Treatment of Sago Wastewater Using Hybrid Anaerobic Reactor

J. Rajesh Banu,1 Sudalyandi Kaliappan1* and Dieter Beck2

1Centre for Environmental Studies (CES), Anna University, Chennai – 600025, Tamilnadu State, India 2Indo-German Project, Centre for Environmental Studies, Freiheits Str. 195, 42853 Remscheid, Germany

Sago, tapioca starch, is manufactured by over 800 small-scale units located in the Salem district of the State of Tamilnadu, South India. These units generate large quantities of high-strength wastewater requiring elaborate treatment prior to disposal. The present study is an attempt to treat the sago wastewater using a hybrid reactor, which combines the advantages of both fixed-film and up-flow anaerobic sludge blanket systems. A hybrid reactor with a volume of 5.9 L was operated at organic loading rates varying from 10.4 to 24.6 kg COD/m3d. After 120 d of start-up, an appreciable decrease in COD and efficient removal of solids were evident. The COD removal varied from 91 to 83%. While the removal of total solids was in the range of 56 to 63%, that of volatile solids varied from 67 to 72%. The methane production during the study period was

in the range of 0.11 to 0.14 L CH4/g COD-d and the percentage was from 55 to 67%. The ideal organic loading rate (OLR) was determined on the basis of tolerance of the reactor towards higher organic loading rate and it was found to be 23.4 kg COD/m3d. The findings of the study open new possibilities for the design of low-cost and compact on-site treatment systems with very short retention periods.

Key words: sago effluent, tapioca starch, hybrid anaerobic reactor, biogas

Introduction wastewaters (Tilche and Vieira 1991). Over the years, hybrid reactors have been used to treat wastewaters Sago, the edible starch globules processed from the from the sugar industry (Coates and Colleran 1990), tubers of tapioca (Mannihot utillisema), is the staple diet pharmaceutical units (Hentry et al. 1996), distilleries of middle income populations in India. Processing of (Shivayogimath and Ramanujam 1999) and domestic tapioca requires 20,000 to 30,000 L of water per tonne sectors (Elmitwalli et al. 2002a,b). Studies on the treat- of sago. In addition, it produces an equal quantity of ment of sago wastewater using hybrid reactors are scarce highly organic, foul-smelling, acidic wastewater (Murthy and hence the present study was undertaken. and Patel 1961; Sastry and Mohan 1963). Various anaerobic technologies including conventional anaerobic Materials and Methods treatment (Pescod and Thanh 1977; Pugalendhi 1996; Saroja and Sastry 1972; Sastry et al. 1964; Tongkasane The schematic diagram of the hybrid reactor is illus- 1970), high-rate anaerobic treatment such as anaerobic trated in Fig. 1. The laboratory-scale hybrid reactor was filters (Khageshan 1998) and fluidized beds (Saravanane fabricated using PVC tube with an internal diameter of et al. 2001) have been used to treat sago wastewater. 11 cm and an overall height of 88 cm. The working vol- The conventional treatment options are not very efficient ume of the reactor was 5.9 L. A gas headspace of 1.5 L owing to high concentrations of solids present in the was maintained above the effluent line. A screen was wastewater. Possibilities of overcoming many of the placed at a height of 60 cm to restrict the floating pack- functional drawbacks of high-rate anaerobic reactors ing material. Two hundred plastic cut rings measuring through modified designs have been discussed by many 1 cm in diameter and 2 cm in height were used as carrier authors. For instance, restricting the supporting material material. A peristaltic pump (Miclins, Mode PP 20) was to the top 25 to 30% of the reactor volume has used for feeding wastewater into the reactor. The efflu- increased the efficiency of reactors (Guiot and Van den ent pipeline in turn was connected to a water seal to pre- Berg 1984, 1985). This would help further realize the vent the escape of gas. The gas outlet was connected to a advantages of both fixed-film and up-flow sludge blan- wet gas meter (Ritter, Model TG 05). ket treatment. This type of reactor is often called a hybrid reactor and is considered more stable for the Seed and Inoculation treatment of a series of soluble or partially soluble Contents of the rumen of a cow immediately after * Corresponding author; [email protected] slaughter were collected from a slaughterhouse in

Downloaded from http://iwaponline.com/wqrj/article-pdf/41/1/56/230287/wqrjc0410056.pdf by guest on 28 September 2021 High-Rate Anaerobic Treatment of Sago Wastewater 57

increased from 1000 to 1050 mL/h (24.6 kg COD/m3d). The efficiency of the treatment was evaluated in terms of removal of total solids (TS), volatile solids (VS), COD and generation of biogas.

Chemical Analysis

Chemical oxygen demand (COD), volatile fatty acids as acetate (VFA), total alkalinity, total solids (TS), volatile solids (VS) and total Kjeldhal nitrogen (TKN) of the raw and treated wastewater were analyzed following APHA 3- (1998). Anions such as phosphate (PO4 ), sulphate 2- - (SO4 ) and chloride (Cl ) were analyzed employing ion exchange chromatography (Dionex, model DX-120) after filtering the samples through a 0.45-µm filter. The eluent was a combination of 3.5 mM bicarbonate and 1 mM carbonate, and the flow rate was 1.2 mL/min with an injection volume of 25 µL. Methane content in the biogas was measured by gas chromatography (Chemito, GC 1000) equipped with flame ionization detector (FID). The column used was Proapak Q.

Results and Discussion

Start-up Phase

The sago wastewater was analyzed for various physico- Fig. 1. Schematic diagram of the anaerobic hybrid reactor. chemical characteristics and the results are furnished in Table 1. To obtain lower organic load, the sago waste- Aduthotti, Chennai, India, and were transported to the water was appropriately diluted with distilled water. Fig- laboratory in an oxygen-free container. The contents ure 2 presents the loading pattern and performance of were strained using cheesecloth and the liquor was used the reactor during the start-up phase. The initial organic as the digester inoculum (Ezeonu and Okaka 1996). To loading rate (OLR) applied during start-up was 0.81 kg accelerate the start-up, 40% (v/v) slurry was mixed with COD/m3d at a HRT of 59 h. This HRT was preferred to the feed as recommended in earlier studies (Hickey et al. prevent the washout of inoculated biomass (Hickley et 1991; Shapiro and Switzenbanum 1984). al. 1991). When the OLR was increased in a stepped manner to 8.3 kg COD/m3d, over a period of 120 d, the Sago Effluent biogas generation also increased and reached a maximum of 9.7 L/d at an OLR of 8.3 kg COD/m3d. Synthetic sago wastewater was prepared following Khageshan (1998). TABLE 1. Physicochemical characteristics of sago Start-up Phase wastewater

a During the start-up, the reactor was operated with the Parameter Values wastewater having a chemical oxygen demand (COD) of pH 7.2 2000 mg/L. The initial hydraulic retention time (HRT) was Total solids 7645 59 h. It was gradually decreased to 5.9 h by increasing the Suspended solids 1405 Volatile solids 1834 flow rate from 100 to 1000 mL/h over a period of 120 d. COD 5750 BOD 4400 Treatment Phase TKN 180 Chloride 231 After the start-up, the reactor was operated by varying Sulphate 87 Phosphate 60 the influent COD at a constant flow rate of 1000 mL up Potassium 15 to its maximum value of 5750 mg/L (23.4 kg COD/m3d). After attaining the maximum COD, flow rate was aAverage of triplicate. All values except pH are in mg/L.

Downloaded from http://iwaponline.com/wqrj/article-pdf/41/1/56/230287/wqrjc0410056.pdf by guest on 28 September 2021 58 Banu et al.

Fig. 2. Loading pattern and biogas production during the treatment of sago wastewater.

The removal of COD from the medium increased and Lettinga 1994). VFA as acetate in the wastewater with OLR (Fig. 3), a finding which is in conformity with decreased further and was in the range of 540 to an earlier report on the treatment of sago wastewater 345 mg/L, indicating a healthy anaerobic environment (Saravanane et al. 2001). The VFA as acetate in the and marked methanogenic activity. Overall, the reactor wastewater on day 1 was 758 mg/L and it fell to 578 performed very satisfactorily due to a rich consortium of mg/L on day 18. Higher levels of VFA as acetate in the physiologically active microorganisms in the seed slurry wastewaters during the initial phases are known to indi- during the start-up. It is known that the selection of seed cate the prevalence of acid fermentation (van Hanndel material plays a crucial role in minimizing time required

Fig. 3. Influence of OLR on COD removal and VFA accumulation during the treatment of sago wastewater.

Downloaded from http://iwaponline.com/wqrj/article-pdf/41/1/56/230287/wqrjc0410056.pdf by guest on 28 September 2021 High-Rate Anaerobic Treatment of Sago Wastewater 59

for initial biofilm establishment (Bull et al. 1983; Salki- 590 mg/L. Buildup of VFA was found to be influenced noja-Salonen et al. 1983). by the digestion period and the OLR. The maximum concentration of 2900 mg/L was recorded at an OLR of Treatment Phase 24.6 kg COD/m3d. VFA, a recognized intermediate during the anaerobic digestion (Wang et al. 1999; Ahring The initial OLR applied during this phase (beginning and Angelidaki 1997) is considered a central parameter from day 121 indicated by an arrow in Fig. 2) was for anaerobic treatment (Ahring and Angelidaki 1995; 10.4 kg COD/m3d. It was increased in a stepped manner Pind et al. 1999, 2002). Accumulation of VFA as acetate to 24.6 kg COD/m3d, over a period of 235 d. The incre- in the medium is due to the sudden increase of flow rate ment between successive OLR was about 2.1 kg from 1000 (23.4 kg COD/m3d) to 1050 mL/h (24.6 kg COD/m3d. The gas generation increased as the OLR COD/m3d). Gerardi (2003) stated that sudden increase increased, reaching a maximum of 30.7 m3/d at an OLR of flow rate of an anaerobic reactor would affect the of 23.4 kg COD/m3d on day 232. Beyond this loading, conversion of VFA to gaseous products, which leads to the gas production decreased with the increase in OLR accumulation of VFA in the medium. A marked decrease (24.6 kg COD/m3d). However, when the OLR was in COD removal from 83 to 69% when the OLR was reverted to 23.4 kg COD/m3d, the gas production exhib- increased from 23.4 to 24.6 kg COD/m3d bears evidence ited an increasing trend. The present observations are of the impact of VFA accumulation. A VFA concentra- not in conformity with the published works. For tion of over 2500 mg/L has been reported to cause sour- instance, Subrahmanyam and Sastry (1989) reported ing of the UASB reactor during the treatment of syn- that during the treatment of sago wastewater using an thetic wastewater (Fang et al. 1994). To prevent the anaerobic filter (AF) with granite stones as the filter souring of the reactor, the flow was reverted to medium, the gas production decreased at an OLR of 1000 mL/h and the corresponding OLR of 23.4 kg 16 kg COD/m3d. Similar decrease in gas production COD/m3d. The reactor then started exhibiting signs of beyond an OLR of 11.64 kg COD/m3d, during the treat- recovery and as discussed earlier, the COD removal and ment of synthetic sago wastewater using anaerobic filter gas production increased as the VFA concentration in with basalt stone chips as filter medium has also been the medium decreased. Similarly, treating distillery reported (Khageshan and Govindan 1995). The higher wastewater by reverting the flow rate was achieved by loading rate tolerated by the system in the present study Shivayoimath and Ramanujam (1999). can be attributed to the structural modifications made to Figure 4 indicates normalized methane production the hybrid reactor and the filter material used. In con- versus biogas production during the study period. The trast to the present observations, a case of a fluidized normalized methane production varied from 0.11 to 3 3 bed reactor tolerating a high OLR of 60.5 kg COD/m d 0.14 m CH4/kg COD-d, during the stable operation during the start-up phase of treatment of synthetic sago period and the corresponding biogas production in the wastewater is available (Saravanane et al. 2001). Inter- estingly, the efficiency of this reactor in terms of COD removal (82%) is lower than the value recorded in the start-up phase of the present study (92%). Further comparison between these two studies is prohibited by the lack of data on HRT applied in the study referred to. Figure 3 illustrates the influence of OLR on COD removal and VFA as acetate concentration in the medium during the treatment of sago wastewater. At an OLR of 10.4 kg COD/m3d (indicated by the arrow) the COD removal was 91%. As the OLR increased, the COD removal exhibited a gradual decrease. At an OLR of 23.4 kg COD/m3d the COD removal was 83% and as mentioned elsewhere in the discussion, the gas production was maximized at this OLR. A similar trend of decrease in COD removal with increase in OLR was observed by Khageshan (1999) for treating synthetic sago wastewater using an anaerobic filter. The lowest COD removal of 69% was recorded at an OLR of 24.6 kg COD/m3d. During the stable operational phase (OLR: Fig. 4. Influence of OLR on normalized methane production 3 10.4–23.4 kg COD/m d) of the reactor, the VFA (as and percentage of methane production during the treatment acetate) levels in the medium ranged from 425 to of sago wastewater.

Downloaded from http://iwaponline.com/wqrj/article-pdf/41/1/56/230287/wqrjc0410056.pdf by guest on 28 September 2021 60 Banu et al.

range of 60 to 67%. The observed values in this study depicted in Fig. 5B. Under stable operating conditions are comparable to the values of 59 to 63% reported by (up to an OLR of 23.4 kg COD/m3d) the removal of TS Khageshan (1998). Similarly, there is a sudden drop in and VS in the wastewater was in the range of 61 to 56% methane production when OLR was raised from 23.4 to and 70 to 65%, respectively. The TS and VS removal 24.6 kg COD/m3d (normalized methane reduced from efficiency of the reactor decreased drastically when the 3 3 0.14 to 0.06 m CH4/kg COD-d and methane content OLR was increased from 23.4 to 24.6 kg COD/m d, the reduced from 61 to 56%). respective values for TS and VS being 54 and 61%. The pH of the treated wastewater was in the range Table 2 presents the utilization pattern of nutrients of 7.4 to 8.1 up to an OLR of 23.4 kg COD/m3d. This is (organic nitrogen and phosphorus) and the removal of indicative of the satisfactory functioning of the reactor. sulphate during the treatment of sago wastewater. From A pH less than 6.8 and greater than 8.3 is known to Table 2 it is evident that throughout the study, microbial cause souring of the reactor during anaerobic digestion utilization of organic nitrogen was markedly higher than (Stronach et al. 1986). The pH of the wastewater that of phosphorus. Higher utilization of nitrogen in dropped to 6.7 when the OLR was increased to 24.6 kg comparison to phosphorus during the anaerobic treat- COD/m3d from 23.4 kg COD/m3d (Fig. 5A). ment of sago wastewater has been reported by others as Total alkalinity of the medium increased from well (Subrahmanyam and Sastry 1989). Utilization of 1510 mg/L at an OLR of 10.9 kg COD/m3d to organic nitrogen fell from 39 to 34% when the OLR 3120 mg/L at an OLR of 24 kg COD/m3d. The increase was increased from 10.4 to 23.4 kg COD/m3d. Similarly, in alkalinity with increase in OLR is due to the increase the phosphorus utilization fell from 22% at an OLR of in feed rate at higher OLR (Stronach et al. 1986). An 10.4 kg COD/m3d to 15% at an OLR of 23.4 kg increase in OLR beyond 24 kg COD/m3d led to a COD/m3d. The decrease in nutrient utilization at higher decrease in alkalinity in the wastewater (Fig. 5A). OLRs can be attributed to a higher flow rate of the Decrease in alkalinity is due to the accumulation of wastewater and the consequent reduction in contact time VFA (Stronach et al. 1986). Alkalinity is known to be a between nutrients and microbes as well as the nutrient critical buffering factor for neutralizing VFA during washout. The removal of sulphate in the wastewater was methanogenisis (van Hanndel and Lettinga 1994). in the range of 84 to 80%. The chloride concentration in Methane production and alkalinity may be correlated, the effluent remained unaffected during the treatment. and this correlation may be used as an indicator of an unstable digester. A decrease in methane production Conclusion and a decrease in alkalinity indicate toxicity occurring in methane-forming bacteria. Obviously, low alkalinity Anaerobic treatment of sago wastewater using the affects the buffering capacity and leads to a drop in pH hybrid reactor appears to be a promising option as it (Gerardi 2003). could be operated at a considerably higher OLR of 23.4 The determination of VS is useful in control of kg COD/m3d, which is twice the loading rate suggested wastewater treatment plant operation as it offers a for the treatment in an anaerobic filter. Further, it is pos- rough approximation of the amount of organic matter sible to achieve higher COD removal (83%) and consid- present in the solid fraction of wastewater (APHA erable generation of biogas (30.7 m3/d) at a HRT as low 1998). Influence of OLR on the removal of TS and VS as 5.9 h. The results are significant, especially in the con- from the sago wastewater during the treatment phase is text of tropical developing countries where wastewater

Fig. 5. Influence of OLR on pH, alkalinity (A), TS and VS (B) removal during the treatment of sago wastewater.

Downloaded from http://iwaponline.com/wqrj/article-pdf/41/1/56/230287/wqrjc0410056.pdf by guest on 28 September 2021 High-Rate Anaerobic Treatment of Sago Wastewater 61

TABLE 2. Utilization of nitrogen, phosphorus and sulphate during the treatment of sago wastewater

Nitrogen Phosphorus Sulphate OLR(kg Influent Effluent Removal Influent Effluent Removal Influent Effluent Removal COD/m3d) (mg/L) (mg/L) (%) (mg/L) (mg/L) (%) (mg/L) (mg/L) (%) 10.4 80.1 ± 2 48.9 ± 2 39 26.7 ± 0.2 20.8 ± 0.1 22 38.7 ± 1 6.2 ± 0.2 84 12.4 95.8 ± 2 59.4 ± 1 38 31.9 ± 0.1 24.9 ± 0.2 22 46.3 ± 1 7.9 ± 0.2 83 14.5 111.4 ± 3 68.0 ± 2 39 37.1 ± 0.2 29.0 ± 0.1 22 53.9 ± 2 9.7 ± 0.2 82 16.6 127.1 ± 3 78.8 ± 2 38 42.4 ± 0.1 34.3 ± 0.2 19 61.4 ± 1 9.8 ± 0.2 84 18.7 142.7 ± 2 88.5 ± 3 38 47.6 ± 0.2 39.0 ± 0.2 18 69.0 ± 2 13.1 ± 0.2 81 20.7 158.4 ± 5 101.4 ± 3 36 52.8 ± 0.2 43.8 ± 0.3 17 76.6 ± 1 14.5 ± 0.2 81 22.8 174.1 ± 2 111.4 ± 4 36 58.0 ± 0.3 48.7 ± 0.2 16 84.1 ± 1 16.0 ± 0.2 81 23.4 176.9 ± 4 116.7 ± 3 34 59.0 ± 0.5 50.1 ± 0.3 15 85.5 ± 2 17.1 ± 0.3 80

treatment reactor design with low HRT and high OLR 705–717. In Proceedings of the 39th Annual Purdue would afford economic and technological viability. Industrial Waste Conference, Lafayette, Indiana. Guiot SR, Van den Berg L. 1985. Performance of an upflow References anaerobic reactor combining a sludge blanket and a filter treating sugar waste. Biotech. Bioeng. 27:800–806. Ahring BK, Angelidaki I. 1997. Monitoring and controlling Hentry MP, Donlon BA, Lens PN, Colleran EM. 1996. Use the biogas process, p. 40–45. In Proceedings at 8th of anaerobic hybrid reactor for treatment of synthetic International Conference on Anaerobic Digestion, pharmaceutical wastewaters containing organic sol- Sendai, Japan. vents. J. Chem. Tech. Biotech. 66:251–264. Ahring BK, Angelidaki I. 1995. Volatile fatty acids as indi- Hickey RF, Wu WM, Viega MC, Jone R. 1991. Start up, cators of process imbalance in anaerobic digestion. operation monitoring and control of high rate anaero- Appl. Microbiol. Biotechnol. 43:559–565. bic treatment systems. Water Sci. Technol. 24:207–256. APHA. 1998. Standard methods for the examination of Khageshan P. 1998. Studies on upflow anaerobic filter for water and wastewater, 20th edition. Victor Graphics, treating sago wastewater. Ph.D. thesis, Anna Univer- Maryland, U.S. sity, Chennai, India. Bull MA, Sterritt RM, Lester JN. 1983. The influence of Khageshan P, Govindan VS. 1995. Anaerobic filter for COD, hydraulic, temperature, and pH shocks on sta- treatment of sago wastewater, p. 248–252. In Proceed- bility of an unheated fluidized bed reactor. J. Chem. ings of 4th National Symposium on Environment, Tech. Biotechnol. 33:221–230. Chennai, India. Coates J, Colleran E. 1990. Effect of initial agitation on the Murthy YS, Patel MD. 1961. Treatment and disposal of start up and operational performance of anaerobic sago wastes. Central Public Health Engineering hybrid reactors treating a synthetic feed. Process Research Institute, India. Biochem. 10:162–171. Pescod MB, Thanh NC. 1977. Treatment alternatives for Elmitwalli TA, Kim LT, Grietje Z, Lettinga G. 2002a. wastewaters from the tapioca starch industry. Water Treatment of domestic sewage in a tow step anaerobic Sci. Technol. 9(3):63–574. filter/anaerobic hybrid system at low temperature. Pind FP, Angelidaki I, Ahring BK. 1999. The use of VFA Water Res. 36:2225–2232. measurement as process indicators in anaerobic reactors Elmitwalli TA, Sklyar V, Zeeman G, Lettinga G. 2002b. treating manure, p. 41–44. In II International Sympo- Low temperature pretreatment of domestic sewage in sium on Anaerobic Digestion of Solid Waste, Barcelona. an anaerobic hybrid or an anaerobic filter reactor. Pind FP, Angelidaki I, Ahring BK. 2002. Dynamics of the Biores. Technol. 52:233–239. anaerobic process: effects of volatile fatty acids. Ezeonu FC, Okaka ANC. 1996. Process kinetics and diges- Biotechnol. Bioeng. 82(7):791–801. tion efficiency of anaerobic fermentation of brewer’s Pugalendhi S. 1996. Biotechnological investigation of cas- spent grains. Water Sci. Technol. 31(1):7–12. sava peelings for biomethanation. Ph.D. thesis, Tamil Fang HHP, Guohua L, Jinfu Z, Bute S, Guowei G. 1994. Nadu Agricultural University, Coimbatore, India. Treatment of brewery effluent by UASB process. J. Salkinoja-Salonen MS, Nyns EJ, Sutton PM, Van den Berg Environ. Eng. Div. ASCE 116:454–460. L, Wheatley AD. 1983. Starting up of an anaerobic Gerardi H. 2003. The microbiology of anaerobic digester. fixed film reactor. Water Sci. Technol. 15(8):305–308. John Wiley & Sons, New Jersey. Saravanane R, Murthy DVS, Krishniash K. 2001. Anaero- Guiot SR, Van den Berg L. 1984. Dynamic performance of bic treatment and biogas recovery for sago wastewater an anaerobic reactor combining an upflow sludge blan- management using a fluidized bed reactor. Water Sci. ket and a filter for the treatment of sugar waste, p. Technol. 44(6):141–147.

Downloaded from http://iwaponline.com/wqrj/article-pdf/41/1/56/230287/wqrjc0410056.pdf by guest on 28 September 2021 62 Banu et al.

Saroja K, Sastry CA. 1972. Report on treatment of sago Subrahmanyam YV, Sastry CA. 1989. Attached versus sus- wastes. National Environmental Engineering Research pended biomass activity in upflow anaerobic filters. Institute, Nagpur, India. J. Institut. Engineers 9:27–32. Sastry CA, Mohan Rao GJ. 1963. Anaerobic digestion of Tilche A, Vieira SMM. 1991. Discussion report on reactor industrial wastes. Indian J. Environ. Health 5(3):20–25. design of anaerobic filters and sludge bed reactors. Sastry CA, Murahari R, Saroja K. 1964. Studies on the Water Sci. Technol. 24:193–206. treatment of sago mill wastewater. In Seminar on Envi- Tongkasane C. 1970. Anaerobic treatment of tapioca starch ronmental Pollution, Kerala, India. wastewater. Water Wastewater Treat. 13:392–398. Shapiro M, Switzenbanum MS. 1984. Initial anaerobic van Hanndel C, Lettinga G. 1994. Anaerobic sewage treat- biofilm development. Biotechnol. Lett. 6:729–734. ment: a practical guide for regions with a hot climate. Shivayogimath CB, Ramanujam TK. 1999. Treatment of Wiley Publishers, U.K. distillery spent wash by hybrid UASB reactor. Bio- Wang QH, Kuninobu M, Ogawa HI, Kato Y. 1999. Degra- process Eng. 21:255–259. dation of volatile fatty acids in highly efficient anaero- Stronach SM, Rudd T, Lester JN. 1986. Anaerobic diges- bic digestion. Biomass Bioenergy 16:407–416. tion process in industrial wastewater treatment. Springer-Verlag, Germany. Received: February 9, 2004; accepted: February 27, 2006.

Downloaded from http://iwaponline.com/wqrj/article-pdf/41/1/56/230287/wqrjc0410056.pdf by guest on 28 September 2021