP probe signal stabilized at 3 h of the cycle while com- plete anaerobic conditions were maintained. When the anaerobic phase ended (at time = 6 h), part of the biomass was purged (step 3) and the su- pernatant of the reactor was settled (step 4). Then, the supernatant account- ing for 70 % of R1-HET volume was transferred to the interchange reactor R4-INT (step 5). The following step of R1-HET (step 6) was the filling of the D NO2- NO2- & NO3- reactor with the supernatants from R2-AUT (from the previous cycle) and [g N m-3] NO3- R3-PRE. Then, the anoxic phase started (step 7). At the beginning of the an- oxic phase, the PO3− 4 concentration (Fig. 3A) decreased to 24.1 g P m−3 (compared to that the end of the aerobic phase of R2-AUT) due to: i) struvite precipitation decreased the P concentration of the supernatant of R3-PRE returned to R1-HET, ii) slightly decrease on the P concentration Time [days] due to P-uptake in R2-AUT (Fig. 3B) and iii) decrease on P concentration during the filling process of R1-HET due to anoxic conditions. The NH+ 4 Fig. 2. SCEPPHAR pilot plant influent and effluent profiles for ammonium (A), concentration considerably decreased due to the dilution of R1-HET, phosphate (B) and COD (C) and effluent profiles for nitrite and nitrate (D) during since the NH+ 4 concentration in the supernatant from R2-AUT returned to the studied period. Filled symbols represent influent concentration while void R1-HET was negligible. During the anoxic phase, the remaining NO− 2 was symbols represent effluent concentration. Dashed line represents discharge limit for TN (A), P (B) and COD (C). used by DPAO as the electron acceptor. In the presented cycle, NO− 2 con- centration at the beginning of the anoxic phase was already negligible, meaning that all NO− 2 was denitrified during the anoxic filling period of R1-HET. Therefore, the observed P-uptake during the anoxic phase was −3 82 ± 24 % and 76 ± 18 %, respectively. The averaged solids concentra- low (2.0 g PO3− 4 -P m ). Soluble COD and NH+ 4 remained constant during tion in R1-HET and R2-AUT were similar: 2.45 ± 0.28 and 2.39 ± the anoxic phase. The next step of R1-HET reactor was the aerobic phase 0.39 gVSS·m−3 in R1-HET and R2-AUT, respectively, and 2.80 ± 0.40 and (step 8). This cycle showed successful P-uptake, obtaining 0.4 g PO3− 4 -P 2.86 ± 0.54 gTSS·m−3 in R1-HET and R2-AUT respectively. m−3 at the end of the aerobic phase (DOSP = 3 g O2 m−3). NH+ 4 was On day 500, the cycle was shortened to 8 h and the pilot plant removal oxidised to 0.5 g N m−3 and NO− − 3 was not observed in R1-HET. NO2 con- − −3 efficiencies for TN, P and COD were maintained (Fig. 2), while the plant centration at the end of aerobic phase was 7.9 g NO2 -N m (Fig. 3A). ORP volumetric capacity increased by 50 %. Shortcut N-removal was achieved probe signal sharply increased with the presence of oxygen as electron ac- during the operational period III: the effluent concentration of NO− 2 was al- ceptor (Fig. 4D). When the aerobic phase ended, aeration and stirring ways higher than that of NO− 3 (Fig. 2D). Shortcut N-removal was success- were turned off, supernatant was settled (step 9) and discharged to the ef- fully attained by applying two control strategies simultaneously at a DO fluent (step 10), fully accomplishing discharge limits. set point (DOSP) of 3 g O2 m−3. Both strategies aimed at reducing NOB ac- In R2-AUT, the cycle started with the reactor filling with the superna- tivity while maintaining high AOB activity: i) the real-time control length of tant of R1-HET (step 1 of R2-AUT in Fig. 1). Once the reactor was filled, the aerobic phase in R2-AUT using the on-line NH+ 4 measurement and ii) the second step was the purge of the reactor (step 2). The aerobic phase operating with a selected sludge retention time (SRT) to remove NOB (step 3) started after an idle phase at 0.63 h of the cycle. Temperature while retaining AOB. The first strategy is based on stopping the aeration was constant during the aerobic phase (Fig. 4C), with an average value of − when NH+ 4 is depleted. Then, NO2 should be accumulated due to the 24.2 °C. At a DOSP of 3 g O2 m−3, all the NH+ − 4 was oxidised to NO2 rather lack of oxygen and NOB growth would be limited (Guo et al., 2009; than to NO− 3 , showing that there was a reduced NOB activity during this Marcelino et al., 2011). The second strategy relied in operating with a se- cycle. The NH+ 4 online sensor showed a constant AOR during the aeration −3 −1 lected SRT to remove NOB faster than their growth rate (Jubany et al., of the reactor, with a value of 8.5 g NH+ 4 -N m h (R2 = 0.99). The 2009), Eq. (6): DO control system opened the aeration valve at the beginning of the aerobic phase (Fig. 4F) to rapidly achieve the desired DOSP and gradually closed it μNOB,app < SRT  1 < μAOB,app (6) considering the decrease on the NH+ 4 concentration. In addition, the aero- bic length control turned off the aeration after 2.4 h of aerobic phase where μNOB,app and μAOB,app are the apparent growth rate of NOB and AOB, (maximum duration 3 h), when the ion-selective NH+ 4 probe signal was −3 respectively, i.e. the apparent specific growth rate minus the apparent 3 g NH+ 4 -N m . The sharp drop in the N2O profile (Fig. 3C) was a conse- decay rate for each bacteria. This optimal SRT value depends on the opera- quence of the stop of aeration, which eliminated N2O stripping and thus tional conditions that affect the apparent growth rate (i.e. DO, T and pH). its emissions. pH slightly decreased due to nitritation (Fig. 4B). The redox 5 B. Solís et al. Science of the Total Environment 849 (2022) 157626 40 300 Feeding Settling and extraction Anaerobic Anoxic Aerobic [g N m-3 and g P m -3 ] R1-HET Purge, settling and CODS [g O2 m -3 ] N and P species 30 200 interchange 20 100 10 A 0 0 0 2 4 6 8 10 12 40 300 Aerobic Idle Idle Idle [g N m-3 and g P m-3 ] Settling, interchange R2-AUT CODS [g O2 m-3] N and P species 30 200 and purge NH4+ NH4 20 NO2- NO2 NO3- NO3 PO4 100 10 PO43- NH4 NH4+sensor sensor COD COD B 0 0 0 2 4 6 8 10 12 20 50 Aerobic phase Aerobic phase R2-AUT R1-HET N2 O emission rate 40 CH4 emission rate 15 [g N-N2O d -1 ] [g CH4 d -1 ] N2O N2O emission R2-AUTR2-AUT 30 10 N2O N2O emission R1-HETR1-HET CH4 CH4emission R2-AUTR2-AUT 20 CH4 CH4emission R1-HETR1-HET 5 10 C 0 0 0 2 4 6 8 10 12 cycle time [h] Fig. 3. Example of SCEPPHAR cycle operation (A and B) and GHG profiles (C) obtained for R1-HET and R2-AUT with a cycle length configuration of 12 h (cycle belonging to day 410 of operation). potential increased with the DO and the NO− 2 accumulation as electron ac- and CH4 emitted per cycle and the corresponding emission factors ceptors (Fig. 4D). Fig. 3B reveals some P-removal in R2-AUT (4.9 g PO3− 4 -P (Eqs. (4) and (5)) obtained for the configurations of 8 (n = 13) and 12 h m−3), probably due to interchange of some biomass between reactors and (n = 21). In addition, Figs. 3C and S1C show the N2O and CH4 emission the long idle period of R2-AUT, acting as a post anaerobic zone (Larriba profiles for typical SCEPPHAR cycles of 12 and 8 h length. et al., 2020; Wang et al., 2012). The rest of the R2-AUT steps were the set- The GHG emissions were monitored for a total of 43 cycles with 12- tling and discharge to R1-HET (steps 4 and 5, respectively, Fig. 1). hour configuration and 18 cycles for the 8-hour configuration. The cy- Finally, the R3-PRE reactor received 0.15 m3 of supernatant from R4- cles in Table 2 were those with the usual operation at a DOSP of 3 g O2 INT with high content of PO3− 4 and NH+ 4 . P precipitation took place after m−3. Overall, the N2O emissions in R2-AUT were higher than those in the air sparging to increase pH and the addition of magnesium solution R1-HET because most of the NH + 4 was nitrified in R2-AUT (Table 2). (19 g Mg2+ m−3 as MgCl2). Typically, P concentration decreased around The N2O emissions (around 0.7 g N2O-N per cycle) and N2O-EF (around 70 % (Larriba et al., 2020) in R3-PRE. Finally, the supernatant of R3-PRE, 1.0 %) were low compared to similar SBR systems (Massara et al., 2017) with low content in PO3− + 4 and NH4 was sent to R1-HET (Fig. 1). for both cycle length configurations. The SCEPPHAR configuration was An example of a fully monitored SCEPPHAR cycle for an 8 h cycle config- capable to maintain low emissions despite NO− 2 being accumulated in uration on day 557 is shown in Figs. S1 and S2 of SI section. This cycle was the reactors. The measured N2O emissions and N2O-EF were approxi- operated at the same DOSP of 3 g O2 m−3 (Fig. S2E of SI). The profiles ob- mately the same operating at 12 and 8 h since both configurations tained in Fig. S1A of SI section show that the pilot plant was able to obtain had the same aerobic phase length and similar average temperatures a high effluent quality with reduced anaerobic and anoxic phases length in (21.8 ± 2.6 °C for 8 h and 22.3 ± 0.8 °C for 12-hour configuration). R1-HET and the observations were similar to those obtained on day 410 Figs. 3C and S1C show how N2 O emissions began when NH + 4 was (Figs. 3 and 4). However, although NH+ 4 concentration obtained at the end being nitrified and the emissions decreased to zero when both aeration of aerobic phase of R2-AUT was below discharge limits (2.9 g NH+ 4 -N and reactor mixing stopped. N2O emissions from R1-HET were higher m−3), the AOR achieved (7.1 g NH+ 4 -N m −3 −1 d ) was lower than that on than those from R2-AUT, which can be attributed to the higher aeration day 410 and the aerobic phase was extended to the maximum length value flowrate required in R1-HET to achieve the same DOSP, increasing N2O allowed (3 h). This fact is probably due to VSS concentration being 14 % stripping. In any case, the factors affecting these N2O emissions are fur- lower in the 8-hour cycle period than in the 12-hour cycle period. ther examined in Section 3.4, where the different DO control strategies implemented are reported. 3.3. Overall GHG emissions of the pilot plant Regarding CH4 emissions, lower variability on the emissions were mea- sured compared to N2O emissions, since the aeration strategy has no effect During part of period III, N2O and CH4 emissions were monitored dur- on the CH4 production. Table 2 shows that CH4 emissions measured during ing the aerobic phases of the reactors. Table 2 shows the average N2O the operation of the pilot plant were low, with <1.0 g CH4 emitted per cycle 6 B. Solís et al. Science of the Total Environment 849 (2022) 157626 1.4 8 1.2 7 6 1 Reactor Level [m] R1-HET level 5 0.8 pH [-] R2-AUT l evel 4 0.6 R4-INT l evel 3 R1-HET pH 0.4 R2-AUT pH 2 0.2 1 A B 0 0 0 2 4 6 8 10 12 0 2 4 6 8 10 12 Cycle time [h] Cycle time[h] 40 500 35 400 R1-HET ORP 30 Temperature [ºC] R2-AUT ORP 300 25 ORP [mV] 20 200 15 100 10 R1-HET Temp. R2-AUT Temp. 0 5 0 Ambient Tem p. C D -100 0 2 4 6 8 10 12 0 2 4 6 8 10 12 Cycle time [h] Cycle time [h] 4 12 3.5 R1-HET DO R1-HET DO SP E F 10 R2-AUT DO 3 Air Flowrate [m 3 h -1] R1-HET Air Flowrate R2-AUT DO SP 8 DO [g O 2 m-3] 2.5 R2-AUT Air Flowrate 2 6 1.5 4 1 2 0.5 0 0 0 2 4 6 8 10 12 0 2 4 6 8 10 12 Cycle time [h] Cycle time [h] Fig. 4. Online monitored variables for an example of SCEPPHAR cycle operation with a cycle length configuration of 12 h (cycle belonging to day 410 of operation). and a CH4-EF lower than 0.20 %. CH4 emissions were clearly higher in R2- emissions were higher in R2-AUT compared to R1-HET due to the sequence AUT than in R1-HET and the emission rates showed a peak shape (Figs. 3C of wastewater treatment in the SCEPPHAR configuration (Fig. 1): air is and S1C). On one hand, the CH4 emission rate had an initial peak since CH4 sparged to the wastewater for the first time in the aerobic phase of R2- was not produced during aerobic conditions but the dissolved CH4 was only AUT. Therefore, the emitted CH4 in R2-AUT was probably introduced dis- stripped from the liquid phase to the gas phase. On the other hand, CH4 solved in the influent wastewater, while a slight amount might be produced Table 2 Average N2O and CH4 emissions and emissions factors measured for the cycle length configurations of 12 (n = 21) and 8 (n = 13) hours. Standard deviation is presented. Cycle length → 12 h 8h Reactor → R1-HET R2-AUT Total R1-HET R2-AUT Total N2O [g N2O-N] 0.19 ± 0.18 0.44 ± 0.15 0.64 ± 0.24 0.27 ± 0.23 0.42 ± 0.37 0.69 ± 0.43 N2O-EF [%] 0.28 ± 0.26 0.64 ± 0.22 0.93 ± 0.34 0.39 ± 0.33 0.61 ± 0.53 1.00 ± 0.62 CH4 [g CH4] 0.25 ± 0.23 0.57 ± 0.19 0.82 ± 0.30 0.06 ± 0.03 0.36 ± 0.34 0.42 ± 0.34 CH4-EF [%] 0.05 ± 0.04 0.11 ± 0.04 0.16 ± 0.06 0.01 ± 0.01 0.07 ± 0.06 0.08 ± 0.06 7 B. Solís et al. Science of the Total Environment 849 (2022) 157626 during the anaerobic phase in R1-HET. Therefore, the existing lower CH4 the dissolved N2O value, the higher the N2O emission peak. The indepen- emissions in R1-HET during the aerobic phase were due to the 30 % of re- dency from the DOSP on the peak emission rate is due to the aeration control actor volume that was not interchanged. Table 2 also reveals that the aver- system fully opened the air distribution valve at the beginning of the aero- aged CH4 emissions from the 12-hour configuration were 50 % higher than bic phase for the three DOSP (Fig. 5B). Both gas and liquid N2O measure- those measured through the 8-hour configuration since, in the first case, the ments (Fig. 5C and D) show that N2O was continuously produced during anaerobic phase was 55 % longer. That supports the hypothesis that part of the pilot plant cycle, since dissolved N2O was being accumulated and the CH4 emitted in R2-AUT could be produced during the anaerobic phase N2O was continuously emitted. A strong correlation between the stripping of R1-HET. of dissolved N2O and the DOSP was found; dissolved N2O increased when operating at lower DOSP (i.e. with lower aeration flowrate) and the N2O 3.4. Effect of the aeration strategy on N2O emissions emissions decreased due to a reduction on the mass transfer coefficient. However, the impact of the stripping effect was not relevant in these cycles This section shows the effect of the different aeration strategies on N2O since most of the N2O emission was found at the beginning of the aerobic emissions. During this period, some changes were made on the pilot plant phase during the emission peak, where the air flowrate was the same for cycle configuration: i) the cycle length was extended to 12 h, ii) the aerobic the three DOSP. Finally, Fig. 5D shows that the dissolved N2O concentration phase length of the R2-AUT was incremented to 7.4 h and iii) the aerobic decreased due to stripping to the gas phase when ammonium was depleted phase length control through the NH+ 4 ion selective probe was deactivated. and NO− 2 was accumulated for the cycles operated at 2 and 3 g O2 m −3 . These changes were made to allow a more detailed study of N2O emissions Table 3 shows that no correlation was found between the operating DO during nitrification in R2-AUT. In addition, a liquid N2O probe was and the total N2O emissions due to the high variability observed on the installed in R2-AUT to monitor the dissolved N2O concentration in order N2O emissions even at the same DOSP: the average N2O emissions from to better understand the effect of the different aeration strategies on both the cycles operated at DO of 2 g O2 m−3 were higher than those obtained N2O production and emission. when operating at a DO of 1 g O2 m−3 and of 3 g O2 m−3. In addition, The aeration strategies implemented are divided into: i) Different DOSP, the variability found in N2O emissions for the cycles operating at the ii) intermittent aeration and iii) steps on the DOSP. This section reports only same DOSP was similar to that calculated for the cycles operated at different the results related to R2-AUT reactor, where dissolved N2O and N2O gas DOSP. Therefore, in our case, the DO concentration seems to have no effect emission were measured. During these pilot plant experiments, the opera- on the N2O emissions of the pilot plant, probably because the same AOR tion of R1-HET was the same as showed in Table S1 of SI section and the re- was achieved at different DO concentrations. actor was operated with a constant DOSP of 3 g O2 m−3 in the aerobic phase. 3.4.2. Intermittent aeration The second set of experiments was conducted to assess the effect of in- 3.4.1. Different DOSP termittent aeration on the overall N2O production and emission. The inter- The first set of experiments was conducted to assess the effect of differ- mittent aeration was implemented as an on/off controller where the reactor ent DOSP on the overall N2O production and emission. The cycles were op- was aerated at a constant air flowrate and aeration was stopped when the erated at three different DOSP: 1, 2 and 3 g O2 m−3. Table 3 shows the DO measurement increased above 2 g O2 m−3 and was turned on again obtained average N2O emissions and N2O-EF in R2-AUT. Fig. 5 shows the for DO below 1 g O2 m−3. This on/off control was maintained throughout profiles of DO, air flowrate, N2O emission, soluble N2O concentration and the aerobic phase of R2-AUT. The pilot plant was operated with on/off aer- NH+ 4 concentration for three cycles at a DOSP of 1, 2 and 3 g O2 m −3 . ation control under two different air flowrates: 12.5 and 5.0 m3 h−1 (100 Fig. 5A shows that the DO in R2-AUT could be controlled properly at and 40 % of the maximum air flowrate, respectively). Fig. 6 shows the pro- different DOSP and Fig. 5E shows that full NH+4 oxidation was achieved re- files of DO, air flowrate, N2O emission, dissolved N2O concentration and gardless of the DOSP. The oscillations in DO concentration and air flowrate NH+ 4 concentration for two typical cycles with on/off control. measured at the end of the aerobic phase (Fig. 5A and B) were due to the Fig. 6A shows the variability of the DO concentration for both imple- low linearity of the valve in the low range of actuation (valve opening mented air flowrates through intermittent aeration. The DO ranged be- below 25 %), which makes it difficult for the PI controller to maintain a sta- tween 0.6 and 2.8 g O2 m−3 for the high air flowrate and between 0.6 ble DO concentration. The AOR was very similar among the cycles: 3.48, and 2.0 g O2 m−3 when the pilot plant was operated at 40 % of the maxi- 3.72 and 3.57 g N m−3 h−1 (R2 > 0.98) for DOSP of 1, 2 and 3 g O2 m−3, mum air flowrate and NH+ 4 was not depleted. DO values outside the respectively, showing that the AOR was not dependent on DO in the range 1–2 g O2 m−3 were due to the dynamics of the DO sensor and the con- range of values tested. Regarding N2O emissions, a different trend in N2O trol valve, even though the on/off controller was sending the command to emission was observed in each cycle when compared to the examples of close the aeration when the measured DO was 2 g O2 m−3 and was sending pilot plant cycles in Figs. 3 and S1. Fig. 5C shows that a N2O emission the command to open it when DO was 1 g O2 m−3. The DO concentration at peak was obtained for the three cycles at the beginning of the aerobic high airflow rate achieved 2 g O2 m−3 at 10 min of aerobic phase, while it phase of R2-AUT regardless of the DOSP. Then, it decreased rapidly to lasted 2 h in the experiment operated with the low air flowrate of 5 m3 h−1 lower constant emission rate until NH+ 4 depletion. This initial emission (Fig. 6B). The duration of the on phases decreased in time for both air peak accounts from 20 % up to 60 % of the total N2O emission measured flowrates meanwhile the oxygen uptake rate decreased due to lower NH+ 4 per cycle for the cycles operated at 2 and 3 g O2 m−3 (Fig. 5) and could concentration. The same NH+ 4 concentration was measured for both cycles be correlated to the dissolved N2O: a similar peak was detected in the at the beginning of the aerobic phase and both cycles achieved full ammo- three cycles on the dissolved N2O profile (Fig. 5D). Moreover, the higher nia oxidation to nitrite. The same AOR was achieved for both cycles (2.70 g N m−3 h−1, R2 = 0.99), showing that the AOR was not significantly differ- ent although the average DO levels were slightly different (1.4 vs 1.8 g O2 m−3 for low and high flowrate). The same fact was observed in results of Table 3 Section 3.4.1. The NH+ 4 concentration decreased linearly over the time, Average N2O emissions and N2O-EF measured in R2-AUT per cycle for different DO without any strong variation in the AOR due to the on/off aeration cycles. set-points. n represents the number of cycles operated at each DO set-point. Stan- Regarding the N2O results, a peak of dissolved N2O was measured in the dard deviation is presented. reactor at the beginning of the aerobic phase for both cycles (Fig. 6D), as in DO set-point [g O2 m–3] N2O emission [g N2O-N] N2O-EF [%] the cycles operated at different DOSP (Fig. 5, Section 3.4.1). This peak of dis- 1 (n = 7) 0.34 ± 0.27 0.50 ± 0.39 solved N2O caused a N2O emission peak at the beginning of the aerobic 2 (n = 6) 0.53 ± 0.30 0.76 ± 0.43 phase (Fig. 6C), however, the emission decreased to negligible levels in 3 (n = 34) 0.43 ± 0.48 0.62 ± 0.70 the off phases since stripping was suppressed (high flowrate scenario). 8 B. Solís et al. Science of the Total Environment 849 (2022) 157626 4 12 A B 10 Air florwrate [m3 h-1 ] 3 DO [g O 2 m -3 ] 8 2 6 4 1 2 0 0 0 2 4 6 8 0 2 4 6 8 Time [hours] Time [hours] 120 4 C D N2 O emission [g N2 O-N d -1 ] 3.5 N2 O soluble [g N2 O-N m -3 ] 100 DO DO == 11 g O2 m-3 3 80 DO DO == 22 g O2 m-3 2.5 DO DO == 33 g O2 m-3 60 2 1.5 40 1 20 0.5 0 0 0 2 4 6 8 0 2 4 6 8 Time [hours] Time [hours] 30 E NH4+ concentration [g N m-3 ] 25 20 DOSP = 1 g O2 m-3 DOSP = 2 g O2 m-3 15 DOSP = 3 g O2 m-3 10 5 0 0 2 4 6 8 Time [hours] Fig. 5. Selected monitored variables for three examples of SCEPPHAR operation with extended aeration phase length in R2-AUT using different DO set-points. (A) DO, (B) air flowrate, (C) N2O emission rate, (D) soluble N2O concentration and (E) ammonium online concentration. This behaviour was not observed in the cycle operated at a low air flowrate N2O. The averaged R2-AUT N2O-EF of all the cycles operated with intermit- because the emission peak finished before the first off phase. The dissolved tent aeration strategy (21 cycles) was 0.40 ± 0.21 %. N2O measurements shows that there was no accumulation of N2O in the liq- uid phase during all the aerobic phase for both cycles. The net N2O produc- 3.4.3. Steps on the DOSP tion seemed null since: i) the dissolved N2O concentration decreased Finally, the last set of experiments was carried out to assess the effect of constantly once the initial peak was stripped to the gas phase, and ii) the varying the DOSP during the same aerobic phase of R2-AUT on the N2O N2O liquid concentration remained constant during the initial peak during stripping and emissions. Two experiments were conducted: in the first the off phase of the cycle operated with the high air flowrate (zoom of one, the DOSP was increased every 2 h in a stepwise manner, testing values Fig. 6D). A possible explanation is that simultaneous nitrification- of 1, 2 and 3 g O2 m−3. In the second experiment, the same strategy was ap- denitrification was taking place and N2O was simultaneously produced plied but with values of 3, 2 and 1 g O2 m−3. The third step of both exper- and consumed. The R2-AUT N2O-EF measured for the cycles operated at iments lasted until the end of the aerobic phase (3.4 h). Fig. 7 shows the 12.5 and 5.0 m−3 h−1 were 0.40 and 0.48 %, respectively (Fig. 6). How- profiles of DO, air flowrate, N2O emission, dissolved N2O concentration ever, no correlation was found between the air flowrate and the emitted and NH+ 4 for both experiments. 9 B. Solís et al. Science of the Total Environment 849 (2022) 157626 4 14 A B 3.5 12 Air florwrate [m3 h -1 ] 3 10 DO [g O 2 m-3 ] 2.5 8 2 6 1.5 4 1 0.5 2 0 0 0 2 4 6 8 0 2 4 6 8 Time [hours] Time [hours] 60 4 C D N2O emission [g N 2O-N d-1] N2 O soluble [g N2O-N m-3] 3.5 50 1.4 3 1.2 40 QQ=air50% = 12.5 3 h-1 vlv.mop. 1 2.5 QQ=air30% = 5.0vlv. m3op. h-1 0.8 30 2 0.6 0.4 1.5 20 0.2 1 0 0 0.2 0.4 0.6 0.8 1 10 0.5 0 0 0 2 4 6 8 0 2 4 6 8 Time [hours] Time [hours] 25 E NH4+ concentration [g N m -3 ] 20 2 Low air flowrate 15 High air flowrate 10 5 0 0 2 4 6 8 Time [hours] Fig. 6. Selected monitored variables for two examples of SCEPPHAR operation with extended aeration phase length in R2-AUT and using on/off aeration control at 5.0 and 12.5 m3 h−1 of air flowrate. (A) DO, (B) air flowrate, (C) N2O emission rate, (D) soluble N2O concentration and (E) ammonium online concentration. Fig. 7 shows that the DO PI control loop reacted fast to setpoint changes, contrast with other experiments. Both the emission and dissolved N2O pro- modifying the aeration to reach effectively the new setpoint. On the first ex- files seemed to have a strong dependency on the air flowrate applied to R2- periment, the aeration control system opened the valve up to 90 % for short AUT. In fact, N2O emissions depend on the air flowrate and the actual liquid periods in order to reach the increased setpoints. On the second experiment, N2O concentration, while the liquid N2O concentration depends on both the the aeration control system closed the air control valve down to 10 % (and air flowrate (i.e. gas-liquid N2O transfer because of stripping) and the DO con- even fully closed) after decreasing the setpoint. Similar NH+ 4 concentrations centration (i.e. different biological N2O production pathways). On the one were measured for both experiments at the beginning of the aerobic phase. hand, the sudden opening of the air valve linked to a DOSP increase caused The AOR of the increasing DO experiment remained constant among the a sudden increase on the N2O emissions, which decreased after the DO −3 −1 DO steps (3.02 g NH+ 4 -N m h , R2 = 0.98) but the AOR of the decreasing reached the desired setpoint. These emissions led to a decrease in the soluble experiment decreased from 3.9 (R2 = 0.80) to 3.3 (R2 = 0.90), and 2.3 g N2O concentration. On the other hand, in the second experiment, the N2O −3 −1 NH+4 -N m h (R2 = 0.83) in parallel with the new DOSP of 3, 2 and 1 g emissions decreased and the dissolved N2O concentration increased every −3 O2 m , respectively (Fig. 7E). Interestingly, neither an initial peak of soluble time the DOSP was decreased as the air control valve was closing. Finally, N2O nor N2O emission was measured in any of the DO step experiment, in N2O was continuously produced in both experiments since the accumulation 10 B. Solís et al. Science of the Total Environment 849 (2022) 157626 3.5 14 A B 3 12 Air florwrate [m3 h-1] 2.5 10 DO [g O 2 m -3 ] 2 8 1.5 6 1 4 0.5 2 0 0 0 2 4 6 8 0 2 4 6 8 Time [hours] Time [hours] 20 4 C D N2 O emission [g N2 O-N d-1 ] 18 N2O soluble [g N2 O-N m-3 ] 3.5 16 QDecreasing = 50% vlv.DO op. 3 14 QIncreasing = 30% vlv. DOop. 2.5 12 10 2 8 1.5 6 1 4 2 0.5 0 0 0 2 4 6 8 0 2 4 6 8 Time [hours] Time [hours] 30 E NH4 + concentration [g N m-3 ] 25 20 Decreasing DO 15 Increasing DO 10 5 0 0 2 4 6 8 Time [hours] Fig. 7. Selected monitored variables for two examples of SCEPPHAR operation with extended aeration phase length in R2-AUT and different DO setpoints for the PI controller during the cycle. (A) DO, (B) air flowrate, (C) N2O emission rate, (D) soluble N2O concentration and (E) ammonium online concentration. of dissolved N2O increased in both experiments and a constant N2O emission 4. Discussion was measured. The accumulation of dissolved N2O seemed to be correlated with the air flowrate, since the soluble N2O accumulation increased in 4.1. Comparison of GHG emissions with other WRRFs every DOSP change in the second experiment: decreasing the air flowrate caused a decrease of N2O stripping. The N2O-EF measured in these tests Several monitoring campaigns to quantify N2O emissions have been was higher (1.3 % when DO was increased and 1.1 % when DO was de- conducted at different full-scale WWTPs. The main objective of the moni- creased) than that achieved through intermittent aeration, because even toring campaigns was to quantify the N2O emissions and later to determine though no initial N2O emission peak was measured, N2O was continuously the factors affecting these emissions (Ahn et al., 2010; Daelman et al., 2013; produced during the cycles. Kampschreur et al., 2009; Law et al., 2012; Massara et al., 2017). These 11 B. Solís et al. Science of the Total Environment 849 (2022) 157626 studies have shown huge variations in the N2O emissions in WWTPs even Another interesting observation was that the AOR was neither de- under similar conditions, demonstrating that it was very difficult to find a pendent on the DO concentration, as the same apparent AOR was clear trend. Recent long term monitoring campaigns have been able to cap- found during the implementation of different DOSPs during nitrification ture N2O seasonal emission dynamics (Daelman et al., 2015; Gruber et al., in R2-AUT. However, the AOR changed during the long-term operation 2020; Kosonen et al., 2016) showing significant seasonal variability with an of the pilot plant under different conditions, showing a high variability N2O emission factor range between 0 and 3 % of the influent N load for from 8.5 to 2.3 g N m−3 h−1. Therefore, the change on the apparent most of the municipal full-scale WWTPs, The variation on the N2O-EF has AOR could be most probably attributed to a decrease in the fraction of been attributed to the different configurations and operational conditions autotrophic organisms in the sludge. The effect of temperature on in each WWTP, as well as to the inherent dynamic conditions of the plants, AOR was not assessed, since all the presented cycles were operated at but probably some unknown factors are still to be discovered. Therefore, it a similar temperature. Only in the cycle operating in a decreasing was difficult to assess the real causes of the high N2O-EF variations and, DOSP step-wise manner (Fig. 7), the AOR was found to decrease with thus, it was difficult to design and implement novel mitigation strategies the decrease of the DOSP, although it could also be related to the de- for the plant. crease in ammonium concentration along the cycle. This study reported the long-term operation and monitoring of the Another issue regarding the high variability of the N2O emissions mainstream SCEPPHAR configuration in a pilot scale treating real was the detected N2 O emission peak at the beginning of the aerobic municipal wastewater. The fact that the pilot plant operated within phase in R2-AUT, that can represent up to the 60 % of the total N2 O shortcut N-removal was a priori unattractive in terms of N2O emissions, emitted per cycle. Yu et al. (2010) studied the NO and N2O emissions since the accumulation of NO− 2 is a major cause of N2O emissions. How- in a pure culture of AOB and found high N2O production linked to tran- ever, the average N2O-EF calculated during the long-term operation of sient conditions (from anoxic conditions to aerobic conditions) when the whole pilot plant was 0.97 ± 0.70 %, which is in the same range NH+ 4 had been accumulated. They concluded that the N2O production for most of municipal WWTPs. This shows the capability of the main- by nitrifying biomass under transient conditions was attributed to a stream SCEPPHAR configuration to obtain a high effluent quality shift in metabolism from low specific activity towards the maximum index while reducing aeration costs and COD requirements without specific activity (Yu et al., 2010). Therefore, the measured N2O emission increasing N2O emissions with respect to conventional nitrification/de- peak at the beginning of the aerobic conditions might be linked to tran- nitrification. The most likely reason for this low N2O-EF value is the fact sient conditions, since the aerobic phase of R2-AUT begins after the fill- that this system was able to avoid nitratation while operating at a high ing of the reactor with the supernatant of R1-HET reactor, with high DOSP, thanks to the use of the aeration phase control. The combination NH+ 4 concentration. This peak was not measured in all the cycles and, of high DO (mitigates N2O production) and relatively low airflow rate therefore, it seems to have a dependency on the conditions of the nitri- (mitigates N2O emissions) was beneficial. fying biomass before the aerobic phase. Others works have also found Regarding CH4 emissions, the range in municipal WWTPs is typically an effect of the transient conditions on the N2O emissions in full-scale from 0.02 to 1.13 % of the inlet COD (Daelman et al., 2013; Ribera- WWTPs (Ahn et al., 2010; Ribera-Guardia et al., 2019). Guardia et al., 2019; Rodriguez-Caballero et al., 2014). The average When describing the influence of the DOSP on N2O emissions, it should CH4 -EF measured in this study was 0.12 ± 0.08 % which is in the be noted that a higher DOSP needs a higher airflow rate in the reactor lower band of the values reported. Regarding the possible production (Fig. 5B). The average air flowrate increased from 3.35 m3 h−1 (DOSP = of CH4 during the anaerobic phase of R1-HET, textbook knowledge 1) to 4.08 m3 h−1 (DOSP = 2) and 4.78 m3 h−1 (DOSP = 3) i.e. 22 % states that methanogenesis is favoured under mesophilic conditions (T and 43 % higher for 2 and 3 with respect to 1. This aeration flowrate in- around 37 °C), strict anaerobic conditions and non-limiting acetate con- crease favours N2O stripping and may, at least in a short-term, increase centration for acetoclastic methanogens. However, it was experimen- N2O emissions. The operation of the pilot plant through different DOSPs tally observed that CH4 emissions decreased by 50 % when the in R2-AUT revealed the effect of the stripping between liquid and gas anaerobic phase was shortened by 55 % suggesting that a non- phases on the N2O liquid accumulation and the N2O measured emission. negligible amount of the emitted CH4 was produced during the anaero- Fig. 5 shows that when the pilot plant was operated at a low DO (i.e. with bic phase of the R1-HET cycle. The presence of methanogens in the low aeration flowrate), the accumulation of N2O in the liquid increased. planktonic biomass can probably be discarded, as this sludge undergoes This increase could be attributed to the negative effect that decreasing the aerobic conditions during part of the cycle, while methanogens are DO has on N2O production through the nitrifier denitrification pathway strictly anaerobic. However, the presence of some methanogens in the (Massara et al., 2018, 2017; Peng et al., 2015) or to a decrease in the trans- possible biofilm of the reactor walls and in the sediments that can be ac- fer rate between liquid and gas phase (i.e. the N2O mass transfer coefficient cumulated at the bottom of the reactor could explain this slight methane decreased due to a decrease in the air flowrate). Hence, although the N2O production. The CH 4 present at the end of the anaerobic phase is production seems to increase when operating at lower DO, the mass trans- stripped/emitted during the posterior aerobic phase. fer between phase decreases. The accumulated N2O at the end of the aero- bic phase of R2-AUT does not have a negative effect on the overall N2O 4.2. Factors affecting N2O emissions emissions of the cycle since, once the aerobic phase of R2-AUT is finished, the supernatant of the reactor is sent to R1-HET to perform the anoxic N2O emissions occurred throughout all the long-term operation. High var- phase, where N2O is reduced to N2. iability was detected not only due to the applied operational changes but also Therefore, a possible mitigation strategy is to operate at a low DO when operating the plant under the same operational conditions. The DOSP in (down to 1 g O2 m−3), because less N2O is emitted due to limited mass R2-AUT did not seem to have any effect on the overall N2O emissions, since transfer rate between phases and the increased accumulation of N2O will no correlation was found between the DO concentration in R2-AUT and the be subsequently denitrified in R1-HET, and not emitted. This strategy shifts measured emissions during the operation at different DOSP between 1 and the conventional paradigm for mitigating N2O emissions, as it focusses on 3 g O2 m−3 (Fig. 5 and Table 3). This fact is in disagreement with previous preventing N2O transfer from liquid to the gas phase rather than focusing experimental reports where a decrease in the DO level led to an increase of on N2O production, since the potential dissolved N2O would be denitrified the N2O emissions (Massara et al., 2017; Peng et al., 2015). A possible expla- in a subsequent step. nation to this is that the lowest DOSP implemented was 1 g O2 m−3, which Finally, the aeration strategy implemented that emitted less N2O was probably was not low enough to trigger the increase of N2O production. In operating the plant through intermittent aeration. The results of the inter- this sense, Peng et al. (2015) measured the highest N2O emissions among dif- mittent aeration showed a decrease of 40 % on the N2O-EF of R2-AUT ferent NO−2 accumulations at 0.35 and 0.85 g O2 m −3 through a range of ap- compared to normal operation of the pilot plant (0.40 ± 0.21 % vs. −3 plied DOs from 0.35 to 3.5 g O2 m . 0.64 ± 0.22 %). A possible explanation is that simultaneous nitrification 12 B. Solís et al. Science of the Total Environment 849 (2022) 157626 and denitrification occurred during the off cycles of the aeration since the Appendix A. Supplementary data denitrifying bacteria can denitrify even in a micro-aerobic environment (Massara et al., 2017). Therefore, the initial peak emissions observed in Supplementary data to this article can be found online at https://doi. some experiments (Fig. 5c), attributed to transient conditions, could be org/10.1016/j.scitotenv.2022.157626. mitigated implementing intermittent aeration strategies leading to simulta- neous nitrification and denitrification. Rodriguez-Caballero et al. (2015) also proposed a cycle configuration with intermittent aeration to mitigate References the N2O emissions from a full-scale SBR treating municipal wastewater. Ahn, J.H., Kim, S., Park, H., Rahm, B., Pagilla, K., Chandran, K., 2010. N2O emissions from acti- vated sludge processes, 2008–2009: results of a national monitoring survey in the United 5. Conclusions States. Environ. Sci. Technol. 44, 4505–4511. https://doi.org/10.1021/es903845y. APHA, 1995. Standard Methods for the Examination of Water and Wastewater. 19th Editi. ed. This study assesses the plant performance and GHG emissions of the American Public Health Association, Washington, DC. USA. http://www. standardmethods.org. mainstream SCEPPHAR novel WRRF configuration at pilot scale and Chen, H., Zeng, L., Wang, D., Zhou, Y., Yang, X., 2020. Recent advances in nitrous oxide pro- under real environmental conditions and discusses the effect that different duction and mitigation in wastewater treatment. Water Res. 184, 116168. https://doi. aeration strategies have on the N2O production and emissions. The main org/10.1016/j.watres.2020.116168. findings are: Cieślik, B., Konieczka, P., 2017. A review of phosphorus recovery methods at various steps of wastewater treatment and sewage sludge management. The concept of “no solid waste generation” and analytical methods. J. Clean. Prod. 142, 1728–1740. https://doi.org/ - Successful removal efficiencies of C, N and P were achieved for a long- 10.1016/j.jclepro.2016.11.116. term in the pilot plant under shortcut N-removal operating the pilot Conthe, M., Lycus, P., Arntzen, M.Ø., Ramos da Silva, A., Frostegård, Å., Bakken, L.R., Kleerebezem, R., van Loosdrecht, M.C.M., 2019. Denitrification as an N2O sink. Water plant at 8-hour and 12-hour configuration. Res. 151, 381–387. https://doi.org/10.1016/j.watres.2018.11.087. - GHG emissions (N2O and CH4) showed high variability and were in the Cordell, D., Rosemarin, A., Schröder, J.J., Smit, A.L., 2011. Towards global phosphorus secu- low range of typical emission factors measured in conventional full- rity: a systems framework for phosphorus recovery and reuse options. Chemosphere 84, 747–758. https://doi.org/10.1016/j.chemosphere.2011.02.032. scale WWTPs, even with high nitrite accumulation, which a priori was Daelman, M.R.J., Van Voorthuizen, E.M., Van Dongen, L.G.J.M., Volcke, E.I.P., Van unattractive in terms of N2O emissions. Loosdrecht, M.C.M., 2013. Methane and nitrous oxide emissions from municipal waste- - Operating the R2-AUT of the pilot plant at different DOSPs did not show water treatment - results from a long-term study. Water Sci. Technol. 67, 2350–2355. a clear trend on the N2O-EF of the pilot plant, within the DO ranges ap- https://doi.org/10.2166/wst.2013.109. Daelman, M.R.J., van Voorthuizen, E.M., van Dongen, U.G.J.M., Volcke, E.I.P., van plied (1 to 3 g O2 m−3). However, the operation at the lowest value re- Loosdrecht, M.C.M., 2015. Seasonal and diurnal variability of N2O emissions from a sulted in the lowest N2O emissions, probably due to the fact that the full-scale municipal wastewater treatment plant. Sci. Total Environ. 536, 1–11. https:// lower the DO value, the lower the aeration needs and the lower mass doi.org/10.1016/j.scitotenv.2015.06.122. EEC Council, 1991. Council directive of 21 may 1991 concerning urban waste water treat- transfer from the liquid to the gas. Dissolved N2O accumulated in the ment (91/271/EEC). Off. J. Eur. Communities (OJ L 135, 30.5.1991), 40–52. http:// liquid gas and was denitrified in a posterior phase. data.europa.eu/eli/dir/1991/271/oj. - A peak of N2O emission was found in many cycles of the pilot plant, at- Gruber, W., Villez, K., Kipf, M., Wunderlin, P., Siegrist, H., Vogt, L., Joss, A., 2020. N2O emis- sion in full-scale wastewater treatment: proposing a refined monitoring strategy. Sci. tributed to the transient conditions of AOB, at the beginning of the aer- Total Environ. 699, 134157. https://doi.org/10.1016/j.scitotenv.2019.134157. obic phase of the R2-AUT operation. Guisasola, A., de Haas, D., Keller, J., Yuan, Z., 2008. Methane formation in sewer systems. - The aeration strategy implemented that most mitigated the N2O emis- Water Res. 42, 1421–1430. https://doi.org/10.1016/j.watres.2007.10.014. Guo, J.H., Peng, Y.Z., Wang, S.Y., Zheng, Y.N., Huang, H.J., Ge, S.J., 2009. Effective and ro- sions in R2-AUT was the intermittent aeration, reducing the N2O emis- bust partial nitrification to nitrite by real-time aeration duration control in an SBR sions by 40 % compared to normal operation of the plant. treating domestic wastewater. Process Biochem. 44, 979–985. https://doi.org/10. 1016/j.procbio.2009.04.022. Gutierrez, O., Sudarjanto, G., Ren, G., Ganigué, R., Jiang, G., Yuan, Z., 2014. Assessment of pH CRediT authorship contribution statement shock as a method for controlling sulfide and methane formation in pressure main sewer systems. Water Res. 48, 569–578. https://doi.org/10.1016/j.watres.2013.10.021. Borja Solís was the person in charge of conducting all the analytical IPCC, 2014. In: Pachauri, R.K., Meyer, L.A. (Eds.), Climate Change 2014: Synthesis Report. Contribution of Working Groups I, II and III to the Fifth Assessment Report of the Inter- work and to write the first draft of the document: formal analysis; governmental Panel on Climate Change. IPCC, Geneva, Switze. http://hdl.handle.net/ methodology; writing – original draft. Albert Guisasola was in charge 10013/epic.45156.d001. of the Conceptualization, Data curation, Formal analysis, Supervision, Jubany, I., Lafuente, J., Baeza, J.a, Carrera, J., 2009. Total and stable washout of nitrite oxi- Visualization and review & editing the document. Maite Pijuan was in dizing bacteria from a nitrifying continuous activated sludge system using automatic con- trol based on Oxygen Uptake Rate measurements. Water Res. 43, 2761–2772. https:// charge of the Conceptualization, Data curation, Formal analysis & doi.org/10.1016/j.watres.2009.03.022. editing the document. Juan Antonio Baeza was in charge of Conceptual- Kampschreur, M.J., Temmink, H., Kleerebezem, R., Jetten, M.S.M., van Loosdrecht, M.C.M., ization; Data curation; Formal analysis; Funding acquisition; Project 2009. Nitrous oxide emission during wastewater treatment. Water Res. 43, 4093–4103. https://doi.org/10.1016/j.watres.2009.03.001. administration; Supervision, Visualization and review & editing. Kosonen, H., Heinonen, M., Mikola, A., Haimi, H., Mulas, M., Corona, F., Vahala, R., 2016. Ni- trous oxide production at a fully covered wastewater treatment plant: results of a long- Declaration of competing interest term online monitoring campaign. Environ. Sci. Technol. 50, 5547–5554. https://doi. org/10.1021/acs.est.5b04466. Larriba, O., Rovira-Cal, E., Juznic-Zonta, Z., Guisasola, A., Baeza, J.A., 2020. Evaluation of the The authors declare that they have no known competing financial inter- integration of P recovery, polyhydroxyalkanoate production and short cut nitrogen re- ests or personal relationships that could have appeared to influence the moval in a mainstream wastewater treatment process. Water Res. 172, 115474. https://doi.org/10.1016/j.watres.2020.115474. work reported in this paper. Law, Y., Ye, L., Pan, Y., Yuan, Z., 2012. Nitrous oxide emissions from wastewater treatment processes. Philos. Trans. R. Soc. B Biol. Sci. 367, 1265–1277. https://doi.org/10.1098/ Acknowledgements rstb.2011.0317. Mannina, G., Butler, D., Benedetti, L., Deletic, A., Fowdar, H., Fu, G., Kleidorfer, M., McCarthy, D., Steen Mikkelsen, P., Rauch, W., Sweetapple, C., Vezzaro, L., Yuan, Z., Willems, P., Borja Solís is grateful to the PIF scholarship from the Universitat 2018. Greenhouse gas emissions from integrated urban drainage systems: where do we Autònoma de Barcelona. B. Solís, A. Guisasola and J. Baeza are members stand? J. Hydrol. 559, 307–314. https://doi.org/10.1016/j.jhydrol.2018.02.058. of the GENOCOV research group (Grup de Recerca Consolidat de la Marcelino, M., Wallaert, D., Guisasola, A., Baeza, J.A., 2011. A two-sludge system for simulta- neous biological C, N and P removal via the nitrite pathway. Water Sci. Technol. 64, Generalitat de Catalunya, 2017 SGR 1175, www.genocov.com) and M. 1142. https://doi.org/10.2166/wst.2011.398. Pijuan is member of the ICRA-TECH consolidated research group (2017 Massara, T.M., Solís, B., Guisasola, A., Katsou, E., Baeza, J.A., 2018. Development of an ASM2d- SGR 1318). This work was funded by the SMART-Plant project (Scale-up N2O model to describe nitrous oxide emissions in municipal WWTPs under dynamic condi- tions. Chem. Eng. J. 335, 185–196. https://doi.org/10.1016/j.cej.2017.10.119. of low-carbon footprint Material Recovery Techniques, EUH2020, grant Massara, T.M.T.M., Malamis, S., Guisasola, A., Baeza, J.A., Noutsopoulos, C., Katsou, E., 2017. agreement 690323). A review on nitrous oxide (N2O) emissions during biological nutrient removal from 13 B. Solís et al. Science of the Total Environment 849 (2022) 157626 municipal wastewater and sludge reject water. Sci. Total Environ. 596–597, 106–123. Rodriguez-Caballero, A., Aymerich, I., Poch, M., Pijuan, M., 2014. Evaluation of process condi- https://doi.org/10.1016/j.scitotenv.2017.03.191. tions triggering emissions of green-house gases from a biological wastewater treatment sys- Mavinic, D.S., Turk, O., 1987. Benefits of using selective inhibition to remove nitrogen from tem. Sci. Total Environ. 493, 384–391. https://doi.org/10.1016/j.scitotenv.2014.06.015. highly nitrogenous wastes. Environ. Technol. Lett. 8, 419–426. https://doi.org/10. Vasilaki, V., Massara, T.M., Stanchev, P., Fatone, F., Katsou, E., 2019. A decade of nitrous 1080/09593338709384500. oxide (N2O) monitoring in full-scale wastewater treatment processes: a critical review. Mayer, B.K., Baker, L.A., Boyer, T.H., Drechsel, P., Gifford, M., Hanjra, M.A., Parameswaran, Water Res. 161, 392–412. https://doi.org/10.1016/j.watres.2019.04.022. P., Stoltzfus, J., Westerhoff, P., Rittmann, B.E., 2016. Total value of phosphorus recovery. Wang, D.-B., Li, X.-M., Yang, Q., Zheng, W., Wu, Y., Zeng, T., Zeng, G., 2012. Improved bio- Environ. Sci. Technol. 50, 6606–6620. https://doi.org/10.1021/acs.est.6b01239. logical phosphorus removal performance driven by the aerobic/extended-idle regime Peng, L., Ni, B.J., Ye, L., Yuan, Z., 2015. The combined effect of dissolved oxygen and nitrite with propionate as the sole carbon source. Water Res. https://doi.org/10.1016/j. on N2O production by ammonia oxidizing bacteria in an enriched nitrifying sludge. watres.2012.04.036. Water Res. 73, 29–36. https://doi.org/10.1016/j.watres.2015.01.021. Wunderlin, P., Mohn, J., Joss, A., Emmenegger, L., Siegrist, H., 2012. Mechanisms of N2O pro- Remy, C., Jossa, P., 2015. Life cycle assessment of selected processes for P recovery from sew- duction in biological wastewater treatment under nitrifying and denitrifying conditions. age sludge, sludge liquor, or ash. Deliverable 9.2 of P-REX project. In: Remy, C (Ed.), Sus- Water Res. 46, 1027–1037. https://doi.org/10.1016/j.watres.2011.11.080. tainable Sewage Sludge Management Fostering Phosphorus Recovery and Energy Yu, R., Kampschreur, M.J., van Loosdrecht, M.C.M., Chandran, K., 2010. Mechanisms and spe- Efficiency. Kompetenzzentrum Wasser Berlin gGmbH. http://publications.kompetenz- cific directionality of autotrophic nitrous oxide and nitric oxide generation during tran- wasser.de/pdf/Remy-2015-821.pdf. sient anoxia. Environ. Sci. Technol. 44 (4), 1313–1319. https://doi.org/10.1021/ Ribera-Guardia, A., Bosch, L., Corominas, L., Pijuan, M., 2019. Nitrous oxide and methane es902794a. emissions from a plug-flow full-scale bioreactor and assessment of its carbon footprint. J. Clean. Prod. 212, 162–172. https://doi.org/10.1016/j.jclepro.2018.11.286. Rodriguez-Caballero, A., Aymerich, I., Marques, R., Poch, M., Pijuan, M., 2015. Minimizing N2O emissions and carbon footprint on a full-scale activated sludge sequencing batch re- actor. Water Res. 71, 1–10. https://doi.org/10.1016/j.watres.2014.12.032. 14