PII: S0043-1354(01)00041-0

Wat. Res. Vol. 35, No. 14, pp. 34413447, 2001 # 2001 Elsevier Science Ltd. All rights reserved Printed in Great Britain 0043-1354/01/$ - see front matter

PERFORMANCE OF UASB REACTOR TREATING LEACHATE FROM ACIDOGENIC FERMENTER IN THE TWO-PHASE ANAEROBIC DIGESTION OF FOOD WASTE
H. S. SHIN1*, S. K. HAN1, Y. C. SONG2 and C. Y. LEE3
1

Department of Civil Engineering, Korea Advanced Institute of Science and Technology, 373-1 Kusongdong, Yusong-Ku, Taejon, 305-701, South Korea; 2 Division of Civil and Environmental Engineering, Korea Maritime University, Dongsam-Dong 1, Youngdo-Ku, Pusan, 606-791, South Korea and 3 R&D Team, Institute of Technology, Engineering & Construction Group, Samsung Corporation, 428-5 Gongse-ri, Giheung-eup, Yongin-city, Kyunggi-Do, 449-900, South Korea (First received 1 June 2000; accepted in revised form 9 January 2001)

Abstract}This study was conducted to investigate the performance of the upow anaerobic sludge blanket (UASB) reactor treating leachate from acidogenic fermenter in the two-phase anaerobic digestion of food waste. The chemical oxygen demand (COD) removal eciency was consistently over 96% up to the loading rates of 15.8 g COD/l d. The methane production rate increased to 5.5 l/l d. Of all the COD removed, 92% was converted to methane and the remaining presumably to biomass. At loading rates over 18.7 g COD/l d, the COD removal eciency decreased due to sludge otation and washout in the reactor, which resulted from short HRT of less than 10.6 h. The residual propionate concentration was the highest among the volatile fatty acids (VFA) in the euent. The specic methanogenic activity (SMA) analysis showed that the VFA-degrading activity of granule was the highest for butyrate, and the lowest for propionate. Typical granules were found to be mainly composed of microcolonies of Methanosaeta. The size distribution of sludge particles indicated that partially granulated sludge could maintain the original structure of granular sludge and continue to gain size in the UASB reactor treating leachate from acidogenic fermenter. # 2001 Elsevier Science Ltd. All rights reserved Key words}UASB reactor, leachate, acidogenic fermenter, food waste, COD removal eciency, methanogenic activity

INTRODUCTION

The generation of municipal solid waste (MSW) amounts to 44,583 t/d in Korea, of which 26.5% is food waste from restaurants, dinning halls, markets and households (MOE, 2000). Food waste is the main source of decay, odor and leachate in collection and transportation due to the high volatile solids (8595%) and moisture content (7585%). Most food waste has been landlled together with other wastes, resulting in various problems such as emanating odor, attracting vermin, emitting toxic gases, contaminating groundwater and wasting landll capacity. Interest in anaerobic digestion has, therefore, increased for the ecient management of food waste because it has advantages of volume reduction and methane recovery as well as waste stabilization. It is considered that two-phase anaerobic digestion is more eective in the anaerobic degradation of

*Author to whom all correspondence should be addressed. Tel.: +82-42-869-3613; fax: +82-42-869-3610; e-mail: hangshin@kaist.ac.kr

particulate substrates, of which the rate-limiting step is hydrolysis and liquefaction. Pohland and Ghosh (1971) suggested a two-phase anaerobic process using two separate reactors, one for hydrolysis/acidication and the other for acetogenesis/methanogenesis. The microorganisms were separated physically to make use of the dierences in their growth kinetics. In order to accomplish phase separation, several techniques were developed such as kinetic control, leaching beds, membrane separation and pH control (Fox and Pohland, 1994; Ince, 1998). The two-phase process permits selection and enrichment of dierent bacteria in each digester by independently controlling the digester operating conditions. In this study, the MUlti-step Sequential batch Two-phase Anaerobic Composting (MUSTAC) process was newly devised as an ideal method for treating food waste as shown in Fig. 1. The MUSTAC process consists of two main parts: ve leaching beds for hydrolysis, acidication and posttreatment, and an upow anaerobic sludge blanket (UASB) reactor for methane recovery. Feedstock for this experiment was food waste collected from a

3441

3442

H. S. Shin et al.

dining hall. The distribution of grains, vegetables and meats in the waste was 40.5, 44.2 and 15.3%, respectively. The leaching beds were operated in a rotation mode with a two-day interval between degradation stages as shown in Table 1. Acidied products in leachate from four leaching beds were converted to methane in the UASB reactor. Each leaching bed was operated in a sequential batch mode. Rumen microorganisms (5% v/v) exhibiting enhanced cellulolytic activity were inoculated into the reactor to improve the low eciency of acidogenic fermentation (Gijzen, 1987; Song, 1995; Shin et al., 2000). After 6 h of acclimation, dilution water was provided to the leaching bed in order to transfer the acidied products to the UASB reactor. Dilution rate (D rate; d1 ) was dened as 1=y (hydraulic retention time; d), or Q (owrate of dilution water; l=d)/V (eective reactor volume; l). The proper control of dilution rate (2:1 ! 0:7 d1 ), depending on the state of the fermentation, could eliminate environmental constraints in the fermentation (Shin et al., 2000). The dierent sizes of shaded portion in the leaching beds (Fig. 1) indicated the volume reduction of food waste according to the degradation stages. Acidogenic fermentation of 8 days was reasonable con-

sidering operation time and eciency (Shin et al., 2000), which was followed by the post-treatment. The residues were dewatered in the leaching bed for 6 h and then 15 l/min of air was introduced through the bottom of the reactor for 42 h. This study was conducted to investigate the performance of the UASB reactor treating leachate from acidogenic fermenter in the two-phase anaerobic digestion of food waste. The key operational parameters were examined for the process development including the eciency of chemical oxygen demand (COD) removal at high loading rates and the characteristics of the granules, such as the specic methane production rate (SMPR), the specic methanogenic activity (SMA), the microstructure and the particle size distribution.

MATERIALS AND METHODS

Experimental set-up The UASB reactor used in this study was 41.0 l in working volume (lower part: 780 mm high by 200 mm ID; upper part: 400 mm high by 280 mm ID). Five evenly distributed sampling ports were installed along the reactor

Table 1. Operating method of ve leaching beds in the MUSTAC processa Day 12 34 56 78 910 1112
a

Reactor 1 Stage Stage Stage Stage Stage Stage 1 2 3 4 5 1

Reactor 2 } Stage Stage Stage Stage Stage

Reactor 3 } } Stage Stage Stage Stage

Reactor 4 } } } Stage 1 Stage 2 Stage 3

Reactor 5 } } } } Stage 1 Stage 2

1 2 3 4 5

1 2 3 4

Stage 1: Acidication of grains (D rate=2.1 d1), Stages 2 & 3: Acidication of vegetables & meats (D rate=0.7 d1), Stage 4: Stabilization (D rate=0.7 d1), and Stage 5: Post-treatment.

Fig. 1. Schematic diagram of the MUSTAC process for food waste.

Performance of UASB reactor wall. The temperature of the reactor was maintained at 378C by water-jacket. Sludge inoculation and feed wastewater Partially granulated sludge (8.5 l) from an anaerobic plant treating brewery wastewater was inoculated into the UASB reactor. The volatile suspended solids (VSS) concentration of the seed sludge was 108.1 g/l. The biomass was periodically removed from the reactor to avoid excess biomass accumulation, and also for the analyses of SMA, SEM and the particle size distribution. Leachate from acidogenic fermenter treating food waste was continuously fed into the UASB reactor by a peristaltic pump. As shown in Table 2, the soluble COD of leachate was about 7000 mg/ l, in which the percentage of acetate, propionate, butyrate, valerate and caproate was 26, 18, 35, 17 and 4% on COD basis, respectively. The pH was about 6.6, and the alkalinity was about 3000 mg/l as CaCO3. Loading rate The initial COD loading rate of the reactor was 1.8 g/l d, which corresponded to 3.9 d of hydraulic retention time (HRT). The loading rate was increased stepwise by reducing HRT, when the COD removal eciency exceeded 95% and the methane production rate was consistent (within 5%) for three consecutive days. The COD removal eciency was determined based on the soluble COD after ltering the sample through glassber lter paper (Whatman GF/C). Fang and Chui (1993a) reported that the COD removal eciency of the UASB reactor was mainly dependent on the COD loading rate and was not sensitive to either the HRT or the COD of the wastewater alone. Specic methanogenic activity The SMA analysis was performed in duplicates in serum vials (125 ml) based on the reported method (Hwang and Cheng, 1991). Sludge samples were taken for the SMA analysis on day 150, when the reactor was operated at 6.0 g COD/l d with over 96% COD removal and without detectable volatile fatty acids (VFA) in the euent. The methanogenic activity of the granule (100 mg VSS) was measured for a specic substrate (2000 mg COD/l). In this study, acetate, propionate, butyrate, valerate, caproate and original substrate (leachate) were used individually as the substrate. Microbial examination and the particle size distribution The microbial examination was conducted by using scanning electron microscopy (SEM). The instruments and the sample preparation procedures were as reported previously (Fang and Chui, 1993b). At the end of this study, the sludge particles were sequentially separated using six stainless-steel sieves (U.S. Mesh No. 10, 14, 18, 35, 60 and 140 with corresponding sieve openings of 2.00, 1.41, 1.00, 0.50, 0.25 and 0.105 mm). The sludge samples for the particle size distribution were taken from the second port from the bottom of the reactor.

3443

Table 2. Characteristics of leachate from acidogenic fermenter treating food waste Parameter Total COD Soluble COD Total suspended solids Volatile suspended solids Total VFA (as acetic acid) VFA composition (COD basis) Acetate Propionate Butyrate Valerate Caproate PO4-P TKN pH Alkalinity Units mg/l mg/l mg/l mg/l mg/l % % % % % mg/l mg/l mg/l as CaCO3 Value 66008600 60008000 350620 300550 59207910 26 18 35 17 4 80150 150250 6.46.8 25003500

200 and 2508C, respectively. The initial temperature of the column was 808C for 5 min and increased gradually by 108C/min, reaching the nal temperature of 1308C. Gas production was determined using a wet gas meter. For biogas composition analyses, the GC (GowMac series 580) was equipped with a thermal conductivity detector (TCD) and a 2 m 2 mm inside diameter (ID) stainless-steel column packed with Porapak Q (80/100 mesh). The temperatures of injector, detector and column were kept at 80, 90 and 508C, respectively. COD and TKN were measured using the closed reux, titrimetric method and the Kjeldahl method, respectively (APHA, 1992). PO4-P was analyzed by ion chromatograph (Dionex DX-120). The parameters such as total suspended solids (TSS), VSS, pH and alkalinity of the euent were determined according to Standard Methods (APHA, 1992).
RESULTS AND DISCUSSION

COD removal eciency Figure 2 shows the eciency of COD removal, the biogas production rate and HRT throughout the study. HRT was decreased stepwise from 3.89 to 0.33 d. The COD removal eciency was consistently over 96% up to HRT of 0.44 d, which corresponded to the loading rate of 15.8 g COD/l d and a food-tomicroorganism (F/M) ratio of 0.8 g COD/g VSS d. The pH in the euent was maintained at a constant level of 7.57.6. The maximum biogas production rate was 279.0 l/d at the COD loading rate of 15.8 g/ l d. The experimented biogas production was compared with the theoretical value as shown in Fig. 2(b). The theoretical biogas production rate was calculated as 0:3974 l CH4 CH4 content of biogas: g COD

CODremoved the conversion efficiency of CODremoved to CH4

Analytical procedures VFA was measured using a gas chromatography (GC; Hewlett Packard model 5890 A). The GC was equipped with a 10 m 0.53 mm HP-FFAP fused-silica capillary column and a ame ionization detector (FID), using helium as a carrier gas. The temperatures of injector and detector were

The maximum COD loading rate (15.8 g/l d) was comparable to the maximum VFA converting activity of 14.0 g COD/l d in the UASB reactor treating synthetic mixed VFA (acetate : propionate : butyrate=2 : 1 : 1 as COD) (Fang et al., 1995).

3444

H. S. Shin et al.

Fig. 2. Performance of UASB reactor: (a) COD removal eciency; (b) biogas production rate; (c) HRT.

Fig. 3. Methane production rate, residual VFA and solid concentration in the euent at various COD loading rates.

Thereafter, the COD removal eciency was drastically reduced to about 70% at 0.37 d HRT (18.7 g COD/l d), and to about 55% at 0.33 d HRT (21.4 g COD/l d). Figure 3 shows the methane production rate, residual VFA and solid concentration in the euent at each COD loading rate. The methane content of the biogas was measured to be 7781%. The methane

production rate increased linearly with COD loading rates, until reaching the maximum value of 5.5 l/l d at 15.8 g COD/l d. Up to the loading rates of 15.8 g COD/l d, the euent had less than 50 mg COD/l of acetate, 100 mg COD/l of propionate, 30 mg COD/l of valerate and negligible quantities of butyrate and caproate. VSS concentration was maintained below 150 mg/l. On

Performance of UASB reactor

3445

the other hand, the residual acetate, propionate, butyrate, valerate and caproate increased sharply to the levels of 450, 660, 60, 290 and 80 mg COD/l at 18.7 g COD/l d, and further to 700, 1,100, 80, 410 and 110 mg COD/l at 21.4 g COD/l d. The VSS concentration in the euent also increased sharply to 241 mg/l at 18.7 g COD/l d, and to 383 mg/l at 21.4 g COD/l d. This meant that the decrease of COD removal eciency was due to sludge otation and washout in the reactor, which resulted from short HRT of less than 10.6 h. The concentrations of residual butyrate and caproate were relatively low in the euent. This indicated that even-numbered carbon fatty acids were degraded more easily than odd-numbered ones, and that their degradation to acetate was not a rate-limiting step. The accumulation of residual acetate was caused by the degradation of other fatty acids to acetic acid. Residual acetate was not quickly converted to methane due to excessive washout of methanogens. The propionate concentration in the euent was higher than any other acid. Among VFA, propionate is known to have the lowest tolerance level for the anaerobic bacteria. When an anaerobic treatment system is overloaded, propionate tends to accumulate in the reactor and its removal is dicult during recovery. The degradation of propionate to acetate is thermodynamically infeasible unless the by-product hydrogen is removed by the hydrogen-consuming bacteria (Boone and Xun, 1987). These indicated that converting propionate to acetate was the rate-limiting step.

substrate utilization rate (SSUR) with a slope of 0.92, until reaching the maximum of 0.70 g methane COD/ g VSS d at the SSUR of 0.76 g COD/g VSS d. The SMPR means the methanogenic activity of the granules under the specic operating conditions of the reactor. The SMPR for the biomass could be estimated from the methane production rate and the total biomass in the reactor as each gram of methane was equivalent to 4 g of COD. The slope indicated that of all the COD removed, 92% was converted to methane and the rest 8% was presumably converted to biomass. Since the biomass in the reactor was measured to have a COD/VSS ratio of 1.41, the sludge yield was estimated to be 0.057 g VSS/g COD (=0.08/1.41), which was comparable to the yield values on acetate (0.0100.054 g VSS/g COD), shortchain fatty acids except acetate (0.0250.047 g VSS/g COD) and long-chain fatty acids (0.040.11 g VSS/g COD) as reported by Pavlostathis and GiraldoGomez (1991).

Specic methanogenic activity Table 3 summarizes the SMA of the granules measured in serum vials using six dierent substrates, i.e., acetate, propionate, butyrate, valerate, caproate and original substrate (leachate). The SMA is an indicator for evaluating the methanogenic activity of the biomass under a condition in which the supply of substrate is not a limiting factor. Corresponding studies on the SMA of UASB granules treating various types of wastewater were also shown for comparison. Each gram of the VFA-degrading granules was capable of producing 0.81, 0.69, 1.10, 0.87, 0.99 and 0.94 g of methane COD per day from acetate, propionate, butyrate, valerate, caproate and original substrate, respectively. This indicated that

COD balance and sludge yield Figure 4 shows that the specic methane production rate (SMPR) increased linearly with the specic

Fig. 4. Specic methane production rates at various specic substrate utilization rates.

3446

H. S. Shin et al.
Table 3. Specic methanogenic activity of granules

Original substrate fed to sludge

SMA (g CH4 COD/g VSS d) using the following substrate Acetate Propionate 0.69 0.35 0.52 1.78 0.41 Butyrate 1.10 0.81 0.61 1.32 } Valerate 0.87 } } } } Caproate 0.99 } } } } Original subtrate 0.94 1.03 0.85 } }

Reference

Leachate Mixed VFAb Sucrose Propionate Sugar

a b

0.81 1.17 1.20 1.89 0.90

This study Fang et al. (1995) Fang et al. (1994a, b) Grotenhuis et al. (1991) Dolng and Mulder (1985)

From acidogenic fermenter treating food waste. Acetate : propionate : butyrate=2 : 1 : 1 as COD.

the VFA-degrading activity of granule was the highest for butyrate, and the lowest for propionate. This is consistent with the observation that, at loading rates over 18.7 g COD/l d, high residual propionate was found in the euent while there was little residual butyrate. The SMA in other studies showed that the VFA-degrading activity of granule was the highest for acetate. However, in this study, the SMA values for valerate and caproate as well as butyrate were higher than that for acetate. This meant that the degradation of C4C6 fatty acids was dominant in the treatment of leachate from acidogenic fermenter for food waste. Microstructure and particle size distribution of granules Although layered microstructure was reported for granules treating wastewater of sucrose (MacLeod et al., 1990) and brewery (Fang et al., 1994a), granules from this study did not exhibit any patterned structure. The microorganisms were distributed throughout the entire granule. Figure 5 shows the anaerobic microorganisms in the sludge granules. Typical granules were found to be mainly composed of microcolonies of Methanosaeta, which had a width of 0.70.8 mm and a length of 3.07.0 mm. Methanosaeta is an acetoclastic bacterium using acetate solely as substrate. With a very low half-rate constant of 30 mg COD/l (Gujer and Zehnder, 1983), Methanosaeta outcompetes other methanogenic bacteria when the acetate concentration is low, as in the mixed liquor of the UASB reactor and in the interior of the granule. The syntrophic association between acetogens (cocci) and methanogens (rods) allowed the rapid removal of hydrogen, which would otherwise hinder the propionate degradation. The dry weight percent for dierent size ranges of granular sludge is shown in Table 4. The original seed sludge had 64.3% of particles smaller than 1.4 mm. At the end of this study, 75.1% of the sludge particles were larger than 1.4 mm in the UASB reactor. This indicated that partially granulated sludge could maintain the original structure of granular sludge and continue to gain size in the UASB reactor treating leachate from acidogenic fermenter.

Fig. 5. SEM picture of typical sludge granule: (a) Methanosaeta; (b) Syntrophic microorganisms.
Table 4. Size distribution of sludge particles before and after this study Size (mm) % Dry weight in each size range Granulesinitial >2.0 1.42.0 1.01.4 0.51.0 0.250.5 0.10.25 50.1 6.5 29.2 47.1 10.6 3.6 2.7 0.3 Granulesnal 10.7 64.4 17.2 6.3 1.0 0.4 0.0

CONCLUSIONS

1. The UASB process showed that the COD removal eciency was consistently over 96% up to the loading rates of 15.8 g COD/l d. The methane

Performance of UASB reactor

3447

production rate increased to 5.5 l/l d. The COD removal eciency deteriorated at loading rates over 18.7 g COD/l d due to sludge otation and washout in the reactor, which resulted from short HRT of less than 10.6 h. At these loading rates, high residual propionate was found in the euent, indicating that propionate degradation was a rate-limiting step. 2. Each gram of VFA-degrading granules in the reactor had a daily maximum capacity of converting 0.70 g of COD to methane at the specic substrate utilization rate of 0.76 g COD/g VSS d. Of all the COD removed, 92% was converted to methane and the rest presumably to biomass with an average sludge yield of 0.057 g VSS/g COD. 3. The SMA using acetate, propionate, butyrate, valerate and caproate, individually, as substrate were 0.81, 0.69, 1.10, 0.87 and 0.99 g methane COD/g VSS d. This indicated that the VFA-degrading activity of granule was the highest for butyrate, and the lowest for propionate. 4. Typical VFA-degrading granules were composed of microcolonies of Methanosaeta. The size distribution of sludge particles indicated that partially granulated sludge could maintain the original structure of granular sludge and continue to gain size in the UASB reactor treating leachate from acidogenic fermenter.
Acknowledgements}This work was supported by grant No. 95-1-13-01-01-3 from a research program of academic and industrial cooperation of the Korea Science and Engineering Foundation.

REFERENCES

APHA (1992) Standard Methods for the Examination of Water and Wastewater. APHA, Washington, DC, USA. Boone D. R. and Xun L. (1987) Eects of pH, temperature, and nutrients on propionate degradation by a methanogenic enrichment culture. Appl. Environ. Microbiol. 53(7), 15891592. Dolng J. and Mulder J. W. (1985) Activity measurements as a tool to characterize the microbial composition of methanogenic environment. J. Microbiol. Meth. 4, 112.

Fang H. H. P. and Chui H. K. (1993a) Maximum COD loading capacity in UASB reactors at 378C. J. Environ. Engng. 119(1), 103119. Fang H. H. P. and Chui H. K. (1993b) Microstructural analysis of anaerobic granules. Biotechnol. Tech. 7(7), 479482. Fang H. H. P., Chui H. K. and Li Y. Y. (1994a) Microbial structure and activity of UASB granules treating dierent wastewaters. Water Sci. Technol. 30(12), 8796. Fang H. H. P., Chui H. K., Li Y. Y. and Chen T. (1994b) Performance and granule characteristics of UASB process treating wastewater with hydrolyzed proteins. Water Sci. Technol. 30(8), 5563. Fang H. H. P., Li Y. Y. and Chui H. K. (1995) UASB treatment of wastewater with concentrated mixed VFA. J. Environ. Engng. 121(2), 153160. Fox P. and Pohland F. G. (1994) Anaerobic treatment applications and fundamentals: substrate specicity during phase separation. Water Environ. Res. 66, 716724. Gijzen H. J. (1987) Anaerobic digestion of cellulosic waste by a rumen derived process. Ph.D. thesis, University of Nijmengen, The Netherlands. Grotenhuis J. T. C., Smit M., Plugge C. M., Yuansheng Xu, van Lammeren A. A. M., Stams A. J. M. and Zehnder A. J. B. (1991) Bacteriological composition and structure of granular sludge adapted to dierent substrates. Appl. Environ. Microbiol. 57(7), 19421949. Gujer W. and Zender A. J. B. (1983) Conversion process in anaerobic digestion. Water Sci. Technol. 15, 127167. Hwang P. C. and Cheng S. S. (1991) The inuence of glucose supplement on the degradation of catechol. Water Sci. Technol. 23(79), 12011209. Ince O. (1998) Performance of a two-phase anaerobic digestion system when treating dairy wastewater. Water Res. 32(9), 27072713. MacLeod F. A., Guiot S. R. and Costerton J. W. (1990) Layered structure of bacterial aggregates produced in an upow anaerobic sludge bed and lter reactor. Appl. Environ. Microbiol. 56(6), 15981607. MOE (2000) Homepage of Ministry of Environment (http:// www.me.go. kr). Pavlostathis S. G. and Giraldo-Gomez E. (1991) Kinetics of anaerobic treatment. Water Sci. Technol. 24(8), 3559. Pohland F. G. and Ghosh S. (1971) Developments in anaerobic treatment processes. Biotechnol. Bioengng. Symp. 2, 85106. Shin H. S., Han S. K., Song Y. C. and Hwang E. J. (2000) Biogasication of food residuals. Biocycle 41(8), 8286. Song Y. C. (1995) High-rate methane fermentation of the organic solid waste. Ph.D. thesis, KAIST.