UASB reactor for domestic wastewater treatment at low temperatures: a comparison between a classical UASB and hybrid UASB-lter

reactor
B. Lew, S. Tarre, M. Belavski and M. Green
Faculty of Civil and Environmental Engineering, Technion 32000- Haifa, Israel (E-mail: benilew@tx.technion.ac.il; shelly@tx.technion.ac.il; michaelb@tx.technion.ac.il; agmgreen@tx.technion.ac.il) Abstract The performance of an upow anaerobic sludge blanket (UASB) reactor and a hybrid UASB-lter reactor was investigated and compared for the treatment of domestic wastewater at different operational temperatures (28, 20, 14 and 10C) and loading rates. For each temperature studied a constant CODt removal was observed as long as the upow velocity was lower than 0.35 m/h. At these upow velocities similar removals were observed for both reactor types at 28 and 20C, 82 and 72% respectively. However, at 14 and 10C the UASB reactor showed a better COD removal (70% and 48%, respectively) than the hybrid reactor (60% and 38%). COD removal resulted from biological degradation and solids accumulation in the reactors. At 28C, a constant 200 g sludge mass was observed in both reactors and COD removal was attributed to biological degradation only. At lower temperatures, solids accumulation was observed in addition to biological degradation with an increase in reactor sludge as the temperature decreased. The decrease in biological degradation at lower temperatures was offset by solids accumulation and explains the similar overall COD removal efciency observed at 28C, 20C and 14C. The decrease in temperature was also followed by an increase in the efuent TSS concentration in both reactors. At 14 and 10C a lower efuent TSS concentration and better performance was observed in the UASB reactor. Keywords Anaerobic treatment; domestic wastewater; hybrid UASB-lter; temperature effect; UASB

Water Science and Technology Vol 49 No 1112 pp 295301 IWA Publishing 2004

Introduction

Anaerobic processes are an attractive technology for wastewater treatment. The high costs of aeration and sludge handling associated with aerobic sewage treatment are dramatically lower as no oxygen is needed and the production of sludge is 320 times lower. However, its use is usually limited to high strength industrial wastewater with soluble substrates. Domestic wastewater has typically low concentrations of COD, resulting in relatively small methane production that is insufficient to heat the reactor to more favorable mesophilic temperatures. Moreover, the relatively high concentration of particulate matter present in domestic wastewater requires an initial hydrolysis step, which is significantly affected by temperature and is usually the rate-limiting step in sub-tropical climate regions. In tropical countries UASB reactors for domestic wastewater have found wide acceptance. There are several full-scale plants already in operation in Colombia, Brazil, Indonesia, India and Egypt and COD removals above 70% have been observed by several authors (Souza and Foresti, 1996; Chernicharo and Cardoso, 1999; Kalogo and Verstraete, 2000). The effluent quality at these installations is reported to be 140 mg COD/l, 75 mg BOD/l and 30 mg TSS/l. At low temperatures, the low hydrolysis rate and a decrease in the degradable organic matter fraction were found to cause the deterioration of the overall anaerobic reactor performance (Elmitwalli et al., 2001). UASB COD removals of ~65% at 20C and of 5565% at 1317C were observed by several authors (Lettinga et al., 1981; Grin et al., 1983; Vieira

295

and Souza, 1986; Elmitwalli et al., 1999; Seghezzo et al., 2000). A decrease in the effluent quality was also observed, together with a decline in the gas production rate. Agrawal et al. (1997) observed a 78% decrease in the gas production rate when the temperature was reduced from 27C to 10C. The low gas production coincided with a 25% lower COD removal at 10C than at 27C, indicating suspended solids accumulation in the reactor. One possible way to improve the performance of a UASB reactor at low temperatures is to provide surface area for biomass attachment and growth in the reactor volume above the sludge blanket (Tilche and Vieira, 1991). This can be accomplished by replacing the typical gas/solids separator of the classical UASB reactor with filter media. Elmitwalli et al. (1999) compared the performances of a hybrid UASB-filter and a classical UASB reactor at 13C. The hybrid UASB-filter reactor reached 64% COD removal, a 4% better removal than the classical UASB. A better colloidal fraction removal was attributed to the attached biomass on the filter. This project concentrates on the effect of temperature on the performance of a UASB and a hybrid UASB-filter reactor for domestic wastewater treatment, at different upflow velocities, hydraulic retention times (HRT) and organic loading rates.
Material and methods

296

B. Lew et al.

Two reactors were constructed of plexiglass with a working volume of 5.3 litres (8.0 cm diameter, 107.5 cm high), with four sampling ports placed at different heights and an inflow manifold at the bottom of the reactors to ensure even influent distribution. The hybrid UASB-filter reactor had 4.0 cm diameter plastic filter rings (100 m2/m3 specific surface area) filling the top half of the reactor instead of the standard UASB gas/solid separator to prevent biomass washout (Figure 1). The reactors were filled with 2 litres of granular sludge taken from a full scale UASB reactor treating food wastewater. The TSS concentration in the reactors immediately after filling was 130 g TSS/l. The reactors were fed with domestic wastewater after primary sedimentation from the Neve Shaanan neighborhood, Haifa. The wastewater can be classified as a medium strength domestic wastewater (Metcalf and Eddy, 1991). In both reactors effluent samples and methane production readings were taken one hydraulic retention time after inflow sampling. All samples were tested on a regular basis for pH, BOD, TSS, VSS, CODt, CODs and VFA. Every 20 days 50 ml sludge samples were taken from the two lower sample ports and were tested for SVI, TSS and VSS. All analyses were performed according to Standard Methods for the Examination of Water and Wastewater (APHA, 1995). Effluent COD was characterized in two ways: well mixed samples were used for CODt and samples allowed to settle for 10 minutes were used for CODs. COD removal was calculated on the basis of influent CODt. The volatile fatty acids concentration was measured using the five-point titration method (Moosbrugger et al., 1993). The methane gas produced was collected and measured as described by Haandel and Lettinga (1994). The dissolved methane in the effluent was calculated according to Henrys law and added to the methane gas actually measured. During start-up, the reactors were operated at 28C with a retention time of 24 hours to allow sludge adaptation to domestic wastewater. Afterwards, the retention time was gradually shortened with the corresponding increase in organic load. At every new retention time, the reactor was allowed to adjust to the higher load. After the performance reached steady state for a number of days, the retention time was shortened further. When the experimental regime was completed at 28C, the same procedure was carried out for reactor temperatures of 20C, 14C and 10C. The reactor was operated at a temperature of 28C for 6 months and 2 months for each of the other temperatures.

Methane Methane

Effluent

Effluent

B. Lew et al.

Influent

Influent

(a)

(b)

Figure 1 Laboratory scale (5.3 litres); (a) UASB reactor equipped with a gas/solid separator; (b) hybrid UASB-Filter reactor equipped with plastic lter rings

Results and discussion

During the start-up period, both reactor types operated efficiently and the effluent TSS was constant with an 86.6% removal. The CODt removal increased with time, starting at 79.2% and reaching 88.5% at the end of the period. The effluent pH was very similar to the influent pH indicating that VFA were not accumulating in the reactors. At the conclusion of the start-up period, the reactors were operated at 28C for six months and the hydraulic retention time (HRT) was gradually reduced from 24 hours to 1 hour only, with the corresponding increase in organic load from 0.2 to 31.8 g COD/l/day. For HRTs from 24 to 3 hours (0.040.35 m/h upflow velocity), a stable effluent COD concentration was observed in both reactors, even with the high fluctuation in influent CODt concentration (2001,300 mg COD/l). The average total and settleable effluent COD were 102 (43) and 81 (26) mg/l respectively, corresponding to a CODt removal of about 82% and a CODs removal of about 83.5%. The effluent COD concentrations observed even at long HRT (24 hours) are probably of a non-biodegradable COD nature. Similar effluent COD concentrations and COD removal were observed by Vieira and Garcia (1992) (83 mg COD/l/day and 65%), Souza and Foresti (1996) (58 mg COD/l/day and 86%), El-Gohary and Nasr (1999) (145 mg COD/l/day and 77%) and Kalogo and Verstraete (2000) (97.4 mg COD/l/day and 71%) for the same temperature. For HRTs shorter than 3 hours (corresponding to upflow velocities and influent organic load higher than 0.35 m/h and 5.0 g COD/l/day, respectively) the effluent quality deteriorated for both reactor types. While at longer HRTs (from 3 to 24 hours) the COD removal was always constant (around 82%), it decreased to 54.3% for CODt and 73.8% for CODs at 2 hours HRT and 37.9% for CODt and 59.2% for CODs at 1 hour HRT for both reactors. The decrease in CODs removal rate can be explained by too short HRT to complete COD degradation. Results of effluent VFA concentration support this observation: at longer HRT the effluent VFA concentration was close to zero, however, at short HRTs (shorter than 3 hours) the effluent VFA concentration increased when VFA was present in the influent. A maximum COD removal of between 6.0 and 8.8 g COD/l/day was observed at inuent loadings between 16.4 and 23.8 g COD/l/day. At the highest organic load of 31.8 g COD/l/day, a decrease to 5.4 g COD removed per litre per day was observed in both reactors. The decrease in CODt removal can be explained by the sharp increase in efuent TSS from about 20 mg/l, at HRTs longer than 3 hours, to about 100 mg/l and 220 mg/l, at 2 and 1 hour HRT respectively (Figure 2). It appears that the higher upow velocity caused inuent suspended solids washout, as opposed to a gradual increase that would be expected if the retention time was limiting. Batch tests also suggested that the efuent TSS originated from the

297

inuent because of the absence of gas production from the efuent particulate material. When the reactors were submitted to two different inuent TSS concentrations (150 mg TSS/l and 300 mg TSS/l) at different HRTs, the same upow velocity limitation (0.35 m/h) was observed. In both cases, both reactors demonstrated a linear relationship between TSS removal (82%) and TSS loads for HRTs longer than 3 hours, even though the inuent TSS load was much higher for the second case. At shorter HRTs, the TSS removal began to decrease in both cases, leading to the conclusion that the high upow velocity was the limiting rate for both CODt and TSS removal, and not the retention time. The critical upow velocity where CODt removal decreased is similar to the range reported by other authors, 0.33 to 0.67 m/h (Collivignarelli et al., 1991; Lettinga and Hulshoff Pol, 1991; Maaskant et al., 1991; Vieira and Garcia, 1992; Chernicharo and Cardoso, 1999; Kalogo and Verstraete, 2000). The same upflow velocity limitation was observed at all temperatures studied (28, 20, 14 and 10C), with an almost constant CODt removal for lower upflow velocities in both reactors. For each temperature studied the CODt removal at HRT longer than 3 hours decreased with the decrease in the temperature (Table 1). Usually, decreasing the temperature negatively affects the kinetics and therefore increasing the HRT results in removal efficiency improvement. However, the HRT did not improve COD degradation at upflow velocities lower than 0.35 m/h. These results suggest a much lower biodegradability at lower temperatures of a range of compounds comprising the heterogeneous composition of domestic wastewater. In addition, it is possible that lower temperatures induced shifts in the bacterial population with predominantly slower kinetics. The COD removal rates observed at 20C, 14C and 10C are in accordance with results observed by others (Lettinga et al., 1981; Grin et al., 1983; Vieira and Souza, 1986; Elmitwalli et al., 1999; Seghezzo et al., 2000). Although COD removal was very similar for both reactors at 28 and 20C, the results at 14 and 10C showed a much better COD removal for the classical UASB reactor. These results cannot be explained by bacteria activity, since methane production rate was very similar for both reactors at all temperatures studied.
Upflow velocity (m/h) Effluent TSS Concentration (mg/l)
0.05 600 500 400 300 200 100 0 24 21 18 15 12 9 6 3 0 0.10 0.15 0.35 1.05

298

B. Lew et al.

UASB Hybrid

Hydraulic Retention Time (h)

Figure 2 Efuent TSS concentration as a function of HRT and Upow Velocity at 28C in the UASB and hybrid UASB-Filter reactors Table 1 Maximum COD removal (%) at different temperatures
28C 20C 14C 10C

Classical UASB Hybrid UASB + lter

82 82

72 72

70 60

48 38

The methane production rate showed a similar pattern at 28C, 20C and 14C. At these temperatures the methane production increased with increasing influent COD organic load, up to 10.0 g COD/l/day for 28C and 5.0 g COD/l/day for 20C and 14C. At higher loads a constant methane production of 1,100, 225 and 85 ml CH4/l/day was observed for 28C, 20C and 14C, respectively. At 10C a very low methane production was observed for all the inflow organic loads, 25 ml CH4/l/day. The calculated temperature activity coefficient based on the maximum methane production was 1.220 (R2 = 0.95), which was in the upper range of reported values for anaerobic processes (Gujer and Zehnder, 1983; Haandel and Lettinga, 1994). Moreover, the experimental temperature activity coefficient is higher than the one observed for methanogenic bacteria in batch experiments with granular sludge, 1.066 (R2 = 0.92). Based on these results, it can be concluded that the hydrolytic bacteria were more affected by temperature changes than the methanogenic bacteria. As seen, the maximum methane production rate showed a decrease with temperature, as with the CODt removal. However, the decrease in the maximum methane production was much more pronounced than the decrease in the CODt removal. This phenomenon can be explained by suspended solids entrapment (accumulation) in the reactor, which resulted in better COD removal, but did not improve methane production. The phenomenon of suspended solids accumulation in the reactors was not apparent at 28C. The sludge mass in both reactors during summer conditions (28C) had a constant value of 200 g and a constant SVI value of 10 ml/g. In contrast, at lower temperatures solids of different characteristics accumulated on top of the typically black granular sludge. Solids with a brown floc-like appearance, with no gas production in a batch test conducted in the laboratory and with a higher SVI (38.8 ml/g) value. Seghezzo et al. (2000) studying a UASB reactor at 15C also observed lower methanogenic activity in the upper part of the sludge blanket. As the temperature decreased and the rate of hydrolysis decreased, the daily increase in the reactor sludge was greater. The increase in the accumulation of particulate matter in the reactors roughly equaled the decrease in the biological degradation rate with temperature and explains the similar overall COD removal efficiencies obtained at 28C, 20C and 14C for each reactor, as observed in Table 1. Due to the higher SVI value, the flocculent type sludge had a greater sensitivity to upflow velocities than the granular sludge, explaining the accumulation of the particulate material on top of the typical bacterial granules and the increase in the effluent TSS concentration at upflow velocities higher than 0.35 m/h (Figure 2). Moreover, the sensitivity to upflow velocities together with the increase in the TSS accumulation in the reactor with the decrease in the temperature led to an increase in the effluent TSS concentration at longer HRTs with the decrease in temperature for both reactors (Table 2). Indeed, a much higher effluent TSS concentration was observed in the hybrid UASB-filter reactor at 14 and 10C than in the classical UASB. This higher effluent TSS concentration can explain the better performance of the classical UASB at lower temperatures (14 and 10C) observed in Table 1.

B. Lew et al.

Table 2 Efuent TSS concentration (mg/l) at upow velocities lower than 0.35 m/h, at different temperatures
28C 20C 14C 10C

Classical UASB Hybrid UASB + lter

17 17

25 25

60 80

120 150

299

During the transition from winter to summer conditions (10C to 28C) the methane production initially increased and then gradually declined over time, reaching values similar to the previous summer (28C) for both reactors. In contrast, CODt removal at the beginning of the new warmer season was very low (59%) in comparison to the previous summer. The low CODt removal suggests that the high gas production observed was due to the degradation of the accumulated suspended matter and not the influent organic matter. An improvement in CODt removal was observed with time, reaching again 82% after 40 days. The improvement in CODt removal together with the reduction in the gas production indicates that the accumulated matter was exhausted and influent degradation reached the values observed in the first summer season. No biofilm growth was observed on the filters in the hybrid reactor at the end of the experiments. The filters were removed from the reactor and some suspended solids accumulation was observed on it, however, these solids formed a very fragile layer, which was easily removed.
Conclusions

300

B. Lew et al.

Both reactor designs gave similar performance. At summer conditions (2028C) COD removal rates above 72% can be obtained in both reactors. At lower temperatures (1410C) when the bacterial activity is lower, solids accumulation in the reactor is more pronounced with better solids retention in the classical UASB. In both reactors, the accumulated sludge from the winter is subsequently digested in the following summer, which is evidenced by a large gas production at the beginning of the new warm season. Based on the results of this research project it can be concluded that the classical UASB reactor and the hybrid UASB-filter reactor can be a good alternative for domestic wastewater treatment even in temperate climates. The hybrid UASB reactor containing filter rings showed no advantage and at lower temperatures performed slightly worse than the conventional UASB reactor.
References
Agrawal, L.K., Ohashi, Y., Mochida, E., Okui, H., Ueki, Y., Harada, H. and Ohashi, A. (1997). Treatment of raw sewage in a temperate climate using a UASB reactor and the Hanging Sponge Cubes process. Wat. Sci. Tech., 36(67), 433440. APHA (1995). Standard Methods for the Examination of Water and Wastewater. 19th edn, APHA, AWWA, WPCF, Washington DC, USA. Chernicharo, C.A.L. and Cardoso, M.R. (1999). Development and evaluation of a partitioned upflow anaerobic sludge blanket (UASB) reactor for the treatment of domestic sewage from small villages. Wat. Sci. Tech,, 40(8), 107113. Collivignarelli, C., Urbini, G., Farneti, A., Bassetti, A. and Barbaresi, U. (1991). Economic removal of organic and nutrient substances from municipal wastewaters with full-scale UASB fluidized and fixed bed reactors. Wat. Sci. Tech., 24(7), 9095. El-Gohary, F.A. and Nasr, F.A. (1999). Cost-effective pre-treatment of wastewater. Wat. Sci. Tech., 39(5), 97103. Elmitwalli, T.A., Zandvoort, M.H., Zeeman, G., Bruning, H. and Lettinga, G. (1999). Low temperature treatment of domestic sewage in upflow anaerobic sludge blanket and anaerobic hybrid reactors. Wat. Sci. Tech., 39(5), 177185. Elmitwalli, T.A., Zeeman, G. and Lettinga, G. (2001). Anaerobic treatment of domestic sewage at low temperature. Wat. Sci. Tech., 44(4), 3340. Grin, P.C., Reresma, R. and Lettinga, G. (1983). Anaerobic treatment of raw sewage at lower termperatures. In Proc. European Symposium on Anaerobic Wastewater Treatment, Noordwijkerhout, The Netherlands, 335347. Gujer, W. and Zehnder, A.J.B. (1983). Conversion processes in anaerobic digestion. Wat. Sci. Tech., 15(89), 127167.

Haandel, A.C. van and Lettinga, G. (1994). Anaerobic sewage treatment: A practical guide for regions with a hot climate. J. Wiley and Sons Inc. Kalogo, Y. and Verstraete, W. (2000). Technical feasibility of the treatment of domestic wastewater by a CEPS-UASB system. Env. Tech., 21, 5565. Lettinga, G., Roersma, R., Grin, P., de Zeeuw, W., Hulshof Pol, L., van Velsen, L., Hovma, S. and Zeeman, G. (1981). Anaerobic treatment of sewage and low strength wastewater. Proceedings of the Second International Symposium on Anaerobic Digestion, Travemunde, Germany, 271291. Lettinga, G. and Hulshoff, Pol, L.W. (1991). UASB process design for various types of wastewaters. Wat. Sci. Tech., 24(8), 87107. Maaskant, W., Magalhaes, C., Maas, J. and Onstwedder, H. (1991). The upflow anaerobic sludge blanket (UASB) process for the treatment of sewage. Env. Pollution, 1, 647653. Metcalf and Eddy (1991). Wastewater Engineering: Treatment Disposal and Reuse. Metcalf and Eddy, Inc. 3rd Edition, McGraw-Hill. Moosbrugger, R.E., Wentzel, M.C., Ekama, G.A. and Marais, G.V.R. (1993). Alkalinity measurement: part 1 a 4 pH point titration method to determine the carbonate weak acid/base in an aqueous carbonate solution. Water SA, 19(1), 1122. Seghezzo, L., Castaneda, M.L., Trupiano, A.P., Gonzalez, S.M., Guerra, R.G., Torrea, A., Cuevas, C.M., Zeeman, G. and Lettinga, G. (2000). Anaerobic treatment of pre-settled sewage in UASB reactors in subtropical regions (Salta, Argentina). Preprints, VI Latin-American Workshop and Seminar on Anaerobic Digestion, Recife, Brazil. pp. 713. Souza, J.T. and Foresti, E. (1996). Domestic sewage treatment in an upflow anaerobic sludge blanketsequencing batch reactor system. Wat. Sci. Tech., 33(3), 7384. Tilche, A. and Vieira, S.M. (1991). Domestic report on reactor design of anaerobic filters and sludge bed reactors. Wat. Sci. Tech., 24(8), 193206. Vieira, S.M.M. and Souza, M.E. (1986). Development of technology for the use of the UASB reactor in domestic sewage treatment. Wat. Sci. Tech., 18(12), 109121. Vieira, S.M.M. and Garcia Jr., A.D. (1992). Sewage treatment by UASB reactor. Operation results and recommendations for design and utilization. Wat. Sci. Tech., 25(7), 143157.

301

B. Lew et al.