The device was upgraded based on an existing collection tank structure, using concrete partitions placed within the original tank. The plane dimensions were 5.0 × 5.0 m², with a tank depth of 4.7 m, as shown in Figure 1Figure 1. The designed wastewater treatment capacity was 3 m³/h. From front to back, the compartments were: screen tank, inlet tank, hydrolysis-acidification tank, primary denitrification tank, partial nitrification tank, secondary denitrification tank, anaerobic phosphorus removal tank, aerobic phosphorus removal tank, effluent sedimentation tank, and disinfection tank (See Table 1).

The tanks were separated by concrete walls. Wastewater passed through a screen mesh with 50 mm × 50 mm openings. Aeration pipes were installed at the bottom of the anaerobic hydrolysis-acidification tank, primary denitrification

Figure 1. Sewage tank test flow chart.

Table 1. The design parameters of each pool of the test pool.

tank, secondary denitrification tank, and anaerobic phosphorus removal tank for backwashing the immobilized fillers. Three sets of fine bubble aerators (3 sets × 4 units, diameter 300 mm) were installed at the bottom of the partial nitrification tank to supply oxygen and provide mixing. Based on online monitoring by a fluorescence-based dissolved oxygen probe, intermittent aeration was achieved by controlling the Roots blower to automatically start/stop within a preset DO range. One set of fine bubble aerators (1 set × 2 units, diameter 300 mm) was installed at the bottom of the aerobic phosphorus removal tank to supply oxygen for aerobic phosphorus-removing microorganisms. The internal recycle ratio was adjusted by the flow rate of the internal recycle pump at the end of the partial nitrification tank. Sludge was recycled (and also wasted) by a sludge return pump. The treatment system had no discharge pipeline. With effluent COD ≤ 30 mg/L, and according to the “Standards for Irrigation Water Quality” (GB 5084-2021), which requires COD ≤ 200 mg/L for irrigating dryland crops, the treated effluent met agricultural irrigation standards. Currently, all treated effluent was used for nearby woodland irrigation (See Table 2).

Table 2.Test operation parameters.

According to the “Division of Climatic Seasons” (GB/T 42074-2022), the winter start date for 2024 was determined to be November 4. Taking this date (November 1, 2024) as the start of the low-temperature period, statistics were collected until March 31, 2025, totaling 149 days. The water temperature in the aerobic nitrification tank ranged from 8˚C to 19.3˚C.

The test water was collected via the rural drainage network system, passed through the screen mesh in the test tank for pre-treatment, and then directly entered the raw water tank. It was then lifted by an inlet pump into the hydrolysis-acidification tank. As the test water was rural wastewater, it exhibited the characteristics of low C/N ratio and unstable flow. During the 150-day test period, the average influent NH 4 + -N concentration was 33.4 mg/L, and the average COD concentration was 150 mg/L.

The embedded immobilized fillers used in the experiment were all prepared using polyvinyl alcohol (PVA) gel as the carrier. The specific preparation process was as follows: A 20% (w/w) PVA solution cooled to about 35˚C was mixed proportionally with sludge enriched and cultured in an industrial fermenter. Subsequently, diatomaceous earth, CaCO₃, and powdered activated carbon were added and stirred evenly. The mixed material was processed by a laboratory-designed forming device or specific molds into hollow ring structures with a diameter of 10 mm and a thickness of 3 mm. Then, cross-linking and solidification were performed in a saturated boric acid solution, followed by stabilization treatment in a Na₃PO₄ solution. Finally, the formed fillers were loaded into polypropylene plastic balls with a diameter of 80 mm to create immobilized filler balls for the experiment.

To enable confident interpretation of the reported averages, a water-quality sampling plan was implemented. Samples were collected at each process node as illustrated in Figure 1Figure 1 (e.g., influent, effluent of each functional tank), rather than at fixed time intervals, to capture the performance of individual units along the treatment train. All samples were grab samples. For quality assurance/quality control (QA/QC), the dissolved oxygen (DO) probe was calibrated weekly using the manufacturer’s two-point calibration method. Duplicate samples were analyzed for 10% of the total samples, and the relative percent difference was consistently below 5%. Blank samples (deionized water) were analyzed daily to check for contamination. The method detection limits for NH 4 + -N , NO 2 − -N , NO 3 − -N , and COD were 0.025 mg/L, 0.003 mg/L, 0.02 mg/L, and 5 mg/L, respectively. The practical quantification limits were set at ten times the detection limits.

The immobilized embedded fillers used in the experiment were all immobilized using polyvinyl alcohol (PVA) gel. The detailed preparation procedure is as follows: The PVA-embedded fillers were fabricated by mixing a PVA solution (20%, w/w) with concentrated seed sludge having a moisture content of 95%. CaCO₃ (18 g/L) and powdered activated carbon (38 g/L) were added separately and stirred. The mixture was then processed into a hollow ring structure (diameter: 10 mm, thickness: 3 mm) using an in-house designed and developed embedding molding apparatus. Cross-linking and stabilization were subsequently performed in saturated H₃BO₃ solution and Na₃PO₄ solution. Finally, the embedded fillers were packed into polypropylene plastic spheres with a diameter of 80 mm to obtain the final filler spheres used in the experiment.

The immobilized fillers were loaded into hollow plastic floating balls with a diameter of 80 mm. The bulk volume of the fillers within each ball accounted for approximately 40% - 50% of the effective volume of the ball. This loading rate was maintained consistently across all functional tanks (hydrolysis-acidification, denitrification, and nitrification tanks).

To prevent clogging and maintain long-term biological activity, the filler balls were backwashed regularly. The backwashing was performed using a combination of air and water scouring for approximately 40 minutes per event, with a frequency of once every one to two weeks, depending on the observed pressure drop and visual inspection of biofilm accumulation.

Total Inorganic Nitrogen (TIN) was calculated as the sum of NH 4 + -N , NO 2 − -N and NO 3 − -N . The “total nitrogen removal contribution (65%)” for the primary denitrification tank was calculated as the percentage of TIN removed in that tank relative to the total TIN removed by the entire system (including assimilation and denitrification in all tanks).

Water quality indicators at the inlet and outlet of the partial nitrification and denitrification reactors were measured periodically. According to relevant standards in “Water and Wastewater Monitoring and Analysis Methods” (Fourth Edition): NH 4 + -N was measured by Nessler’s reagent spectrophotometry; NO 2 − -N by N-(1-naphthyl)-ethylenediamine spectrophotometry; NO 3 − -N by ultraviolet spectrophotometry. COD was measured using a Lianhua Technology COD rapid analyzer [19], and DO was measured using a Hach HQ30D portable DO meter.

NAR (Nitrite Accumulation Rate) and AOR (Ammonia Oxidation Rate, mg/(L·h)) were calculated using the following formulas:

where: NAR: Nitrite Accumulation Rate (%); AOR: Ammonia Oxidation Rate (mg/(L·h)); NO 2 − -N eff : Effluent nitrite nitrogen concentration (mg/L); NO 3 − -N eff Effluent nitrate nitrogen concentration (mg/L).

On March 31, 2025, the last day of the system’s low-temperature operation, embedded fillers were taken as representative samples from the hydrolysis-acidification tank, primary denitrification tank, and partial nitrification tank to analyze changes in microbial community structure during long-term operation. Samples were sent to Shanghai Majorbio Bio-pharm Technology Co., Ltd. for microbial diversity analysis. Analysis process: 1) Total DNA was extracted using the Omega Soil DNA Kit; 2) PCR amplification was performed using primers targeting the V3-V4 region of the 16S rRNA gene (338F: 5’-ACTCCTACGGGAGGCAGCAG-3’; 806R: 5’-GGACTACHVGGGTWTCTAAT-3’); 3) Amplified products were purified and subjected to high-throughput sequencing on the Illumina MiSeq platform; 4) Sequencing data were processed on the Majorbio Cloud Platform for OTU clustering, species annotation, and diversity analysis.

As shown in Figure 2Figure 2, this is a photo taken at the scene.

To study the low-temperature resistance of the immobilized biological fillers in actual rural wastewater during winter, Figure 4Figure 4 shows the changes in influent and effluent water quality. Among the collected water samples, the lowest temperature reached 8˚C. Figure 4Figure 4 demonstrates the good treatment performance of the wastewater treatment tank: average NH 4 + -N removal rate was 93.43%, average NO 2 − -N accumulation rate was 82.8%, average COD removal rate was 82.25%, average effluent NH 4 + -N was 0.65 mg/L, and average effluent COD was 24.97 mg/L. This indicates that the biological fillers maintained good biological activity under winter low-temperature conditions. The effluent quality met and surpassed the Class IV standard of the “Environmental Quality Standards for Surface Water” (GB3838-2002), and exceeded the Beijing local discharge standard Level A (DB11/1612-2019).

Figure 4. Change of water quality in and out of the system.

3.1.1. Denitrification Performance in Winter

Figure 5Figure 5 shows a cross-sectional view of the nitrification tank. According to Figure 5Figure 5, the internal structure of the nitrification tank can be seen more intuitively, and it serves as an equipment diagram for this experiment.

According to the aforementioned standard, on day 1 when the water temperature was still 19.3˚C, the NH 4 + -N concentration in the nitrification tank effluent was 0.32 mg/L, the nitrite accumulation rate (NAR) was 83.88%, and the system effluent NH 4 + -N was 0.1 mg/L, far below the Class IV water standard (of China’s GB 3838-2002). As the temperature decreased, the system effluent NH 4 + -N in the partial nitrification tank did not show significant fluctuations. This might be because the AOB population remained dominant as the water temperature slowly decreased, consistent with the research by W Ai et al. [21], which could enhance resistance to temperature. However, on day 55 when the water temperature was 12.5˚C, the system effluent NH 4 + -N was 0.67 mg/L, and NAR decreased to 79.44%. By day 74, the average system effluent NH 4 + -N was 0.52 mg/L > 0.5 mg/L, and the average system NH 4 + -N removal rate decreased from 98.57% to 97.11%.

Figure 5. Section diagram of nitrification tank.

Although the system NH 4 + -N removal rate remained relatively stable, the NAR during this period dropped to a minimum of 76.04%. This could be due to temperature inhibiting AOB activity. However, the embedded fillers provided a stable environment for bacterial growth, maintaining biological activity and mitigating the impact of temperature reduction. Moreover, multiple microbial groups worked synergistically, giving the entire system high stability and resistance to temperature shocks. Simulation experiments on the embedded fillers brought back from the site showed they could still maintain a high NAR. During this period, the system experienced sudden flow changes and cooling. However, by day 111, the system effluent NH 4 + -N returned to 0.55 mg/L, indicating normal effluent. Although the water temperature dropped to 8˚C, based on stable effluent quality, the system’s treatment capacity was higher compared to the activated sludge process [22]. As shown in Figure 6Figure 6, after the Spring Festival, the influent NH 4 + -N concentration gradually increased, reaching a maximum of 55.43 mg/L. However, despite the continuous rise in influent NH 4 + -N , the NH 4 + -N concentrations in the partial nitrification tank and system effluent remained relatively stable without significant fluctuations.

The overall system’s ammonia oxidation rate (AOR) was relatively low, only

Figure 6. Nitrification tank inlet and outlet water nitrogen concentration changes.

4.915 mg/(L·h), but it achieved a relatively stable partial nitrification process. This might be due to: 1) The embedded immobilized fillers used were partial nitrification fillers enriched with AOB, and MIT significantly reduced biomass loss, maintaining a high AOB abundance, thus enabling partial nitrification; 2) Under low DO conditions, the O₂ mass transfer driving force is smaller. This contradicts the research by Yang Qing et al. [23], which suggested high DO could achieve stable partial nitrification, but studies by Blackburne et al. and Zhang Gongliang et al. [24][25] achieved high NAR accumulation by controlling low DO. Although both AOB and NOB activity decrease under low DO conditions, AOB has a lower oxygen half-saturation coefficient than NOB [26], giving AOB a competitive advantage over NOB under low DO. According to the microbial community analysis in Section 4, the dominant genus was Nitrosomonas (10.82%), a key AOB that maintained high activity even under low-temperature conditions [27]. The intermittent aeration and low DO operation mode employed by the process effectively suppressed nitrite-oxidizing bacteria (NOB), inhibiting their proportion to 1.93%, compensating for the decreased metabolic activity caused by low DO, and enabling a stable partial nitrification phase.

Overall, in the low-temperature environment, this phenomenon was mainly attributed to the high biomass maintained by the immobilization technology and the synergistic promotion of microbial activity by suitable substrate concentration and DO environment. This indicates that the embedded biological fillers possess good resistance to shock loads, effectively buffering influent water quality changes, thereby enhancing the overall operational efficiency and stability of the system.

3.1.2. Denitrification Performance in Winter

On day 1, the NO x − -N concentration in the primary denitrification tank effluent was high. At this time, the DO in the nitrification tank was in the range of 3.4 - 4.4 mg/L, and the DO in the recycle flow to the primary denitrification tank was around 2.0 mg/L, which disrupted the anoxic environment for denitrification, leading to poor denitrification performance. The effluent NO x − -N concentration reached 7.19 mg/L. By day 4, the DO in the denitrification tank was reduced by placing the recycle pump at the bottom end of the nitrification tank (below the aerators). On day 7, the effluent NO x − -N concentration from the denitrification tank significantly decreased to 3.51 mg/L. On one hand, because the recycle liquid carried less DO, it helped maintain the anoxic environment in the denitrification tank; on the other hand, due to the low C/N ratio of the wastewater, organic matter in the water could not be fully utilized, resulting in residual NO x − -N in the effluent [28].

Average influent COD = 76.59 mg/L, recycle liquid NO x − -N = 14.33 mg/L, effluent NO x − -N = 2.65 mg/L. The theoretical C/N ratio for reducing NO 3 − -N is 2.86, and for reducing NO 2 − -N it is 1.71 [29]. In this experiment, due to the low NH 4 + -N concentration resulting in fewer electron acceptors provided by the recycle liquid, and the fact that denitrifying bacteria grow thicker biofilms at low temperatures, which might block electron acceptor channels, the effluent NO x − -N concentration from the primary denitrification tank reached 6.19 mg/L on day 107. This was caused by the accumulation of activated sludge and inadequately intercepted sludge from the influent. After backwashing and sludge removal, the system quickly returned to normal.

To enhance the nitrogen removal capability of the primary denitrification tank, the impact of adjusting the internal recycle ratio on denitrification effectiveness was investigated. Under a 100% internal recycle ratio, the effluent NO x − -N concentration was 1.61 mg/L, and effluent TIN was 2.06 mg/L. However, residual COD still existed. From day 36, the internal recycle ratio was increased to enhance nitrite/nitrate supply, strengthen mass transfer effects, and stimulate the activity of denitrifying bacteria [30], thereby improving nitrogen removal efficiency and internal carbon source utilization. As shown in Figure 7Figure 7, a significant decrease in effluent NO x − -N was observed on day 39. To verify whether this result was caused by the recycle ratio adjustment, the internal recycle ratio was restored to 100%. As a result, an increase in effluent NO x − -N was observed again on day 43. This indicates that under low-temperature conditions, increasing the internal recycle ratio can effectively enhance the system’s denitrification capacity.

To further verify the impact of different internal recycle ratios, the internal recycle ratio was increased to 150% starting from day 49. During this phase, the average effluent NO x − -N concentration from the primary denitrification tank was 2.32 mg/L, slightly lower than the 2.97 mg/L under the 100% recycle ratio condition. However, the average water temperature during the operation period with a 150% recycle ratio was lower, which exerted some inhibitory effect on the metabolic activity of denitrifying bacteria. Nevertheless, because the system employed embedding immobilization technology, which effectively maintained

Figure 7. COD and Nitrogen removal effect of the denitrification tank.

microbial performance, the effluent NO x − -N could still be maintained at a relatively low level.

In summary, under low-temperature operating conditions, the scenario with a 150% internal recycle ratio was superior in terms of nitrogen removal effect compared to maintaining a 100% internal recycle ratio, more effectively enhancing the overall denitrification capacity and carbon source utilization efficiency of the system.

To observe changes in pollutant concentrations along the process, samples were taken at points along the flow path. The average concentration changes of pollutants over 150 days along the process are shown in Figure 8Figure 8. The transformation processes of pollutants showed similar overall trends in different stages.

After passing through the hydrolysis-acidification tank, the COD concentration decreased significantly, while the NH 4 + -N concentration slightly increased. This phenomenon is highly consistent with the functions of the microorganisms enriched in this functional tank. The dominant phyla in the embedded fillers of this tank were Bacillota (19.50%) and Bacteroidota (11.29%) (Figure 9(a)). These two groups are efficient hydrolysis-acidification bacteria. From the reaction process perspective, complex organic matter in the wastewater first undergoes hydrolysis by extracellular enzymes of hydrolytic bacteria, producing small molecule organic substances that can be utilized by acidogenic bacteria. These substances are further converted into VFAs [31]. This directly led to the decrease in COD and provided carbon sources for subsequent denitrification. Simultaneously, nitrogenous organic matter released ammonia nitrogen during degradation, undergoing ammonification. At the species level, unclassified_k__norank_d__Bacteria, norank_f__Bacteroidetes_vadinHA17, Rossellomorea, and Trichlorobacter were the dominant species, accounting for 14.67%, 5.22%, 5.41%, and 4.51% respectively. They play important roles in treating antibiotics, increasing VFAs in the system, enhancing nitrogen and phosphorus removal capacity, removing iron and

Figure 8. Average concentration variation of pollutants along the process in the system during low-temperature operation period. (a) Variation of pollutant concentrations along the process in Stage I; (b) Variation of pollutant concentrations along the process in Stage II; (c) Variation of pollutant concentrations along the entire process.

manganese from wastewater, degrading cellulose, and improving hydrolysis-acidification efficiency [32]-[35].

The core function of the primary denitrification tank is to use VFAs produced in the hydrolysis-acidification tank as carbon sources to reduce nitrate/nitrite from the recycle flow of the nitrification tank to nitrogen gas. From Figure 8Figure 8, it can be seen that the effluent COD and NO x − -N concentrations from this unit decreased significantly and simultaneously. In Stage II (R = 150%), this unit’s removal effect on NO x − -N was stronger than in Stage I (R = 100%). Its denitrification performance is inseparable from the special microbial community in this tank. The dominant genus in the embedded fillers of this tank was Acinetobacter, accounting for 31.10% (Figure 9(b)). Acinetobacter is a versatile denitrifying bacterium that can not only use nitrate/nitrite for denitrification but also directly utilize some organic nitrogen sources. This explains why the total nitrogen (TN) loss in this unit is not only from denitrification but may also come from bacterial assimilation.

Figure 9. Microbial community and population abundance in functional pool. (a) Phylum level; (b) Genus level.

It can be observed that a decrease in nitrate nitrogen concentration was detected in the nitrification tank. According to 16S rRNA microbial community analysis: besides Nitrosomonas, the tank contained relatively high abundances of norank_f__JG30-KF-CM45 (6.46%), norank_f__Chitinophagaceae (6.09%), and unclassified_k__norank_d__Bacteria (9.43%), which are non-classical nitrifying bacteria. These bacterial groups can participate in the decomposition of nitrogenous organic matter and possess denitrification capabilities [36]. Therefore, the TIN in the nitrification tank effluent was slightly lower than that of the primary denitrification tank effluent, showing a downward trend in the graph.

When comparing system conditions with recycle ratios of 100% and 150% (Figure 9(a), Figure 9(b)), it was found that the TIN was closer to NH 4 + -N at a 100% recycle ratio. This was due to: 1) Prolonged low temperatures between days 50 - 150 led to reduced biological activity. The overall biological activity in Stage II was lower than in Stage I, resulting in incomplete denitrification in the secondary denitrification process; 2) There were certain fluctuations in water quality after the Spring Festival during Stage II. This led to residual NO x − -N during Stage II, but the effluent TIN still met discharge standards.

For the overall system, under low-temperature conditions, maintaining low DO (3 - 5 mg/L) in the nitrification tank can stabilize AOB growth, and intermittent aeration in a “supply on demand” manner can suppress NOB growth. Simultaneously, increasing the recycle ratio can enhance the denitrification capacity of the primary denitrification tank by providing more substrate. The low DO from the nitrification tank in the recycle liquid to the denitrification tank also helps maintain low DO, preserving the anoxic state in the denitrification tank, creating conditions for denitrifying bacteria growth, and enhancing the overall system’s nitrogen removal performance.