磷是一种生命体必须的元素并被广泛应用于农业以及工业生产中，但它也是一种不可再生资源。生活污水向自然水体中排放时，磷是引起水体富营养化的重要因素，同时这也是磷资源从人类社会流失的一个途径。因此从生活污水中回收磷既可以减少对磷矿石的需求，又可以避免过多的磷在自然水体中造成污染。利用城市生活污水的处理过程中采用的强化生物除磷技术(Enhanced biological phosphorus removal，EBPR)可以将磷富集在活性污泥中。富磷的EBPR污泥在厌氧条件下可以释磷得到高浓度的含磷上清液，从而可以进行磷回收。这种方法具有更加稳定的磷回收率以及较少的药剂成本，是现阶段从经济性以及能耗方面来看最可行的基于城镇生活污水的磷回收技术。但由于较高的磷回收会引起EBPR工艺污泥含磷率的下降，从而影响EBPR系统的稳定性。因此如何在获得高回收率条件下保证EPBR系统的稳定运行需要进一步研究。 本研究旨在开发一种城市污水同时脱氮除磷与磷回收耦合工艺以及相应的运行策略。在保证主流脱氮除磷工艺稳定运行的同时尽可能提高系统的磷回收率。针对所开发的AAO-SBSPR(Anaerobic-Anoxic-Aerobic／Sequencing Batch Side-stream Phosphorus Recovery Process)磷回收工艺首先提出了基于系统磷平衡的运行策略。其次在模拟废水和实际废水的条件下对该工艺和运行策略进行验证，并最终在中试规模的基础上验证新工艺的可行性。此外，研究还采用了ASM2d模型对磷回收工艺进行建模和校正，所得到的参数可以进一步用于优化磷回收工艺的设计与控制。 在以模拟废水为进水的小试AAO-SBSPR反应器中，验证了基于磷平衡的泥龄(Sludge retention time，SRT)与磷回收率相关联的运行方式的可行性。所提出的基于系统磷平衡的公式可以较好的反映出SRT、污泥含磷率以及磷回收三者之间的关系。其中SRT延长对系统污泥含磷率的影响要小于磷回收对污泥含磷率的影响，且在高SRT或进水磷浓度下系统的磷回收潜力更高。此外，采用侧流排出剩余污泥的方式可以进一步提高系统的磷回收潜力。在SRT=35 d的条件下最高回收率可以达到75％，此时系统的污泥含磷率仍有0.043㎎P／㎎VSS，高于一般活性污泥系统0.02㎎P／㎎VSS的含磷率。采用实际废水的小试AAO-SBSPR工艺运行结果表明，由于进水磷浓度较低，基于系统磷平衡的运行方式可以保证在60％的磷回收率前提下将系统的污泥含磷率维持在0.033㎎P／㎎VSS并稳定运行。 此外，模拟和实际废水为进水的AAO-SBSPR工艺小试反应器运行结果表明，磷回收不影响对COD以及氨氮的去除，AAO工艺与AAO-SBSPR工艺对二者的去除效果没有明显不同。同时，AAO-SBSPR系统的磷去除率相较于AAO工艺同样没有明显变化，即使在SRT较高的工况下。其原因在于部分磷通过化学沉淀(回收)的方式去除，提高了主流系统的可处理的磷负荷。而当系统以AAO-SBSPR工艺运行时，TN的处理效果明显提高。主要原因在于以化学沉淀(回收)的方式减少了生物除磷对COD的去除以及AAO-SBSPR工艺对反硝化吸磷作用明显的促进节省了碳源而被用于反硝化过程，因此提高了系统的TN去除效果。 同时，批次实验结果表明对于以模拟废水为进水的小试AAO-SBSPR反应器而言，氨利用率(Ammonia utilization rate，AUR)以及(Nitrite utilization rate，NUR)随着SRT增加而增加。在较高的SRT以及磷回收率的工况条件下，系统中的活性污泥仍具有较好的EBPR活性。但同时观察当污泥含磷率降低后可以明显发现聚磷菌(Polyphosphate accumulating organisms，PAOs)的代谢方式向聚糖菌(Glycogen accumulating organisms，GAOs)的代谢方式转变，分解更多的糖原用于吸收乙酸，同时合成更多的聚羟基脂肪酸酯(PHA)。 从高通量测序结果可知AAO-SBSPR工艺与AAO工艺的微生物种群结构明显不同，AAO-SBSPR工艺的微生物群落具有更高的多样性与丰度。对于小试AAO-SBSPR反应器而言，都观察到了PAOs的种群丰度随着SRT的延长都有所提高且没有观察到常见的GAOs在系统中的增殖。而对AOB和NOB而言可以观察到其相对丰度在SRT=50 d的系统中有较为明显的提高。 采用实际废水的中试AAO-SBSPR工艺的运行结果表明，在延长系统SRT至35 d后可以实现45.0％的磷回收率。此时释磷池上清液PO₄³⁻-P浓度平均为49.7㎎／L，而系统污泥含磷率为0.049㎎P／㎎VSS，仍有进一步提高磷回收率的潜力。但是中试反应器中由于PAOs富集程度低于小试反应器，这使得其污泥EBPR活性较小试有所降低。但批次试验结果表明中试系统以AAO-SBSPR工艺运行时，反硝化吸磷速率占好氧吸磷速率的比例上升。此外，中试AAO-SBSPR反应器的TP的去除并不稳定。这主要是受到温度、进水负荷、水质波动以及系统较长的好氧池停留时间的影响，使得GAOs在系统随温度的升高而大量增殖所致。 采用ASM2d模型对AAO-SBSPR工艺小试反应器的运行数据进行模拟，结果表明只需要通过调整有限的参数就可以使模拟得到的结果与实测数据较为吻合，特别的是对于长SRT数据的模拟也取得了较好的结果。这说明可以利用ASM2d模型模拟AAO-SBSPR工艺并进行运行参数的优化与控制策略的开发。 关键词：脱氮除磷，磷回收，磷平衡，微生物群落结构，数学模拟

 Phosphorus is an essential element of living organisms and is widely used in agricultural and industrial production. Meanwhile, phosphorus is also a non-renewable resource. Municipal wastewater which contains a lot of phosphorus will cause eutrophication if directly discharged into the natural water bodies. Meanwhile, discharge of the phosphorus to the natural water bodies along with the municipal wastewater is also the main pathway for the loss of the phosphorus from human society. However, if we recover phosphorus from municipal wastewater, it can reduce the world's demand for phosphate rock and in the meantime prevent the pollution caused by directly discharging the wastewater into the natural water bodies. If Enhanced biological phosphorus removal (EBPR) technology can be used in the treatment of municipal wastewater, and it can enrich the phosphorus into the activated sludge. For the EBPR sludge, phosphorus release can be achieved by providing Volatile Fatty Acids, (VFAs) under anaerobic conditions and therefore, supernatant with high phosphorus concentration can be obtained for further recovery. This method has more stable phosphorus recovery rate and less chemical cost, which is the most feasible way of recovering phosphorus from municipal wastewater in terms of economy and energy consumption. However, it is reported that higher phosphorus recovery rate will decrease the phosphorus content of the activated sludge, which will affect the performance of the EBPR system. In this condition, further research is needed to find out the operation parameter to achieve higher phosphorus recovery rate while maintaining the stable operation of the EBPR process. Therefore, the aim of this study is to propose a novel biological nutrient removal process to recover phosphorus from municipal wastewater and develop an operation strategy in order to ensure the stable nitrogen and phosphorus removal efficiencies of the process while increasing the overall phosphorus recovery rate of the process. The feasibility of the proposed phosphorus recovery process, AAO-SBSPR (Anaerobic-Anoxic-Aerobic / Sequencing Batch Side-stream Phosphorus Recovery) and the operation strategy were assessed by feeding synthetic and real wastewater in lab-scale as well as pilot scale reactors. In addition, the ASM2d model was also used to simulate the results of the lab-scale AAO-SBSPR reactors, and the calibrated parameters obtained can be further used to optimize the design and operation parameters control of the AAO-SBSPR process. During the operation of the lab-scale AAO-SBSPR reactors fed with synthetic municipal wastewater, base on the phosphorus balance of the system, the sludge retention time (SRT) was coupled with the phosphorus recovery rates of the reactors. Meanwhile, a formula was also proposed and it can accurately reflect the relationship between SRT, phosphorus content in activated sludge and phosphorus recovery rate. It is also found that the effect of SRT extension on the phosphorus content of the system sludge is less than the phosphorus recovering from the system, and the phosphorus recovery potential of the system with long SRT or high influent phosphorus load is higher. Moreover, discharging excess sludge from the side-stream phosphorus release reactor can further increase the phosphorus recovery potential as its lower phosphorus content. The maximum recovery rate can reach 75% with SRTof 35 d while the phosphorus content of the activated sludge decreased to 0.043 ㎎P/㎎VSS, which was still higher than the phosphorus content of 0.02 ㎎P/㎎VSS in the conventional activated sludge system. The results of the lab-scale AAO-SBSPR reactors fed with real wastewater showed that due to the lower phosphorus concentration in the influent, the operation strategy based on the phosphorus balance of the system can still steadily achieve 60% of the phosphorus recovery rate while the phosphorus content was 0.033㎎P/㎎VSS. The performance of the COD and ammonia removal were not affected by the phosphorus recovery, and the removal efficiencies of the COD and ammonia in AAO-SBSPR reactors were comparable to the ones observed in AAO reactors according to the results of the lab-scale reactors fed with synthetic wastewater. This phenomenon was also observed in the operation of the lab-scale reactors fed with real municipal wastewater. Meanwhile, the phosphorus removal efficiencies of the AAO-SBSPR process were comparable to the ones observed in the AAO process, even under the condition of high SRT. The reason may be that part of the phosphorus was removed by chemical precipitation (recovery), which increased phosphorus removal capacity of the mainstream system. The increase of the TN removal efficiencies of the AAO-SBSPR process was observed during the operation of the lab-scale reactors fed with both synthetic and real municipal wastewater. The main reasons were that, on one hand, the chemical precipitation (recovery) reduced COD required by biological phosphorus removal, where the saved the COD can be used in the denitrification process. On the other hand, the promotion of denitrifying phosphorus uptake was observed during the operation of the AAO-SBSPR process. The results of batch experiments showed that for the lab-scale AAO-SBSPR process fed with synthetic wastewater, ammonia utilized rate (AUR) and nitrite utilized rate (NUR) increased with the increase of SRT. Under the conditions of higher SRT and phosphorus recovery rate, the activated sludge in the system still had good EBPR activity. Meanwhile, it can be observed that when the phosphorus content of sludge was reduced, the metabolic pathway of PAOs was shifted where more glycogen was consumed to provide energy for uptake of acetic acid, and more polyhydroxyalkanoates (PHA) were synthesized. The results of high-throughput sequencing analysis showed that the microbial community structures of AAO-SBSPR process and AAO process were different with each other, and the richness and diversity of the microbial community of AAO-SBSPR process were higher than those of AAO process. For the lab-scale AAO-SBSPR reactors, it was observed that the relative abundance of PAOs increased with the extension of SRT and no growth of the GAOs in the systems. Moreover, the increase of the relative abundance of ammonia-oxidizing bacteria (AOB) and nitrite oxidizing bacteria (NOB) was observed in the AAO-SBSPR process with SRT of 50 The results of the pilot-scale AAO-SBSPR reactor showed that 45.0% of the phosphorus recovery rate could be achieved after extending the SRT to 35 d. In this condition, the concentration of PO₄³⁻-P in the supernatant was 49.7 ㎎/L on average, and the phosphorus content in activated sludge was 0.049 ㎎P/㎎VSS, which indicated that further increase of the phosphorus recovery rate was possible at the expense of lowering phosphorus content in activated sludge. In addition, the enrichment of PAOs in the pilot reactor was lower than that of the lab-scale reactors, which resulted in the lower EBPR activity of the system. However, the ratio of denitrifying phosphorus uptake rate to aerobic phosphorus uptake rate was observed in the batch test or pilot test reactor, which indicates that the AAO-SBSPR process promoted denitrification. Moreover, unlike the lab-scale reactors, the removal performance of TP of the pilot scale reactor was unstable. This was mainly caused by the high temperature, variation of the influent load, and the long aerobic hydraulic retention time of the system, which resulted in the proliferation of the GAOs, especially at a higher temperature as indicated by the results of high-throughput sequencing analysis. The ASM2d was used to simulate the operation data of the lab-scale AAO-SBSPR reactors and the results showed that the simulation results can fit the measured data well by only adjusting limited parameters, including for the simulation of long SRT data. This indicated that the AAO-SBSPR process can be simulated using the ASM2d model and the optimization of the operation parameters and the development of the control strategy can be carried out on the basis of the calibrated model. Keywords: Phosphorus and nitrogen removal, Phosphorus recovery, Phosphorus balance, Microbial community structure, Mathematical modeling