Advanced Treatment of Digested Restaurant Wastewater Using a Combination of Anaerobic/OxicUnit, Fenton, and a Biological Aerated Filter in Pilot-Scale Treatment

Biological treatment is the most economical and practical wastewater treatment method; however, for highly concentrated organic wastewater, such as digested restaurant wastewater, a single biological treatment method does not meet the discharge requirements. Herein, an anaerobic/oxic-Fenton-biological aerated filter (A/O-Fenton-BAF) system was employed to treat digested restaurant wastewater with a high concentration of organic compounds in a pilot-scale experiment. The degradation process and mechanism of chemical oxygen demand (COD), total nitrogen (TN), NH4+-N, NO3−-N, and dissolved organic nitrogen (DON) in each stage of the process were analyzed. Gas chromatography–mass spectrometry and fluorescence spectrum characteristics were also studied. The average removal rate of both COD and NH4+-N in the entire process was 98%. The removal rates of COD, TN, NH4+-N, and DON reached 78.5%, 66.0%, 95.3%, and 51%, respectively, using the A/O unit. Although Fenton was ineffective in the removal of nitrogenous organic and inorganic substances, the fluorescence spectra and GC–MS showed that the nitrogen-containing organic compounds of macromolecules were transformed into small molecules after the Fenton reaction and could be removed by the BAF unit. The removal rate of DON was up to 24.3% in the Fenton + BAF process, which reduced the TN concentration in the effluent. The dominant species in all biological processes were nitrifying and organic matter-decomposing bacteria. This study provides key data for the design of a full-scale system for treating digested restaurant wastewater.


Introduction
China has more than a dozen cities with populations reaching up to the tens of millions owing to rapid urbanization. Furthermore, large agglomerations of urban residents have resulted in the expansion of the catering industry within these cities, which is growing by approximately 18% per year, twice as fast as the gross domestic product. This sudden growth has triggered complex environmental problems mainly pertaining to kitchen waste. According J. Yin to statistics, the annual output of kitchen waste in China is approximately 1.2 × 10 9 t, and the trend of rapid growth is being maintained (Gao et al., 2022). To manage the ensuing large amounts of kitchen waste, China has designated several pilot cities to focus on testing the harmless treatment of kitchen waste resources for the centralized treatment of kitchen waste. In the process of kitchen waste treatment, a large amount of kitchen wastewater is generated after the solid-liquid separation, the reduction of salinity, and the removal of high chemical oxygen demand (COD), ammonia nitrogen, and oil and grease from most of the oils and fats (Kaushik et al., 2015;Joonveob et al., 2016;Geng & Wang, 2019). Currently, most kitchen wastewater is treated via anaerobic digestion (Lee et al., 2016); however, the digested wastewater still contains high COD concentrations, NH 4 + -N, and total nitrogen (TN) and would be a significant burden for local wastewater plants. The wastewater must meet the standards established as per municipal wastewater management regulations before it enters the municipal wastewater treatment plant. Therefore, the efficient treatment of kitchen digested wastewater is important to reduce the potential harm caused by the digested wastewater and resources consumed in the treatment of kitchen waste. To date, most technical studies on kitchen wastewater have used small-scale actual kitchen wastewater or simulated kitchen wastewater, including membrane bioreactors, anaerobic digestion, electroflocculation, and advanced oxidation technologies (Feng et al., 2016;Han et al., 2014;Zhu et al., 2019). Furthermore, most studies focused mainly on solid kitchen waste treatment, and few studies have reported on the large-scale treatment of actual kitchen digested wastewater (Pham et al., 2015). Some researchers in China have used the relatively expensive membrane process treatment method, for example, Tang et al., (2017aTang et al., ( , 2017b used the combined process of pretreatment + Membrane Bio-Reactor (MBR) + ozone advanced oxidation to treat digested restaurant wastewater in Hebei Province, and determined an average COD effluent value of 296 mg/L. Zhu et al. (2019) employed a process that combined upflow anaerobic sludge blanket (UASB) reactor with A 2 /O-membrane bioreactor (A 2 /O-MBR) to treat kitchen digested wastewater. They found that when the average concentrations of influent COD and ammonia nitrogen were 2590 mg/L and 300 mg/L, the effluent concentrations were 76 mg/L and 117 mg/L, respectively. Additionally, Jiang et al. (2020) conducted a pilot study on the use of an anaerobic membrane bioreactor to treat digested restaurant wastewater. They found that the increase of sludge and dissolved microbial products accelerated membrane contamination and led to the decrease of the treatment effect. Based on such research, there are still technical limitations on understanding how to treat large-scale digested restaurant wastewater both cost-effectively and harmlessly.
Digested restaurant wastewater contains high concentrations of organic matter, nitrogen, and phosphorus, which can be easily degraded by microorganisms. Hence, biological nitrogen-and phosphorus-removal processes, such as the anaerobic/oxic (A/O) and biological aerated filter (BAF) technologies, are the most simple and effective methods to treat digested restaurant wastewater (Gabarró et al., 2019;Gao et al., 2017;Zeng et al., 2018aZeng et al., , 2018b. Such processes have the advantages of low operational cost, low susceptibility to sludge filamentous expansion, strong resistance to shock loading, and ease of automation and management. In our preliminary experiments, we attempted to achieve discharge standards of digested restaurant wastewater using only biological treatment, especially under the conditions of high TN concentrations; however, the results were not encouraging. Advanced oxidation processes offer considerable advantages for the degradation of highly concentrated organic wastewater (Klavarioti et al., 2009;Li et al., 2022;Sirés et al., 2014). In 1894, the French scientist Fenton presented a novel process-now called the Fenton process-for analyzing reduced organic substances and oxidants. H 2 O 2 decomposes under the catalytic effect of Fe 2+ to produce -OH, which oxidizes and decomposes organic matter into intermediates with lower molecular weight through electron transfer and other pathways. Furthermore, Fe 2+ is oxidized to Fe 3+ to produce flocculation precipitation, which can remove a large amount of organic matter. The Fenton process, as an advanced oxidation technology, is widely used in the deep treatment of organic wastewater because of its simple operation and rapid reaction time (Wookeun et al., 2015). In practical engineering, a small amount of the Fenton reagent is often used for the pretreatment of industrial wastewater to partially oxidize the hard-todegrade organic substances in the wastewater and change their biochemical properties, solubility, and coagulation performance for subsequent treatment (Bae et al., 2015;Li et al., 2016aLi et al., , 2016bWu et al., 2017).
Based on previous research, a combined A/O-Fenton-BAF process was developed in this study. The aim of this study was to effectively reduce nutrients, such as COD and nitrogen, in the wastewater and improve the effluent water quality. The performance of the combined process for COD and ammonia nitrogen removal from digested restaurant wastewater was investigated, and the organic constituents in each section were identified using gas chromatography-mass spectrometry (GC-MS) and the excitation-emission matrix (EEM) technique. Functional bacterial populations in A/O and BAF reactors were analyzed to investigate the biological characteristics in the combined process. The results of the present study provide a new engineering perspective for the treatment of digested municipal restaurant wastewater.

Wastewater and Inoculated Activated Sludge
The wastewater used in the present study comprised mixed liquor that was discharged from an anaerobic digestion tank in a kitchen waste treatment plant in Changsha, Hunan Province (South China). The general characteristics of the raw wastewater were as follows: COD, 3000-8000 mg L −1 ; BOD 5 , 1200-3300 mg L −1 ; NH 4 + -N, 600-1500 mg L −1 ; TN, 1200-2000 mg L −1 ; and pH, 7.0-8.2. The activated sludge inoculated in the pilot A/O unit and BAF unit was also obtained from the aeration tank of the same waste treatment plant.

Experimental System
The pilot-scale unit consisted of process units such as a raw water storage tank, A/O unit (8.9 m 3 ), BAF unit (0.5 m 3 ), Fenton reactor, and discharge tank with supporting piping and electrical control devices. The configuration of the pilot-scale unit is shown in Fig. 1, and the set of equipment used for pilot testing is shown in Fig. 2. The flow rate of the A/O was 40 L/h, and the hydraulic retention time of A/O and BAF was 9 days and 10 h, respectively. The dissolved oxygen (DO) in value A and O and the BAF unit was approximately 0.3-0.5, 1.5-2.5, and 3.0-5.0 mg/L, respectively.

Analytical Methods
The pilot unit was operated for 6 months, and the influent and effluent waters of each functional unit were regularly collected, analyzed, and tested during the stabilization period. The DO and pH of wastewater in the reactor were recorded daily using a portable DO/pH meter (HQ20; Hach, Loveland, USA). The COD was measured using a COD speed measuring device (CTL-12; Huatong Inc., Chengde, China). The NH 4 + -N concentration was measured using a spectrophotometer (8453; Agilent, Palo Alto, USA) according to Nessler's reagent spectrometry. The TN and NO 3 --N concentrations were measured using alkaline potassium persulfate-ultraviolet spectrophotometry and the phenol-two sulfonic acid photometric method (State Environmental Protection Administration of China, 2002), respectively. Dissolved organic nitrogen (DON) was calculated by subtracting total inorganic nitrogen content (TIN) from TN content (Kong et al., 2016).

Anaerobic/Oxic Setup
First, 9 m 3 of sludge was inoculated in the A/O process. As the inoculated activated sludge was obtained from the kitchen waste treatment plant, undiluted original wastewater was fed into the A/O unit. The flow rate was 21 L/h, A/O temperature was 25 °C, and sludge reflux ratio was 100%. The COD, MLSS, SV, and SVI of the A/O system were monitored, and the microbial growth phase in the reactor was investigated. When the effluent COD was stable and activated sludge was normal, the startup process lasted 30 days.

BAF Setup
BAF startup was performed simultaneously with the A/O system startup. Inoculated activated sludge was the same as that in the A/O system. After 7 days of aeration, water was continuously fed in the BAF, and the flow rate was gradually increased (5, 10, and 15 L/h). After 18 days, the inlet water flow rate was adjusted to 20 L/h, and the aeration rate was 173 L/h. After 40 days, the COD and nitrogen removal rate remained stable, and the microscopic examination showed rotifers, nematodes, clostridia, and a small amount of filamentous bacteria.

Fenton Experiments
Fenton performance has been mainly investigated according to variations in the initial pH, concentration of H 2 O 2 and Fe 2+ , and reaction time (Li et al., 2016a(Li et al., , 2016bXu et al., 2017). In this study, an L 16 (4 4 ) orthogonal experiment was designed, and the results were analyzed. Thereafter, a single-factor test was conducted. The optimal reaction conditions were as follows: H 2 O 2 concentration, 100 mg/L; Fe 2+ concentration, 300 mg/L; initial pH, 5; and reaction time, 60 min.

Determination and Analysis of Microbial Community Structure
Sludge was obtained from the A/O and BAF systems of the pilot plant, and the sludge used as a comparison sample was also obtained from the aeration tank in the combined kitchen wastewater treatment station. All samples were transported on dry ice to Biotech Bioengineering (Shanghai) Co. for characterizing the variations in microbial communities using Illumina-Miseq sequencing. The total DNA was extracted and amplified using an OMEGA kit (Omega Bio-Tek, Norcross, GA, USA). The primers used for polymerase chain reaction were fused with the V3-V4 universal primers of the Miseq sequencing platform. The V3-V4 region of the 16S rRNA genes was amplified using primer pair 5′-ACT CCT ACG GGA GGC AGC AG-3′ and 5′-GGAC-TACHVGGG TWT CTAAT-3′.

Gas Chromatography-Mass Spectrometry
The organic compounds in the waters were analyzed using electron impact gas chromatography-mass spectrometry (EI-GC-MS7890A/5977C; Agilent, Palo Alto, USA; EI mode at 70 eV). The GC conditions were as follows: inlet temperature, 280 °C; gas interface temperature, 250 °C; carrier gas flow rate, 1.5 mL/min; injection volume, 1 µL; and shunt ratio, 1:1. Samples were heated initially at 40 °C, maintained for 4 min, and then heated to 270 °C at a rate of 10 °C/min and maintained for 20 min. The samples were then heated to 290 °C at a rate of 10 °C/min and maintained for 10 min. The MS conditions were as follows: ion source temperature, 230 °C; quadrupole temperature, 150 °C; EI ionization, 70 eV; and full scan, 35-550 da.

Spectrum Analysis
3DMM fluorescence spectroscopy was employed to measure the concentrations of organic compounds using a Hitachi F-4600 fluorescence spectrometer (F-4600; Hitachi, Ltd, Tokyo, Japan). All samples were tested with the subsequent scanned emission wavelength (Em) range of 220-500 nm and the excitation wavelength (Ex) range of 200-500 nm with a step size of 10 nm; the scanning speed of the spectrum was 12,000 nm/min, and the bandwidth of emission and excitation was 5 nm.

COD Removal Performance
When the system was stably operated, the combined process was operated for 134 days according to the set operating conditions; the activated sludge performance of the system was stable, and the effluent quality was relatively stable (Fig. 3). COD in the influent of the pilot plant fluctuated between 3600 and 7400 mg/L, with an average concentration of 5568 mg/L. The average COD in the effluent was 121 mg/L, which satisfied the concentration limit of the Comprehensive Wastewater Discharge Standard of China (COD concentration < 150 mg/L), and the COD in some periods was < 100 mg/L. The average COD removal rate reached 98%.
In the A/O unit, the COD decreased from an average of 5568 to 1195 mg/L (Fig. 4). The highest COD removal efficiency was achieved, with a removal rate of approximately 78.5%. Furthermore, the CODremoval rates were 69.8% and 66.3% in the Fenton and BAF units, respectively. According to the degradation effect of COD during the process shown in Fig. 4, the strong oxidation effect of the Fenton process resulted in high COD removal in the BAF unit. This combined process can accomplish wastewater treatment goals more effectively than a single biological treatment process.

NH 4 + -N Removal Performance
The influent NH 4 + -N concentration of the pilot plant fluctuated between 660 and 1280 mg/L, with an average concentration of 943 mg/L. The average concentration of effluent NH 4 + -N was 12.6 mg/L after   (Fig. 5), which met the most stringent standard of the combined wastewater discharge standard of China (NH 4 + -N concentration < 15 mg/L). However, it was > 15 mg/L on days 105-125, which may be because of the increase in the activated sludge temperature during operation. Days 105-125 were within the highest average summer temperature period, during which the water temperature of the A/O unit reached 35-37 °C. The high temperature reduced the activity and proliferation rate of nitrifying bacteria and decreased the NH 4 + -N-removal rate. The NH 4 + -N-removal rate quickly increased again to a stable state after the adjustment of device temperature.

Nitrogen-Removal Process
The change in the concentration of N along the process is shown in Fig. 6 The average concentration of NH 4 + -N in the effluent of the BAF unit was 12.6 mg/L. Both TN and DON concentrations decreased substantially in the effluent compared to those measured for the influent. This may be because of the strong oxidation effect of Fenton, which changed the structure of nitrogenous organic matter and converted it to substances suitable for biochemical treatment to be removed by BAF. Although the NO 3 − -N concentration in the BAF effluent was reduced, it still reached 234 mg/L, accounting for 92.5% of the effluent TN. The DON accounted for 6% of the effluent TN.

Microbial Population Diversity Analysis
High-throughput sequencing of microorganisms in the A/O anoxic zone, aerobic zone, and BAF reactor using the Miseq platform yielded 2252, 3133, and 3712 operational taxonomic units, respectively (Table 1), and the coverage indices of the three samples were 95%, 96%, and 96%, respectively. The results indicated that the bacterial library created using this high-throughput sequencing technology contained most bacteria in the system, and the sequencing results represented the

Analysis of Microbial Structures and Functional Population
The taxonomic distribution at the phylum level is summarized in Fig. 7a. Bacteroidetes, Proteobacteria, Chloroflexi, and Planctomycetes, which are commonly found in biological wastewater treatment systems (Ji et al., 2019;Zeng et al., 2018aZeng et al., , 2018bZhao, 2018), dominated all three samples. Bacteroidetes members are involved in refractory organic matter catabolism and polysaccharide metabolism (Huang et al., 2019;Tang et al., 2017aTang et al., , 2017b. Chloroflexi members mainly use carbohydrates in wastewater as nutrient recharge and are also common companion members in anaerobic ammonia oxidation systems. Proteobacteria and Planctomycete members are considered denitrifying functional bacteria (Chen et al., 2013;Hester et al., 2018;Kragelund et al., 2007;Reino et al., 2016;Torresi et al., 2018;Yang et al., 2017). To further explore the evolution of the microbial population, the taxonomic distribution at the class, order, and family levels is summarized in Fig. 7b-d. The abundance and species of the dominant bacteria in the two installations in the A/O process did not differ substantially because of the sludge reflux factor; however, there were some differences in the BAF unit. At the family level, Comamonadaceae, Chitinophagaceae, and Cryomorphaceae were the main microbes in the A/O and BAF units, whereas Anaerolineaceae and Planctomycetaceae were only relatively abundant in the BAF unit. Both bacterial members have a denitrification function. Anaerolineaceae, as a representative family of the green bacterium, have the function of not only denitrification but also degradation of carbohydrates and organic matter (such as amino acids) produced by the decomposition of other bacteria (Chaffron et al., 2010;Han et al., 2019;Li et al., 2009). Here, the analysis of the structural composition of microbial populations revealed that the dominant species in the combined process had the function of degrading COD and TN.

Gas Chromatography-Mass Spectrometry
The samples of raw wastewater and effluent of each unit were analyzed using GC-MS, and the organic composition and change pattern in each unit were obtained as shown in Fig. 8. Fifty-three organic species were detected in the feed water of the A/O unit, comprising several alcohols, ketones, acids, esters, and other long-chain organic compounds as well as aromatic heterocyclic organic compounds and nitrogen-containing heterocyclic compounds, among which p-methyl phenol, as an edible species, was abundant in the kitchen wastewater. After the A/O unit, the COD and nitrogen concentrations substantially decreased. The organic matter in the effluent decreased. Only 12 organics were detected, of which, halogenated alkanes, alkanes, and olefins were the most abundant. There was no major change in the type and content of organics in the Fenton effluent; however, the percentage change in organic matter was calculated based on peak areas. Fenton only changed the structure of certain haloalkane organics and may have changed the structure of some halogenated alkane organics. The number of organic substances in the BAF effluent increased to > 20, and some organic substances such as 3-methyl-2-butanone and 2-chloro-2-methylbutane, which were predominant in water, were mostly degraded, and the organic matter of the hexadecane to eicosane series increased. Microbial metabolism in the biological treatment process produces a variety of organic metabolites such as urea, amino acids, DNA, peptides, and fulvic and humic acids (Cowie & Hedges, 1992). The increase in organic species in the effluent should be related to abundant microbial metabolic activities in the BAF. These metabolic activities produce low-molecular microbial metabolites.

Excitation-Emission Matrix Fluorescence Spectroscopy Characteristics
The EEM spectra of the influent and effluent of different units are presented in Fig. 9. The influent had clear peaks A (220-240/320-360 nm) and B (270-285/320-350 nm). In addition, in region C (250-300/425-450 nm), the fluorescence intensity was high, which was similar to that of A; however, no fluorescence peak appeared (Fig. 9a). According to the region division study of the fluorescence spectrum of EEMs, the A peaks may represent the matter generated by low-excitation wavelength tryptophan-like substances such as tyrosine, tryptophan, and aromatic proteins; the B peaks may represent the matter produced by tryptophan-like substances with high excitation wavelengths, such as tryptophan and its analogs and complexine and its analogs; and the C region may relate to the matter produced by the excitation of humic and fulvic acids and their analogs. Figure 9b shows the EEM profile of the A/O effluent, where the A and B peaks representing protein-like substances disappear and the fluorescence is enhanced in the C region. This means that the organic matter in the water changed after A/O biochemical treatment. Figure 9c, d shows the EEM profiles of the Fenton and BAF effluents, respectively. The fluorescence intensity decreased after Fenton treatment, whereas the fluorescence was invisible after the BAF unit. This indicates that Fenton effectively changes the structure of protein and humic acid substances and facilitates the subsequent biochemical treatment.

Conclusions
The targeted goal of this work was to treat actual digested restaurant wastewater stably and efficiently. Thus, the study had limitations in terms of the choice of treatment process. To achieve this, the present study carried out an innovative pilot experiment in which the combined A/O-Fenton-BAF process was operated for 6 months. Based on the experimental results, three primary conclusions were drawn. (1) The combined process exhibited effective removal of the studied contaminants from digested restaurant wastewater. The process operated stably, and when the influent COD and NH 4 + -N concentrations were 3600-7400 and 660-1280 mg/L, respectively, the average effluent COD was 121 mg/L, and the average NH 4 + -N concentration was < 15 mg/L. (2) The removal rates of COD, TN, NH 4 + -N, and DON by the A/O unit were found to reach 78.5%, 66%, 95.3%, and 51%, respectively. Although nitrogenous organic and inorganic substances were not effectively removed by Fenton process, the fluorescence spectra and GC-MS analyses showed that the nitrogen-containing organic compounds of macromolecules were transformed into smaller molecules after the Fenton reaction and could be removed by the BAF unit. The maximum-removal rate of DON was 24.3% in the Fenton + BAF process, which reduced the concentration of TN in the effluent. NO 3 − -N was the main component of TN in the combined process effluent. (3) Finally, the microbial analysis of the combined process showed that the dominant species in the biological process were nitrifying and organic matter-decomposing bacteria, which was in good agreement with the pollutant treatment objectives. Moreover, it was determined that the increase in DON species in the combination process effluent was related to the abundance of microbial species in the BAF unit, and the metabolites of the microorganisms may be the main cause of DON production. Compared with existing kitchen wastewater treatment processes, the proposed method of this study can both reduce the cost of the membrane process and improve the removal effect of contaminants in kitchen wastewater in a stable and efficient manner. The findings of the present study have direct implications for improving the designs of treatment techniques for digested municipal restaurant wastewater.
Author Contribution All authors contributed to the study conception and design. Material preparation, data collection, and analysis were performed by JY, JJ, and QT. The first draft of the manuscript was written by JY, and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Conflict of Interest
The authors declare no competing interests.