Coking wastewater is generated in the production of coke, coal gas, tar and other coke by-products. The coking wastewater includes inorganic pollutants such as ammonia, cyanogen, sulfocyan, heterocycle compounds and polycyclic aromatic compounds such as phenol, oils, naphthalene, pyridine, quinoline and anthracites, which are difficult to biodegradation under aerobic conditions (Ganczarczyk, 1972). Discharge of coking wastewater to the environment may cause severe contamination to it and also threaten the normal life of human beings, so it is necessary to treat the coking wastewater in order to reduce its harm to the environment.

The conventional biological treatment of coking wastewater is not efficient enough in providing the required quality criteria (Sotton et al., 1999; Yu et al., 1997).

Considerable success in wastewater treatment was achieved by addition of special microorganisms and immobilized microorganisms in laboratory-scale treatment of coking wastewater (Huang et al., 1995; Wang et al., 2002). However, the maintenance of the dominant species’ bioactivity and regular supplementation of immobilized microorganisms in practical application is a problem, and their feasibility need to be confirmed.

Pretreatment was often used to make coking wastewater more amenable to biological treatment (Ganczarczyk, 1980) and the common pretreatment approach included equalization and storage, ammonia stripping, chlorination and air flotation. And recently, for treatment of hard to biodegrade wastewater, combinations of anaerobic and aerobic treatments were effective and less expensive, in which A/O (anaerobic-aerobic) and A²/O (anaerobic-anoxic-aerobic) were major methods. In these systems, the anaerobic processes serve as pretreatment for partial degradation of some refractory and inhibitory organic compounds to lighten the load of subsequent anoxic and aerobic processes. So far, A/O and A²/O are getting popular for treatment of coking wastewater (Zhang et al., 1998; Wen et al., 1991; Li et al., 2001; 2003). Studies showed that the anaerobic treatment could greatly improve the biodegradability of the coking wastewater, and created good condition for subsequent aerobic treatment.

ASBR is a high rate anaerobic process (US Pat. No. 5 185 079) developed by Dague and co-workers at Lowa State University (Dague et al., 1992; Sung and Dague, 1992). The promising feature of the ASBR process is that granular biomass can be achieved, and in this way higher biomass can be maintained in the reactor with efficient biomass setting and a long solids retention time (SRT) (Dague et al., 1966; 1970). There are many reports on studies of ASBR (Herum and Dague, 1993; Wirtz and Dague, 1996; Sung and Dague, 1995) but few of them were on the treatment of actual wastewater (Timur and Özturk, 1997; 1999).

The main purpose of this paper was to study the pretreatment of coking wastewater using ASBR, and the optimization of the ASBR process operating parameters such as the ratio of filling time to reacting time (t _f/t _r), the mixing intensity (MI) and the intermittent mixing mode (IMM) by orthogonal test. For elaborating the experiments plan, we use the method of orthogonal test for three factors at three levels in this research.

Table 1 Characteristics of the coking wastewater

Fig. 1 Schematic of ASBR

The ASBR operating principles were relatively simple as described in a previous publication (Sung and Dague, 1992). The reactor sequence included four steps: filling, reacting, settling and decanting. During the filling and reacting steps, the reactor was intermittently mixed through biogas circulation. During the settling step, mixing was shut off to allow biomass solids separation. The decanting step was implemented after sufficient solids separation had occurred. The decant volume was equal to the volume fed during the filling step.

In the initial test, 5 L anaerobic granular biomass was seeded; 7 L sucrose solution at concentration of 7 200 mg/L and appropriate quantity of N and P nutrients with the COD: N: P ratio of 300:7:1 introduced into the ASBR reactor, which resulted in biomass concentration of 21.5 g mlvss/L. The system was operated for 20 d under the following conditions: pH of 7.0~8.0, fill influent of 5 L sucrose solution at concentration of 7 200 mg/L and cycling period of 24 h (4 h filling period, 19 h reacting period, 0.5 settling period, 0.5 decant period).

BMP was determined at 35 °C. Seeded culture of 100 ml acclimated anaerobic granular biomass and 350 ml coking wastewater (with the COD of 800 mg/L) were introduced into a 500 ml serum bottle anaerobically purged with carbon dioxide and nitrogen gases. Some background inorganic nutrients were introduced into airproofed (with rubber cap) serum bottles (Speece and McCarty, 1964) (Table 2). To eliminate error introduced by the gasses generated by the self-decomposition of anaerobic biomass, control test was needed to correct the error. The control test was prepared according to the same procedure prescribed above, but it was done without any addition of coking wastewater and acclimated anaerobic granular biomass.

Table 2 Background inorganic nutrients added in BMP test

The gas produced in the anaerobic reactions includes CO₂ and CH₄. Because CO₂ production does not represent COD reduction under anaerobic conditions, it is necessary to eliminate CO₂. In this study, 0.1 mol/L sodium hydroxide solution was used as a CO₂ absorbent, so the gas produced in the anaerobic reactions contains CH₄ only. Gas production from all bottles was measured periodically at incubation temperature (35 °C) by the liquid displacement method.

The typical operation of ASBR with fast filling time (Dague et al., 1992; Sung and Dague, 1992) leads to a low ratio of fill time to reaction time (t _f/t _r). But this operating strategy may cause the acid formation problem observed by Suthaker et al.(1991). Considering the toxicity of the coking wastewater and the accumulation of acid resulting from the fast filling, the filling time was extended properly and three levels of the t _f/t _r was selected as 0.3, 0.5 and 1.0 in this study. As to the mixing intensity, sufficient intensity should be selected to ensure uniform conditions throughout the reactor and effective contact of the wastewater with the granulated biomass to improve the mass transfer flux. While excessive intensity of mixing may shear the anaerobic granular biomass and result in poor solids separation. Usually the mixing was supplied by biogas recirculation or mechanical agitation (when biogas cannot insufficiently provide suitable mixing) (Sung and Dague, 1995; Zaiat et al., 2001; Rodrigues et al., 2003). It was reported that intermittent mixing rather than continuous mixing resulted in superior performance (Dague et al., 1970). Therefore, intermittent mixing supplied by biogas was applied in this study.

In the experiments plan we use the method of orthogonal test for three factors at three levels. By levels we mean the values taken by the factors. Table 3 lists the factors to be studied and the assignment of the corresponding levels. The orthogonal array L₉ (3⁴) shown in Table 4 was chosen with 9 rows corresponding to the number of tests and 4 columns at three levels.

Table 3 Assignment of the levels to the factorsa

Table 4 Orthogonal array L9 (34) in this testa

Size classification was achieved by wet sieving in perspex cylinders (Ø 65 mm×100 mm) equipped with a sieve plate. After filling the sieve column with mineral medium and placing the sample on top, it was flushed with nitrogen gas. During gentle intermittent swirling the granules settle throng the sieve plates according to their size.

Fig. 2 BMP determination gas production curve

BMP was referenced either to sample volume (m³ CH₄/m³ sample), sample mass (m³ CH₄/kg sample) or sample organic content (m³ CH₄/kg COD) (Oven et al., 1979). The latter method permits direct transfer of results into percent organic matter converted to methane. Since 395 ml methane at 35 °C is equivalent to 1 g of COD removed from the wastewater during the anaerobic process, there is a stoichiometric relationship which allows calculation of the COD reduction in the liquid phase. In the sample test of the BMP assay, 350 ml coking wastewater was assayed with 100 ml of anaerobic granular biomass inoculums. After 130 d of incubation, 248.6 ml CH₄ was produced in all. As to the control test, there were 100 ml anaerobic granular biomass inoculums without the addition of coking wastewater. After only 58 d of incubation, the gas production stopped, and 202.3 ml CH₄ was produced in all. Therefore, the net CH₄ production was 46.3 ml, and the BMP result was 0.165 m³ CH₄/kg COD.

In theory, 350 ml coking wastewater with COD of 800 mg/L used in the BMP assay should produce 110.6 ml CH₄ at 35 °C under anaerobic conditions based on the stoichiometric relationship mentioned above. Therefore, the percent coking wastewater converted to methane was 41.9%.

The BMP gas production curve in Fig.2 showed that in the first 58 d, the gas production rate of the control test was higher than that in the coking wastewater sample test. This showed that the coking wastewater inhibiting effect on anaerobic biomass inoculums, but that the anaerobic microorganisms adapted to coking wastewater gradually after long time acclimation. Acclimation of 225 d was conducted before the BMP assay, which showed that the gas production rate was still low. The gas production lasted for 130 d, which was far more than the 30~60 d suggested for BMP assay. From the result of the BMP assay it could be concluded that the biodegradability of the coking wastewater was rather low. Therefore, long time acclimation of the inoculated anaerobic granular biomass to the coking wastewater is essential for satisfactory treatment.

Table 5 The orthogonal array for COD removal efficiency

Table 5 can be used to analyze the optimal level for every factor. For the factor t _f/t _r, K ₁₂>K ₁₃>K ₁₁. It shows that level 2 (t _f/t _r was 0.5) was the optimal level for the factor of t _f/t _r. For the factor MI, K ₂₂>K ₂₃>K ₂₁. It indicates that level 2 (mixing intensity was 0.025 L biogas/(L reactor volume·min)) was the optimal level for the factor of mixing intensity. For the factor IMM, K ₃₃>K ₃₂>K ₃₁. It shows that level 3 (intermittent mixing mode was 100 s/45 min) was the optimal level for the factor intermittent mixing mode.

In Table 5, S_j is the quadratic sum of dispersion factor j. From S_j we can determine the importance of different factor. Therefore, the relation S ₁>S ₃>S ₂ indicates that the factor t _f/t _r is the most important parameter affecting treatment efficiency.

In addition, R_j the range of K_{j l}, and the value of R_j reflect the importance of the factor j. Here R ₁>R ₃>R ₂, which indicates that the factor of t _f/t _r was the most important parameter affecting treatment efficiency. The same result of analysis was obtained by using S_j (the quadratic sum of dispersion), so the result from this orthogonal test was credible. As for the intermittent mixing mode, the best value of 100 s/45 min was the biggest of three values (30 s/20 min, 30 s/30 min and 100 s/45 min). So higher IMM values were used for conducting further study which revealing poor effect of sedimentation on biomass in the reactor. Therefore, t _f/t _r of 0.5, mixing intensity of 0.025 L biogas/(L reactor volume·min) and intermittent mixing mode of 100 s/45 min, were the optimal conditions for the orthogonal test.

A stable operation of three months was carried out under the optimal conditions stated above. In ASBR, when the influent coking wastewater COD concentration was 796~1304 mg/L, influent ammonia nitrogen concentration was 230~668 mg/L, and influent pH was 7.2~8.6, COD removal rate can steadily reach 38%~50%, which agrees with the BMP determination result. Furthermore, the BOD₅ and COD of the effluent and influent of the coking wastewater to the ASBR were measured, and the sample measured for BOD₅ and COD contained no SS. The ratio of BOD₅/COD, COD removal efficiency and biomass in the stable operation are given in Table 6. The results indicated that the average ratio of BOD₅/COD to the influent was 0.27, while the average ratio of BOD₅/COD to the effluent was up to 0.58, which showed that the biodegradability of the coking wastewater increased greatly after the pretreatment using ASBR. During the stable operation, there was higher biomass to be maintained in the reactor and solids retention time (SRT) was more than 150 d, so there was very little biomass overflowing out of the reactor (Table 6).

Table 6 The result in the stable ASBR operation

During the stable ASBR operation, there was a steady-going run and the COD removal efficiency was as high as 38%~50% even with the fluctuating influent of coking wastewater. According to Table 6, the BOD₅/COD ratio increased greatly. All these results were due to the higher granular biomass maintained in the ASBR, and the granular biomass had good sedimentation property which resulted in the long SRT inductive to treatment efficiency. The different bacteria on the granular biomass are spaced farther a part than the bacteria of the suspended biomass, and can slightly metabolize the intermediate products of the reaction.

The predominant bacteria attached to the seeded granular biomass were Methanosarcina both on the surface of the particle and in the middle of the sludge (Tian, 2001). After ASBR anaerobic pretreatment of coking wastewater, the microorganisms on the granular biomass in the ASBR was observed using SEM (scanning electron microscope) and fluoroscope. The surface of the granular biomass in the ASBR was densely occupied by many small cavities (Fig.3a), which were the passages formed due to the release of gas generated by internal bacteria and to inlet’s substrate entering into the biomass. The observed result showed that the main body of granular biomass was Methanosaeta in the middle part of the granular sludge (Fig.3b). A small mass of Methanosarcina was found in the outer hole of the granular biomass (Fig.3c). The top parts of Fig.3b and Fig.3c are amplification of the bottom parts of the pictures. Methanosaeta and Methanosarcina have different metabolization characteristics (μ _max and k _s), and Methanosaeta has lower substrate affinity than Methanosarcina as mentioned elsewhere (Sekiguchi et al., 1999; Wu et al., 1996). Many substances in coking wastewater in the influent of the ASBR resist biodegradation and the concentration of influent was low, which were favorable for Methanosaeta instead of Methanosarcina to be the dominant bacteria in the ASBR as observed in this study.

Fig. 3 SEM images of granular biomass in ASBR (a) Surface of the granular biomass; (b) Methanosaeta in the middle part of the granular sludge, the up part is magnification (×6) of the lower part; (c) Methanosarcina in the outer hole of the granular biomass, (more ...)

The following conclusions were derived from the results of the tests described above:

In this study, the acclimation lasted for 225 d. The long time acclimation of the inoculated anaerobic granular biomass was essential for inducing the microorganisms to adapt to the coking wastewater without disrupting granular biomass activity.

The BMP assay indicated that 0.165 m³ CH₄/kg COD had been converted into methane in coking wastewater under anaerobic condition. The result of the BMP assay showed that the maximum mechanization of the coking wastewater under anaerobic condition was 41.9%.

Analysis of the orthogonal test results showed that when the coking wastewater was pretreated using ASBR, the optimal t _f/t _r, the mixing intensity and the intermittent mixing mode were 0.5, 0.025 L biogas/(L reactor volume·min) and 100 s/45 min respectively. The cycle of ASBR operation was 24 h, of which the filling time was 7.17 h, the reacting time was 14.33 h, the settling time was 2.0 h and the decant time was 0.5 h. When the OLR ranged from 0.37 to 0.54 g COD/(L·d), the removal efficiency of COD reached 38~50%. The SRT was more than 150 d.

There was a great increase of the biodegradability of coking wastewater after its pretreatment with ASBR, and the ratio of BOD₅/COD also increased from 0.27 in influent to 0.58 in effluent.

At the end of treatment of coking wastewater using ASBR, the predominant bacteria on the granular biomass was Methanosaeta instead of Methanosarcina that dominated in the seeded granular biomass.