An A2O-MBR system for simultaneous nitrogen and phosphorus removal from brewery wastewater

Anaerobic/Anoxic/Oxic – Membrane BioReactor (A2O-MBR) system was used to enhance simultaneous removal of nitrogen and phosphorus from brewery wastewater. The A2O unit containing microorganisms with short solids retention time (SRT) was employedmainly for removal of organic matter and phosphorus together with denitrification. The MBR containing microorganisms with long SRT was employed mainly for nitrification of NH4-N and recirculation of NO3-N. The model of A2O-MBR system made from polyacrylic with the capacity of 49.5 liters was operated with hydraulic retention times decreased from 24, 18 to 12 hours corresponding to organic loading rates increased from 0.50, 0.75 to 1.00 kg COD/m3 .day. The results showed that the model not only treated organic matter well but also nearly completely removed both nitrogen and phosphorus. For all three loading rates, chemical oxygen demand (COD) concentration decreased significantly in the anaerobic and anoxic compartments of the A2O unit, indicating that most of organic matter was utilized in the anaerobic and anoxic compartments for phosphorus release and denitrification, respectively. Nitrification in the MBR was almost perfectly completed, with average NH4-N removal efficiencies of over 98%. Denitrification in the anoxic compartment happened as much as possible. Demands for the development of PAOs, which were responsible for enhanced biological phosphorus removal (EBPR) processes, could be provided. For loading rate of 0.75 kg COD/m3 .day, treatment efficiencies of COD, NH4-N, total nitrogen (TN) and total phosphorus (TP) of the model were the highest as 95.4, 99.2, 86.7 and 84.6%, respectively. Output values of these parameters were within the limits of Vietnam National Technical Regulation on Industrial Wastewater (QCVN 40:2011/BTNMT), columnA. Themodel of A2O-MBR systemwas capable of achieving effluentswith very low nitrogen and phosphorus concentrations from brewery wastewater.


INTRODUCTION
By Vietnam Beer Alchohol Beverage Association, Vietnamese people consumed nearly 4.1 billion liters of beer in 2017. Currently, there are approximately 129 brewery production facilities across the country with the installed capacity of 4.8 billion litres of beer. Along with this consumption, serious problems with environmental pollution may be caused by a huge amount of brewery wastewater. This amount of wastewater must be treated before discharge into environment. To brewery wastewater, a combination anaerobic-aerobic treatment system has been used and traditional aerobic biological treatment processes such as activated sludge (suspended growth) or biological filter (attached growth) are often implemented [1][2][3][4] . However, these processes have not yet treated thoroughly nitrogen and phosphorus from brewery wastewater to meet QCVN 40:2011/BT-NMT, column A. Anaerobic/Anoxic/Oxic (A 2 O) process commonly used in wastewater treatment is able to remove or-ganic matter together with nitrogen and phosphorus with its own inherent advantages such as short hydraulic retention time (HRT), high pollutant removal efficiency and good shock loading capacity 5,6 . The process consists of three anaerobic, anoxic, oxic compartments and one settling tank which are arranged in sequence with nitrate circulating flow from the oxic compartment to the anoxic compartment and sludge circulating flow from the settling tank to the anaerobic compartment. In this process, nitrification by nitrifiers occurs in the oxic compartment; denitrification by denitrifiers in the anoxic compartment; absorption of β -polyhydroxybutyrate (PHB) for phosphate release by Phosphorus Accumulating Organisms (PAOs) in the anaerobic compartment and then oxidation of PHB for phosphorus accumulation in the oxic compartment. Excess sludge discharge occurs in the settling tank 7 . However, A 2 O process is a single sludge process with the only line for excess sludge discharge at the settling tank so there has been limitation to satisfy a proper SRT for both nitrifiers and PAOs in the oxic compartment of A 2 O process 8,9 . On the other hand, nitrifiers need long SRT and PAOs need short SRT. To solve this problem, incorporation of a biological reactor into A 2 O unit, so-called A 2 O -Biological Reactor system, for simultaneous nitrogen and phosphorus removal has been attempted in the past decade [8][9][10] . The A 2 O unit containing microorganisms with short SRT is employed mainly for removal of organic matter and phosphorus together with denitrification. The biological reactor containing microorganisms with long SRT is employed mainly for nitrification of NH 4 + -N and recirculation of NO 3 − -N. Recently, membrane bioreactor (MBR) is an attractive process that has been increasingly used for advanced biological wastewater treatment. With membrane filtration replacing secondary clarification, MBR possesses a number of merits such as biomass enrichment, perfect nitrification, small footprint, ensured sludge-effluent separation, easy manipulation of HRT and SRT, and excellent effluent quality with little organic and solid contents [12][13][14][15] . Thus, MBR was selected as Biological Reactor in the combined system because of the capacity to achieve enhanced nitrification rate and produce high quality effluent 16,17 .
In this study, an A 2 O-MBR system was used to evaluate the effects of loading rate on the combined system's simultaneous nitrogen and phosphorus removal performance via continuous flow by treating real brewery wastewater. The role of MBR in the combined system and its contribution to organic matter, nitrogen and phosphorus removal were also investigated.

System operating conditions
The wastewater treatment experiment was conducted in four phases in the laboratory at room temperature (∼ 25 • C). In the short initial phase, so-called phase 0, seed sludge was given to 50% volume of the model with MLSS concentration about 5000 mg/L. Influent wastewater with average COD concentration of 500 mg/L diluted with tap water was pumped into the model. Organic loading rate was increased little by little from 0.1 to 0.3 kgCOD/m 3 .day. The phase 0 ended when COD removal efficiency remained stable at above 80%. There was no sludge discharged except sampling to keep large amounts of biomass. In the next three phases according to overall treatment performance in relation to the different loading rates, denoted as 1, 2 and 3, respetively, raw wastewater was pumped continuously with wastewater flow rates increased from 49.5 to 99.0 liters/day corresponding to HRTs decreased from 24 to 12 hours and organic, nitrogen, phosphorus loading rates increased from 0.5 to 1.0 kgCOD/m 3 .day, 0.08 to 0.16 kgTN/m 3 .day, 0.014 to 0.028 kgTP/m 3 .day, respectively as in Table 1.
Excess sludge was discharged from the A 2 O unit and MBR to maintain SRTs from 5 to 7 days and from 45 to 60 days, respectively. Trans-membrane pressure (TMP) was used as an indicator of membrane fouling and monitored continuously by a data logging manometer. When TMP reached 40 kPa, membrane washing was performed physically and chemically following the guidelines of the manufacturer. In the phases 0, 1, 2 and 3, the membrane module was physically washed on a daily basis for 15 min. During the entire period of experiment, the TMP was maintained below 40 kPa. Therefore, the membrane module was not cleaned chemically.

Analytical methods
The samples were collected at the input and output positions of the experimental system. They were also collected in three compartments of the A 2 O unit.
For each loading rate, the model was operated for 45 days to achieve a steady-state condition and the samples were collected over a 3-day period during these days. For determination of the overall treatment performance in terms of organic and nutrient removals, the parameters of wastewater such as COD, suspended solid (SS), Total Kjeldahl Nitrogen (TKN),

RESULTS -DISCUSSION
Organic removal efficiency COD concentrations at different positions in the model were revealed in Figure 2 for loading rates of 0.50, 0.75 and 1.00 kgCOD/m 3 .day. The results showed that COD concentration decreased significantly in the anaerobic and anoxic compartments. The decline could be attributed mainly by the dilution and uptake. About 40% of COD was utilized in the anaerobic compartment by PAOs and 40% of COD was consumed in the anoxic compartment by denitrifiers 10,20 . It changed slightly in the oxic compartment and the MBR. The additional organic removal was attributable to the step of membrane filtration which is beneficial to keep a higher COD removal efficiency 21,22 . Accumulation of PO 4 3− -P by PAOs happened mostly in the oxic compartment. Nitrification of NH 4 + -N by nitrifiers happened mostly in the MBR. Before wastewater flowed into the MBR, large amount of COD in wastewater was removed. It   was considered to be advantageous for the nitrification because of non-inhibitory effects. Therefore, the growth of nitrifiers was favourable and the nitrification was enhanced as well. COD removal efficiencies at various loading rates of the model were represented in Figure 3. For loading rates of 0.50, 0.75 and 1.00 kg COD/m 3 .day, average COD removal efficiencies of the model were 94.1, 95.4 and 92.3%, respectively. It could be seen that COD removal efficiency reached the highest value at the proper loading rate of 0.75 kgCOD/m 3 .day. For these three loading rates, output values of COD were within the limits of QCVN 40:2011/BTNMT, column A. COD removal at different loading rates depended on nitrogen and phosphorus removal mutually through treatment performance.

Phosphorus removal efficiency
Phosphorus concentrations at different positions in the model were revealed in Figure 8 for loading rates of 0.50, 0.75 and 1.00 kgCOD/m 3 .day. The results showed that TP concentration increased to the maximum level in the anaerobic compartment when PAOs released phosphate by utilizing 40% of COD in wastewater as mentioned above. Conditions that favor the growth of PAOs and anaerobic phosphorus release could be provided. TP concentration decreased in the anoxic compartment by the dilution of the return effluent flow from the MBR. In addition, TP concentration also decreased significantly in the anoxic compartment due to its uptake by Denitrifying Phosphorus Accumulating Organisms (DPAOs), which could use nitrate and/or nitrite rather than oxygen as an electron accepter when exposed to an anoxic environment. In the oxic compartment, TP was further accumulated by PAOs to reach complete biological phosphorus removal. Yongzhi Chen et al., 2011 also showed that DPAOs played an important role in removing almost entirely phosphorus from wastewater when treating domestic wastewater by an A 2 O-BAF system 9 . Phosphorus removal efficiencies at various loading rates of the model were represented in Figure 9. For loading rates of 0.50, 0.75 and 1.00 kgCOD/m 3 .day, average TP removal efficiencies of the model were 74.6, 84.6 and 73.5%, respectively. Phosphorus removal efficiency also reached the highest values at the proper loading rate of 0.75 kgCOD/m 3 .day. For all three loading rates, output values of TP were within the limits of QCVN 40:2011/BTNMT, column A. In relation to the results obtained above, the more COD removal or cell growth is, the more phosphorus removal is.

Membrane fouling
Membrane fouling in MBR was inevitable. The TMP in the MBR of the model was monitored continuously to evaluate the membrane fouling during the entire running period. The TMP was in the range of 10 -33 kPa and the flux was from 6.4 to 12.8 L/m 2 .h (LMH). The membrane fouling rate in the MBR correlates well with the MLSS concentration 23 . Figure 10 and Figure 11 show the variations of TMP and MLSS concentration during 135 days of operation. The MLSS concentration initially increased from around 5600 mg/L to nearly 6000 mg/L on day 60 and was maintained for the remaining days of running. When the flux was 6.4 LMH in the phase 1, the TMP was in the range of 10 -16 kPa for 45 days. During the phase 2, the flux was kept at 9.6 LMH. The TMP increased gradually with  time to 26 kPa on day 90. After the phase 2, the flux increased again to 12.8 LMH in the phase 3. The TMP increased almost linearly and reached about 33 kPa on day 135. As mentioned above, the membrane fouling could be alleviated to a certain degree by the intermittent operation of the membrane (2 min rest in every 10 min operation), air bubbling and backflushing.