0 Introduction

Sludge bulking is a frequent phenomenon in the activated sludge system with severe environmental and economic consequences ^[1]. It is mainly associated with excessive growth of filamentous bacteria. The commonest factors to induce "filamentous bulking" are long sludge retention time (SRT), low dissolved oxygen (DO), low organic load, and so on^[2-3]. Except for filamentous bulking, there is another type of sludge bulking termed "viscous bulking", which is determined by sludge bulking without filamentous bacteria. Viscous bulking is usually formed by excessive extracellular polymeric substances (EPS)^[4], and often occurs when the nutrients is deficient or the organic load is high^[5]. When viscous bulking occurs, the viscosity of sludge flocs increases significantly^[6]. Unlike filamentous bulking, viscous bulking does not occur frequently, and adding flocculant is considered to be an effective method to prevent viscous bulking ^[7].

So far, most research about sludge bulking is focused on how to eliminate it but not to make use of it. Recently, Guo et al.^[8] established a low energy-saving wastewater treatment system. In the system, limited or slight filamentous bacteria were stimulated by the low DO. By utilizing the characteristics of filamentous bacteria such as larger specific surface area and stronger ability to degrade low concentration substrates, this system can improve treatment performance while saving energy consumption. It presented a novel pathway to treat filamentous sludge bulking. In some special cases, sludge bulking can be utilized. However, few researches have been reported in this new area, especially for viscous bulking.

The aim of this research was to investigate the characteristics of nutrients removal performance in viscous bulking state. The kinetics of nitrification between normal sludge and viscous bulking sludge were compared. It is also hoped this research can provide a new perspective on viscous bulking.

1 Material and Methods 1.1 Experimental Design

The experiments were operated using synthetic wastewater which contained (per liter)CH₃COONa·3H₂O (663.80 mg), NH₄Cl (166.90 mg), KH₂PO₄ (18.80 mg), CaCl₂·2H₂O (40 mg), NaHCO₃ (375 mg), MgSO₄ (80 mg), and micro nutrients solution^[9] (0.30 mL). The corresponding water quality was 330 mg/L COD, 45 mg/L NH₄⁺-N, and 4.30 mg/L PO₄^3--P.

The study was carried out using one sequencing batch reactor (SBR) with a height of 700 mm, diameter of 200 mm, and working volume of 12 L. Sampling ports with a 10 cm interval were set in the vertical direction of the reactor, as shown in Fig. 1. Each cycle contained feed period, anaerobic period, aerobic period, and settle/discharge/decant period (Table 1). At the end of each aerobic period, excess sludge was discharged to control SRT. At the end of settle period, supernatant was discharged to control hydraulic retention time (HRT). Temperature was controlled at 21 ± 1 ℃ by temperature controller. Inoculated sludge was obtained from an anaerobic/ anoxic/aerobic (AAO) reactor in our laboratory (SVI and MLSS were around 90 mL/g and 2000 mg/L, respectively), and synthetic wastewater was treated with the same components as above. This research consisted of a sequence of three experimental phases. In the first phase, inoculated sludge was acclimated to the synthetic wastewater under low COD load; in the second phase, COD load increased with the increase of volume exchange ratio to induce viscous sludge bulking; in the third phase, activated sludge under viscous bulking was operated under low COD load. Operation procedures are highlighted in Table 1.

1.2 Kinetic Analysis

The nitrification rates are fitted by the Monod equation^[10]:

$ V=-\frac{\mathrm{d} C}{\mathrm{~d} t}=V_{m} X \frac{C}{K_{s}+C} $

(1)

where V is the nitrification rates of NH₄⁺-N and NO₂^--N, mg/(L·h); X is the MLSS, mg/L; V_m is the maximum specific nitrification rates of NH₄⁺-N and NO₂^--N, mg/(g·h); K_s is the half-saturation constant, mg/L.

Otherwise, the nitrification rate could also be expressed in the following form:

$ V=-\frac{\mathrm{d} C}{\mathrm{~d} t}=k C^{n} $

(2)

where n is the reaction order; k is the reaction rate constant. Actually, many nitrification kinetics occurring in biological nitrogen removal systems is close to a first-order model (n=1)^[10].

1.3 Calculatory Procedures

The SND efficiency and NO₂^--N accumulation ratio (NAR) were calculated using Eq.(3) and Eq.(4)^[11]:

$ \mathrm{SND}_{\text {efficiency }}=\frac{\mathrm{NH}_{4}^{+}-\mathrm{N}_{\text {(oxidized) }}-\mathrm{NO}_{x}^{-}-\mathrm{N}_{\text {(produced) }}}{\mathrm{NH}_{4}^{+}-\mathrm{N}_{\text {(oxidized) }}} \times 100 \% $

(3)

$ \mathrm{NAR}=\frac{\mathrm{C}_{\mathrm{NO}_{2}{ }^{-}-\mathrm{N}}}{\mathrm{C}_{\mathrm{NO}_{2}{ }^{-}-\mathrm{N}}+\mathrm{C}_{\mathrm{NO}_{3}{ }^{-}-\mathrm{N}}} \times 100 \% $

(4)

where NH₄⁺-N_(oxidized) is the amount of oxidized NH₄⁺- N, mg; NO_x^--N_(produced) is the amount of produced NO_x^--N, mg; $ \mathrm{C}_{\mathrm{NO}_{2}^{-}-\mathrm{N}}$ and $ \mathrm{C}_{\mathrm{NO}_{3}^{-}-\mathrm{N}}$ are the concentrations of NO_x^--N.

1.4 Analytical Methods

Liquid samples were filtered with a 0.45 μ m filter. The analyses of COD, PO₄^3--P, NH₄⁺-N, NO₂^--N, NO₃^--N, MLSS, SVI, and MLVSS were measured according to the standard methods ^[12]. Each water sample was tested three times. Microorganism morphology was observed with an OLYPUS BX51 optical microscope. Viscosity was tested using an NDJ-79 viscosimeter. EPS were extracted by heat extraction method^[13]. Polysaccharides were analyzed by the Dubois method^[14]. Protein was analyzed by the modified Lowry method^[15]. Distribution of particle size was tested using HIAC Royco 9703 particle size analyzer. Temperature, pH, DO, and ORP were recorded by WTW inoLab Oxi level 2. Correlation between different variables was analyzed by SPSS software (11.0).

2 Results and Discussion 2.1 Variations of Activated Sludge Characteristics

EPS, MLSS, and sludge settleability fluctuated during the experiment. In this experiment, EPS mainly consisted of polysaccharides and proteins. The ratio of EPS to MLSS can be used to measure the amount of EPS secreted per unit mass of activated sludge. It correlated well with SVI (r²=0.92), as shown in Fig. 2. In the first experimental phase, the average SVI and EPS/MLSS were 98.30 mL/g and 18.68 mg/g, respectively. In the third experimental phase, they changed to 345.36 mL/g and 27.76 mg/g, respectively. Meanwhile, the average viscosity and MLVSS/MLSS increased from 1.21 mPa/s and 0.82 to 1.51 mPa/s and 0.88, respectively. The morphology of activated sludge in each experimental phase was observed by microscope. Only little filamentous bacteria were present. Therefore, the viscosity sludge bulking was confirmed^[16]. Particle size distribution was tested in the 100^th and the 250^th cycle, and the results are shown in Fig. 3. Compared with normal sludge, the viscous bulking sludge had a larger particle size. The polysaccharides and proteins contained in EPS are all hydrophilic substances. Researches have demonstrated that EPS could contribute to aggregate sludge flocs, so the viscous bulking sludge with more EPS are prone to form larger particle size^[17].

2.2 Nutrients Removal Performance 2.2.1 Variations in nutrient concentrations

The concentration variations of nitrogen- containing compounds are displayed in Fig. 4. In the first phase, the average NO₃^--N and NH₄⁺-N were 9.05 and 0.91 mg/L, respectively, and no NO₂^--N was accumulated. In the second phase, NH₄⁺-N increased greatly with a decrease in NO₃^--N. In the third phase, the average NH₄⁺-N reduced to 0.87 mg/L with significant NO₂^--N accumulation. It should be noted that at the first several cycles of the third phase, NH₄⁺-N dropped down to lower than 2 mg/L immediately. In such a short time, ammonia-oxidizing bacteria (AOB) could not proliferate significantly. It seemed AOB was fallow in high COD loading state. Similar phenomenon was also reported by van den Akker^[18]. Theoretically, biological phosphorus removal is divided into two processes: anaerobic phosphorus release and aerobic phosphorus uptake. Compared with the PO₄^3--P concentrations at each experimental phase, it was found that the amount of released PO₄^3--P and uptaken PO₄^3--P both decreased during viscous bulking formation process, as shown in Fig. 5. Mass balance calculations for both N and P were carried out. The results showed that in first phase (normal sludge) 24.17% of TN and 10.20% of TP from the influent existed in the effluent, while in the third phase (viscous bulking sludge), 14.38% of TN and 17.50% of TP were residual.

2.2.2 Cyclic analysis

Cyclic analysis of pH, DO, ORP, and nutrients was conducted at the 123^th cycle and 381^th cycle (Fig. 6 and Fig. 7). In the anaerobic period, NO₂^--N, NO₃^--N, and ORP decreased, while pH did not drop until the denitrification was completed. The turning point was called "nitrate knee", which could exactly indicate the end of denitrification^[19]. Compared with normal sludge, nitrate knee of viscous bulking sludge occurred earlier. During the anaerobic period, PO₄^3--P increased from 1.06 and 3.45 mg/L to 21.04 and 14.04 mg/L, respectively.

During the aerobic period, because nitrification produced acid, pH decreased gradually in normal sludge system (Fig. 6). When NH₄⁺-N was depleted, pH switched from decreasing to increasing. The turning point on the pH profile was called "ammonia valley". After the ammonia valley, DO rapidly increased. By the end, NH₄⁺-N was basically nitrified to NO₃^--N with no accumulation of NO₂^--N, and PO₄^3--P had been uptaken to 0.24 mg/L. In contrast, in viscous bulking system, because the sludge adsorbed large amount of organics, pH gradually increased with the conversion of organics to CO₂ at the beginning of aerobic period, and the acidity decreased when CO₂ was blown off^[19]. As the nitrification process began at the 55^th min, NH₄⁺-N and pH decreased gradually just as those occurred in normal sludge. The difference was obvious accumulation of NO₂^--N occurred. At the end, PO₄^3--P had been uptaken to 0.25 mg/L.

2.2.3 Nutrients removal characteristics

The ideal COD/TN to sustain conventional biological nitrogen removal was considered to be 6.00-8.00^[20]. In this experiment, the COD/TN ratio was 7.67. In addition to denitrification, assimilation, phosphorus release, and respiration also utilized COD. In the first and third phases, the effluent NH₄⁺-N concentrations were limited. The NO_x^--N was mainly NO₃^--N in the first phase, while it changed to NO₂^--N in the third phase. The average nitrite accumulation ratios were 2.26% and 78.02%, respectively, as shown in Fig. 8. It has been reported that the sludge flocs with bigger particle size were prone to form low DO micro-environment inside ^[21], and the AOB is more adapted to low DO ^[19], so the growth advantage of AOB over nitrite oxidation bacteria (NOB) may be the cause of nitrite accumulation. In addition to NO₂^--N accumulation, obvious SND phenomenon occurred in viscous bulking state. The average efficiency of SND increased from 7.92% in the first phase to 27.31% in the third phase. The reason may be that compared with NO₃^--N, denitrifying NO₂^--N to N₂ can save 40% of carbon source^[22-23].

EPS acted as a determining factor in the transmission and accumulation of PO₄^3--P between phosphorus-accumulating bacteria (PAO) and bulk liquid^[24]. It was found that EPS/MLSS correlated well (r² = -0.90) with the mass of released PO₄-P^3-(m_p), as shown in Fig. 9. This indicated that EPS secreted in viscous bulking sludge could reduce the amount of released PO₄^3--P. The reason may be that the Ca²⁺ in EPS could crystallize with PO₄^3--P^[25]. Zhou et al.^[26] found that 15.70% of the phosphorus in biological phosphorus removal process was removed by EPS storage. In this experiment, the discrepancy between specific phosphorus uptake (V_mpu) and specific phosphorus release (V_mpr) showed obvious positive correlation with EPS/MLSS (r²=0.71), which confirmed that EPS played the role of container in phosphorus removal process.

2.3 Nitrification Kinetics

The oxidation rate of NH₄⁺-N (NO₂^--N) was measured through observing the change in NH₄⁺-N (NO₃^--N) over time. The measured values were fitted by Eq.(1) to determine the parameters of V_m and K_s^[8]. The parameter values of nitrification kinetics in different sludge states are shown in Table 2.

The R² of Monod equation in each state is close to 1.00, which means that Monod equation is basically followed. For normal sludge, the First-order equation can well describe nitrification kinetics. The reason was that under the dilution effect of feed period, the nitrogen concentration was in low level^[27]. In contrast, for viscous bulking sludge, R² of the First-order equation were only around 0.70. This might be because diffusion in viscous bulking flocs could also affect nitrification rate ^[28]. It was found the V_m of NH₄⁺-N between normal sludge and viscous bulking sludge were similar, which agreed well with the nitrification performance (Fig. 4). But the V_m of NO₂^--N dropped from 24.69 to 1.20 mg/(g·h), and the lower V_m of NO₂^--N correlated well with NO₂^--N accumulation in the third phase (Fig. 4).

2.4 Evaluation of Nutrients Removal Characteristics in Viscous Bulking

In general, the main determining factors of the above results might be floc particular size and EPS. For example, research by Han et al.^[21] showed at the internal center of flocs, DO concentration would decrease by 10%-55% with the floc size increasing 2.50 times. Same phenomenon was also observed in the transmission of NH₄⁺-N and NO_x^--N. Similarly, Li et al.^[29] showed that DO only partially penetrated from the flocs surface. Therefore, low DO concentration in the floc was the main cause of NO₂^--N accumulation in viscous bulking state. In this experiment, not only the viscous bulking sludge secreted more EPS, but also the average percentages of polysaccharides in EPS increased by 22.36% compared with normal sludge. Lin et al.^[30] found that struvite could be accumulated in EPS. In the synthetic wastewater, NH₄Cl and MgSO₄ were both added, so all the necessary components to synthesize struvite existed. It should be noted that COD load and C/N ratio are very important factors determining nutrients removal performance. In this experiment, during the first and third experimental phases, the two factors were similar. So their effects on nutrients removal could be negligible.

3 Conclusions

This research explored the characteristics of nutrients removal in viscous sludge bulking state. In terms of nitrogen removal, the maximum specific oxidation rate of NH₄⁺-N was similar to normal sludge, while obvious accumulation of NO₂^--N occurred during the aerobic period. The impact of bigger particle size and more EPS on transfer resistance was considered as the main cause of NO₂^--N accumulation. In terms of phosphorus removal, the storage effect of EPS was more obvious than normal sludge. Its mechanism needs to be further researched.