### 1. Introduction

### 2. Materials and Methods

### 2.1. Equipment

### 2.2. Water Characteristics

### 2.3. Experimental Set-up

_{4}

^{+}-N, permanganate index was of 5–8 mg/L. Finally the operation should be continued until the reactor operation was stable. Samples were taken to measure ammonia and permanganate index of the effluent at the same time everyday during the start-up process. When the removal rate of ammonia reached more than 60%, the reactor was considered successful. The entire start-up phase lasted about 30 d.

### 2.4. Analytical Methods

### 3. Results and Discussion

### 3.1. Effect of pH on Pollutants Removal

^{3}/(m

^{2}·h). Fig. 2 shows the removal rates for pollutants at different pH.

_{4}

^{+}-N gradually increased as it became closer to the optimal pH for the growth of nitrifying bacteria. Nitrite bacteria are suitable to live in neutral or weakly alkaline environment. When the environmental pH is less than 6, the metabolism of nitrite bacteria will be seriously affected and even threatened to stagnate. Otherwise, ambient pH is greater than 9.6, the bacteria will also be unable to survive and play biological activities due to the alkaline condition [16]. As nitrifying bacteria are suitable for living in weakly alkaline environment, the ammonia nitrogen purification effect of the reactor is excellent in the pH range of 7.1–8.0. Although nitrification was affected to a certain extent due to the increase of pH in the range of 8.1–8.5, it still showed adaptability to pH.

### 3.2. Effect of Air Gas Ratio (AWR) on Pollutants Removal

^{3}/(m

^{2}·h). These results suggest that AWR impacts the ammonia and permanganate index removal. DO in the water increases as AWR increases, BAF should maintain a certain concentration of DO to facilitate bacterial grow for water purification, otherwise, the rate of biochemical reactions will be affected by the limitations. Therefore, it is very important to choose the appropriate air-water ratio for the stable operation of the biological aerated filter. Fig. 3 shows the removal rates for two pollutants at different AWR from 0.5:1.0–2.0:1.0.

### 3.3. Effect of Hydraulic Load Ratio (HLR) on Pollutants Removal

^{3}/(m

^{2}·h). Increasing the HLR from 3 to 6 m

^{3}/(m

^{2}·h), removal efficiency dropped to 38.8%. Due to long HRT, bacteria had sufficient time to degrade and remove pollutants, and BAF showed good purification efficiency for NH

_{4}

^{+}-N. When HLR increased, although HRT is relatively reduced, the erosion ability and activity of the biofilm was enhanced due to the doubling of the HLR. With the HLR further increasing, the contact time between NH

_{3}-N and nitrifying bacteria in the biofilm decreases and the water flow has a certain hydraulic shear effect on the biofilm, leading to a significant decrease in the removal rate of NH

_{4}

^{+}-N.

^{3}/(m

^{2}·h). Further increase in HLR, reactor purification effect of permanganate index continued to decline, this was because the high HLR, with staying time was too short, insoluble organic matter degradation was directly discharge, and high filtration speed lead to biofilm washed, the biofilm thickness was beyond the scope purification effect, worsen water quality.

### 3.4. Response Surface Test Results and Model Analysis

_{1}), AWR (X

_{2}) and HLR(X

_{3}) were taken as independent variables, and the removal rate of NH

_{4}

^{+}-N (Y

_{1}) and removal rate of permanganate index (Y

_{2}) were taken as response values. Response surface Box-Behnken central combination test (using response surface software Design Expert 8.0.5) was used to optimize the BAF. Table 2 shows the factors, horizontal combinations, and response values of the test. Multiple regression analysis was shown in Eq. (1) and Eq. (2) which were the equations of removal rate of NH

_{4}

^{+}-N and removal rate of permanganate index. The correlation coefficients R

^{2}of the equations were 0.9902 and 0.9934, indicating that the regression model had a good fit with the actual situation and could be used in the optimization test of operational conditions of BAF. Meanwhile, the models (Y

_{1}and Y

_{2}) have a high significance (P < 0.0001).

##### (1)

$$\begin{array}{c}{\text{Y}}_{1}=-16.29706+20.10533{\text{X}}_{1}+3.28222{\text{X}}_{2}+\\ 0.80267{\text{X}}_{3}+0.13333{\text{X}}_{1}{\text{X}}_{2}-0.024{\text{X}}_{1}{\text{X}}_{2}+\\ 0.30667{\text{X}}_{2}{\text{X}}_{3}-1.384{{\text{X}}_{1}}^{2}-1.62222\hspace{0.17em}{{\text{X}}_{2}}^{2}-0.346\hspace{0.17em}{{\text{X}}_{3}}^{2}\end{array}$$##### (2)

$$\begin{array}{c}{\text{Y}}_{2}=-20.13189+9.6626{\text{7X}}_{1}+5.93556{\text{X}}_{2}-0.67{\text{X}}_{3}-\\ 0.37333{\text{X}}_{1}{\text{X}}_{2}+0.064\hspace{0.17em}{\text{X}}_{1}{\text{X}}_{3}+0.16\hspace{0.17em}{\text{X}}_{2}{\text{X}}_{3}-\\ 0.64{{\text{X}}_{1}}^{2}-1.15556{{\text{X}}_{2}}^{2}-0.32\hspace{0.17em}{{\text{X}}_{3}}^{2}\end{array}$$_{4}

^{+}-N and the removal rate of permanganate index was made, and the results were shown in Fig. 5.

_{4}

^{+}-N increased first and then decreased with the increasing of pH. However, from the contour line, the interaction between pH and AWR was not obvious. According to Fig. 5(b), HLR was fixed, the removal rate of NH

_{4}

^{+}-N increased first and then decreased with the increasing of PH, and the interaction was not obvious. From Fig. 5(c), HLR was constant, the removal rate of NH

_{4}

^{+}-N increased first and then tended to be stable with the increase of AWR, and there was an interaction.

^{3}/(m

^{2}•h) at this time, the removal rate of NH

_{4}

^{+}-N and permanganate index reached 59.76%, and 18.16%.

_{4}

^{+}-N and permanganate index can achieve long-term stability and meet the class water quality standard III in “Environmental Quality Standard for Surface Water” as city water source.

### 4. Conclusions

^{3}/(m

^{2}·h), under these conditions, the removal rate of NH

_{4}

^{+}-N is 59.2%, permanganate index is 17.8%.