Aerobic granular sludge (AGS) in continuous-flow reactors (CFRs) has attracted significant interest, with notable progress in research and application over the past two decades. Cumulative studies have shown that AGS-CFRs exhibit comparable morphology, settleability, and pollutant removal efficiency to AGS cultivated in sequencing batch reactors, despite their smaller particle sizes. Shear force and selection pressure are the primary drivers of granulation. While not mandatory for granulation, feast/famine conditions play a crucial role in ensuring long-term stability and nutrient removal. Additionally, bioaugmentation can facilitate the granulation process. Furthermore, this paper comprehensively assesses the application of AGS-CFRs in full-scale wastewater treatment plants (WWTPs). Currently, AGS-CFRs have been implemented in nine WWTPs, encompassing two distinct processes. Hydrocyclone-based densified activated sludge significantly enhances sludge density, settleability, and biological phosphorus removal efficiency, thus increasing treatment capacity. The microaerobic–aerobic configuration with internal separators can induce granulation, ensuring long-term stability, eliminating the need for external clarifiers, and reducing land and energy requirements. This review demonstrates the high potential of AGS-CFRs for intensifying existing WWTPs with minimal retrofitting needs. However, further research is required in granulation mechanisms, long-term stability, and nutrient removal to promote the widespread adoption of AGS.

  • Over 100 papers on AGS in CFRs were systematically reviewed.

  • Full-scale applications of conventional AGS in CFRs were assessed.

  • Long-term stability, particularly in the cold season, was the primary challenge.

  • Further research on granulation and nutrient removal mechanisms is required.

In order to achieve a sustainable development between the rapidly growing urban population and limited resources, modern wastewater treatment plants (WWTPs) face the daunting challenge of increasing treatment capacity while meeting more stringent effluent discharge standards and reducing their energy demand and footprint. Aerobic granular sludge (AGS) is a specific form of biofilm arisen from microbial self-aggregation, exhibiting excellent settleability, high biomass retention, and the capability to simultaneously remove organics, nitrogen, and phosphorus (de Kreuk et al. 2007). Compared to conventional activated sludge, AGS offers a significant reduction in footprint, reaching up to 50%, along with substantial energy savings of up to 40% (Pronk et al. 2015; Ekholm et al. 2022). Therefore, AGS has been recognized as a promising wastewater treatment technology that can minimize energy and footprint demands (Winkler & van Loosdrecht 2022). Over 100 WWTPs have validated the feasibility of AGS in sequencing batch reactor(s) (SBR(s)). However, as most existing WWTPs operate continuously, AGS can only be widely applied if they can be maintained in continuous-flow reactor(s) (CFR(s)).

Research conducted in SBRs demonstrated that shear force, selection pressure, feast/famine conditions, etc. play critical roles in aerobic granulation (van Dijk et al. 2022). However, there are several inherent challenges to creating these conditions in CFRs, including: (i) feast/famine conditions: most CFRs used for laboratory studies are completely mixed and full-scale facilities usually use long tanks. The substrate concentration gradients in reactors are too low to create feast/famine conditions. (ii) Selection pressure: short settling time is commonly used in SBRs to selectively waste biomass with lower settling velocity and consequently increase AGS proportion. However, secondary clarifiers in existing WWTPs are typically designed with a long settling time of 2–4 h to obtain clear effluent. (iii) Shear force: SBRs typically use column-type reactors, which can generate significant shear forces to promote granulation. However, most full-scale CFRs have flat geometries with lower shear forces. Additionally, traditional sludge return equipment mechanically crushes AGS, making aerobic granulation in CFRs more difficult.

To address the issues mentioned earlier, extensive research efforts have been dedicated to the investigation of AGS-CFRs over the past two decades. Nonetheless, there is a deficiency in evaluating the effectiveness of proposed startup strategies and reactor configurations. Furthermore, previous reviews have predominantly focused on the application status of AGS-SBRs or anammox-supported AGS (Franca et al. 2018; Hamza et al. 2022), rather than addressing AGS-CFRs. Therefore, a timely review of studies on AGS-CFRs is imperative. In this study, a search utilizing the keywords ‘aerobic granular sludge’ OR ‘aerobic granule’ OR ‘aerobic granulation’ AND ‘continuous’ was conducted in Scopus database, yielding a total of 247 outcomes. After applying the filtering function of the Scopus search engine and manually scrutinizing titles and abstracts, a selective review was conducted on over 100 research papers specifically relevant to conventional AGS within CFRs. The objectives are to assess the research and applications of AGS-CFRs, identify existing technological gaps, propose future research directions, and evaluate the feasibility of AGS-CFRs technology in intensifying existing WWTPs.

In 2006, a workshop proposed the definition of AGS as microbial aggregates that do not coagulate under reduced hydrodynamic shear and settle significantly faster than activated sludge (de Kreuk et al. 2007). The minimum size is set at 0.2 mm, which can be adjusted per case/granule type, provided the other requirements are satisfied. Moreover, extensive research suggests the adoption of a sludge volume index (SVI30) below 60 mL/g and an SVI5/SVI30 ratio approaching 1.0 as quantitative indicators for AGS (Kent et al. 2018).

Among the different types of AGS, conventional AGS exhibits compact structures, clear boundaries, high settling velocities of 15–40 m/h, and abilities to simultaneously remove organic matter and nutrients within a single reactor (Kent et al. 2018). Another type is nitrifying/short-cut nitrifying AGS, which shares similar morphological and settling properties with conventional AGS but primarily performs nitrification or short-cut nitrification (Zhu et al. 2018). Red-color anammox-supported AGS achieves autotrophic nitrogen removal and consists of an inner anaerobic layer of anammox bacteria and an outer aerobic layer containing nitrifying bacteria (Yang et al. 2023). Filamentous aerobic granules (FAGS) are characterized by the entanglement of filamentous bacteria, exhibiting a loose, fluffy, and irregular structure with an SVI30 value exceeding 100 mL/g (Song et al. 2023). Under suitable conditions, FAGS can undergo a gradual transformation into conventional AGS. The final type is synergetic algal–bacterial granules, which rely on a mutualistic symbiosis between algae and bacteria (Si et al. 2022). This review primarily focuses on conventional AGS, and unless explicitly stated otherwise, the term AGS refers specifically to conventional AGS throughout the article.

In CFRs, well-cultivated AGS has been characterized by spherical shape, compact structures, clear outline boundaries, and smooth surfaces, with typical colors of brown, brown–yellow, or white. The characteristics data of AGS-CFRs, as reported in the available literature, are presented in Table 1.

Table 1

Reported laboratory and pilot-scale investigations on AGS-CFRs

Configuration and scaleWastewaterHRTOLRInoculumStartup timeStable timeAGS characteristics
MLSSEPSTreatment performanceReferences
ColorSizeSVI30
Upflow sludge bed reactor with three-phase separators (0.89 L) Synthetic wastewater 10 h 1–3 kg/(m3·day) Activated sludge and anaerobic digested sludge 27 days 23 days Brown 1.5–3.1 mm 50–120 mL/g    Chen et al. (2009)  
Column-type reactor with a submerged membrane (3.4 L) Synthetic wastewater 27 h 7 kg/(m3·day) Mature AGS  216 days Brown–yellow 1.9 mm  10.2 g/L  R-COD = 89% Juang et al. (2010)  
Column-type reactor (0.8 L) Synthetic wastewater 2.2–19 h 6–39 kg/(m3·day) Mature AGS chemically modified with MgCO3  220 days      R-COD = 20–95% Lee & Chen (2015)  
Column-type reactor with a submerged filter module (4 L) Synthetic wastewater 6.7 h 1.4 kg/(m3·day) Mature AGS  32 days  0.5 mm  5 g/L 135.0 mg/g MLVSS R-COD = 88.5%, R- = 96.5% Li et al. (2012)  
Upflow fluidized bed reactor (3.6 L) Synthetic wastewater 6–7.2 h 3–6 kg/(m3·day) Activated sludge 91 days About 60 days   85 mL/g 3 g/L  R-COD > 90%, R- > 90% Kishida et al. (2012)  
Modified airlift fluidized bed reactor (5.6 L) Synthetic wastewater 2–3 h 1.2–5 kg/(m3·day) Activated sludge 20 days 12 days  0.6 mm 40 mL/g   E-COD = 50 mg/L Zhou et al. (2013a, 2013b
Biofilm 25 days 55 days   35–60 mL/g   R-COD = 90%, R- = 60%, R-TN < 30% Yang et al. (2014)  
Cylindrical CSTRs with demi-submerged effluent discharge tube (R1–3 L, R2–6 L) Diluted pig slurry 1.6 h 2.4–12 kg/(m3·day) R1-activated sludge, R2- pig slurry 227 days  White 4–6.8 mm SVI10 = 127 mL/g   R-COD = 30–80%, R-TN = 10–15% Morales et al. (2012)  
Connected an airlift reactor, a settling tank, a dynamic membrane tank and a sludge selection tank with a sieve in series Synthetic wastewater →septic tank wastewater 13 h 0.5 kg/(m3·day) Mature AGS  50 days  0.1–1.0 mm (60%)    R-COD = 83%, R- = 73%, R-TN = 67%, R-TP = 60% Liu et al. (2012)  
Connected an airlift reactor (7.5 L) and a sludge selection tank with a sieve in series Synthetic wastewater 9 h 1 kg/(m3·day) Activated sludge 7 days 30 days Yellow–white 1.0–3.0 mm (>80%) 35 mL/g 3 g/L  E-COD < 50, E- < 1, E-TN < 5, E-TP < 1 mg/L Liu et al. (2014)  
CSTRs with internal settling zone (6 L) Swine wastewater 3 days 0.4 kg/(m3·day) Activated sludge More than 2 years   0.1 mm     Mota et al. (2014)  
Reverse flow baffled reactor (120 L) 30% municipal and 70% industrial wastewater 5.5–16.4 h 0.5–1.6 kg/(m3·day) Activated sludge 21 days 130 days  0.13 mm 33 mL/g 4.3 g/L 540 (PN) + 47 (PS) mg/g VSS R-BOD5 = 90–94%, R- = 97–100% Li et al. (2015)  
Aeration tank with a two-zone sedimentation tank (26.8 L) 25% municipal and 75% industrial wastewater 18 h 016 kg/(m3·day) Activated sludge 62 days 90 days  0.1 mm 26 mL/g 5 g/L 71 (PN) + 0.4 (PS) mg/g VSS E-COD < 65, E- < 1.4 mg/L Zou et al. (2018)  
Aeration tank with a two-zone sedimentation tank (140 L) Municipal wastewater 16 h 0.27–0.53 kg/(m3·day) Dewatered sludge; 18 days 60 days  0.66 mm <40 mL/g 3.6 g/L 65.3 mg/g MLVSS R-COD = 83–93%, R- = 85–100%, R-TN = 12–22% Xu et al. (2020)  
AAO process with a two-zone sedimentation tank (202 L) Municipal wastewater 10–13.5 h 0.61–1.05 kg/(m3·day) Activated sludge 40 days 75 days  0.2 mm 47.5 mL/g 3.5 g/L 126.2 (PN) + 62.7 (PS) mg/g VSS E-COD = 22.8, E-TN = 3.5, E-TP = 0.2 mg/L Liu et al. (2020)  
CSTR Synthetic wastewater 8 h 1.4 kg/(m3·day) Activated sludge with or without filamentous bacteria 21 days 70 days  0.18–1.25 mm 100 mL/g 3.5 g/L  E-COD < 40, E- < 1 mg/L Chen et al. (2015)  
Modified internal loop airlift reactor (4.75 L) Synthetic wastewater 8 h 3 kg/(m3·day) Mature AGS  67 days  1.2 mm 2 mL/g 6.7 g/L  R-COD > 97% Liu et al. (2015)  
CSTR with a hollow fiber membrane (14 m3Municipal wastewater 6–12 h 0.6–1.2 kg/(m3·day) Activated sludge  110 days  0.6 mm (60%) 50 mL/g  45 (PN) + 33 (PS) mg/g VSS  Sajjad et al. (2016)  
CSTR with an internal settling zone (3 L) Synthetic wastewater   Activated sludge  35 days  0.36 mm 91 mL/g  72 (PN) + 31 (PS) mg/g SS R-TOC = 62%, R-TN = 83%, R-TP = 64% Sajjad & Kim (2015)  
CSTR with an internal settling zone (10 L) Synthetic wastewater   Mature AGS  12 weeks      R-COD = 80–99%, R- = 52–98%, R-TP = 5–87% Manea & Bumbac (2019)  
Dairy wastewater 10 h 3.8 kg/(m3·day) Mature AGS  10 weeks  0.2–1 mm    R-COD = 81–93%, R- = 83–99%, R-TN = 52–82%, R-TP = 65–99% Bumbac et al. (2015)  
Serpentine flow reactor coupled with membrane (7.5 L) Synthetic wastewater 7.5 h 3.8 kg/(m3·day) Mature AGS  30 days  1.3 mm  8 g/L 600 mg/g MLVSS R-COD = 90% Corsino et al. (2016)  
Two-cylindrical reactor with a settling tank (15 L) Synthetic wastewater 6 h 1.2 kg/(m3·day) Mature EBPR-AGS  More than 6 months  0.9 mm 20 mL/g 6 g/L  R-COD = 97%, R-TP = 96% Li et al. (2016a, 2016b
Four SBRs connected in series (4 × 6.16 L) Synthetic wastewater →municipal wastewater   Mature EBPR-AGS     55–67 mL/g 3.1 g/L  R-COD = 82%, R-TN = 71%, R-TP = 31% Li et al. (2019a)  
Modified airlift reactor with an internal settling zone with or without an anaerobic tank (20 L and 35 L) Synthetic wastewater 6–20 h 0.4–1.2 kg/(m3·day) 80% activated sludge and 20% mature EBPR-AGS 30 days 50 days  0.9–1.0 mm 40–54 mL/g 2.1–2.5 g/L  R-COD = 60–90%, R-TN = 26–82%, R-TP = 36–82% Li et al. (2021)  
Cylindrical reactor with a three-phase separator (2 L) Synthetic wastewater 6 h 0.8–3.2 kg/(m3·day) Anaerobic granules 45 days 27 days Dark yellow 1–3 mm    E-COD < 40 mg/L, R- = 100%, R-TN = 93% Sun et al. (2017)  
CSTR coupled with membrane (36 L) Synthetic wastewater 5 h 0.2–0.32 kg/(m3·day) Activated sludge 37 days 73 days  0.2 mm 100 mL/g >7 g/L  R-COD >80%, R-TN >80% Chen et al. (2017)  
CSTR coupled with membrane (20 L) synthetic wastewater 10 h 0.84 kg/(m3·day) activated sludge 28 days 67 days  0.8 mm  8 g/L 32 (PN) + 30 (PS) mg/g VSS R-COD > 90%, R-TN > 85% Dai et al. (2020)  
Modified airlift reactor with an internal settling zone (4.3 L) Synthetic wastewater  0.6–1 kg/(m3·day) 80% bacterium TN-14 cells and 20% activated sludge 40 days 32 days  0.5–2 mm 44–61 mL/g  512 (PN) + (5.6–12.6) (PS) mg/g VSS  Xin et al. (2017)  
Two CSTRs and a clarifier connected in series Synthetic wastewater 3.8–6.5 h 2.2–4.4 kg/(m3·day) Activated sludge     SVI10 = 70 mL/g   R-COD > 95%, R- = 15–30% Cofré et al. (2018)  
Five CSTRs and a clarifier connected in series (60 L) Municipal wastewater 6 h 1.5 kg/(m3·day) Activated sludge    0.28 mm  <0.5 g/L  E-COD = 25, E-TN = 11, E-PO43−-P = 0.1 mg/L Devlin & Oleszkiewicz (2018)  
Anaerobic/oxic reactor, and a clarifier connected in series (49 L) Synthetic wastewater →municipal wastewater  0.19–0.36 kg/(m3·day) Activated sludge    0.3 mm (50%) 102 mL/g 1.8–4.0 g/L  R-COD = 92%, R- = 99%, R-TN = 80%, R-TP < 40% Jahn et al. (2019)  
Modified airlift reactor with an internal settling zone (22 L) Synthetic wastewater 9 h 0.8 kg/(m3·day) Mature EBPR-AGS  116 days  0.85 mm 50 mL/g 4 g/L 31 (PN) + 38 (PS) mg/g MLSS R-COD > 90%, R- = 100%, R-TN > 65%, R-TP = 60–65% Li et al. (2020a)  
200 days  0.88 mm 40 mL/g 4 g/L 44 (PN) + 44 (PS) mg/g MLSS R-COD = 95%, R-TN = 87%, R-TP = 82% Li et al. (2019b)  
Modified airlift reactor with an anaerobic zone and internal settling zone (20 L) Synthetic wastewater 9 h 0.8 kg/(m3·day) Mature EBPR-AGS  110 days  0.9 mm 47 mL/g 4.1 g/L  R-COD = 95%, R-TN = 87%, R-TP = 81% Li et al. (2020b)  
CSTR coupled with membrane (15 L) Synthetic wastewater 12 h 1 kg/(m3·day) 20% mycelial pellets and 80% activated sludge 40 days 130 days Yellow-margins and brown-center; brown-margin and black-center 0.7–0.8 mm 78 mL/g 13 g/L 34–38 (PN) + 11–18 (PS) mg/g VSS R-COD = 97–99%, R- = 100%, R-TN = 32–42% Xiao et al. (2022)  
Several CSTRs and a sludge selector Municipal wastewater 6.5 h 1.2 kg/(m3·day) Activated sludge with a d50 = 0.3 mm 90 days   3.4 mm 64 mL/g 2.4 g/L 59 (PN) + 15 (PS) mg/g VSS R-COD = 60%, R- > 90% Sun et al. (2019)  
Several CSTRs, a sludge selector and a conventional clarifier (128 L)     MLVSS ≈ 1 g/L  E-COD < 50, E- < 1 mg/L An et al. (2021)  
Cylinder airlift reactor with an integrated separator (25 L) Real ethylene glycol wastewater 24 h 0.2–2.2 kg/(m3·day) Mature AGS        R-COD = 50%, R- > 95% Qi et al. (2020)  
Modified airlift loop reactor (180 L) Municipal wastewater 10.8 h 0.4–0.7 kg/(m3·day) Activated sludge 61 days 70 days Yellow 2–4 mm 35 mL/g 4 g/L 200 mg/g MLVSS R-COD = 89%, R-TN = 89%, R-TP = 88% Li et al. (2020c)  
Microaerobic–aerobic reactor with internal separators (3,000 m3/day) Municipal wastewater   Activated sludge  More than 2 months  90.1 μm    R- = 98%, R-TN = 85% Yu et al. (2023b)  
A simulated conceptual hybrid granule/floc process in SBRs (3 L) Synthetic wastewater   50% activated sludge and 50% AGS    0.43 mm (35%) 50.8 mL/g   R- = 99%, R-PO43−-P = 96–99% Wei et al. (2021a)  
Upflow fluidized bed reactor with a three-phase separator (3.2 L) Synthetic wastewater 5.6–8 h 2.4–3.4 kg/(m3·day) Homemade spherical dewatered activated sludge with a particle size of 2 mm  120 days  0.5–2 mm (70%) 32 mL/g 16 g/L 209.8 mg/g MLVSS R-COD > 90%, R- > 95%, R-TN > 70%, R-TP > 85% Sun et al. (2023)  
Anaerobic/aerobic process with a clarifier (14 m3Municipal wastewater 7–18 h 0.15–0.41 kg/(m3·day) Activated sludge  About 200 days  0.2 mm (>50%) 75 mL/g 1.6 g/L  R-TOC > 82%, R- > 89%, R-TN > 50%, R-TP > 38% Miyake et al. (2023)  
Configuration and scaleWastewaterHRTOLRInoculumStartup timeStable timeAGS characteristics
MLSSEPSTreatment performanceReferences
ColorSizeSVI30
Upflow sludge bed reactor with three-phase separators (0.89 L) Synthetic wastewater 10 h 1–3 kg/(m3·day) Activated sludge and anaerobic digested sludge 27 days 23 days Brown 1.5–3.1 mm 50–120 mL/g    Chen et al. (2009)  
Column-type reactor with a submerged membrane (3.4 L) Synthetic wastewater 27 h 7 kg/(m3·day) Mature AGS  216 days Brown–yellow 1.9 mm  10.2 g/L  R-COD = 89% Juang et al. (2010)  
Column-type reactor (0.8 L) Synthetic wastewater 2.2–19 h 6–39 kg/(m3·day) Mature AGS chemically modified with MgCO3  220 days      R-COD = 20–95% Lee & Chen (2015)  
Column-type reactor with a submerged filter module (4 L) Synthetic wastewater 6.7 h 1.4 kg/(m3·day) Mature AGS  32 days  0.5 mm  5 g/L 135.0 mg/g MLVSS R-COD = 88.5%, R- = 96.5% Li et al. (2012)  
Upflow fluidized bed reactor (3.6 L) Synthetic wastewater 6–7.2 h 3–6 kg/(m3·day) Activated sludge 91 days About 60 days   85 mL/g 3 g/L  R-COD > 90%, R- > 90% Kishida et al. (2012)  
Modified airlift fluidized bed reactor (5.6 L) Synthetic wastewater 2–3 h 1.2–5 kg/(m3·day) Activated sludge 20 days 12 days  0.6 mm 40 mL/g   E-COD = 50 mg/L Zhou et al. (2013a, 2013b
Biofilm 25 days 55 days   35–60 mL/g   R-COD = 90%, R- = 60%, R-TN < 30% Yang et al. (2014)  
Cylindrical CSTRs with demi-submerged effluent discharge tube (R1–3 L, R2–6 L) Diluted pig slurry 1.6 h 2.4–12 kg/(m3·day) R1-activated sludge, R2- pig slurry 227 days  White 4–6.8 mm SVI10 = 127 mL/g   R-COD = 30–80%, R-TN = 10–15% Morales et al. (2012)  
Connected an airlift reactor, a settling tank, a dynamic membrane tank and a sludge selection tank with a sieve in series Synthetic wastewater →septic tank wastewater 13 h 0.5 kg/(m3·day) Mature AGS  50 days  0.1–1.0 mm (60%)    R-COD = 83%, R- = 73%, R-TN = 67%, R-TP = 60% Liu et al. (2012)  
Connected an airlift reactor (7.5 L) and a sludge selection tank with a sieve in series Synthetic wastewater 9 h 1 kg/(m3·day) Activated sludge 7 days 30 days Yellow–white 1.0–3.0 mm (>80%) 35 mL/g 3 g/L  E-COD < 50, E- < 1, E-TN < 5, E-TP < 1 mg/L Liu et al. (2014)  
CSTRs with internal settling zone (6 L) Swine wastewater 3 days 0.4 kg/(m3·day) Activated sludge More than 2 years   0.1 mm     Mota et al. (2014)  
Reverse flow baffled reactor (120 L) 30% municipal and 70% industrial wastewater 5.5–16.4 h 0.5–1.6 kg/(m3·day) Activated sludge 21 days 130 days  0.13 mm 33 mL/g 4.3 g/L 540 (PN) + 47 (PS) mg/g VSS R-BOD5 = 90–94%, R- = 97–100% Li et al. (2015)  
Aeration tank with a two-zone sedimentation tank (26.8 L) 25% municipal and 75% industrial wastewater 18 h 016 kg/(m3·day) Activated sludge 62 days 90 days  0.1 mm 26 mL/g 5 g/L 71 (PN) + 0.4 (PS) mg/g VSS E-COD < 65, E- < 1.4 mg/L Zou et al. (2018)  
Aeration tank with a two-zone sedimentation tank (140 L) Municipal wastewater 16 h 0.27–0.53 kg/(m3·day) Dewatered sludge; 18 days 60 days  0.66 mm <40 mL/g 3.6 g/L 65.3 mg/g MLVSS R-COD = 83–93%, R- = 85–100%, R-TN = 12–22% Xu et al. (2020)  
AAO process with a two-zone sedimentation tank (202 L) Municipal wastewater 10–13.5 h 0.61–1.05 kg/(m3·day) Activated sludge 40 days 75 days  0.2 mm 47.5 mL/g 3.5 g/L 126.2 (PN) + 62.7 (PS) mg/g VSS E-COD = 22.8, E-TN = 3.5, E-TP = 0.2 mg/L Liu et al. (2020)  
CSTR Synthetic wastewater 8 h 1.4 kg/(m3·day) Activated sludge with or without filamentous bacteria 21 days 70 days  0.18–1.25 mm 100 mL/g 3.5 g/L  E-COD < 40, E- < 1 mg/L Chen et al. (2015)  
Modified internal loop airlift reactor (4.75 L) Synthetic wastewater 8 h 3 kg/(m3·day) Mature AGS  67 days  1.2 mm 2 mL/g 6.7 g/L  R-COD > 97% Liu et al. (2015)  
CSTR with a hollow fiber membrane (14 m3Municipal wastewater 6–12 h 0.6–1.2 kg/(m3·day) Activated sludge  110 days  0.6 mm (60%) 50 mL/g  45 (PN) + 33 (PS) mg/g VSS  Sajjad et al. (2016)  
CSTR with an internal settling zone (3 L) Synthetic wastewater   Activated sludge  35 days  0.36 mm 91 mL/g  72 (PN) + 31 (PS) mg/g SS R-TOC = 62%, R-TN = 83%, R-TP = 64% Sajjad & Kim (2015)  
CSTR with an internal settling zone (10 L) Synthetic wastewater   Mature AGS  12 weeks      R-COD = 80–99%, R- = 52–98%, R-TP = 5–87% Manea & Bumbac (2019)  
Dairy wastewater 10 h 3.8 kg/(m3·day) Mature AGS  10 weeks  0.2–1 mm    R-COD = 81–93%, R- = 83–99%, R-TN = 52–82%, R-TP = 65–99% Bumbac et al. (2015)  
Serpentine flow reactor coupled with membrane (7.5 L) Synthetic wastewater 7.5 h 3.8 kg/(m3·day) Mature AGS  30 days  1.3 mm  8 g/L 600 mg/g MLVSS R-COD = 90% Corsino et al. (2016)  
Two-cylindrical reactor with a settling tank (15 L) Synthetic wastewater 6 h 1.2 kg/(m3·day) Mature EBPR-AGS  More than 6 months  0.9 mm 20 mL/g 6 g/L  R-COD = 97%, R-TP = 96% Li et al. (2016a, 2016b
Four SBRs connected in series (4 × 6.16 L) Synthetic wastewater →municipal wastewater   Mature EBPR-AGS     55–67 mL/g 3.1 g/L  R-COD = 82%, R-TN = 71%, R-TP = 31% Li et al. (2019a)  
Modified airlift reactor with an internal settling zone with or without an anaerobic tank (20 L and 35 L) Synthetic wastewater 6–20 h 0.4–1.2 kg/(m3·day) 80% activated sludge and 20% mature EBPR-AGS 30 days 50 days  0.9–1.0 mm 40–54 mL/g 2.1–2.5 g/L  R-COD = 60–90%, R-TN = 26–82%, R-TP = 36–82% Li et al. (2021)  
Cylindrical reactor with a three-phase separator (2 L) Synthetic wastewater 6 h 0.8–3.2 kg/(m3·day) Anaerobic granules 45 days 27 days Dark yellow 1–3 mm    E-COD < 40 mg/L, R- = 100%, R-TN = 93% Sun et al. (2017)  
CSTR coupled with membrane (36 L) Synthetic wastewater 5 h 0.2–0.32 kg/(m3·day) Activated sludge 37 days 73 days  0.2 mm 100 mL/g >7 g/L  R-COD >80%, R-TN >80% Chen et al. (2017)  
CSTR coupled with membrane (20 L) synthetic wastewater 10 h 0.84 kg/(m3·day) activated sludge 28 days 67 days  0.8 mm  8 g/L 32 (PN) + 30 (PS) mg/g VSS R-COD > 90%, R-TN > 85% Dai et al. (2020)  
Modified airlift reactor with an internal settling zone (4.3 L) Synthetic wastewater  0.6–1 kg/(m3·day) 80% bacterium TN-14 cells and 20% activated sludge 40 days 32 days  0.5–2 mm 44–61 mL/g  512 (PN) + (5.6–12.6) (PS) mg/g VSS  Xin et al. (2017)  
Two CSTRs and a clarifier connected in series Synthetic wastewater 3.8–6.5 h 2.2–4.4 kg/(m3·day) Activated sludge     SVI10 = 70 mL/g   R-COD > 95%, R- = 15–30% Cofré et al. (2018)  
Five CSTRs and a clarifier connected in series (60 L) Municipal wastewater 6 h 1.5 kg/(m3·day) Activated sludge    0.28 mm  <0.5 g/L  E-COD = 25, E-TN = 11, E-PO43−-P = 0.1 mg/L Devlin & Oleszkiewicz (2018)  
Anaerobic/oxic reactor, and a clarifier connected in series (49 L) Synthetic wastewater →municipal wastewater  0.19–0.36 kg/(m3·day) Activated sludge    0.3 mm (50%) 102 mL/g 1.8–4.0 g/L  R-COD = 92%, R- = 99%, R-TN = 80%, R-TP < 40% Jahn et al. (2019)  
Modified airlift reactor with an internal settling zone (22 L) Synthetic wastewater 9 h 0.8 kg/(m3·day) Mature EBPR-AGS  116 days  0.85 mm 50 mL/g 4 g/L 31 (PN) + 38 (PS) mg/g MLSS R-COD > 90%, R- = 100%, R-TN > 65%, R-TP = 60–65% Li et al. (2020a)  
200 days  0.88 mm 40 mL/g 4 g/L 44 (PN) + 44 (PS) mg/g MLSS R-COD = 95%, R-TN = 87%, R-TP = 82% Li et al. (2019b)  
Modified airlift reactor with an anaerobic zone and internal settling zone (20 L) Synthetic wastewater 9 h 0.8 kg/(m3·day) Mature EBPR-AGS  110 days  0.9 mm 47 mL/g 4.1 g/L  R-COD = 95%, R-TN = 87%, R-TP = 81% Li et al. (2020b)  
CSTR coupled with membrane (15 L) Synthetic wastewater 12 h 1 kg/(m3·day) 20% mycelial pellets and 80% activated sludge 40 days 130 days Yellow-margins and brown-center; brown-margin and black-center 0.7–0.8 mm 78 mL/g 13 g/L 34–38 (PN) + 11–18 (PS) mg/g VSS R-COD = 97–99%, R- = 100%, R-TN = 32–42% Xiao et al. (2022)  
Several CSTRs and a sludge selector Municipal wastewater 6.5 h 1.2 kg/(m3·day) Activated sludge with a d50 = 0.3 mm 90 days   3.4 mm 64 mL/g 2.4 g/L 59 (PN) + 15 (PS) mg/g VSS R-COD = 60%, R- > 90% Sun et al. (2019)  
Several CSTRs, a sludge selector and a conventional clarifier (128 L)     MLVSS ≈ 1 g/L  E-COD < 50, E- < 1 mg/L An et al. (2021)  
Cylinder airlift reactor with an integrated separator (25 L) Real ethylene glycol wastewater 24 h 0.2–2.2 kg/(m3·day) Mature AGS        R-COD = 50%, R- > 95% Qi et al. (2020)  
Modified airlift loop reactor (180 L) Municipal wastewater 10.8 h 0.4–0.7 kg/(m3·day) Activated sludge 61 days 70 days Yellow 2–4 mm 35 mL/g 4 g/L 200 mg/g MLVSS R-COD = 89%, R-TN = 89%, R-TP = 88% Li et al. (2020c)  
Microaerobic–aerobic reactor with internal separators (3,000 m3/day) Municipal wastewater   Activated sludge  More than 2 months  90.1 μm    R- = 98%, R-TN = 85% Yu et al. (2023b)  
A simulated conceptual hybrid granule/floc process in SBRs (3 L) Synthetic wastewater   50% activated sludge and 50% AGS    0.43 mm (35%) 50.8 mL/g   R- = 99%, R-PO43−-P = 96–99% Wei et al. (2021a)  
Upflow fluidized bed reactor with a three-phase separator (3.2 L) Synthetic wastewater 5.6–8 h 2.4–3.4 kg/(m3·day) Homemade spherical dewatered activated sludge with a particle size of 2 mm  120 days  0.5–2 mm (70%) 32 mL/g 16 g/L 209.8 mg/g MLVSS R-COD > 90%, R- > 95%, R-TN > 70%, R-TP > 85% Sun et al. (2023)  
Anaerobic/aerobic process with a clarifier (14 m3Municipal wastewater 7–18 h 0.15–0.41 kg/(m3·day) Activated sludge  About 200 days  0.2 mm (>50%) 75 mL/g 1.6 g/L  R-TOC > 82%, R- > 89%, R-TN > 50%, R-TP > 38% Miyake et al. (2023)  

R represents removal efficiency. E represents effluent concentration. CSTR(s) represents continously stirred tank reactor(s). TOC represents total organic carbon. EBPR represents enhanced biological phosphorus removal. MLVSS represents mixed liquor volatile suspended solids.

Physicochemical characterization

Particle size

The size distribution of AGS-CFRs ranges from 0.09 to 6.8 mm. In over 80% of the reviewed literature, the average particle size of AGS-CFRs does not exceed 1 mm, a dimension significantly smaller than that of AGS-SBRs. The discrepancy in particle size between AGS-CFRs and SBRs can be ascribed to the lower substrate concentration under continuously mixed conditions, resulting in a shallower penetration diameter and impeding further increase of granular size (Kent et al. 2018). Notably, the study conducted by Morales et al. (2012) reported the largest particle size of 6.8 mm, likely attributable to an ultra-short hydraulic retention time (HRT) of 1 h and an ultra-high organic loading rate (OLR) of 12 kg/(m3·day). Moreover, a minimum average particle size of 0.09 mm was detected in a pilot scale (3,000 m3/day) microaerobic–aerobic reactor with internal separators, potentially stemming from the low OLR and the absence of selectively wasting flocculent sludge (Yu et al. 2023b).

Settleability

The SVI30 of AGS-CFRs display a wide range from 2 to 100 mL/g. Nearly 80% of the documented SVI30 values fall below 60 mL/g, suggesting comparable settling characteristics of AGS-CFRs and SBRs. The lowest SVI30 value of 2 mL/g was observed in an internal loop airlift reactor operating with synthetic wastewater and supplemented with 19 mg/L of Ca2+, leading to the accumulation of CaCO3 precipitates within the granules (Liu et al. 2015). Conversely, several studies have reported the highest SVI30 value of 100 mL/g (Chen et al. 2015, 2017), primarily attributed to the overgrowth of filamentous bacteria and the absence of selective pressure. Additionally, the reported SVI5/SVI30 values range from 1.0 to 1.3.

Mixed liquor suspended solids

In most studies, mixed liquor suspended solids (MLSS) concentrations typically range from 4 to 6 g/L, comparable to that in SBRs. Nevertheless, some deviations from this range have been observed. Sun et al. (2023) documented a maximum MLSS value of 16 g/L, which was attributed to the utilization of a special inoculum of homemade spherical dehydrated activated sludge with a particle size of 2 mm and an initial MLSS concentration of 20 g/L. Furthermore, membrane bioreactor(s) (MBR(s)), where sludge is completely retained, have been found to yield relatively higher MLSS values (Corsino et al. 2016; Chen et al. 2017). In contrast, lower MLSS values are frequently encountered in CFRs employing clarifiers equipped with mixers or short settling times. For example, Devlin & Oleszkiewicz (2018) reported a decrease in MLSS concentration to less than 0.5 g/L. The results suggest that the MLSS concentration is largely influenced by the reactor configuration, particularly the sedimentation and sludge waste module, rather than the flow regime.

Extracellular polymeric substances

The extracellular polymeric substances (EPS) contents of AGS-CFRs range from 37 to 600 mg/g volatile suspended solids (VSS). Prior investigations in SBRs have revealed that EPS contents tend to ascend during the granulation phase and descend during the stable phase (Liu et al. 2022). The similar trend has also been observed in AGS-CFRs. For instance, Corsino et al. (2016) documented the disintegration of mature AGS during the initial phase when transferred from SBRs to continuous-flow MBRs, resulting in decreases in EPS contents and the ratio of protein to polysaccharide (PN/PS) from 500 to 200 mg/g VSS and 6 to 2, respectively. Subsequently, an intermittent feeding strategy was introduced to generate feast/famine conditions, facilitating the reformation of AGS and consequent increments in EPS contents and PN/PS ratio to approximately 600 mg/g VSS and 4, respectively. However, it is worth noting that a notable proportion of studies have not quantified EPS contents or the PN/PS ratio.

Pollutant removal performance

AGS-CFRs have been proven to be effective in treating various types of wastewaters, including synthetic wastewater, municipal wastewater, mixtures of municipal sewage and industrial wastewater containing certain metal elements and bio-refractory compounds (Li et al. 2015), dairy wastewater (Bumbac et al. 2015), industrial wastewater containing ethylene glycol (Qi et al. 2020), and industrial wastewater containing dinitrophenol, o-cresol, and phenol (Ramos et al. 2016). The removal efficiencies for organic matter and nutrients are outlined as follows.

Organics removal

Successful aerobic granulation has been observed over a wide range of 0.15 to 39.0 kg/(m3·day) in CFRs, suggesting that OLR alone is not a decisive factor for granulation. Additionally, efficient organic matter removal has been consistently achieved, as demonstrated by chemical oxygen demand (COD) removal efficiencies surpassing 90% in most studies.

Ammonia removal

AGS-CFRs have been proven highly effective in ammonia nitrogen removal (), with reported removal efficiencies in most studies surpassing 90% and, in certain cases, nearing 100% (Manea & Bumbac 2019; Xiao et al. 2022). Notably, Kishida et al. (2012) observed that AGS formation occurred exclusively in the absence of the nitrification inhibitor allylthiourea (ATU), suggesting that the proliferation of nitrifying bacteria plays a pivotal role in granulation. While Cofré et al. (2018) documented the dominance of AGS-CFRs after a 29-day startup period with a solid retention time (SRT) of 0.4–2.7 days. This markedly short SRT was insufficient for nitrifying bacteria to thrive, resulting in a relatively low removal efficiency of 15–30%. Despite the disparate conclusions reached by these studies, most research concurs that the cultivation of AGS-CFRs capable of effectively removing ammonia nitrogen is relatively easy.

Total nitrogen removal

Total nitrogen (TN) removal efficiencies observed in AGS-CFRs fall below 50% in several studies, suggesting that denitrification poses greater challenges compared to nitrification. The establishment of stratified structure within granules, consisting of an aerobic outer layer and an anoxic/anaerobic inner layer, is essential for simultaneous nitrification–denitrification (Layer et al. 2020). The relatively lower TN removal efficiencies in AGS-CFRs can be attributed to the smaller particle size, which allows deeper penetration of dissolved oxygen (DO) and hinders the formation of the anoxic/anaerobic inner layer. Various approaches have been attempted to enhance TN removal performance. For instance, Yang et al. (2014) demonstrated that reducing aeration intensity led to an increase in TN removal efficiency from 10 to 30%. However, system instability and sludge bulking occurred, due to the decreased hydraulic shear force. Li et al. (2020c) implemented a dynamic feeding strategy to create feast/famine conditions, achieving a remarkable TN removal efficiency of 89%. Additionally, Yu et al. (2023b) employed separate microaerobic reactors to facilitate simultaneous nitrification and denitrification, followed by aerobic reactors providing adequate shear forces. This configuration yielded satisfactory TN removal performance and ensured the long-term stability of AGS-CFRs. Moreover, the incorporation of an additional anoxic reactor, as demonstrated by Xu et al. (2020), also led to a notable augmentation in TN removal efficiency from 27 to 60%. Sun et al. (2017) developed a modified column-type reactor with a bottom anoxic zone and a top aerobic zone, leading to an exceptional TN removal efficiency of 93%.

Phosphorus removal

The reported total phosphorus (TP) removal efficiencies of AGS-CFRs vary greatly, ranging from 5 to 99%, indicating that achieving effective biological phosphorus removal is challenging. Certain studies have demonstrated notable phosphorus removal performance. For instance, Liu et al. (2014) achieved effluent TP concentrations below 1 mg/L in a column-type upflow reactor. Long et al. (2015) accomplished an average TP removal efficiency of 87% using a two-stage airlift reactor, with a switch in flow direction every 2 h to induce feast/famine conditions. Li et al. (2020a) achieved a stable TP removal efficiency of 60% in a modified airlift reactor. Notably, none of these AGS-CFRs systems incorporated an anaerobic zone. Furthermore, incorporating anaerobic zone into AGS-CFRs systems is commonly adopted to enhance phosphorus removal. For example, Li et al. (2016b) achieved an average TP removal rate of 96% by incorporating an additional anaerobic reactor. Likewise, Liu et al. (2020) employed an anaerobic/anoxic/aerobic (AAO) process to concurrently remove organic matter and nutrients, resulting in satisfactory effluent TP concentrations of 0.2 mg/L.

Overall, AGS-CFRs exhibit similar physicochemical properties and pollutant removal performance to AGS-SBRs.

In the past two decades, significant efforts have been devoted to developing startup strategies and CFR configurations to replicate the granulation conditions demonstrated in SBRs. Table 1 provides a summary of laboratory and pilot-scale investigations, encompassing information on reactor configurations, granulation strategies, characteristics of formed AGS, pollutant removal performance, startup and stable time, and potential challenges.

Hydraulic shear force

Hydraulic shear force is recognized as a critical factor in aerobic granulation. During the startup period, shear force promotes microbial motion and fosters effective collision, thereby facilitating the formation of initial micro-aggregates. Additionally, shear force prompts EPS secretion and enhances surface hydrophobicity, thus increasing micro-aggregates density (Liu et al. 2005; Gao et al. 2011). In the stable stage, shear force removes filamentous bacteria adhering to the AGS surface, thus governing the morphology of mature AGS, and promoting a balanced coexistence between fast- and slow-growing bacteria, thereby contributing to long-term stability.

In line with researches in SBRs, column-type airlift reactors, known for their ability to provide significant hydraulic shear force, have played a leading role in the studies of AGS-CFRs. For instance, Sun et al. (2023) (Figure 1(a)), Kishida et al. (2012) (Figure 1(b)), Sun et al. (2017) (Figure 1(c)), and Zhou et al. (2013a) (Figure 1(d)) successfully cultivated AGS using aerobic upflow fluidized bed reactors. Furthermore, Zhou et al. (2014) directly demonstrated the critical role of shear forces in AGS formation through fluorescence labeling investigations. However, in their study, the long-term stability and denitrification performance of AGS were limited. When DO concentration decreased from 4–5 to 3 mg/L, the TN removal rate increased only from 10 to 30%, but the growth of filamentous bacteria led to the disintegration of AGS (Yang et al. 2014). This indicates that hydrodynamic shear force-driven AGS is significantly influenced by aeration intensity, posing challenges in achieving efficient denitrification and long-term stability. Therefore, it is necessary to integrate hydraulic shear forces with additional granulation strategies.
Figure 1

CFR configurations with high shear force: (a) a typical airlift reactor with three-phase separator (Sun et al. 2023), (b) a modified airlift reactor with baffles (Kishida et al. 2012), (c) a modified airlift reactor with aeration equipment at waist (Sun et al. 2017), (d) an internal loop airlift reactors (Zhou et al. 2013a) (black – reactor; red – water flow; blue – air flow; green – sludge flow; black circle – granules; white circle – air bubbles).

Figure 1

CFR configurations with high shear force: (a) a typical airlift reactor with three-phase separator (Sun et al. 2023), (b) a modified airlift reactor with baffles (Kishida et al. 2012), (c) a modified airlift reactor with aeration equipment at waist (Sun et al. 2017), (d) an internal loop airlift reactors (Zhou et al. 2013a) (black – reactor; red – water flow; blue – air flow; green – sludge flow; black circle – granules; white circle – air bubbles).

Close modal

Physical selective pressure

Physical selection pressure is a stress that directly induces changes in the physical properties of sludge, including settling velocity, particle size, and density.

Settling velocity-based selective pressure

Short settling time is commonly utilized in SBRs to generate settling velocity-based physical selection pressure. However, in CFRs, shortening the settling time to generate such pressure is challenging due to the disturbance caused by the continuously flowing mixture. To overcome this challenge, various strategies and reactor configurations have been developed. Laboratory studies have shown that introducing baffles to separate the settling zone from the aeration zone (Figure 2(a)–2(d)) (Mota et al. 2014; Bumbac et al. 2015; Sajjad & Kim 2015; Xin et al. 2017) or serially connecting external sludge selectors (Figure 2(e)) (Li et al. 2016b) are common strategies to introduce settling velocity-based selection pressure.
Figure 2

CFR configurations with settling velocity-based selective pressure: (a–d) modified airlift reactors with baffles (Mota et al. 2014; Bumbac et al. 2015; Sajjad & Kim 2015; Xin et al. 2017), (e) a double-column reactor connected with a sludge selector (Li et al. 2016b), and (f) a configuration of a two-zone sedimentation tank (Li et al. 2014).

Figure 2

CFR configurations with settling velocity-based selective pressure: (a–d) modified airlift reactors with baffles (Mota et al. 2014; Bumbac et al. 2015; Sajjad & Kim 2015; Xin et al. 2017), (e) a double-column reactor connected with a sludge selector (Li et al. 2016b), and (f) a configuration of a two-zone sedimentation tank (Li et al. 2014).

Close modal

Recently, a novel configuration known as the two-zone sedimentation tank was successfully demonstrated in a pilot scale (Figure 2(f)) (Li et al. 2014). This configuration segregates the fast-settling sludge, which settles in the first sedimentation tank for recycling purposes, while the slow-settling sludge settles in the second sedimentation tank and is discharged as surplus sludge. After a startup period of 13 days, AGS was observed and eventually stabilized with an average diameter of 0.6 mm and an SVI30 value of 44 mL/g. Remarkable removal rates of 90% for BOD5 and 95% for were achieved. Moreover, the direct integration of a two-zone sedimentation tank into the AAO process successfully cultivated AGS with an average diameter of 0.2 mm and an SVI30 value of 47.5 mL/g (Liu et al. 2020). This integration yielded high-quality effluent with concentrations of 22.8 mg/L for COD, 3.5 mg/L for TN, and 0.2 mg/L for TP.

Sludge density/size-based selective pressure

Hydrocyclones can impose density-based selective pressure by retaining sludge with higher density in the underflow while discharging sludge with lower density in the overflow (Xu et al. 2019). Figure 3(a) illustrates the implementation of hydrocyclones within an anoxic–oxic (AO) process (Liu et al. 2017). The investigation revealed an enhancement in the microbial activity, with an increase in specific oxygen uptake rate, nitrate reductase activity, and nitrite reductase activity. However, a decrease in sludge size from 78.8 μm to a range of 15.8–23.3 μm was observed, concomitant with a decline in settleability, as indicated by an increased SVI30 value from 112.2 to 125.2 mL/g. Similar sludge fragmentation resulting in reduced particle size was also observed in SBRs (Xu et al. 2019). These findings suggest that, while hydrocyclones can increase sludge density, the introduced centrifugal and shear forces may lead to the breakdown of microbial aggregates and a decrease in sludge diameter.
Figure 3

CFR configurations with density/size-based selective pressure: (a) hydrocyclone implementation (Liu et al. 2017), (b) sieve implementation (Liu et al. 2014), and (c) sieve implementation in MBR system (Liu et al. 2012).

Figure 3

CFR configurations with density/size-based selective pressure: (a) hydrocyclone implementation (Liu et al. 2017), (b) sieve implementation (Liu et al. 2014), and (c) sieve implementation in MBR system (Liu et al. 2012).

Close modal

Sieves can introduce size-based physical selection pressure. As depicted in Figure 3(b), smaller-sized sludge passes through the sieve and is discharged with the effluent, while larger-sized sludge is retained and periodically returned to the reactor (Liu et al. 2014). In Liu et al.’s (2014) study, the sieve apertures range from 0.1 to 1.0 mm, with smaller apertures utilized during the startup phase and larger apertures during the stable phase. After 14 days of inoculating activated sludge, well-defined AGS was observed, with over 80% of the sludge having diameters ranging from 1.0 to 3.0 mm. The effluent concentrations of COD, , TN, and TP were measured as 50, < 1, < 5, and <0.5 mg/L, respectively. Notably, in this study, shear forces generated by flow and aeration played a critical role in granulation. Furthermore, to address the issue of elevated suspended solid (SS) concentration resulting from the simultaneous discharge of effluent and smaller-sized sludge, an MBR system was employed (Figure 3(c)) (Liu et al. 2012). AGS with a particle size range of approximately 0.1–1.0 mm was obtained. However, filamentous bacteria overgrowth led to the formation of loosely structured AGS.

Feast/famine condition

Feast/famine conditions act as stressors that induce shifts in microbial populations, also known as microbial selection pressure (Kent et al. 2018). During the feast phase, the assimilation rate of readily biodegradable substrates lags their transport rate, resulting in the accumulation of substrates as polymers within slow-growing microorganisms (de Kreuk 2006; Wagner et al. 2015). Subsequently, in the famine phase, these stored polymers serve as a resource for microbial growth. Alternating feast/famine conditions have also been found to stimulate excessive EPS secretion and enhance surface hydrophobicity, thus facilitating microbial aggregation (Gao et al. 2011). Moreover, feast/famine conditions tend to favor the proliferation of slow-growing bacteria such as glycogen-accumulating organisms (GAOs), polyphosphate-accumulating organisms (PAOs), and denitrifying phosphorus-accumulating organisms (DPAOs), which contribute to improving nutrient removal performance and promoting long-term stability (de Kreuk et al. 2005).

In CFRs, conditions of overall or local complete mixing result in low substrate concentration gradients, making it challenging to create feast/famine conditions. Various strategies have been developed in the laboratory to overcome this challenge and reactor configuration modification stands as the prevailing method frequently employed. Li et al. (2021) compared the effectiveness of internally anaerobic-feast (Figure 4(a)) and externally anaerobic-feast (Figure 4(b)) conditions. It was observed that in the externally anaerobic-feast reactor configuration, granulation time significantly decreased from 71 to 31 days, alongside an augmentation in the removal rates of COD, TN, and TP from 60, 26, and 36% to 90, 82, and 82%, respectively. Devlin & Oleszkiewicz (2018) (Figure 4(c)) and Jahn et al. (2019) (Figure 4(d)) successfully cultured AGS by creating front-end feast and back-end famine conditions through the series connection of multiple reactors. Similarly, Sun et al. (2019, 2021) developed a plug-flow reactor consisting of 10 identical CSTRs and a sludge selector with a settling time of only 4 min (Figure 4(e)). They successfully cultivated AGS with an average particle size of 3.4 mm and an SVI30 value of 64 mL/g. The removal rates of COD and exceeded 60 and 90%, respectively. Further research by Sun et al. (2021) revealed that the cultivation of AGS failed when fewer CSTRs were connected, with the corresponding feast/famine ratio exceeding 0.5. Additionally, to address the challenge of biomass washout from the sludge selector during startup, researchers implemented an external clarifier, designed with a surface overflow rate of 1 m/h, collecting floc sludge from the overflow of the sludge selector and returning it to the famine zone (An et al. 2021).
Figure 4

CFR configurations with feast/famine conditions: (a) a modified airlift reactor with an anaerobic zone, (b) a modified airlift reactor with an external anaerobic reactor (Li et al. 2021), (c,d) multiple CSTRs connected in series (Devlin & Oleszkiewicz 2018; Jahn et al. 2019), (e) a plug-flow reactor comprising 10 identical CSTRs and a sludge selector (Sun et al. 2019), (f) a modified airlift loop reactor (Li et al. 2020c), (g) a two-column reactor with a sedimentation column (Long et al. 2015), (h) a plug-flow CFR with serpentine baffles (Li et al. 2015), and (i) a microaerobic–aerobic reactor with internal three-phase separators (Yu et al. 2023b) (red zone – feast condition; green zone – famine condition).

Figure 4

CFR configurations with feast/famine conditions: (a) a modified airlift reactor with an anaerobic zone, (b) a modified airlift reactor with an external anaerobic reactor (Li et al. 2021), (c,d) multiple CSTRs connected in series (Devlin & Oleszkiewicz 2018; Jahn et al. 2019), (e) a plug-flow reactor comprising 10 identical CSTRs and a sludge selector (Sun et al. 2019), (f) a modified airlift loop reactor (Li et al. 2020c), (g) a two-column reactor with a sedimentation column (Long et al. 2015), (h) a plug-flow CFR with serpentine baffles (Li et al. 2015), and (i) a microaerobic–aerobic reactor with internal three-phase separators (Yu et al. 2023b) (red zone – feast condition; green zone – famine condition).

Close modal

In addition to modifying reactor configuration, laboratory researchers have also developed dynamic feeding strategies. For example, Li et al. (2020c) bifurcated the feeding flow into peak and normal flows, creating temporal feast–famine conditions (Figure 4(f)). They successfully cultivated compact AGS with diameters of 2–4 mm and an average SVI30 of 35 mL/g. Long et al. (2015) connected a two-column reactor with a sedimentation column (Figure 4(g)) and reversed the flow direction every 2 h, achieving successful cultivation of AGS with removal rates for COD, , TN, and TP at 97, 88, 90, and 87%, respectively. Similarly, Li et al. (2015) constructed a plug-flow CFR with serpentine baffles (Figure 4(h)) and reversed the flow direction every 2 h. Successful cultivation of AGS was achieved with an average particle size of 0.13 mm and an SVI30 value of 33 mL/g.

Furthermore, at the pilot scale, Yu et al. (2023b) successfully demonstrated the feasibility of a microaerobic–aerobic configuration with internal separators with a treatment capacity of 3,000 m3/day (Figure 4(i)). In this study, feast conditions were created within the microaerobic tanks, enabling the selective enrichment of slow-growing microorganisms, such as DPAOs. In the aerobic tanks under famine conditions, intense aeration and internal circulation generated the necessary hydraulic shear force to promote granulation. Well-defined AGS with an average diameter of 90.1 μm was observed and granules larger than 100 and 200 μm accounted for 47.8 and 9.4%, respectively. The average concentrations of and TN in the effluent were 1.2 and 9.5 mg/L, respectively.

Bioaugmentation

Using mature AGS cultivated in SBRs as inoculum is a commonly employed bioaugmentation method for initiating CFRs. Following an adaptation period, the morphology, structure, and pollutant removal performance can be reinstated. Moreover, inoculating specific bacterial strains has also been proven to accelerate aerobic granulation, reducing startup time. For instance, Xin et al. (2017) achieved aerobic granulation within 40 days by inoculating 20% activated sludge and 80% denitrifying bacteria TN-14, which exhibit autotrophic denitrification, aerobic denitrification, and coagulation functions. Xiao et al. (2022) sped up granulation by at least 65 days by using inoculum comprised of 80% activated sludge and 20% Aspergillus niger 557 mycelial pellets. Sun et al. (2023) skipped the startup period with a special inoculum of homemade spherical dewatered activated sludge with a diameter of 2 mm and a high MLSS concentration of 20 g/L. Moreover, Sun et al. (2017) transformed anaerobic granular sludge into AGS within 45 days.

Another bioaugmentation strategy involves periodically inoculating AGS from sidestream SBRs into mainstream CFRs. Miyake et al. (2023) introduced mixed liquor from the aeration phase and effluent from the settling phase of a sidestream SBR into a mainstream CFR (Figure 5(a)). As a result, the proportion of AGS in the mainstream CFR increased to over 50%. Notably, the particle size of AGS in the sidestream SBR typically exceeded 0.5 mm, while in the mainstream CFR, it was merely 0.1–0.2 mm. This indicates that in mainstream CFRs, the disintegration rate of AGS consistently exceeds its formation rate. To address this issue, Sajjad et al. (2016) proposed a method involving sludge exchange between the sidestream SBR and the mainstream CFR every 10 days, as shown in Figure 5(b). The findings revealed near-identical sizes of AGS in both reactors, approximately 0.6 mm, with SVI5 values approaching the SVI30 values, approximately 50 mL/g.
Figure 5

CFR operations with bioaugmentation: (a) introducing mixed liquor from the sidestream SBR into the mainstream CFR (Miyake et al. 2023) and (b) exchanging sludge between the sidestream SBR and the mainstream CFR every 10 days (Sajjad et al. 2016).

Figure 5

CFR operations with bioaugmentation: (a) introducing mixed liquor from the sidestream SBR into the mainstream CFR (Miyake et al. 2023) and (b) exchanging sludge between the sidestream SBR and the mainstream CFR every 10 days (Sajjad et al. 2016).

Close modal

Moreover, the addition of metal ions or carriers has also been proven to accelerate granulation in CFRs. For instance, Xu et al. (2020) added treated dewatered sludge daily, forming AGS within 18 days. Dai et al. (2020) successfully cultivated AGS 28 days after inoculation with activated sludge by adding carriers. However, the potential cost increase and secondary pollution issues caused by the addition of metal ions or carriers need to be carefully considered.

Membrane bioreactors

Recently, aerobic granulation in MBRs has garnered considerable interest. Studies indicate that AGS-MBRs (Figure 6(a)) develop a biofilm layer with greater porosity, exhibiting a reduced fouling propensity compared to activated sludge (Li et al. 2012). On the other hand, Qiu et al. (2022) observed the formation of spherical AGS in the presence of carriers, while only flocs were cultivated in their absence (Figure 6(b)). Chen et al. (2017) found filamentous bulking followed by the formation of well-defined AGS with smooth surfaces and distinct boundaries in an AO-MBR. This phenomenon could be attributed to the complete retention of sludge by the membrane module, resulting in an extremely high MLSS concentration and filamentous bacteria overgrowth due to a low F/M ratio. Subsequently, under the influence of shear forces, entangled filamentous bacteria formed stable AGS. Additionally, Corsino et al. (2016) observed that initially inoculated AGS disintegrated immediately under continuous feeding conditions, achieving stability only when intermittent feeding was employed in a plug-flow reactor with baffles and an integrated membrane module (Figure 6(c)). These investigations emphasize the necessity of incorporating additional granulation strategies in MBRs.
Figure 6

MBRs for aerobic granulation: (a) an aerated reactor integrated with a membrane module (Li et al. 2012), (b) a segregated MBR system (Qiu et al. 2022), and (c) a plug-flow reactor integrated with a membrane module (Corsino et al. 2016).

Figure 6

MBRs for aerobic granulation: (a) an aerated reactor integrated with a membrane module (Li et al. 2012), (b) a segregated MBR system (Qiu et al. 2022), and (c) a plug-flow reactor integrated with a membrane module (Corsino et al. 2016).

Close modal

Currently, AGS-CFRs have been implemented in nine WWTPs worldwide with three primary objectives: (i) improving sludge settleability to obtain higher-quality effluent or enhance the treatment capacity of the biological treatment system during winter and rainy seasons, (ii) enhancing biological phosphorus removal performance, thereby reducing the addition of chemical phosphorus removal agents and lowering operational cost, and (iii) reducing energy consumption and carbon emissions by improving energy efficiency. Table 2 summarizes data from globally full-scale AGS-CFRs-based WWTPs, including information on wastewater types, pre-existing treatment processes, treatment capacities, startup time, sludge characteristics before and after implementation, effluent quality, achievements, and challenges. As shown in Table 2, the treatment capacities range from 189 to 7.6 × 104 m3/day, incorporating various pre-existing treatment processes, such as Bardenpho, BIOCOS®, Modified Ludzack-Ettinger (MLE), AO, and AAO. This indicates that the upgrade to continuous-flow AGS process in existing wastewater treatment facilities is not hindered by treatment capacity or pre-existing wastewater treatment processes.

Table 2

Reported applications for full-scale WWTPs based on AGS-CFRs

RetrofitPlantWastewaterPre-existing treatment processScaleStartup timeSettleability
Effluent after upgradeAchievementsChallengesReferences
Before upgrade/control groupAfter upgrade /test group
Implementing hydrocyclones James River WWTP Municipal wastewater 4-stage IFAS Bardenpho system 7.6 × 104 m3/day  SVI5 = 292 mL/g, SVI30 = 142 mL/g SVI5 = 248 mL/g, SVI30 = 130–144 mL/g  
  • 1.

    Ferric chloride usage is reduced;

  • 2.

    Biomass density increased from 1.05 to 1.07 g/mL;

  • 3.

    Granule percentage increased from 0 to 24%–40%.

 
SVI did not decrease Welling (2015)  
Kunmig WWTP Municipal wastewater BIOCOS® 600 m3/day 1 month SVI30 = 160 mL/g SVI30 = 60 mL/g E-TP decreased from 3.5–4 to 1.0–2.2 mg/L SVI30 decreased from 160 to 60 mL/g within 5 months.  Sturm (2020)  
Urbanna WWTP Municipal wastewater MLE 189 m3/day  SVI5 = 176 mL/g, SVI30 = 100 mL/g SVI5 = 134 mL/g, SVI30 = 77 mL/g E-COD, , and TP unchanged; E-NOX-N increased Granule percentage increased to about 15–28%. 1. Settleability improved significantly in summer but sludge bulking still occurred in winter;
2. VSS percentage was lower than the control train. 
Brickles (2017)  
James R. Dilorio WRRF Municipal wastewater AAO 6 × 104 m3/day 3 months SVI30 = 184 mL/g SVI30 = 83 mL/g E-TN = 11, E-TP < 1 mg/L Granule percentage increased to 57%. Settleability improved temporarily and deteriorated again in winter. Partin (2019)  
Wroclaw WWTP Municipal wastewater AAO 2.9 × 104 m3/day   Warm season: SVI30 = 61 mL/g; cold season SVI30 ≈ 150 mL/g E-COD < 50, E-TN < 20, E-TP < 1 mg/L Granule percentage increased to 25–30%. 1. Seasonal filamentous outgrowth occurred;
2. Sludge blanket height raised in fall. 
Gemza & Kuśnierz (2022)  
Dijon WRRF Municipal wastewater AAO 50,000 m3/day 7 months Cold season SVI30 = 149 mL/g SVI30 all year below 50 mL/g E-TN = 6.6, E-TP = 0.71 mg/L 1. Granule percentage increased to 40–50%;
2. Biomass density increased
to 1.07 g/mL;
3. Surface loading rates increased to 15–20 kg/(m2·h) and surface overflow rates increased to 1.5–2.4 m/h. 
 Roche et al. (2022)  
Implementing hydrocyclones and an anaerobic
selector 
Robert W. Hite WWTP Municipal wastewater AO 3.3 × 104 m3/day 3 months SVI30 = 117 mL/g SVI30 = 77 mL/g E-COD = 80, E- = 0.2, E-TN = 5.8, E-TP = 0.74 mg/L 1. Granule percentage increased to 32–56%;
2. Improved settleability, led to a treatment capacity increase by 32%. 
 Avila et al. (2021)  
Implementing microaerobic–aerobic reactor with internal separators WWTP in Hebei Province, China Municipal wastewater AAO 2.5 × 104 m3/day 1 month SVI30 = 62.0 mL/g SVI30 = 56.8 mL/g E-COD = 25.5, E- = 0.5, E-TN = 10.1 mg/L 1. Granule percentage increased to29.6%;
2. Area use is reduced by 38.2% as no need for secondary clarifiers. 
 Yu et al. (2023a, 2024
Introducing dry kenaf media and a rotary drum screen The Town of Moorfield Advanced Nutrient WWTP Mixture of 10% municipal and 90% industrial wastewater 5-stage Bardenpho 1.55 × 104 m3/day 5 months SVI30 > 200 mL/g SVI30 < 50 mL/g E- < 5, E- NOX-N < 6, E-TP = 1.6 mg/L Settleability significantly improved It remains a matter of debate whether to categorize kenaf granules as AGS. Wei et al. (2021b)  
RetrofitPlantWastewaterPre-existing treatment processScaleStartup timeSettleability
Effluent after upgradeAchievementsChallengesReferences
Before upgrade/control groupAfter upgrade /test group
Implementing hydrocyclones James River WWTP Municipal wastewater 4-stage IFAS Bardenpho system 7.6 × 104 m3/day  SVI5 = 292 mL/g, SVI30 = 142 mL/g SVI5 = 248 mL/g, SVI30 = 130–144 mL/g  
  • 1.

    Ferric chloride usage is reduced;

  • 2.

    Biomass density increased from 1.05 to 1.07 g/mL;

  • 3.

    Granule percentage increased from 0 to 24%–40%.

 
SVI did not decrease Welling (2015)  
Kunmig WWTP Municipal wastewater BIOCOS® 600 m3/day 1 month SVI30 = 160 mL/g SVI30 = 60 mL/g E-TP decreased from 3.5–4 to 1.0–2.2 mg/L SVI30 decreased from 160 to 60 mL/g within 5 months.  Sturm (2020)  
Urbanna WWTP Municipal wastewater MLE 189 m3/day  SVI5 = 176 mL/g, SVI30 = 100 mL/g SVI5 = 134 mL/g, SVI30 = 77 mL/g E-COD, , and TP unchanged; E-NOX-N increased Granule percentage increased to about 15–28%. 1. Settleability improved significantly in summer but sludge bulking still occurred in winter;
2. VSS percentage was lower than the control train. 
Brickles (2017)  
James R. Dilorio WRRF Municipal wastewater AAO 6 × 104 m3/day 3 months SVI30 = 184 mL/g SVI30 = 83 mL/g E-TN = 11, E-TP < 1 mg/L Granule percentage increased to 57%. Settleability improved temporarily and deteriorated again in winter. Partin (2019)  
Wroclaw WWTP Municipal wastewater AAO 2.9 × 104 m3/day   Warm season: SVI30 = 61 mL/g; cold season SVI30 ≈ 150 mL/g E-COD < 50, E-TN < 20, E-TP < 1 mg/L Granule percentage increased to 25–30%. 1. Seasonal filamentous outgrowth occurred;
2. Sludge blanket height raised in fall. 
Gemza & Kuśnierz (2022)  
Dijon WRRF Municipal wastewater AAO 50,000 m3/day 7 months Cold season SVI30 = 149 mL/g SVI30 all year below 50 mL/g E-TN = 6.6, E-TP = 0.71 mg/L 1. Granule percentage increased to 40–50%;
2. Biomass density increased
to 1.07 g/mL;
3. Surface loading rates increased to 15–20 kg/(m2·h) and surface overflow rates increased to 1.5–2.4 m/h. 
 Roche et al. (2022)  
Implementing hydrocyclones and an anaerobic
selector 
Robert W. Hite WWTP Municipal wastewater AO 3.3 × 104 m3/day 3 months SVI30 = 117 mL/g SVI30 = 77 mL/g E-COD = 80, E- = 0.2, E-TN = 5.8, E-TP = 0.74 mg/L 1. Granule percentage increased to 32–56%;
2. Improved settleability, led to a treatment capacity increase by 32%. 
 Avila et al. (2021)  
Implementing microaerobic–aerobic reactor with internal separators WWTP in Hebei Province, China Municipal wastewater AAO 2.5 × 104 m3/day 1 month SVI30 = 62.0 mL/g SVI30 = 56.8 mL/g E-COD = 25.5, E- = 0.5, E-TN = 10.1 mg/L 1. Granule percentage increased to29.6%;
2. Area use is reduced by 38.2% as no need for secondary clarifiers. 
 Yu et al. (2023a, 2024
Introducing dry kenaf media and a rotary drum screen The Town of Moorfield Advanced Nutrient WWTP Mixture of 10% municipal and 90% industrial wastewater 5-stage Bardenpho 1.55 × 104 m3/day 5 months SVI30 > 200 mL/g SVI30 < 50 mL/g E- < 5, E- NOX-N < 6, E-TP = 1.6 mg/L Settleability significantly improved It remains a matter of debate whether to categorize kenaf granules as AGS. Wei et al. (2021b)  

R represents removal efficiency. E represents effluent concentration.

Hydrocyclones

In 2015, to address issues of poor settleability and unstable phosphorus removal performance, James River WWTP installed eight hydrocyclones, upgrading the pre-existing activated sludge process to AGS-CFRs (Welling 2015). As shown in Figure 7(a), the hydrocyclones receive influent from the return-activated sludge (RAS) pipeline. The designed hydraulic spit was 80% in the overflow and 20% in the underflow, ensuring an equal total suspended solids (TSS) mass between the overflow and underflow. Consequently, biological phosphorus removal performance was improved, and ferric chloride usage was reduced. Meanwhile, the granule percentage increased from 0 to 24–40%, coupled with an elevation in biomass density from 1.05 to 1.07 g/mL. However, sludge settleability did not improve, with the SVI30 value remaining high at 144.1 mL/g, and the aggregate percentage with a critical settling velocity greater than 1.5 m/h significantly reduced from 86 to 47%. This might be attributed to excessive centrifugal and shear forces generated by the hydrocyclones, resulting in disintegration and breakage of aggregates, as demonstrated in laboratory studies. On the other hand, improvements in settleability, along with increased granule percentage and biomass density, were observed across four other full-scale applications at Kunming WWTP (Sturm 2020), Urbanna WWTP (Brickles 2017), James R. Dolorio water resource recovery facility (WRRF) (Regmi et al. 2022), and Wroclaw WWTP (Gemza & Kuśnierz 2022). However, unfortunately, settleability improvements were temporary, and seasonal filamentous bulking occurred. Additionally, noteworthy is that, following the same modifications, the SVI30 value of Dijon WRRF decreased from 149 to below 50 mL/g over the entire year (Roche et al. 2022). Inflow rates reached up to 1.5 times the pre-upgrade level, resulting in surface loading rates and overflow rates increasing to 15–20 kg/(m2·h) and 1.5–2.4 m/h, respectively. Furthermore, at Robert W. Hite WWTP, alongside the installation of hydrocyclones, anaerobic selectors were introduced to enhance selective pressure. The effluent from the gravity thickener with a COD concentration of 600 mg COD/L was directed into the first anaerobic tank, while untreated wastewater was introduced into the second anaerobic tank, as depicted in Figure 7(b) (Avila et al. 2021). The results showed an increase in particle size, a significant improvement in settling velocity, but a decline in granule stability as the F/M ratio of the anaerobic selector decreased from 0.18 to 0.13 mg rbCOD/mg VSS, leading to deteriorated settling performance. The outcomes of these applications suggest that implementing hydrocyclones can enhance settling and biological nutrient removal performance. However, concerns about long-term stability, particularly regarding seasonal filamentous growth, remain a prominent challenge. Additionally, the use of anaerobic selectors with a high F/M ratio is expected to mitigate these concerns.
Figure 7

Process schematic of full-scale WWTPs upgraded with AGS-CFRs process: (a) James River WWTP (Welling 2015), (b) Robert W. Hite WWTP (Avila et al. 2021), (c) a municipal WWTP situated in Hebei Province, China (Yu et al. 2023a), and (d) the Town of Moorfield Advanced Nutrient WWTP (Wei et al. 2021b).

Figure 7

Process schematic of full-scale WWTPs upgraded with AGS-CFRs process: (a) James River WWTP (Welling 2015), (b) Robert W. Hite WWTP (Avila et al. 2021), (c) a municipal WWTP situated in Hebei Province, China (Yu et al. 2023a), and (d) the Town of Moorfield Advanced Nutrient WWTP (Wei et al. 2021b).

Close modal

Microaerobic–aerobic configuration with internal separators

In April 2021, a municipal WWTP in Hebei Province, China, implemented a microaerobic/aerobic configuration with internal separators to upgrade the initial AAO process to a continuous-flow AGS process (Yu et al. 2023a, 2024). Prior to the full-scale application, a pilot trial was conducted using the pre-existing anaerobic tank (Yu et al. 2023b). Subsequently, the pre-existing anoxic tank was converted into a microaerobic tank by installing aerators, while internal separators were incorporated into the pre-existing oxic tank, resulting in a total treatment capacity of 2.5×104 m3/day, as shown in Figure 7(c) (Yu et al. 2023a). Over an approximate month-long startup period, distinct and compact AGS was cultivated. During stable operation, granules exceeding 200 μm constituted 28.9% of the total. The 95th percentile effluent concentrations for COD, , and TN were 35.0, 1.2, and 13.3 mg/L, respectively. Moreover, the system demonstrated excellent settling performance and long-term stability, with SVI5 and SVI30 values of 71.6 ± 11.6 and 51.4 ± 8.2 mL/g, respectively. Notably, the configuration eliminated the need for secondary clarifiers and nitrification liquid backflow, resulting in significant area reduction and energy savings.

Moreover, the Town of Moorfield Advanced Nutrient WWTP, with a treatment capacity of 1.55×104 m3/day and originally employing a five-stage Bardenpho process, underwent an upgrade that introduced dry kenaf media and a rotary drum screen for retaining larger particles from the underflow of secondary clarifiers, accompanied by selective waste of flocs (Figure 7(d)) (Wei et al. 2021b). After a 5-month startup period, settleability markedly improved as the SVI30 decreased from over 200 to 50 mL/g, and biological nutrient removal performance was enhanced, with effluent TN and TP concentrations falling below 11 and 1.6 mg/L, respectively. Over 2 years since the upgrade, 60% of the kenaf biofilm particulates displayed sizes ranging between 600 and 1,400 μm, comparable to the diameter of AGS. However, fluorescence in situ hybridization images revealed that ammonia-oxidizing bacteria and PAOs were randomly dispersed throughout the outer 50–60 μm layer. This random distribution of microorganisms with distinct DO requirements indicated the lack of a stratified structure in the kenaf granules, significantly differing from that of AGS. Moreover, according to the definition of AGS established at the 2006 workshop, AGS is formed without the addition of carrier materials. Consequently, it remains a matter of debate whether to categorize kenaf granules as AGS.

Despite the challenges faced in the full-scale applications of AGS-CFRs, a recent investigation into thirteen EBPR WWTPs unveiled that six of these facilities exhibited a significant predominance of PAO-based granules, surpassing 25% (Wei et al. 2020). Moreover, four of these sites demonstrated an even more substantial dominance, exceeding 50% (Wei et al. 2020). These findings suggest that incorporating AGS-CFRs into existing wastewater infrastructure may be easier than expected.

Future research on AGS-CFRs needs to be conducted to address the challenges that currently impede its application. The following recommendations are proposed:

  • (1)

    Elucidate the mechanisms underpinning aerobic granulation: Presently, the primary research focus of AGS-CFRs lies in replicating favorable granulation conditions learned from SBRs. Various reactor configurations and granulation strategies have been developed, based on one or multiple factors influencing granulation. However, due to the limited understanding of granulation processes within CFRs, many proposed configurations are incompatible with existing shallow infrastructures. Furthermore, the formed AGS frequently encounters unforeseen filamentous bacteria overgrowth and disintegration during a stable operation period. Consequently, a comprehensive investigation of granulation mechanisms within CFRs is critical for effectively guiding full-scale implementation.

  • (2)

    Investigate nutrient removal mechanisms: Several studies have documented the efficient biological phosphorus removal of AGS-CFRs under continuous aeration conditions, which challenges the broadly accepted mechanisms involving anaerobic phosphorus release followed by aerobic phosphorus uptake. It is crucial to determine if an alternative metabolic pathway exists for biological phosphorus removal, or if simultaneous phosphorus release and uptake processes transpire within AGS-CFRs under microenvironmental conditions. Moreover, phosphorus removal is intimately connected to denitrification, as DPAOs can employ as electron acceptors for phosphorus removal. Moreover, given that PAOs frequently exhibit a positive correlation with the settling performance and long-term stability of AGS, the importance of this research is threefold magnified.

  • (3)

    Reduce the attention being paid to larger particle size and higher granulation ratio: Due to the inherently lower substrate concentration in CFRs, substrate penetration into granules is limited, leading to smaller particle sizes than that in SBRs. Moreover, given that the AGS definition highlights that the permissible size limit of 0.2 mm can be adjusted according to specific cases or particle types, the particle size threshold for AGS-CFRs can be appropriately reduced. Additionally, as flocculent sludge exhibits lower mass transfer resistance than granules, a hybrid system of flocs and granules is anticipated to be more efficient and versatile, akin to integrated fixed-film activated sludge (IFAS) systems. Consequently, investigations should emphasize system performance, such as sludge settleability and pollutant removal efficiency, rather than the larger particle size or higher granulation ratio.

  • (4)

    Assess long-term stability. Most existing laboratory studies ceased stable operation within 1 year, which may be insufficient to evaluate long-term stability, as filamentous bacteria overgrowth may not have been discovered. Therefore, future research must extend stable operation durations to ascertain long-term stability.

  • (5)

    Encourage full-scale implementation. As current research is primarily focused on the laboratory scale, further full-scale studies are needed to assess whether the developed reactor configurations and granulation strategies can be compatible with existing wastewater treatment facilities.

Significant efforts have been dedicated to achieving aerobic granulation in CFRs. While high shear force is sufficient to drive granulation, additional strategies are imperative for maintaining long-term stability and efficient nutrient removal. Physical selection pressure has emerged as a critical driving force for granulation, with selection pressures based on settling velocity, sludge density, and size all applicable in CFRs. Feast/famine conditions are vital for ensuring long-term stability and nutrient removal, often in conjunction with shear force and physical selective pressure. Furthermore, bioaugmentation can expedite AGS formation and reduce startup time. Although AGS-MBRs can mitigate membrane fouling, they cannot prevent the overgrowth of filamentous bacteria. Nine WWTPs globally have adopted AGS-CFRs technology. The implementation of hydrocyclones can enhance settling and biological nutrient removal performance; however, long-term stability remains a pressing concern. The microaerobic–aerobic reactor with internal separators exhibits exceptional long-term stability, eliminating the need for secondary sedimentation tanks and thereby reducing footprint and energy consumption. Future research should prioritize understanding the granulation mechanism, biological phosphorus removal pathways, system performance, and long-term stability to facilitate the full-scale application of AGS-CFRs.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.

An
Z. H.
,
Bott
C. B.
,
Angelotti
B.
,
Brooks
M.
&
Wang
Z. W.
2021
Leveraging feast and famine selection pressure during startup of continuous flow aerobic granulation systems to manage treatment performance
.
Environ. Sci.: Water Res. Technol.
7
,
1622
1629
.
Avila
I.
,
Freedman
D.
,
Johnston
J.
,
Wisdom
B.
&
McQuarrie
J.
2021
Inducing granulation within a full-scale activated sludge system to improve settling
.
Water Sci. Technol.
84
(
2
),
302
313
.
Brickles
T. A.
2017
A Comparative Study of the Effects of External Selection on Settleability and Formation of Aerobic Granular Sludge
.
Old Dominion University. Norfolk, VA
.
Bumbac
C.
,
Ionescu
I. A.
,
Tiron
O.
&
Badescu
V. R.
2015
Continuous flow aerobic granular sludge reactor for dairy wastewater treatment
.
Water Sci. Technol.
71
(
3
),
440
445
.
Chen
Y. C.
,
Lin
C. J.
,
Chen
H. L.
,
Fu
S. Y.
&
Zhan
H. Y.
2009
Cultivation of biogranules in a continuous flow reactor at low dissolved oxygen
.
Water Air Soil Pollut.: Focus.
9
,
213
221
.
Chen
C. Q.
,
Bin
L. Y.
,
Tang
B.
,
Huang
S. S.
,
Fu
F. L.
,
Chen
Q. Y.
,
Wu
L. Y.
&
Wu
C. M.
2017
Cultivating granular sludge directly in a continuous-flow membrane bioreactor with internal circulation
.
Chem. Eng. J.
309
,
108
117
.
Cofré
C.
,
Campos
J. L.
,
Valenzuela-Heredia
D.
,
Pavissich
J. P.
,
Camus
N.
,
Belmonte
M.
,
Pedrouso
A.
,
Carrera
P.
,
Mosquera-Corral
A.
&
Val del Río
A.
2018
Novel system configuration with activated sludge like-geometry to develop aerobic granular biomass under continuous flow
.
Bioresour. Technol.
267
,
778
781
.
Corsino
S. F.
,
Campo
R.
,
Bella
G. D.
,
Torregrossa
M.
&
Viviani
G.
2016
Study of aerobic granular sludge stability in a continuous-flow membrane bioreactor
.
Bioresour. Technol.
200
,
1055
1059
.
de Kreuk
M. K.
2006
Aerobic Granular Sludge: Scaling Up a New Technology
.
Delft University of Technology, Delft
.
de Kreuk
M. K.
,
Heijnen
J. J.
&
van Loosdrecht
M. C. M.
2005
Simultaneous COD, nitrogen, and phosphate removal by aerobic granular sludge
.
Biotechnol. Bioeng.
90
,
761
769
.
de Kreuk
M. K.
,
Kishida
N.
&
van Loosdrecht
M. C. M.
2007
Aerobic granular sludge-state of the art
.
Water Sci. Technol.
55
,
75
81
.
Ekholm
J.
,
Persson
F.
,
de Blois
M.
,
Modin
O.
,
Pronk
M.
,
van Loosdrecht
M. C. M.
,
Suarez
C.
,
Gustavsson
D. J. I.
&
Wilén
B. M.
2022
Full-scale aerobic granular sludge for municipal wastewater treatment–granule formation, microbial succession, and process performance
.
Environ. Sci.: Water Res. Technol.
8
,
3138
.
Franca
R. D. G.
,
Pinheiro
H. M.
,
Pronk M
M. C. M.
&
Lourenço
N. D.
2018
Stability of aerobic granules during long-term bioreactor operation
.
Biotechnol. Adv.
36
,
228
246
.
Hamza
R.
,
Rabii
A.
,
Ezzahraoui
F. Z.
,
Morgan
G.
&
Lorhemen
O. T.
2022
A review of the state of development of aerobic granular sludge technology over the last 20 years: Full-scale applications and resource recovery
.
Case Studies Chem. Environ. Eng.
5
,
100173
.
Jahn
L.
,
Svardal
K.
&
Krampe
J.
2019
Comparison of aerobic granulation in SBR and continuous-flow plants
.
J. Environ. Manage.
231
,
953
961
.
Kent
T. R.
,
Bott
C. B.
&
Wang
Z. W.
2018
State of the art of aerobic granulation in continuous flow bioreactors
.
Biotechnol. Adv.
36
(
4
),
1139
1166
.
Kishida
N.
,
Totsuka
R.
&
Tsuneda
S.
2012
Challenge for formation of aerobic granular sludge in a continuous-flow reactor
.
J. Water Environ. Technol.
10
(
2
),
79
86
.
Lee
D. J.
&
Chen
Y. Y.
2015
Magnesium carbonate precipitate strengthened aerobic granules
.
Bioresour. Technol.
183
,
136
140
.
Li
W. W.
,
Wang
Y. K.
,
Sheng
G. P.
,
Gui
Y. X.
,
Yu
L.
,
Xie
T. Q.
&
Yu
H. Q.
2012
Integration of aerobic granular sludge and mesh filter membrane bioreactor for cost-effective wastewater treatment
.
Bioresour. Technol.
122
,
22
26
.
Li
J.
,
Cai
A.
,
Wang
M.
,
Ding
L. B.
&
Ni
Y. J.
2014
Aerobic granulation in a modified oxidation ditch with an adjustable volume intraclarifier
.
Bioresour. Technol.
157
,
351
354
.
Liu
Y.
,
Wang
Z. W.
,
Qin
L.
,
Liu
Y. Q.
&
Tay
J. H.
2005
Selection pressure-driven aerobic granulation in a sequencing batch reactor
.
Appl. Microbiol. Biotechnol.
67
,
26
32
.
Liu
H. B.
,
Li
Y. J.
,
Yang
C. Z.
,
Pu
W. H.
,
He
L.
&
Bo
F.
2012
Stable aerobic granules in continuous-flow bioreactor with self-forming dynamic membrane
.
Bioresour. Technol.
121
,
111
118
.
Liu
H. B.
,
Xiao
H.
,
Huang
S.
,
Ma
H. J.
&
Liu
H.
2014
Aerobic granules cultivated and operated in continuous-flow bioreactor under particle-size selective pressure
.
J. Environ. Sci. (China)
26
(
11
),
2215
2221
.
Long
B.
,
Yang
C. Z.
,
Pu
W. H.
,
Yang
J. K.
,
Liu
F. B.
,
Zhang
L.
&
Cheng
K.
2015
Rapid cultivation of aerobic granular sludge in a continuous flow reactor
.
J. Environ. Chem. Eng.
3
(
4
),
2966
2973
.
Morales
N.
,
Figueroa
M.
,
Mosquera-Corral
A.
,
Campos
J. L.
&
Méndez
R.
2012
Aerobic granular-type biomass development in a continuous stirred tank reactor
.
Sep. Purif. Technol.
89
,
199
205
.
Mota
C. R.
,
Head
M. A.
,
Williams
J. C.
,
Eland
L.
,
Cheng
J. J.
&
de los Reyes
F. L.
III
.
2014
Structural integrity affects nitrogen removal activity of granules in semi-continuous reactors
.
Biodegradation
25
,
923
934
.
Partin
A. K.
2019
Hydrocyclone Implementation at Two Wastewater Treatment Facilities to Promote Overall Settling Improvement
.
Virginia Polytechnic Institute and State University Blacksburg, VA
.
Pronk
M.
,
de Kreuk
M. K.
,
de Bruin
B.
,
Kamminga
P.
,
Kleerebezem
R.
&
van Loosdrecht
M. C. M.
2015
Full scale performance of the aerobic granular sludge process for sewage treatment
.
Water Res.
84
,
207
217
.
Qiu
B. Q.
,
Liao
G. H.
,
Wu
C. D.
,
Dai
C. C.
,
Bin
L. Y.
,
Gao
X. L.
,
Zhao
Y.
,
Li
P.
,
Huang
S. S.
,
Fu
F. L.
&
Tang
B.
2022
Rapid granulation of aerobic granular sludge and maintaining its stability by combining the effects of multi-ionic matrix and bio-carrier in a continuous-flow membrane bioreactor
.
Sci. Total Environ.
813
,
152644
.
Song
T.
,
Zhang
X. L.
,
Li
J.
,
Xie
W. Y.
,
Dong
W. Y.
&
Wang
H. J.
2023
Sulfamethoxazole impact on pollutant removal and microbial community of aerobic granular sludge with filamentous bacteria
.
Bioresour. Technol.
379
,
128823
.
Sturm
B.
2020
Balancing Flocs and Granules for Activated Sludge Process Intensification in Plug Flow Configurations
.
The Water Research Foundation. Project
.
Sun
H. H.
,
Yu
P.
,
Li
Q. L.
,
Ren
H. Q.
,
Liu
B.
,
Ye
L.
&
Zhang
X. X.
2017
Transformation of anaerobic granules into aerobic granules and the succession of bacterial community
.
Appl. Microbiol. Biotechnol.
101
,
7703
7713
.
Sun
Y. W.
,
Angelotti
B.
,
Brooks
M.
&
Wang
Z. W.
2021
Feast/famine ratio determined continuous flow aerobic granulation
.
Sci. Total Environ.
750
,
141467
.
Sun
Z.
,
Zhang
J. M.
,
Wang
J.
,
Zhu
H. X.
,
Xiong
J. H.
,
Nong
G. Y.
,
Luo
M. Q.
&
Wang
J.
2023
Direct startup of aerobic granular sludge system with dewatered sludge granular particles as inoculant
.
J. Environ. Manage.
326
,
116540
.
van Dijk
E. J. H.
,
Haaksman
V. A.
,
van Loosdrecht
M. C. M.
&
Pronk
M.
2022
On the mechanisms for aerobic granulation-model based evaluation
.
Water Res.
216
,
118365
.
Wagner
J.
,
Weissbrodt
D. G.
,
Manguin
V.
,
da Costa
R. H. R.
,
Morgenroth
E.
&
Derlon
N.
2015
Effect of particulate organic substrate on aerobic granulation and operating conditions of sequencing batch reactors
.
Water Res.
85
,
158
166
.
Wei
S. P.
,
Stensel
H. D.
,
Quoc
B. N.
,
Stahl
D. A.
,
Huang
X. W.
,
Lee
P. H.
&
Winkler
M. K. H.
2020
Flocs in disguise? High granule abundance found in continuous-flow activated sludge treatment plants
.
Water Res.
179
,
115865
.
Wei
S. P.
,
Stensel
H. D.
,
Ziels
R. M.
,
Herrera
S.
,
Lee
P. H.
&
Winkler
M. K. H.
2021a
Partitioning of nutrient removal contribution between granules and flocs in a hybrid granular activated sludge system
.
Water Res.
203
,
117514
.
Wei
S. P.
,
Quoc
B. N.
,
Shapiro
M.
,
Chang
P. H.
,
Calhoun
J.
&
Winkler
M. K. H.
2021b
Application of aerobic kenaf granules for biological nutrient removal in a full-scale continuous flow activated sludge system
.
Chemosphere
271
,
129522
.
Welling
C. M.
2015
Improving Settleability and Enhancing Biological Phosphorus Removal Through the Implementation of Hydrocyclones
.
MSc thesis, Virginia Polytechnic Institute and State University
.
Winkler
M. K. H.
&
van Loosdrecht
M. C. M.
2022
Intensifying existing urban wastewater
.
Science
375
(
6579
),
377
378
.
Xin
X.
,
Lu
H.
,
Yao
L.
,
Leng
L.
&
Guan
L.
2017
Rapid formation of aerobic granular sludge and its mechanism in a continuous-flow bioreactor
.
Appl. Biochem. Biotechnol.
181
,
424
433
.
Xu
J. P.
,
Sun
Y. X.
,
Liu
Y.
,
Yuan
W.
,
Dai
L.
,
Xu
W. N.
&
Wang
H. L.
2019
In-situ sludge settleability improvement and carbon reuse in SBR process coupled with hydrocyclone
.
Sci. Total Environ.
695
,
133825
.
Yang
Y.
,
Zhou
D. D.
,
Xu
Z. X.
,
Li
A. J.
,
Gao
H.
&
Hou
D. X.
2014
Enhanced aerobic granulation, stabilization, and nitrification in a continuous-flow bioreactor by inoculating biofilms
.
Appl. Microbiol. Biotechnol.
98
,
5737
5745
.
Yu
C.
,
Wang
K. J.
,
Zhang
K. Y.
,
Fu
X. Y.
,
Liu
R. Y.
&
Zheng
P. P.
2023a
Continuous flow aerobic granular sludge process upgrading a full-scale urban wastewater treatment plant
.
China Water Wastewater.
39
(
13
),
1
8
.
Yu
C.
,
Wang
K. J.
,
Zhang
K. Y.
,
Fu
X. Y.
,
Liu
R. Y.
&
Zheng
P. P.
2023b
Pilot study on treatment of low concentration municipal sewage by continuous flow aerobic granular sludge technology
.
Chinese J. Environ. Eng.
17
(
3
),
713
721
.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Licence (CC BY 4.0), which permits copying, adaptation and redistribution, provided the original work is properly cited (http://creativecommons.org/licenses/by/4.0/).