Evaluating waste-derived crude glycerol (CG) as an external carbon source to enhance biological phosphorus removal (BPR) in wastewater treatment Open Access

Enhanced biological phosphorus removal (EBPR) is a widely employed water resource recovery process crucial for the removal of phosphorus (P) from wastewater, thus mitigating eutrophication in receiving water bodies (Massoompour et al. 2022). EBPR relies on the metabolic activity of polyphosphate-accumulating organisms (PAOs) that take up and store carbon (C) and release P under anaerobic conditions followed by C consumption as an energy source to remove P under aerobic conditions (Zhou et al. 2010). Full-scale implementation of EBPR faces several challenges, including (i) the susceptibility of the PAOs to environmental fluctuations and operational upsets, leading to instability in P removal performance, and (ii) the proliferation of glycogen accumulating organisms (GAOs) and other microorganisms that compete with PAOs for resources, such as C, thereby disrupting the P removal efficiency of the system (Dai et al. 2017; Yu et al. 2021). The addition of a supplemental C source has been proposed and used as an effective solution to ensure sufficient C availability for EBPR (Aghilinasrollahabadi et al. 2024a). The supplemental C source, typically in the form of volatile fatty acids (VFAs), simple sugars, or alcohols, serves as a substrate for PAOs during the anaerobic phase, supporting their growth and activity thus enhancing the overall P removal efficiency (Andalib et al. 2017). However, since these external C sources are commonly manufactured from fossil fuel-derived materials, their use increases the overall C footprint of water resource recovery facilities (WRRF) in addition to diminishing the economic benefits (Melia et al. 2017; Izadi et al. 2020; Aghilinasrollahabadi et al. 2024a). As a comparison, waste-derived C sources, such as crude glycerol (CG), have gained increasing attention as a sustainable alternative. As a byproduct of biodiesel production, CG has the potential to enhance the effectiveness of EBPR in WRRFs, especially those with access to CG from nearby biodiesel production plants, while also promoting the system's sustainability (Kumar et al. 2019).

CG is produced at a rate of approximately 10 per 100 kg of biodiesel produced (Attarbachi et al. 2023). As biodiesel production expands, the accumulation of waste CG poses a growing challenge, as it is often incinerated on-site, contributing to environmental concerns such as greenhouse gas (GHG) emissions (Pagliaro 2017). Considering the low cost and high organic content of CG from biodiesel plants, the utilization of CG as a supplemental C source in EBPR presents a potential solution for enhancing the performance of PAO in WRRFs. Furthermore, utilizing CG as a supplemental C source in EBPR provides a more sustainable alternative to fossil-derived C sources, reducing waste disposal impacts and supporting circular economy principles. Guerrero et al. (2012) showed that the direct application of CG (at 200 mg/L) as a C source with an extended anaerobic retention time (5 h) can achieve up to 90% P removal. Other studies investigated the feasibility of alcohols (mainly methanol and ethanol) as the external C source in the EBPR process and the results showed that the addition of methanol in the EBPR process did not improve the activity of PAOs (Shen & Zhou 2016). However, certain groups of PAOs, such as Ca. Halomonas phosphatis can use ethanol under both anaerobic and aerobic conditions, whereas Defluviicoccus (GAOs) present in the system are incapable of taking up ethanol as a C source because they lack the metabolic pathways required for ethanol assimilation. (Burow et al. 2008; Nguyen et al. 2012). Yuan et al. (2010) investigated the effects of CG as the supplemental C source on EBPR and results showed a P removal efficiency of 30%, compared with 95–99% in the control reactor using synthetic wastewater (50% of COD from acetate, 50% of COD from beef extract, yeast extract). Additionally, the majority of the COD was rapidly consumed without P release, suggesting that the direct addition of glycerol at a COD concentration of 350 mg/L was not suitable for EBPR. In another study, Guerrero et al. (2015) conducted pilot-scale experiments using two configurations (anaerobic/anoxic/aerobic and Johannesburg (anaerobic/anoxic/aerobic and one additional anoxic reactor to denitrify nitrate and nitrite recycled)) to evaluate the feasibility of direct CG addition in preventing EBPR failure caused by nitrite and nitrite. High nitrate levels in the anaerobic reactor can lead to competition between denitrifying ordinary heterotrophic organisms and PAOs for the available carbon source, reducing COD availability for EBPR and compromising phosphorus removal efficiency. They found that controlled CG dosing, based on the P concentration in the aerobic phase, maintained stable P removal (>90%). The system also resisted disturbances, including high ammonium (80 mg-N/L) and nitrite (25 mg-N/L) loading, with P removal efficiencies above 75 and 90%, respectively. These results suggested that CG has the potential to prevent EBPR failure due to insufficient C source presence and to enhance simultaneous P and nitrogen (N) removal by supporting both PAOs and denitrifying microorganisms. Additionally, the co-fermentation of CG and waste-activated sludge (WAS) was reported to increase the VFA production by three times compared to the fermentation of WAS alone (Yuan et al. 2010). Given that CG typically contains 60–80% glycerol along with 20–40% methanol and various impurities such as salts, long-chain fatty acids (LCFAs), and soaps (Kumar et al. 2019), its composition may influence its suitability as a C source in EBPR. These constituents may impact PAO activity and downstream processes through different mechanisms. For instance, LCFAs can adsorb onto microbial cell membranes, reducing permeability and interfering with nutrient uptake and energy metabolism (Chen et al. 2018). Additionally, soaps have been shown to alter microbial morphology and reduce motility, leading to inhibition of growth and metabolic activity in mixed microbial communities (Rahman et al. 2017). These results highlight the potential for further investigation into the application of CG, either as a direct supplemental C source for PAOs or as an indirect C source (co-substrate to enhance C yield from fermentation in side-stream enhanced biological phosphorus removal (S2EBPR)), to improve the performance and stability of biological P removal (BPR) in wastewater treatment.

The application of CG aligns with the principles of sustainable wastewater treatment by reducing reliance on chemical additives and promoting resource recovery from waste streams. However, there is limited information on the effectiveness of waste-derived C sources, such as CG, in enhancing PAO activity for P removal in wastewater treatment. The role of different compositions in CG within the BPR process is unclear, and the best strategy for utilizing CG – whether through direct addition or indirect use via co-fermentation – remains uncertain for wastewater treatment professionals and researchers seeking optimal and cost-effective approaches to enhance P removal. This lack of information hinders the practical implementation of CG for P removal, highlighting a knowledge gap. Therefore, this study aimed to investigate the feasibility of using CG in both direct and indirect methods to enhance P removal in WRRFs. Batch bioassays were conducted to assess the activity of PAOs when CG was directly added as a supplemental C source. Additionally, CG was used as a co-substrate with WAS for fermentation, and the produced effluent was evaluated for its effectiveness in supporting PAO activity. Kinetic parameters of PAOs, including P release rate and C uptake rate in the anaerobic phase and P uptake rate in the aerobic phase, were determined and compared. The contribution of primary components in CG, including glycerol, methanol, and VFAs, to PAO activity was also assessed, with results compared to other commercially available C sources (i.e., MicroC^® 2000 and MicroC^® 3000). The potential generation of GHGs, including N₂O and CO₂, from CG application was monitored. Lastly, the economic benefits of direct and indirect CG addition (via fermentation) were compared using an economic model developed in a previous study (Aghilinasrollahabadi et al. 2024b). The findings from this study provide valuable insights into the potential benefits and challenges of integrating CG into wastewater treatment processes, aiming to enhance P removal and improve the sustainability of modern wastewater treatment systems.

Bioassays were performed in batch mode following previous work by Aghilinasrollahabadi et al. (2024b). Prior to the activity tests, 750 mL of PAO-activated sludge from Parkway was added into the laboratory-scale reactor (1 L) and pre-aerated for 30-min (dissolved oxygen (DO) > 2.0 mg/L) to deplete the remaining soluble chemical oxygen demand (sCOD) (sCOD <20 mg/L) in the sludge. Then, different C sources were added (equivalent to sCOD of 100 mg/L) at the beginning of a 150-min anaerobic phase (DO < 0.2 mg/L), followed by a 300-min aerobic phase (DO > 2.0 mg/L), to monitor P release and uptake by PAOs. The reactors were operated at 20 ± 1 °C with a pH of 7.0 ± 0.2 maintained manually using 0.05 M HCl/NaOH. Mixing and anaerobic/aerobic conditions in the reactors were provided by pumping N₂ gas (>99% purity, Robert Oxygen Company, Inc, Rockville, MD, USA) during the anaerobic phase and air during the aerobic phase. Liquid samples were manually collected at 0, 30, 60, 90, 120, 150, 180, 210, 270, 360, and 450 mins and filtered (0.45 μm, Nylon (VWR International LLC, Radnor, PA, USA)) prior to analysis of orthophosphate (OP) and sCOD (Oehmen et al. 2005). Activity tests of each C source were performed at an initial concentration of 100 mg/L as sCOD in duplicate. This concentration was selected based on preliminary screening and previous studies (Aghilinasrollahabadi et al. 2024b; Doyle et al. 2025), which showed that 100 mg/L creates a C source-unlimited condition sufficient to stimulate PAO activity without causing substrate limitation or inhibition. Acetate was used as the control group for comparison.

The performance of indirect CG addition in enhancing PAO activity was assessed via co-fermentation experiments in batch mode followed by batch bioassays. In the co-fermentation experiments, CG was added as a co-substrate to WAS. Then, fermented sludge was collected and used as a C source in the activity tests. To ensure consistency and reproducibility, all experiments (both direct and indirect CG applications) utilized CG from a single, homogeneous batch. This controlled approach minimized potential variability due to differences in CG quality, such as impurities concentration, thereby ensuring comparability of experimental results across tests and supporting reliable discussion and conclusion in this work.

Fermentation experiments were conducted in 30 laboratory-scale reactors (118 mL) with a working volume of 50 mL following the method developed by Yuan et al. (2011) with modifications. These reactors include three groups: Group 1 with WAS as control (WAS-alone), Group 2 with WAS and CG at 150 mg COD/L (WAS-CG-150), and Group 3 with WAS and CG at 300 mg COD/L (WAS-CG-300). 50 mL of WAS (collected from the Parkway WRRF) was added to each group equivalent to 6,980 ± 30 mg COD/L (Table 2). Two CG concentrations (150 and 300 mg/L as COD) were selected for co-fermentation based on previous EBPR studies (Tayà et al. 2015), representing effective yet non-inhibitory concentrations suitable for enhancing fermentative sCOD production. These concentrations also enabled the preparation of bioassays with sCOD levels comparable to those in the direct addition tests, supporting comparison across both strategies. Fermentation (WAS-alone) and co-fermentation (WAS-CG-150 and WAS-CG-300) tests were performed in duplicate. Two additional bottles from each group were kept sealed during the fermentation to track the produced N₂O and CO₂ from the fermentation tests. All reactors were incubated at 20 ± 1 °C and mixed at 170 rpm using a shaker (Thermo Fisher Scientific, USA). Prior to the experiment, reactors were flushed with helium for 10 min and capped with a rubber stopper to achieve an oxygen-limiting condition (DO < 0.2 mg/L). During fermentation, samples were collected on days 0, 2, 5, and 10 to measure total COD (tCOD), sCOD, and OP, as previous studies indicated that fermentation was completed within 10 days under similar operational conditions (Yuan et al. 2011). Samples were filtered (0.45 μm) before sCOD and OP analysis. While longer fermentation may further enhance VFA production, a maximum duration of 10 days was selected here considering process feasibility and operational constraints.

Fermented sludge collected on day 5 (120-h) from the fermentation reactors was used to assess their effectiveness as a C-source in supporting PAO in the activity tests. Similar to the activity tests described in the previous Section 2.2.1, sludge samples were added into the bioassay reactor containing PAO-activated sludge (2.15 ± 0.0 g-VSS/L) with the volumetric ratio of 1:1 (50 mL fermented sludge +50 mL PAO activated sludge) to achieve the initial sCOD concentration of 100–200 mg/L. The reactors with acetate added as a C source were used as the control group for comparison. Batch bioassays were run in duplicate and liquid samples were collected following the procedure described in Section 2.1.1.

The sCOD concentration was analyzed using COD digestion vials (with a detection limit of 20 mg/L) in HACH colorimeter DR900 following the manufacturer's manual (HACH, Loveland, CO, USA). OP was determined utilizing an ultraviolet–visible spectrophotometer (Agilent, Santa Clara, CA, USA) at 880 nm with a detection limit of 0.01 mg/L, following the ascorbic acid method (Aghilinasrollahabadi et al., 2024b). Total suspended solids and volatile suspended solids (VSS) were measured in duplicate using 2 mL of the sample (with a detection limit of 50 mg/L) following the standard methods (APHA 2017). The production of GHGs, including N₂O and CO₂, in the reactors was measured by injecting 100 μL gas samples into an Agilent 6890 gas chromatograph (Agilent Technologies, Santa Clara, CA, USA) equipped with an HP-PLOT Molesieve column (30 m × 0.32 mm) and thermal conductivity detector. Calibration curves were made by injecting 100 μL of the prepared gas at the concentration range of 1–20% (v/v) in triplicate.

To assess the activity of PAOs, kinetic parameters (including the P release rate (mg-P/g-VSS/h), C uptake rate (mg-COD/g-VSS/h) in the anaerobic phase, P uptake rate (mg-P/g-VSS/h) in the aerobic phase, P-uptake rate/P-release rate (dimensionless), and the total P released to C uptake ratio (mg of P released/mg of COD consumed), were calculated as previously described by Aghilinasrollahabadi et al. (2024b). P release and C uptake rates were calculated by the amount of the OP released to the system and the amount of the COD consumed in the first 120 min, respectively, in the anaerobic phase. The ratio of total P released to C uptake was calculated based on the difference in OP and COD concentrations between t = 0 min and t = 150 min (beginning and end of the anaerobic phase). The P uptake rate was determined based on the OP concentration within the first 120 min of the aerobic phase. The values were normalized by the mass of the VSS and time (Drewnowski & Makinia 2014; Larriba et al. 2020). The results were statistically evaluated using Student's paired t-test with a p-value of 0.05, corresponding to a 95% confidence interval.

An Excel-based economic analysis that was previously established by Aghilinasrollahabadi et al. (2024b) was used to estimate the cumulative costs of P removal using S2EBPR by comparing three scenarios: (1) no CG addition (baseline, current S2EBPR configuration at the Parkway WRRF), (2) direct addition of CG to the reactor basins, and (3) indirect addition of CG to fermenter as co-substrate with WAS. The parameters used for the economic analysis were all based on 2024 conditions, and they are summarized in Table 3. The assessment considered the operational costs for a period of 15 years. Capital costs were not included in the analysis due to specific conditions at the study site, the Parkway WRRF. A sidestream anaerobic reactor was already available, having been established in 2020 through the repurposing of an abandoned secondary clarifier. In addition, no new infrastructure for CG storage or dosing was deemed necessary, as CG is supplied by a biodiesel production facility and can be delivered on-demand within two hours as needed (Table 3). As such, the cost model was simplified to evaluate operational savings associated with chemical usage, specifically alum, MicroC^® 3000, and CG. However, it is important to note that potential capital costs – such as those related to storage tanks, dosing systems, safety systems, and possible fermenter construction or modification – were excluded from this framework. While the results indicate favorable operational savings under the tested conditions, these findings reflect localized assumptions.

The C uptake rate for CG was significantly lower (p< 0.05) compared to all tested VFAs (Figure 2(a)). The measured C uptake rates were 5.9 ± 1.3, 6.0 ± 0.1, and 5.0 ± 0.0 mg-COD/g-VSS/h for acetate, propionate, and butyrate, respectively, compared to 2.2 ± 0.1 mg-COD/g-VSS/h for CG. These results highlight CG's lower effectiveness in supporting anaerobic carbon uptake by PAOs relative to VFAs. This may be attributed to that glycerol, the primary component of CG, is not a preferred substrate for PAOs and requires conversion into intermediates such as succinate before it can be taken up and metabolized by PAOs (Viana et al. 2012; Elahinik et al. 2024). Therefore, PAOs might not utilize CG for energy storage as efficiently as the VFAs. Despite this, PAOs were still able to release P, with a P-release rate/C-uptake rate of 1.6 ± 0.2 mg-P/mg-COD and a total P released/total C consumption of 5.6 ± 0.2 mg-P/mg-COD, both significantly higher (p< 0.05) than the values measured for VFAs. This indicated that although the overall C uptake rate was low compared to tested VFAs, CG remained effective in facilitating P release, which is known as an indicator in the PAO-driven EBPR (Wang et al. 2019). However, the P uptake rate with CG was similar to that of butyrate but was 69 and 42% lower than the rates observed using acetate and propionate, respectively (Figure 2(b)). This reduction in the P uptake rate using CG highlights a potential limitation, as higher P uptake is desirable in the EBPR processes to ensure the effective removal of P from wastewater. Additionally, a lower P uptake might require additional chemical dosing to reduce the P concentration and meet the discharge requirements (Izadi et al. 2020). Nevertheless, the P-uptake/P-release ratio of 0.6 ± 0.0 in the group with CG, an indicator of overall PAO activity, fell within the range observed for acetate and propionate (0.6 ± 0.1 and 0.7 ± 0.0, respectively) as C sources. This suggests that CG can support PAO activity at a level comparable to these well-established VFAs. This may be due to the partial conversion of glycerol into short-chain VFAs or other readily biodegradable compounds by non-PAO members in the microbial community, such as actinobacteria (Elahinik et al. 2024). These conversion products can still activate the PAO pathways involved in P release (Yuan et al. 2010). Moreover, the observed ratio (0.6 ± 0.0) aligns with the reported range of 0.2–0.7 for full-scale WRRFs with EBPR (Gu et al. 2008), indicating that CG has the potential to serve as an alternative C source for enhancing EBPR performance in WRRFs.

These results suggest that while CG may not be as efficient as VFAs in terms of C uptake, P release, and P uptake rates, it could be a feasible alternative when considering overall PAO activity, potentially due to the readily biodegradable organic C (e.g., VFA) content in CG. Therefore, the application of CG as a C source to support PAOs for P removal in WRRFs could be considered for facilities with similar processes to the Parkway WRRF, which have easy access to CG. Compared to fossil-fuel-derived C sources such as VFAs, CG could be an environmentally sustainable solution that helps reduce GHG emissions from chemical manufacturing processes, while also serving as a cost-effective alternative C source as it is a waste-product of biodiesel production (Bansod et al. 2024). A feasibility study evaluating alternative waste-derived C sources based on local availability and facility operational conditions would make it possible for other facilities to determine whether and how to use waste-derived C sources, while also providing valuable information to assess its long-term economic and operational sustainability.

Methanol as a C source to support PAO was studied by Puig et al. (2008), and the results revealed a low P release rate (0.14 mg-P/g-VSS/h), suggesting that methanol is not an effective C source to support the activity of PAOs (Puig et al. 2008). However, Tayà et al. (2013) successfully implemented an EBPR process using methanol as the C source, where methanol-degrading organisms adapted to sustain P removal In another study, Puig et al. (2007) revealed the potential for EBPR using ethanol. However, achieving efficient P removal required 140 days of acclimation, posing a significant limitation for full-scale application. Their results showed a low P release rate (0.59 mg-P/g-VSS/h) without acclimation, which increased to 2.02 mg-P/g-VSS/h after 140 days of acclimation. Based on the results in this study and in the literature, VFAs are the preferred choice for short-term C supplementation, as they exhibit higher P release rates and C consumption rates compared to alcohol (Figures 2(a) and 2(b)). However, alcohol may be suitable for long-term supplementation after a period of acclimation.

The C uptake rate for CG was measured at 0.5 ± 0.0 mg-COD/g-VSS/h, which is significantly lower (p< 0.05) than that for ethanol and PG. This lower C uptake suggests that PAOs may not utilize CG as efficiently as ethanol or PG, likely due to an insufficient C source for PAOs, as methanol – an unfavorable C source for PAOs – makes up a significant portion of the organic C content in CG (20–40 wt%) (Puig et al. 2008). The presence of methanol might not directly promote PAO activity, as indicated by the low C uptake rate shown in Figure 3(a), compared to other C sources. Previous studies examined the use of methanol as an external C source in the EBPR process, finding that it led to a low anaerobic P release of 0.13 mg-P/g-VSS/h compared to acetate (1.91 mg-P/g-VSS/h), propionate (1.71 mg-P/g-VSS/h), and ethanol (0.59 mg-P/g-VSS/h) (Puig et al. 2008). Methanol is not efficiently utilized by PAOs under anaerobic conditions and is generally considered an unsuitable substrate for EBPR (Puig et al. 2008). In full-scale systems where nitrate or nitrite is present, methanol may promote denitrification and favor the growth of denitrifiers, potentially leading to substrate competition with PAOs (Shen & Zhou 2016). However, since our batch bioassays did not include nitrate or nitrite, such competition was unlikely. Therefore, the low C uptake rates observed with both methanol and CG (Figure 3(a)) are more likely attributed to PAOs' limited metabolic capacity to utilize methanol, along with possible inhibitory effects from CG impurities. Despite the low C uptake rate in the presence of CG (0.5 ± 0.0 mg-COD/g-VSS/h), PAOs were still able to release P effectively using CG with a rate of 0.8 ± 0.0 mg-P/g-VSS/h. The feasibility of CG as a potential C source was also demonstrated by a high P-release rate to C-uptake rate ratio of 1.6 ± 0.2 mg-P/mg-COD. Previous results suggested that a P-release to C-uptake ratio higher than 0.5 mol-P/mol-C indicates PAOs are dominant in consumption of available C sources (Zhang et al. 2022). These results indicate that, even with limited C uptake, the P release process remained efficient when CG was applied as a supplemental C source.

The P uptake rate of PAOs using CG (0.5 ± 0.0 mg-P/g-VSS/h) was similar to methanol (0.5 ± 0.0 mg-P/g-VSS/h) and higher than PG (0.2 ± 0.0 mg-P/g-VSS/h) and ethanol (0.1 ± 0.0 mg-P/g-VSS/h), demonstrating another aspect of CG as a potential C source for PAOs (Figure 3(b)) compared to alcohols. The methanol content in CG appears to play a key role, as it contributes to the low C uptake while still enabling effective P release. However, this study does not have sufficient data to confirm this relationship. Impurities such as methanol and LCFAs, both commonly present in CG, may influence PAO metabolism through several mechanisms. Methanol, although not a preferred substrate for PAOs, may support the growth of other heterotrophic organisms such as denitrifiers, thereby increasing competition for available C and indirectly suppressing PAO activity (Zhao et al. 2022). LCFAs, particularly at elevated concentrations, can adsorb onto microbial cell surfaces and disrupt membrane integrity, leading to impaired nutrient transport, and inhibition of key enzymes involved in phosphorus metabolism, such as polyphosphate kinase (Tayà et al. 2015). While specific inhibitory thresholds for methanol and LCFAs on PAOs have not been definitively established, studies have reported methanol concentrations above 14 g/L and LCFA concentrations above 1.5 g/L to cause acute inhibition in microbial processes (Palatsi et al. 2009; Kumar et al. 2021). The CG concentrations applied in this study are below these levels, suggesting that acute toxicity is unlikely. Nonetheless, the possibility of sub-inhibitory or synergistic effects – where methanol and LCFAs together may exert combined stress on PAOs – remains unknown. Previous research by Chen et al. (2024) demonstrated that methanol had a significant inhibitory effect on Anammox bacteria, with a half-maximal inhibitory concentration of 143 mg-COD/L, compared to 435 mg-COD/L for CG. While Anammox and PAO activity differ fundamentally, these findings suggest that methanol, as a primary component of CG, could also influence PAO performance, however, its precise role remains unclear. Further investigation is needed to quantify the role of individual and combined CG impurities on PAO function, particularly in long-term EBPR applications.

These results show the importance of evaluating the quality of CG before its use as a C source in BPR for wastewater treatment. This aligns with previous studies indicating that impurities in CG, such as methanol, salts, soap, and LCFAs, can affect PAO activity and disrupt the microbial community composition in the EBPR process (Viana et al. 2012). Notably, CG quality from full-scale manufacturers can vary significantly due to differences in feedstock, production processes, and purification methods employed by each manufacturer, potentially leading to variations in purity and composition (Kumar et al. 2019; Attarbachi et al. 2023). This further reinforces the need to assess CG quality prior to its application as a C source in BPR systems. Considering the important role of methanol in PAO activity and the potential for using alcohols, such as ethanol, as a C source through pre-acclimation, future research should investigate the long-term application of CG in EBPR. Additionally, as CG is a waste-derived C source, exploring other locally available waste-derived C sources could further enhance sustainable and cost-effective solutions for EBPR in wastewater treatment. This includes exploring strategies to optimize PAO activity using local waste-derived C sources with diverse organic compositions (e.g., alcohols and VFAs) while mitigating the inhibitory effects of impurities. In addition, understanding the synergies between CG and other C sources could enhance its effectiveness in wastewater treatment processes and support CG's application in full-scale WRRFs.

C uptake rate using CG was measured at 0.5 ± 0.0 mg-COD/g-VSS/h, which is lower than that of MicroC^® 2000 (1.4 ± 0.4 mg-COD/g-VSS/h), a glycerol-based C source (containing 70–74 wt% glycerol). Considering the C uptake rate using PG (3.2 ± 0.0 mg-COD/g-VSS/h) (Figure 3(a)), low C uptake using CG compared to MicroC^® 2000 and PG can be explained as previous studies confirmed the inability of PAOs in using methanol (20–40 wt% of CG) as the C source (Puig et al. 2008). Although CG and MicroC^® 2000 have a similar range of glycerol (wt%), the C uptake follows the order of glycerol wt%: PG > MicroC^® 2000 > CG, indicating the role of methanol in reducing the C uptake rate by PAOs. This trend may also reflect differences in C forms and the inhibitory impacts associated with impurities in the C sources. While PG is a purified C source, CG and MicroC^® 2000 contain additional constituents, such as methanol, salts, and LCFAs, that may reduce C bioavailability or interfere with PAO uptake kinetics. The inhibitory impacts from CG impurities have been noted in studies by Chen et al. (2024). Although the exact mechanisms were not explored, the findings suggest that CG impurities – particularly methanol – can suppress microbial activity at elevated concentrations. The P uptake rate was also higher for CG (0.5 ± 0.0 mg-P/g-VSS/h) compared to MicroC^® 2000 (0.3 ± 0.0 mg-P/g-VSS/h) and MicroC^® 3000 (0.2 ± 0.0 mg-P/g-VSS/h). Considering the P uptake and C uptake rates, the results suggest that while MicroC^® 2000 can be utilized faster, it may not support PAO activity, as it did not improve the P uptake rate. This is consistent with the results of the higher P-release/C-uptake ratio of PAOs in the presence of CG (1.6 ± 0.2 mg-P/mg-COD) compared to that of MicroC^® 2000 (0.5 ± 0.2 mg-P/mg-COD). Investigating the impacts of denitrification on BPR using MicroC^® 2000 as a supplemental C source by Begum et al. (2009), revealed similar results. Their results showed that MicroC^® 2000 promoted denitrification, with P release and P uptake rates comparable to those of acetate application as a C source. However, they found that the addition of MicroC^® 2000 was not suitable for BPR as it did not enhance P release. Future research could explore the long-term application of CG in WRRFs, as well as potential combinations with other C sources to optimize both cost and efficiency in P removal processes. Additionally, reactor configurations such as sequencing batch reactors (SBRs) or continuous-flow bioreactors may be considered to evaluate the PAO performance of P removal in BPR systems with waste-derived C sources.

An increase in the sCOD production rate within the first 48 h was observed when CG was added at 300 mg/L as tCOD (Figure 5(b)). This increase highlights the potential of higher CG concentrations to improve the fermentation process, thereby boosting the available C for the mainstream of the EBPR. Comparatively, this result suggests that higher CG levels might be a practical strategy to provide an additional C source for the fermentation effluent stream. Previously, Yuan et al. (2010) investigated the co-fermentation of CG with WAS in a laboratory-scale SBRs, where CG was added at 140 mg COD/L to a fermenter operated at HRT of 3 days. Their results showed that VFA production increased by 200% when using 0.3 g/day of CG and 2.02 g-VSS/L of WAS, reaching 0.36 g-VFA/g-VSS/day compared to 0.12 g-VFA/g-VSS/day in WAS fermentation alone. Additionally, they reported that 67 wt% of the produced VFA from co-fermentation was propionic acid, while fermentation of WAS alone predominantly produced acetate (66 wt%). Their observation highlighted a change in fermentation pathways when CG was used as a co-substrate. However, previous research has not reported the impact of CG addition on the sCOD production rates in the co-fermentation process with WAS. The results from this study offer new insights into how CG can be leveraged to improve soluble C production, potentially leading to a more efficient P removal. Figure 5(c) shows the P release rate during the fermentation and co-fermentation. The final P release of 56.5 ± 0.0, 59.9 ± 0.8, and 61.1 ± 0.8 mg/L were observed for WAS-alone, WAS-CG-150, and WAS-CG-300, respectively. P release during WAS fermentation was previously measured by Yuan et al. (2011) within the range of 50 to 100 mg-P/L at various temperatures (4–24 °C) after 10 days. The results of this study also indicated a significant (p< 0.05) increase in P release at the first 48 hours within the co-fermentation groups (WAS + CG) compared to WAS-alone fermentation, suggesting a potential of P release due to PAO activity under anaerobic conditions in the presence of CG as supplementary C sources. This finding underscores the importance of understanding the microbial community composition, particularly P-related biological processes, such as denitrifying PAOs, within the fermentation process in future investigations. Moreover, to our knowledge, no studies have systematically defined the safe upper concentration limit of CG in co-fermentation processes. Although 300 mg/L CG did not result in observable inhibition in this study, higher dosages may cause substrate overload or fermentation instability due to impurity accumulation (e.g., methanol, salts, LCFAs, or soaps), which can disrupt pH balance, damage cell membranes (and reduce nutrient transport), and impair microbial activity (Venkataramanan et al. 2012; Gao et al. 2016; Kumar et al. 2019). Compared with direct CG addition, co-fermentation may tolerate higher CG concentrations because the resulting fermentation products are typically more biodegradable and contain fewer refractory or inhibitory compounds (Perez-Esteban et al. 2022). Nonetheless, identifying concentration thresholds that ensure safe and effective CG application remains a key research need for full-scale EBPR implementation.

It should be noted that the CG contains impurities such as chloride and sulfates, heavy metals, water, methanol, and LCFAs that might impact the performance of the CG as a supplemental C source in EBPR (Viana et al. 2012). Under fermentation conditions, these impurities can have varying effects on microbial activity. For instance, methanol, which is often present in CG as a residual from the biodiesel production process, can be toxic to microorganisms at high concentrations, potentially inhibiting their metabolic activity (Viana et al. 2012). Similarly, LCFAs, which are derived from the triglycerides used in biodiesel production, can adhere to microbial cell walls, preventing nutrient uptake and causing cell death or flotation of biomass (Kolesarova et al. 2010; Viana et al. 2012). Future research focuses on elucidating the role of impurities in CG, particularly potential inhibitory components, on PAO activity. Additionally, the quality of CG should be carefully assessed for the presence of toxic compounds, such as heavy metals, as these contaminants may impair the quality of final wastewater treatment products, including discharged effluent and biosolids. In a previous study, Jensen et al. (2012) demonstrated that Clostridium pasteurianum could efficiently utilize pretreated CG for butanol production through fermentation. Their findings underscore the importance of pretreatment methods, including carbonation (to remove calcium) and electrodialysis (to remove sodium chloride) to enhance CG fermentation. These preprocessing methods could be used in optimizing CG for use in EBPR, as they reduce the negative impacts of impurities while retaining the organic C content that can be utilized by microorganisms. Therefore, future research should explore the effects of different pretreatment methods on the composition of CG fermentation and co-fermentation and their subsequent impact on microbial activity in EBPR systems.

Previous research has highlighted the role of GHG emissions from WRRFs in offsetting the benefits of biological treatment (Wang et al. 2022). Therefore, GHGs, including CO₂ and N₂O, were monitored during the fermentation experiment. No significant (<1%) generation of these gases was detected in any of the tested groups. These results suggest that adding CG as a co-substrate in side-stream reactors for S2EBPR has minimal impact on GHG emissions under the tested condition.

In addition to economic considerations, the decision to implement CG as an alternative C source for PAOs in full-scale WRRFs must account for practical and operational challenges. These include system complexity, increased operation and maintenance demands, and safe handling and storage of CG. For example, CG storage and dosing infrastructure may require corrosion-resistant materials and safety protocols due to impurities such as methanol; higher CG concentrations may lead to toxicity or inhibition of microbial activity; excess dosing could increase foaming potential or lead to sludge flotation and poor settling; and unfermented compounds or suspended impurities may clog filters or impact downstream processes (Attarbachi et al. 2023). These limitations can vary depending on CG quality, application strategy, and plant-specific conditions. To address these concerns and support successful implementation, three actions are recommended prior to full-scale application: (1) assess the availability and delivery logistics of CG for the specific WRRF; (2) conduct a comprehensive quality evaluation of CG – including impurity screening – whether for direct use or co-fermentation; and (3) perform a techno-economic analysis and life cycle assessment to evaluate overall feasibility, potential tradeoffs, and long-term sustainability.

This study highlights CG's potential as both a direct and indirect C source for P removal in EBPR. Compared to VFAs, CG demonstrated a higher P-release/C-uptake rate and a comparable P- uptake/P-release ratio to acetate and propionate, revealing its benefit for supporting PAO activity. However, its lower C uptake compared to VFAs suggested potential limitations as a primary C source. The presence of methanol in CG appeared to influence its effectiveness, contributing to challenges in C uptake. Additionally, CG improved the fermentation process, increasing the sCOD production rate by 68.7% in 48 h, while also exhibiting higher P uptake and C uptake rates (compared to WAS-alone fermentation), reinforcing its benefits for biological P removal.

Direct CG addition provided an immediate C source, but impurities such as methanol and LCFAs require pretreatment to prevent inhibitory effects. In contrast, co-fermentation gradually produces sCOD while reducing refractory and inhibitory compounds, enhancing sCOD accumulation in the fermenter effluent to further support the activity of PAOs. CG was shown to be a cost-effective and environmentally sustainable option for enhancing EBPR, particularly in facilities with access to CG. However, this study used CG from a single, homogeneous batch. While this ensured consistency across experiments, it highlights the need to evaluate how batch-to-batch variability and impurity levels affect PAO performance. Although indirect CG addition showed promising results, the composition and biological effects of the resulting fermentation products remain insufficiently characterized. The economic analysis focused on operational savings under site-specific conditions and excluded capital investments, which should be considered in future cost assessments. Long-term impacts of CG application on nitrogen removal, sludge characteristics, and effluent quality also require further investigation. Future research should focus on defining the optimal and upper-limit dosages of CG in both direct and indirect applications, evaluating CG pretreatment strategies to reduce inhibitory compounds, and assessing long-term impacts on microbial community dynamics, sludge settleability, and downstream treatment units. Pilot-scale or full-scale demonstrations are recommended to validate batch-scale findings and to develop operational strategies tailored to different WRRF configurations and influent characteristics. The choice of CG application strategy – direct addition or indirect use via fermentation – should consider infrastructure availability (e.g., presence of fermenters), process flexibility (e.g., sufficient HRT), CG quality, and transportation cost.

During the preparation of this work, the authors used OpenAI ChatGPT in order to improve language and readability. After using this tool, the authors reviewed and edited the content as needed and take full responsibility for the content of the publication.

This research was funded by A. James Clark of Engineering at the University of Maryland, College Park, and Washington Suburban Sanitary Commission (WSSC Water) in Laurel, MD. The authors also thank Marya Orf Anderson (UMD), Patricia Dotingco Arcellana (UMD), Liu Jiang (UMD), Frank Schmidt (WSSC Water), and other Parkway WRRF operators for assistance.

K.A. investigated the project, developed the methodology, wrote the article, visualized the process, conceptualized the work. B.V.K. conceptualized the work, developed the methodology, wrote the article, validated the study, supervised the process. C.N. conceptualized the work, wrote the article. Y.S. conceptualized the work and wrote the article. G.L. conceptualized the work, developed the methodology, wrote the article, visualized the process, validated the project, supervised the process.

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

The authors declare there is no conflict.