Low-cost, reliable, and highly efficient removal of COD and total nitrogen from sewage using a sponge-filled trickling filter

India being a country with the highest human population in the world generates a huge flow of domestic sewage. This voluminous sewage output poses two main challenges to administrative bodies: (1) providing sewerage system and centralized sewage treatment infrastructure and (2) operating sewage treatment facilities to achieve disposal norms. As reported by the Central Pollution Control Board (CPCB), the estimated daily sewage generation in India amounts to a staggering 72.368 × 10⁶ m³/d while the operational sewage treatment capacity stands at 26.869 × 10⁶ m³/d, accounting for merely 37% of the total sewage generation in India (CPCB 2021). One of the main reasons for this conspicuous disparity is the absence of economically viable infrastructure to convey sewage from individual houses to treatment plants. Furthermore, the sewage treatment plants in urban areas are already grappling with spatial constraints, limiting their capacity expansion prospect (Katam et al. 2021). Rural areas are not only challenged by the lack of sewage collection and treatment infrastructure but also by the scarcity of skilled personnel to operate treatment facilities. As a result, unrestrained disposal of untreated sewage causes air, soil, rivers, lakes, and groundwater degradation (Singh et al. 2012; Wear et al. 2021).

The challenges posed by sewage overload can be mitigated through the adoption of decentralized wastewater treatment (DWT) approaches. A spectrum of decentralized technologies has been employed for wastewater treatment, encompassing constructed wetland systems, septic tanks, vermifiltration, etc. However, drawbacks exist for each technology. The integrated constructed wetland system, studied by Behrends et al. (2007), demonstrated good chemical oxygen demand (COD) destruction (80–87%) and ammoniacal nitrogen removal (95%). However, key limitations of constructed wetland systems include a large space requirement, longer hydraulic retention time (HRT), lengthy maturity period, limited oxygen availability, and clogging due to recalcitrated sludge (Makopondo et al. 2020). Sabry (2010) reported DWT using upflow septic tanks/baffled reactors (USBRs) with a notable COD removal of 87% after 1 year of operation. USBR systems involve multiple components, such as baffles, plates, electrical controllers, and biogas escaping lines, and their fully efficient performance requires an extended operational period (Santiago-Díaz et al. 2019). Vermifiltration studied by Sinha et al. (2008) achieved 98% BOD₅ removal. Nonetheless, vermifiltration being a surface area-based technology, higher hydraulic loading rates (HLRs) can reduce HRT and subsequently impact system performance. Moreover, earthworms are very sensitive to higher hydraulic loads and/or adverse environmental conditions (Dey Chowdhury et al. 2022). Besides the above technologies, conventional suspended growth (e.g. activated sludge process (ASP)) and attached growth (e.g. trickling filter (TF)) processes offer excellent sewage treatment efficiency. However, the use of ASPs in rural areas is not feasible due to the need for skilled operators and high-energy consumption. On the other hand, attached growth processes are reported to offer superior treatment efficiencies, operational flexibility, microbial density, and lower biomass production and operational expenditure (Abyar & Nowrouzi 2023). Moreover, attached growth systems such as TFs are reported to remove emerging pollutants from domestic wastewater (Jong et al. 2018; Shukla & Ahammad 2023) and perform nitrification to transform ammonia into nitrate (Forbis-Stokes et al. 2018). Thus, a TF may be explored as an ideal DWT system that is of low-cost and easy to install and operate. However, a conventional TF packed with stones requires a strong structure. Furthermore, the stone medium provides a smaller surface area for the immobilization of microbes, leading to poor treatment efficiency. Although plastic media with a high surface area are commercially available, they are primarily manufactured by some select companies, making them costly. Recently, polyurethane (PU) sponge pieces of different shapes have gained the attraction of researchers as low-cost biomass carriers owing to their lightweight and high surface area and porosity. Various versions of downflow hanging sponge (DHS) reactors have been reported for sewage treatment to achieve >90% BOD and COD removal (Machdar et al. 2000; Tandukar et al. 2006; Ismail et al. 2020; Dacewicz & Grzybowska-Pietras 2021). However, DHS reactors involve hanging sponge pieces of different shapes through a string or a curtain (Machdar et al. 2000; Tandukar et al. 2007). Another version of DHS reactor employed randomly packed sponge pieces with each sponge piece covered by a polypropylene plastic net. Yet another DHS reactor used solid cubes made by copolymerization of PU with epoxy resin (Onodera et al. 2014; Nomoto et al. 2018). Modifications such as hanging the sponge media or covering the sponge media with a plastic net require skillful supervision and specialized manufacturing tools, which also limits its application in rural areas and increases the cost of TF systems.

Hence, this study is conducted to systematically investigate the potential of a TF system consisting of randomly packed sponge cubes as a standalone sewage treatment unit. Two types of sponge cubes, namely polyethylene (PE) and PU, were used as the packing media. Removal of COD and total nitrogen (TN) was investigated along the vertical segments of TF systems operated under varied HLRs. Also, COD removals achieved at various HLRs were fed to a simple first-order kinetic model to derive the kinetic parameters that can be used to design field-scale sponge-filled trickling filter (SPTF) systems.

A common feed tank was employed to supply sewage to both reactors. The sewage was introduced into the system through a single-point inlet pipe, positioned at the center of the uppermost segment of each reactor. To regulate the flow of sewage, peristaltic pumps were utilized. To facilitate the collection of treated effluent, dedicated effluent collection tanks were situated at the base of each reactor. Recirculation of treated sewage was carried out using submersible pumps placed in the effluent collection tanks.

The performance of SPTFs was examined for two different packing media. One was a semi-rigid expanded PE foam, also known as hit-loan sponge foam, and the other was a soft PU foam. Both PE and PU foams are known to have a remarkable resistance to a range of weather conditions, exhibiting minimal signs of deterioration over time. Both PE and PU foams were uniformly cut into 12 mm square cubes and randomly packed in each segment of SPTFs.

The apparent density of media was determined in a laboratory by considering the ratio of the dry mass to the volume of the cube. The porosity of sponge media was determined by immersing a sponge cube in a known volume of distilled water under a vacuum for 5 h, with a reduced volume of water representing the percentage of porosity after removal of the immersed media (Beas et al. 2015; Dacewicz & Grzybowska-Pietras 2021). The apparent density value for PE and PU was 20 and 32 kg/m³, respectively. The porosities of PE and PU foams were approximately 58.8 ± 0.2 and 95.9 ± 0.4%, respectively.

The thorough examination of SPTF system's performance at different HLRs required consistency in the characteristics of raw sewage (Nguyen et al. 2010).

To maintain consistency in the experimental set-up, daily quantities of synthetic sewage were prepared to emulate the composition of real sewage, encompassing organic matter, nutrients, and trace elements. A synthetic sewage solution was prepared by adding 33.7 g sugar, 5 g NH₄Cl, 1 g NaCl, 2.6 g NaH₂PO₄·2H₂O, 1.025 g MgSO₄·7H₂O, 5 g peptone, 1 g CaCl₂, and 0.5 g FeCl₃ per 100 L of tap water (Rottlers et al. 1999; Nopens et al. 2001; O'Flaherty & Gray 2013). The characteristics of this synthetic sewage were: 375 ± 25 mg/L COD, 14.56 ± 0.62 mg/L ammonia (NH₄-N), 11.23 ± 0.55 mg/L nitrate (NO₃-N), 5.17 ± 1.25 mg/L PO₄-P, and a pH of 7 ± 0.4.

Activated sludge was collected from a secondary clarifier of an activated sludge plant treating domestic sewage, and diluted with sewage and tap water to the mixed liquor suspended solids (MLSS) concentration of 3,000–3,500 mg/L. First, all the packing media were submerged in the diluted activated sludge and then filled in all segments of the corresponding reactors. To achieve effective immobilization, the diluted activated sludge was continuously recirculated to the uppermost segment of each SPTF system using a submersible pump until the effluent from the bottom-most stage was almost free from turbidity. The initial amounts of biomass immobilized on media in segments 1, 2, 3, and 4 were 10.8, 11.5, 11.2, and 9.6 g-SS/L of media volume in PE-SPTF, and 14.52, 13.48, 14.96, and 11.7 g-suspended solids/L of media volume in PU-SPTF, respectively.

Both reactors were operated continuously under ambient temperature conditions. To prevent microbial growth, the feed tank and the feed pipeline were cleaned daily using an acidic solution.

HLR, HRT, and organic loading rate (OLR) are important operating parameters for TFs (Arthur et al. 2022). The PE-SPTF and PU-SPTF were operated at HLRs of 2, 3, 4, 5, and 6 m/d. The corresponding overall OLRs were 0.58, 0.88, 1.17, 1.46, and 1.75 kg-COD/m³·d, respectively. At these HLRs, the corresponding overall HRTs were 0.6, 0.4, 0.3, 0.24, and 0.2 d, respectively. Owing to superior performance, the PU media reactor was subjected to further evaluation at two additional HLRs of 8 and 10 m/d.

A sample of influent was obtained from the end of the pipeline feeding the raw sewage to the uppermost segment of the reactor. Effluent samples were collected from the bottom of each segment and analyzed for various parameters without filtration.

The pH of the samples was measured using an Analab digital pH meter (India). Analyses of parameters such as COD, ammonia (NH₄-N), and nitrate (NO₃-N) were performed according to methods 5220 C, 4500-NH₃ F, and 4500-NO₃ B, respectively, described in Standard Methods for the examination of water and wastewater (APHA 2017).

It may be noted from Figure 2 that both reactors achieved an excellent COD removal of up to 97%. Also, most of the COD removal occurred in the first segment of each SPTF reactor. In the PE-SPTF, as the HLR increased from 2 to 5 m/d, the COD concentrations in the effluent from the first segment increased from 30.5 ± 1.37 to 41.7 ± 2.12 mg/L. In contrast, the PU-SPTF consistently maintained effluent COD concentrations below 35 mg/L after the first segment at the HLRs of 2–5 m/d. However, at an HLR of 6 m/d, COD increased to 50.8 ± 2.28 mg/L after the first segment. On the other hand, in the PE-SPTF first segment effluent COD increased from 41.7 ± 2.12 to 80.74 ± 2.26 mg/L at an HLR of 6 m/d. Nevertheless, the final treated effluent from both reactors consistently recorded COD levels below 10 mg/L across all the examined HLRs, providing a remarkable COD removal efficiency of about 97%.

The consistent COD removal in SPTFs within the HLRs ranging from 2 to 5 m/d can be elucidated by considering the inherent characteristics of the randomly packed reactors. Notably, such reactors lack a defined flow path. At lower hydraulic rates, the influent sewage may not fully cover the entire media volume and surface. Subsequently, as the hydraulic rate is increased, a more uniform distribution of the influent is achieved (Huamán et al. 2022), resulting in greater coverage of the media surface by the applied sewage. Furthermore, the biomass growth within the media can introduce delays and alterations in the flow pathways (Fleifle et al. 2013) leading to better contact opportunities between the biomass and sewage. Thus, the SPTFs could maintain excellent COD removal efficiencies at higher HLRs and OLRs due to the reasons mentioned above.

The results obtained from both PE and PU media reactors demonstrate notable efficient performance in comparison with prior studies. For instance, Mahmoud et al. (2010) investigated a DHS reactor containing PU sponge cubes wrapped in a plastic net. The authors reported the maximum COD removal of 89% from an initial concentration of 302 mg/L under an organic load of 1.2 kg-COD/m³·d. In a study by Murata et al. (2021) employing a PU foam structural bed reactor with intermittent aeration, the maximum COD removal of 94 ± 5% was achieved at an organic load of 0.184 kg-COD/m³·d. Uemura et al. (2012) studied the DHS system with three different media sizes. The applied organic load was 1.27 kg-COD/m³·d on each media and achieved a COD removal ranging from 82 to 85%. These comparisons suggest that SPTFs with unmodified and randomly packed sponge media and receiving significantly higher COD load (overall OLR: 0.58–1.75 kg-COD/m³·d) provide superior COD removal efficiencies as compared with the reactors involving modified and hanging sponge media.

It was observed that a significant amount of biomass was washed out from PE-SPTF and collected in the effluent collection tank when operated at an HLR of 6 m/d. This may be attributed to lower porosity and hence the lower biomass holding capacity of PE media. Hence, PE-SPTF was not subjected to HLRs greater than 6 m/d; however, PU-SPTF was tested further at HLRs of 8 and 10 m/d.

In the PE-SPTF as shown in Figure 4(a), NH₄-N removal ranged from 68 to 79% within the first segment and reached near complete elimination at the end of the fourth segment for HLRs up to 5 m/d. However, at an HLR of 6 m/d, the removal efficiency slightly decreased to 92.52 ± 2.31%. On the other hand, as illustrated in Figure 4(b), NH₄-N was completely removed at the end of the second segment and onward in the PU-SPTF, irrespective of the HLR.

A careful observation of Figure 4(a) reveals that for PE-SPTF, for the second segment and onward, the decrease in NH₄-N correlates well with the increase in NO₃-N concentration for HLRs of 2–5 m/d. A similar observation can also be made for PU-SPTF (Figure 4(b)) at HLRs of 2–5 m/d. This suggests that nitrification is the main mechanism contributing to ammonia removal in segments 2–4. Since most of the organic load is removed in the first segment, the lower segments will remain fully aerated, which is a prerequisite for nitrification (Onodera et al. 2014). However, it is interesting to note that at an HLR of 6 m/d, NO₃-N is almost completely removed, and NH₄-N is partially (∼50%) removed in the first segment. Specifically, in the first segment of the PE-SPTF, NO₃-N removal increased from 47.29 ± 4.24% at an HLR of 2 m/d to 99.74 ± 0.13% at an HLR of 6 m/d. Similarly, in the PU-SPTF, NO₃-N removal in the first segment increased from 41.25 ± 2.06% at an HLR of 2 m/d to 96.36 ± 3.01% at an HLR of 6 m/d. The higher HLR will lead to a higher OLR (in this case, ∼7.0 kg-COD/m³·d on the first segment) which may trigger anoxic conditions leading to consumption of NO₃-N as an electron acceptor. Nevertheless, the NH₄-N concentration would remain unaffected under anoxic conditions. However, the simultaneous removal of NO₃-N and NH₄-N in the first segment in the current study indicates the possibility of anaerobic oxidation of NH₄-N (ANAMMOX) in the presence of NO₂-N by the ANAMMOX organisms (Mac Conell et al. 2015). No other plausible explanation exists for the simultaneous and significant removal of NO₃-N and NH₄-N. It is worth noting that no specific conditions were intentionally maintained or provided in SPTF systems for the simultaneous removal of NO₃-N and NH₄-N. It may be understood that the first segment receiving the highest organic load will be partially aerobic. Such conditions facilitate the partial oxidation of NH₄-N to NO₂-N in the first segment (Guo et al. 2010; Wang et al. 2010; Kumar et al. 2016; Bressani-Ribeiro et al. 2018) which in the presence of remaining NH₄-N will be oxidized by ANAMMOX organisms to N₂. The presence of ANAMMOX organisms in wastewater treatment plants and aerobic, anaerobic, and anoxic sludge is already reported (Ding et al. 2017; Mirza et al. 2021). Since the biomass used for the start-up was collected from a sewage treatment plant, the presence of ANAMMOX organisms in both SPTF systems is plausible. Consistent with the results of NO₃-N and NH₄-N removal discussed above, the TN (TN = NO₃-N + NH₄-N) removal increased with an increase in the HLR. The maximum TN removals observed in the PE-SPTF and PU-SPTF were 93.56 ± 1.36 and 92.24 ± 0.66%, respectively, at an HLR of 6 m/d. Since, almost the entire TN removal occurred in the first segments of PE- and PU-SPTFs, we calculated the volumetric TN removal rate based on the volume of media in the first stage. The volumetric TN removal rates in the PE-SPTF at HLRs 2, 3, 4, 5, and 6 m/d were 0.1, 0.16, 0.23, 0.28, and 0.45 kg-TN/m³·d. The corresponding volumetric TN removal rates in the PU-SPTF were 0.11, 0.21, 0.29, 0.38, and 0.37 kg-TN/m³·d. These results reveal that the volumetric TN removal rates in the PU-SPTF at HLRs 2–5 m/d are significantly higher than that in the PE-SPTF. The smaller volumetric TN removal rate in the PU-SPTF at 6 m/d may be attributed to a greater anaerobicity at a higher COD load which will adversely affect oxidation of ammonia to nitrite.

In the previous studies employing TFs containing hanging or covered sponge media, NH₄-N removal was found to be adversely affected due to the increase in organic load (Tawfik et al. 2011; Nomoto et al. 2018; Ismail et al. 2020; Murata et al. 2021). In contrast to these observations, the data from our study reveal that NH₄-N removal increased with the increase in the OLR from 0.58 to 1.46 kg-COD/m³·d (HLR: 2–5 m/d), mainly due to ANAMMOX activity in the first segments of both the reactors. The residual NH₄-N in the effluent of the first segments was removed by nitrification in the subsequent segments of both reactors. Thus, the NH₄-N removal in PE- and PU-SPTFs was unaffected by organic loading. As explained in the above paragraphs and shown in Figure 5, more than one mechanism of NH₄-N and TN removal is possible in PE- and PU-SPTF systems containing randomly packed unmodified sponge media, leading to a greater removal of TN.

The TN removal obtained in our study is comparable to or better than that reported in the published literature. Murata et al. (2021) found the highest TN removal of 74 ± 7% in a PU foam structural bed reactor with periodic mechanical aeration (Murata et al. 2021). Using a third-generation DHS reactor, Mahmoud et al. (2010) reported 72, 90, and 99% removal of ammonia at different HRTs of 2, 4, and 6 h, respectively. The authors also noted that out of the above removals, about 41, 31, and 19% of the initial ammonia remained unaccounted, i.e. could not be related to the generation of nitrate, indicating ammonia removal by the ANAMMOX process. Using TFs with rigid sponge media, Onodera et al. noted 30 and 28% TN removal (Onodera et al. 2013, 2014). Bundy et al. (2017) used a partially submerged DHS reactor operated at an OLR of 0.4 kg-COD/m³·d, HRT of 0.6 d, and influent TN concentration of 42 mg/L, and reported 78% TN removal. A detailed comparison of COD and TN removal obtained in this study with the published reports is given in Table 1.

Since COD removal in segments other than the first segment was very small, the depth of media in Equation (2) was taken as 0.3 m being the packed depth of the first segment. The value of ‘n’ for randomly packed sponge cubes is not reported in the published literature to the best of our knowledge. Since the values of cannot be calculated without knowing ‘n’, we used various values of ‘n’ to calculate and reported the one that provided the best fit for the plotted data offering the R² value closest to 1.

The SPTF system operates as a self-aerated system, offering a distinct advantage in terms of energy consumption when compared with a conventional ASP or any other aerobic biological process. In case of the SPTF system, power is primarily consumed for the pumping of sewage and recirculation of effluent (if any) to the top of the reactor, whereas the ASP system requires power for the aeration system and recirculation of biomass. A comparison was made between the power consumption of the SPTF system and the ASP for the identical COD removal performance. The influent COD concentration of 350 mg/L and a sewage flow equivalent to an HLR of 6 m/d were considered for comparison (under these conditions, COD removal in PU-SPTF and PE-SPTF was 97%). Detailed calculations with relevant data are shown in Supplementary material. The power requirements were found to be 0.222 × 10⁻³ kWh/L for the ASP and 0.0078 × 10⁻³ kWh/L for the SPTF system. It is worth noting that the additional power consumption for continuous sludge recirculation in the ASP is not considered in the above values. Considering CO₂ generation of 0.82 kg/kWh in India (CEA 2022), the CO₂ emission contribution by SPTF and ASP will be 6.4 and 182 mg/L of wastewater treated, respectively. These results highlight the extraordinary energy efficiency exhibited by the SPTF system, demanding a mere 3–4% of the energy when compared with a conventional ASP for an identical COD removal efficiency.

• From initial 350 mg/L COD and ∼27 mg/L TN in raw sewage, both the SPTF systems achieved <2 mg/L of TN and <10 mg/L of COD in the effluent at an HLR of 6 m/d, overall OLR of 1.75 kg-COD/m³·d, and HRT of 0.2 d, thereby conforming to the prevailing sewage disposal standards in India.

• The simultaneous removal of and NH₄-N in the first segment of PE-SPTF and PU-SPTF suggested a significant ANAMMOX activity. The TN removal rate in the first segments of PE- and PU-SPTFs increased with the increase in the HLR and OLR.

• As compared with the PE-SPTF, the PU-SPTF consistently achieved superior COD and TN removal. Based on the kinetic parameters, the equations for COD removal in first segment, related to the HLR were: and for PE-SPTF and PU-SPTF reactors, respectively. These equations were validated with the predicted COD removal correlating with the actual COD removal with R² > 0.97.

• Under similar conditions, as compared with a conventional aerobic biological treatment system such as the activated sludge system, the energy consumption per liter of sewage treated in the PU-SPTF was almost 25 times less.

• The SPTF system with randomly packed sponge media emerges as a cost-effective, reliable, and straightforward solution for DWT.

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

The authors declare there is no conflict.