Bioresource Technology 162 (2014) 266–272

Contents lists available at ScienceDirect

Bioresource Technology

journal homepage: www.elsevier .com/locate /bior tech

Performance of two-stage vegetable waste anaerobic digestiondepending on varying recirculation rates

http://dx.doi.org/10.1016/j.biortech.2014.03.1560960-8524/� 2014 Elsevier Ltd. All rights reserved.

⇑ Corresponding author. Address: Key Laboratory of Clean Utilization Technologyfor Renewable Energy in Ministry of Agriculture, College of Engineering, ChinaAgricultural University, No. 17 Qinghuadonglu, Haidian District, Beijing 100083, PRChina. Tel.: +86 10 62737852; fax: +86 10 62737885.

E-mail address: wushubiao@gmail.com (S. Wu).

Zhuang Zuo a, Shubiao Wu b,⇑, Wanqin Zhang a, Renjie Dong b

a College of Agronomy and Biotechnology, China Agricultural University, Beijing 100193, PR Chinab Key Laboratory of Clean Utilization Technology for Renewable Energy in Ministry of Agriculture, College of Engineering, China Agricultural University, Beijing 100083, PR China

h i g h l i g h t s

� The performance of VW anaerobic digestion under varying RR was investigated.� The appropriate RR could improve acidogenesis in acidogenic reactor.� Recirculation influenced COD and VS removal efficiency of the whole system.� Biogas production from the whole two stage system was improved by increasing RR.

a r t i c l e i n f o

Article history:Received 20 January 2014Received in revised form 27 March 2014Accepted 28 March 2014Available online 5 April 2014

Keywords:Vegetable wasteTwo-stage processAnaerobic digestionBiogas productionRecirculation rate

a b s t r a c t

Vegetable waste, which characterized by high moisture content, was evaluated as a substrate for biogasproduction. The effects of recirculation rate (RR) on the performance of two-stage anaerobic digestionwere investigated. The system was operated at an organic loading rate of 1.7 g VS/L/d with varying RRs(0, 0.6, 1, and 1.4). Results demonstrated that volumetric biogas production rates in acidogenic reactorincreased from approximately 0.27 L/L/d to 0.97 L/L/d, when pH is increased from approximately 5.1 to6.7. These indicate that recirculation of alkaline effluent from the methanogenic reactor helps create afavorable condition for biogas production in the acidogenic reactor. The decrease in chemical oxygendemand (COD) concentrations from approximately 21,000 mg/L to 6800 mg/L was also observed in theacidogenic reactor. This condition may be attributed to dilution under recirculation. The dynamicsbetween hydrolysis and methanogenesis under recirculation indicated that mass transfer capacitybetween two-stage reactors improved.

� 2014 Elsevier Ltd. All rights reserved.

1. Introduction

Vegetable waste (VW) is generated in the processes ofharvesting, transportation, processing, marketing and storage(Sahu, 2004). VW is made up of highly biodegradable organicmatter content and can be beneficially used for renewable energyproduction by anaerobic digestion (Bouallagui et al., 2005). A majorlimitation of the anaerobic digestion of VW in a one-stage systemis the accumulation of volatile fatty acids (VFA), which results in arapid decrease in pH, potentially inhibiting methanogens activity(Bouallagui et al., 2009; Jiang et al., 2012). In such a case, a numberof studies introduced remedial measures, e.g., co-digestion with

other organic wastes to improve buffering capacity (Callaghanet al., 2002). However, co-digestion depends on feedstock availabil-ity and transportation efficiency and may be limited in vegetableproduction areas situated far from livestock farms (Poeschl et al.,2010). The two-stage system is another technology designed toimprove the overall process stability. The advantage of such sys-tem lies in the buffering of the organic loading rate (OLR) in thefirst stage, which in turn allows for a more constant feeding ratein the methanogenic stage (Bouallagui et al., 2004; Schievanoet al., 2012).

The two-stage system has been widely suggested for enhancingdigestion performance, employing a process configuration thatutilizes reactors for acidification and methanogenesis, allowingfor the optimization of both processes (Boe and Angelidaki, 2009;Kafle and Kim, 2011). However, a limited amount of data isavailable in the literature on the two-stage anaerobic digestion ofVW. There are problems associated with the operation of anaerobictreatment processes due to the low specific growth rate and

Biogas

Vegetablewaste

Liquid

Biogas

Methanogeniceffluent

AcidogenicReactor

Methanogenic

Fig. 1. Scheme of acidogenic reactor and methanogenic reactor used for thetreatment of vegetable waste.

Z. Zuo et al. / Bioresource Technology 162 (2014) 266–272 267

sensitivity to changes in operating conditions of some of the meth-anogens involved. It is also important to control the reaction path-way towards the selective production of desired end-products. Theconcentration and composition of VFAs which are highly depen-dent on the substrate type and process conditions (Li and Yu,2011). Therefore, the operation and control of the anaerobic diges-tion process should be directed to ensure process efficiency andbiogas production improvement (Schievano et al., 2012). The efflu-ent recirculating from the methanogenic reactor to the acidogenicreactor is one of the important operations and control methods,which exerts considerable influence on the overall process perfor-mance. The two-stage system with recirculation has recently beensuccessfully operated with a variety of solid organic waste, such asbiowaste, and food waste. It is reported that recirculation couldaccelerate the rate of soybean meal degradation in a rotationaldrum fermentation system (Chen et al., 2007). The recirculationmakes better gas yields where the pH reached an optimal valuethanks to the buffering capacity of the recycle stream (Cavinatoet al., 2011). In addition, Kobayashi et al. (2012) suggest that therecirculation of active methanogenic sludge had an inhibitive effecton the hydrogen production and supplemented the NH4

+ in the firststage reactor. Results of previous studies showed that the maineffect of the recycle stream is the alleviation of the problem oflow pH level caused by a high VFA level in the acidogenic reactor.Moreover, recirculation also affects the microbial ecology, hydrau-lic regime, and other characteristics of the operating systembecause of the sludge exchange between the two digesters, inwhich anaerobic microorganisms share nutrients and intermedi-ates (Song et al., 2004). Recirculation leads to excess VFA in the aci-dogenic reactor transfers to the methanogenic reactor. Therecirculation in turn transfers effluent from methanogenic reactorrich in alkaline and methanogens to the acidogenic reactor. Therecirculation could balance the VFA, alkalinity, and methanogensin both the reactors and thus improve the efficiency of the system(Kafle and Kim, 2011). However, the effects of recirculation in theanaerobic digestion of rapidly degrading materials such as VWwith varying recirculation rates (RR) remain unclear.

In this study, the anaerobic system was used based on the char-acteristics of VW as raw material for biogas production. The condi-tions in the process were optimized for biomethanization of VWthrough the variation in recirculation rates. The objective of thisstudy was to determine the process performance in terms of biogasproduction, methane content, pH, chemical oxygen demand (COD),VFA and its composition. The effects of RR on the characteristics ofanaerobic digestion in a two-stage process were identified.

Table 1The system operation conditions with different RRs.

Period Days Feed (L/d) COD inlet (g/L) Recirculation (L/d) RR

I 0–46 0.5 42.3 0 0II 47–90 0.5 42.3 0.3 0.6III 91–134 0.5 42.3 0.5 1IV 135–178 0.5 42.3 0.7 1.4

2. Methods

2.1. Reactors set-up and operation

The schematic diagram of laboratory-scale two-stage system isshown in Fig. 1. The hydrolysis–acidification process was carriedout in a completely stirred anaerobic reactor (CSTR, diameter16 cm, height 25 cm) and the process of methane fermentationwas performed in a fixed-bed biofilm reactor (diameter 16 cm,height 30 cm). The working volume of the reactors were 3 L and4 L, respectively. The stirring of CSTR was conducted automaticallyfor 10 min every hour at a speed of 70 rpm throughout the entireexperimental period. The fixed-bed reactor packed with activatedcarbon fiber (Yongtong Environmental Science and TechnologyCompany, Jiangsu, China) as the biofilm carrier which is preferen-tially for Methanomicrobiales adhered. Four cylindrical activatedcarbon fiber textiles (inner diameter was 6 cm; height was20 cm; thickness was 2 mm) were bundled together using a stain-less steel wire and placed in the reactor. Two ports were fitted at

the top and bottom of the reactor walls for feeding andwithdrawing.

The acidogenic reactor and methanogenic reactor were oper-ated in semi-continuous feeding mode. Firstly, 7 L sewage sludgewas added to the two reactors separately, and then diluted VWwas fed to acidogenic reactor once a day by the draw-and-fillmethod. The output of the acidogenic reactor was fed directly tothe methanogenic reactor. The overall hydraulic retention time(HRT) of the system was 14 d. The system was operated at a verylow OLR of 0.5–1.7 g volatile solids (VS)/L/d for the start-up period.

After stabilization of the start-up period, the substrates werefed for digesters with OLR of 1.7 g VS/L/d (3.4 g COD/L/d) for fourperiods. HRTs of the acidogenic reactor and methanogenic reactorwere fixed at 6 d and 8 d, respectively, without recirculation. Partof the effluent from the methanogenic reactor was recycled intothe acidogenic reactor was introduced from periods II to IV, andthe varying RRs of 0.6, 1 and 1.4 were investigated. The RR wasdefined as the ratio of the returned flow rate to that of the baseinlet flow rate (Lee et al., 2010). The system was operated at a con-stant mesophilic temperature of 37 ± 2 �C. Gas bags were used forbiogas collection from the reactors and recorded every day. Con-trolling parameters for the experiment were presented in Table 1.

2.2. Feedstock and inoculum

The VW used in this study was obtained from a market inBeijing (116.46�E, 39.92�N), China. It mainly included leafy wastematerials (e.g. cabbage, Chinese cabbage, lettuce). The raw VWwas shredded and stored in a refrigerator at �20 �C before feeding.The general chemical properties of the VW are summarized inTable 2. The seed sludge used as inoculum for the reactors was col-lected from a mesophilic anaerobic digester in XiaohongmenWastewater Treatment Plant, Beijing, China. The total solids (TS)content and volatile solids (VS) content were 5.4 ± 0.4% and2.9 ± 0.4%, respectively.

Table 2Characteristics of the feed vegetable waste.

Item Unit Average values

Total solids (TS) % 5.4 ± 0.6Volatile solids (VS) % 4.9 ± 0.6Total COD mg/g (Humid weight) 87 ± 10.5

Element compositionsC %TS 41.5 ± 0.2H %TS 5.1 ± 0.1N %TS 3.8 ± 0.1O %TS 39.6 ± 0.2

Biomass compositionsCellulose %TS 18.5 ± 0.2Hemicellulose %TS 7.9 ± 0.3Lignin %TS 3.8 ± 0.2

268 Z. Zuo et al. / Bioresource Technology 162 (2014) 266–272

2.3. Analytical methods

TS and VS were tested according to the standard methods(APHA, 1998). pH was measured using a portable Orion 3-StarpH meter. Biogas volume was measured from a gasbag by a wet-type gas flow meter (LML-1, China), and methane content was ana-lyzed by a biogas analyzer EHEIM visit 03 (Messtechnik Eheim,Germany). The effluent COD was determined by the HACH colorim-eter (DRB-200, USA), according to its standard calibration andoperation. Ethanol and VFAs (acetic acid, propionic acid,iso-butyric acid, butyric acid, iso-valeric acid, valeric acid andcaproic acid) were quantified with a gas chromatograph(Shimadzu, GC-2010 Plus, Kyoto, Japan) using a flame ionizationdetector and a capillary column type rtx-wax with extra puritynitrogen as the carrier gas at a flow of 40 mL/min, with a split ratioof 30. The column with initial temperature 60 �C (2 min holdingtime) ramp = 10 �C/min to 140 �C, ramp = 20 �C/min to 230 �C(5 min holding time). The temperatures of injector and detectorwere 230 and 250 �C, respectively. First, samples were filteredthrough 0.22 lm filters. Then the samples were acidified with for-mic acid to a pH less than 2.5, in order to convert the fatty acids totheir undissociated forms (i.e. acid forms). The concentrationswere summed as total VFA (TVFA). The TVFA/alkalinity was definedas a ratio between TVFA and alkalinity ratio which was based onthe Nordmann method (Kafle and Kim, 2011). Spectrophotometricdetermination of ammonium-nitrogen (NH4

+-N) was conductedaccording to Phenate Methods (4500-NH3 F) (APHA, 1998). Theelemental compositions of raw waste was analyzed by elementalanalyzer (Flash EA112, Thermo Electron Co., USA).

2.4. Statistical analysis

Differences in the biogas yield and methane yields amongresults were evaluated by using single factor analysis of variances(ANOVA) in Excel software 2010 (Zhang et al., 2014).

3. Results and discussion

3.1. Process performance of two-stage vegetable waste anaerobicdigestion

The two-stage reactors were operated initially in a non-recircu-lation mode for 46 d at an OLR of 1.7 g VS/L/d (3.4 g COD/L/d). AsFig. 2 shows, in period I, the average daily biogas production andmethane content in the acidogenic reactor were about 0.73 L and12.6%, and the volumetric biogas and methane production rateswere about 0.27 L/L/d and 0.04 L/L/d. After introducing recircula-tion to the two-stage VW anaerobic digestion system, noticeablevariations of methane production in the acidogenic reactor were

observed. Daily biogas production and methane content increasedin proportion to the increase of RR. From periods I to IV, daily bio-gas production increased to about 3.0 L/d, while methane contentincreased gradually to about 24.0%. The volumetric biogas andmethane production rates were about 0.97 L/L/d and 0.22 L/L/d.Moreover, the pH value, approximately 5.1 during period I,increased to 6.8 during period IV (Fig. 3). The increased methaneproduction reflected the higher activity of methanogenic bacteriain the acidogenic reactor, the result of recycled active methanogensfrom the methanogenic reactor as well as the pH adjustment. Fromperiods I to II, the TVFA/alkalinity ratio in the acidogenic reactordropped from about 11.5 to 2.9. This indicated that recirculationcould be helpful to increase the pH and alkalinity levels in the aci-dogenic reactor. In addition, constant methane production andTVFA/alkalinity ratio (<0.5) in the methanogenic reactor from peri-ods I to II were achieved. However, from periods III to IV, TVFA/alkalinity ratio in the methanogenic reactor increased to about0.6 (Fig. 3). TVFA/alkalinity ratio was used as an indicator of theprocess stability, and a ratio higher than 0.5 indicated a potentialimminent failure of the methanogenic process (Wan et al., 2011).Simultaneously, daily biogas production decreased on averagefrom 5.7 L/d to 4.9 L/d, and methane content from approximately68.6% to 62.4% in methanogenic reactor. The volumetric methaneproduction rates were decreased from about 0.94 L/L/d to 0.78 L/L/d. This decreased methane production in the methanogenic reac-tor may be attributed to the transfer of excess TVFA from the aci-dogenic reactor, which resulted in an accumulation of TVFA anddeactivation of methanogens. The results of biogas productionproved that effluent recirculation with optimizing recirculationrate helps create a favorable condition for methane production.The results were in accordance with the report of Aslanzadehet al. (2013) who suggested that a higher methane productionwas achieved in the recirculation system comparing to non-recir-culation system.

VFAs were the main soluble metabolic products of VW in theacidogenic reactor. As shown in Fig. 4, the major VFAs detectedin this study were acetic and butyric acids. Low TVFA concentra-tion in the acidogenic reactor during the non-recirculation opera-tion period may be attributed to either the low pH in theacidogenic reactor or the consumption of the alkalinity (Mohanet al., 2007). The TVFA concentration increased from about4700 mg/L to a peak of 5700 mg/L with an RR of 0.6, and thendecreased gradually to approximately 4560 mg/L (Fig. 4). Thisdemonstrated that effluent recirculation at an appropriate RRcould improve acidogenesis in the acidogenic reactor. Furthermore,recirculation causes considerable dilution in acidogenic reactors.At the start of the recirculation in period II, the COD concentrationin the acidogenic reactor dropped from approximately 20,700 mg/Lto 10,700 mg/L (Fig. 5). This decrease in COD was due to dilutioncaused by effluent recirculation and stimulated biogas production.Moreover, decreased COD concentration and increased TVFA con-centration during period II caused the increase of TVFA/COD ratio.In the methanogenic reactor, TVFA concentration was maintainedbelow 400 mg/L during periods I and II. TVFA concentration inthe methanogenic reactor increased to approximately 3000 mg/Lwith higher VFAs flux influent from the acidogenic reactor duringperiods III and IV. Based on individual VFAs profiles, their accumu-lation can be given mainly as acetic acid. This agrees with theresults reported by others on acetate concentrations between 1.5and 3.3 g/L following organic overloading (Mshandete et al.,2004). An accumulation of VFAs gave a corresponding a low buffer-ing capacity which TVFA/alkalinity ratio was about 0.6 (Fig. 3). Anaccumulation of VFAs gave a corresponding drop in pH because thesystem had a low buffering capacity. Moreover, the effluent COD inmethanogenic reactor increased gradually from about 1200 to6700 mg/L (Fig. 5). Such results indicated that the optimized RR

Dai

ly b

ioga

s pr

oduc

tion

(L/d

)

0

1

2

3

4

5

6

7Period I Period II Period III

Met

hane

con

tent

(%)

0

10

20

30

40

50

60

70

80

Acidogenic reactor Methanogenic reactor

Period IV

Time (d)

0 20 40 60 80 100 120 140 160 180

Volu

met

ric b

ioga

s pr

oduc

tion

rate

(L/L

/d)

0.0

0.5

1.0

1.5

2.0

Time (d)

0 20 40 60 80 100 120 140 160 180

Volu

met

ric C

H4 p

rodu

ctio

n ra

te (L

/L/d

)

0.0

0.2

0.4

0.6

0.8

1.0

1.2

Period I Period II Period III Period IV

Fig. 2. Daily biogas production and methane content of two-stage reactors.

Time (d)

0 20 40 60 80 100 120 140 160 180

pH

4.0

4.5

5.0

5.5

6.0

6.5

7.0

7.5

8.0

8.5

TVFA

/alk

alin

ity ra

tio

0

2

4

6

8

10

12

14

pH Acidogenic reactor A1pH Methanogenic reactor M1TVFA /alkalinity ratio A1TVFA /alkalinity ratio M1

Period I Period II Period III Period IV

Fig. 3. Profiles of pH and TVFA/alkalinity ratio of two-stage reactors with different RRs.

Z. Zuo et al. / Bioresource Technology 162 (2014) 266–272 269

VFA

conc

entra

tions

(mg/

L)

0

2000

4000

6000

8000

TVFA

con

cent

ratio

ns (m

g/L)

0

2000

4000

6000

8000

Acetic acid Propionic acid Butyric acid Iso-butyric acid Valeric acid Iso-valeric acid Caproic acid Ethanol TVFA

Time (d)

0 20 40 60 80 100 120 140 160 180

VFA

conc

entra

tions

(mg/

L)

0

1000

2000

3000

4000

TVFA

con

cent

ratio

ns (m

g/L)

0

1000

2000

3000

4000

Acidogenic reactor

Methanogenic reactor

Period I Period II Period III Period IV

Fig. 4. Variation of VFA concentrations in two-stage reactors with different RRs.

270 Z. Zuo et al. / Bioresource Technology 162 (2014) 266–272

could improve hydrolysis in acidogenic reactor. However, the VFAbalance in the overall two-stage treatment process would bealtered by the high RR. Mshandete et al. (2004) had report thathigh recirculation rate provided rapid liquid mixing betweentwo-stage reactors and this fast diffusion represents a reductionin the liquid phase mass transfer resistance around the beadswhich reduce the thickness of the laminar liquid layer past thebeads leading to decreased system efficiency. On the other hand,high RR did not provide enough time for VFAs to be degraded bythe microorganisms and hence accumulation of these took placein the methanogenic reactor. Furthermore, a relatively high recir-culation rate would cause disproportionate increase in the effec-tive loading rate of the methanogenic reactor and interactivelygive rise to a progressive increase in the organic output concentra-tion and decline in the performance (Yu et al., 2000).

The C/N ratios for VW (10.9) are lower than the numbers(20–30) suggested in literature for stable operation of the digester(Yen and Brune, 2007; Li et al., 2011). This can be explained by thefact that VW is easy biodegradable and released fast ammonium.

The NH4+-N concentration of reactors gradually increased from

approximately 700 mg/L to 1400 mg/L (Fig. 6). Although someNH4

+-N may be consumed by the growth of microorganisms, onlysmall amounts of NH4

+-N were removed from the system. Thisincrease in ammonium concentration could be the result of thestimulated the hydrolysis of protein, which may be enhanced byincreasing the RR. The methanogens were not evidently influencedwith NH4

+-N 1000–2500 mg/L (Duan et al., 2012), since this resultcoincided with the alkalinity levels in the acidogenic reactorincreased with RR (Fig. 3). The alkalinity levels increased because(1) acid is removed to methanogenic reactor or consumed bymicroorganisms; (2) protein is further degraded to ammonia, lead-ing to alkalinity.

The VS content in the methanogenic reactor increased to 5.9 g/Lwas noticeable in the period IV where COD removal decreased to86.8% (Table 3). This indicates that RR of 1.4 influence the perfor-mance of two-stage anaerobic digestion. The washout of biomasshappen during the period IV, which demonstrated in previousstudies (Erdirencelebi, 2011). The biomass lost in the effluent

Time (d)

0 20 40 60 80 100 120 140 160 180

CO

Dcr

(mg/

L)

0

5000

10000

15000

20000

25000

30000

Acidogenic reactorMethanogenic reactor

Period I Period II Period III Period IV

Fig. 5. Time profiles of effluent COD in two-stage reactors with different RRs.

Time (d)

0 20 40 60 80 100 120 140 160 180

NH

4+ -N

(mg/

L

0

500

1000

1500

2000

Acidogenic reactor, A1 Methanogenic reactor, M1 Average A1 Average M1

Period I Period II Period III Period IV

Fig. 6. Time profiles of ammonia-nitrogen concentrations in two-stage reactors (A1 and M1) with different RRs.

Table 3Summary of experimental results from reactors at different RRs.

Period VS content (g/L) The system

Acidogenic reactor Methanogenic reactor Biogas yield (L/g VS) Methane yield (L/g VS) COD removal (%) VS removal (%)

I 7.0 ± 0.5 2.2 ± 0.3 0.50 ± 0.01A 0.29 ± 0.01A 97.1 ± 0.2 91.0 ± 1.2II 4.5 ± 0.6 2.0 ± 0.5 0.55 ± 0.04B 0.33 ± 0.02B 95.0 ± 0.8 91.1 ± 2.4III 7.4 ± 0.8 1.7 ± 0.3 0.57 ± 0.04B 0.33 ± 0.03B 93.3 ± 2.1 90.9 ± 2.8IV 10.4 ± 0.3 5.9 ± 0.4 0.66 ± 0.09C 0.31 ± 0.05B 86.8 ± 3.2 77.2 ± 1.8

Data represent the mean values of samples taken during last 10 d at steady-state for each periods.The COD and VS removal were calculated by the difference between influent and effluent concentration of the system.A,B,C Means different superscript letters differ (P < 0.05) in columns.

Z. Zuo et al. / Bioresource Technology 162 (2014) 266–272 271

272 Z. Zuo et al. / Bioresource Technology 162 (2014) 266–272

was returned to the system by recycling in our study, but the deci-sion lies between overloading of methanogenic reactor or applyinglonger retention time in the system as a beneficial precaution tolessen the effects of washout.

3.2. Stoichiometry of the methane fermentation of vegetable waste

Organic matters in raw wastes can be represented withformulation of CaHbOcNd. According to the results of elementalcompositions of VW is summarized in Table 1. Organic matter inraw VW is represented with formulation of C12.74H18.79O9.12N.The theoretical methane production (TMP) can be estimatedusing Eqs. CaHbOcNd + (4a � b � 2c + 3d)/4H2O ? (4a + b � 2c �3d)/8CH4 + (4a � b + 2c + 3d)/8CO2 + dNH3 (Lin et al., 2011), TMPfor VW can be estimated as about 0.41 L CH4/g VS. As describedin Table 2, biogas yield increased from 0.50 to 0.66 L/g VS. Theenhanced biogas yield was mainly the recompense from the acido-genic reactor. Meanwhile, CH4 yield was apparently enhanced byrecirculation during periods II and III, with RRs of 0.6 and 1.0.The CH4 yield were achieved 0.33 L CH4/g VS which was about80.5% of TMP. Given the differences in biogas yield and CH4 yieldof two-stage VW anaerobic digestion at various RRs, single factorANOVA was introduced to evaluate data. ANOVA results of biogasyield showed that the non-recirculation and recirculation were sig-nificantly (P < 0.05). However, no significant differences wereobserved in RRs of 0.6 and 1.0. The difference in CH4 yield ofnon-recirculation and recirculation systems was significant, how-ever, there was a lack of significant difference with RRs of 0.6,1.0, 1.4.

4. Conclusions

j The appropriate RR could be beneficial for a two-stage vegetablewaste anaerobic digestion.

j The improved methane production in the acidogenic reactorwas an effect of the recirculation of effluent from the methano-genic reactor.

j Hydrolysis in the acidogenic reactor improved with an RR of 0.6.Recirculation can lead to dilution in the acidogenic reactor andalso influence COD and VS removal efficiency of the wholesystem.

j With an increased RR from 0 to 1.4, the biogas yield of thewhole two-stage system also increased from 0.50 L/g to0.66 L/g VS.

Acknowledgements

This work was supported by grants from the ‘‘Key Technologieson Energy Cycle Regulation in Typical Agricultural System(2012BAD14B03)’’ and National Key Technology Research andDevelopment Program of China during the 12th Five-Year Plan Per-iod (Grant No. 2012BAD47B02) of the Chinese Ministry of Scienceand Technology. We likewise greatly appreciate the critical andconstructive comments from the anonymous reviewers, whichhave helped improve this manuscript.

References

APHA, 1998. Standard Methods for the Examination of Water and Wastewater,20th ed. American Public Health Association, Washington, DC, USA.

Aslanzadeh, S., Rajendran, K., Jeihanipour, A., Taherzadeh, M.J., 2013. The effect ofeffluent recirculation in a semi-continuous two-stage anaerobic digestionsystem. Energies 6, 2966–2981.

Boe, K., Angelidaki, I., 2009. Serial CSTR digester configuration for improving biogasproduction from manure. Water Res. 43, 166–172.

Bouallagui, H., Torrijos, M., Godon, J., Moletta, R., Ben Cheikh, R., Touhami, Y.,Delgenes, J., Hamdi, M., 2004. Two-phases anaerobic digestion of fruit andvegetable wastes: bioreactors performance. Biochem. Eng. J. 21, 193–197.

Bouallagui, H., Touhami, Y., Ben Cheikh, R., Hamdi, M., 2005. Bioreactor performancein anaerobic digestion of fruit and vegetable wastes. Process Biochem. 40, 989–995.

Bouallagui, H., Lahdheb, H., Ben Romdan, E., Rachdi, B., Hamdi, M., 2009.Improvement of fruit and vegetable waste anaerobic digestion performanceand stability with co-substrates addition. J. Environ. Manage. 90, 1844–1849.

Callaghan, F., Wase, D., Thayanithy, K., Forster, C., 2002. Continuous co-digestion ofcattle slurry with fruit and vegetable wastes and chicken manure. BiomassBioenergy 22, 71–77.

Cavinato, C., Bolzonella, D., Fatone, F., Cecchi, F., Pavan, P., 2011. Optimization oftwo-phase thermophilic anaerobic digestion of biowaste for hydrogen andmethane production through reject water recirculation. Bioresour. Technol.102, 8605–8611.

Chen, L., Jiang, W., Kitamura, Y., Li, B., 2007. Enhancement of hydrolysis andacidification of solid organic waste by a rotational drum fermentation systemwith methanogenic leachate recirculation. Bioresour. Technol. 98, 2194–2200.

Duan, N., Dong, B., Wu, B., Dai, X., 2012. High-solid anaerobic digestion of sewagesludge under mesophilic conditions: feasibility study. Bioresour. Technol. 104,150–156.

Erdirencelebi, D., 2011. Treatment of high-fat-containing dairy wastewater in asequential UASBR system: influence of recycle. J. Chem. Technol. Biotechnol. 86,525–533.

Jiang, Y., Heaven, S., Banks, C., 2012. Strategies for stable anaerobic digestion ofvegetable waste. Renew. Energy 44, 206–214.

Kafle, G.K., Kim, S.H., 2011. Sludge exchange process on two serial CSTRs anaerobicdigestions: process failure and recovery. Bioresour. Technol. 102, 6815–6822.

Kobayashi, T., Xu, K.Q., Li, Y.Y., Inamori, Y., 2012. Effect of sludge recirculation oncharacteristics of hydrogen production in a two-stage hydrogen–methanefermentation process treating food wastes. Int. J. Hydrogen Energy 37, 5602–5611.

Lee, D.Y., Ebie, Y., Xu, K.Q., Li, Y.Y., Inamori, Y., 2010. Continuous H2 and CH4

production from high-solid food waste in the two-stage thermophilicfermentation process with the recirculation of digester sludge. Bioresour.Technol. 101, S42–S47.

Li, W.W., Yu, H.Q., 2011. From wastewater to bioenergy and biochemicals via two-stage bioconversion processes: a future paradigm. Biotechnol. Adv. 29, 972–982.

Li, Y., Park, S.Y., Zhu, J., 2011. Solid-state anaerobic digestion for methaneproduction from organic waste. Renew. Sustain. Energy Rev. 15, 821–826.

Lin, J., Zuo, J., Gan, L., Li, P., Liu, F., Wang, K., Chen, L., Gan, H., 2011. Effects of mixtureratio on anaerobic co-digestion with fruit and vegetable waste and food wasteof China. J. Environ. Sci. 23, 1403–1408.

Mohan, S.V., Babu, V.L., Bhaskar, Y.V., Sarma, P., 2007. Influence of recirculation onthe performance of anaerobic sequencing batch biofilm reactor (AnSBBR)treating hypersaline composite chemical wastewater. Bioresour. Technol. 98,1373–1379.

Mshandete, A., Murto, M., Kivaisi, A., Rubindamayugi, M., Mattiasson, B., 2004.Influence of recirculation flow rate on the performance of anaerobic packed-bedbioreactors treating potato-waste leachate. Environ. Technol. 25, 929–936.

Poeschl, M., Ward, S., Owende, P., 2010. Prospects for expanded utilization of biogasin Germany. Renew. Sustain. Energy Rev. 14, 1782–1797.

Sahu, P., 2004. Statistical analysis of vegetable production in India, China, and theworld. J. Veg. Crop Prod. 10, 3–9.

Schievano, A., Tenca, A., Scaglia, B., Merlino, G., Rizzi, A., Daffonchio, D., Oberti, R.,Adani, F., 2012. Two-stage vs single-stage thermophilic anaerobic digestion:comparison of energy production and biodegradation efficiencies. Environ. Sci.Technol. 46, 8502–8510.

Song, Y.C., Kwon, S.J., Woo, J.H., 2004. Mesophilic and thermophilic temperature co-phase anaerobic digestion compared with single-stage mesophilic- andthermophilic digestion of sewage sludge. Water Res. 38, 1653–1662.

Wan, C., Zhou, Q., Fu, G., Li, Y., 2011. Semi-continuous anaerobic co-digestion ofthickened waste activated sludge and fat, oil and grease. Waste Manage. 31,1752–1758.

Yen, H.W., Brune, D.E., 2007. Anaerobic co-digestion of algal sludge and waste paperto produce methane. Bioresour. Technol. 98, 130–134.

Yu, H., Wilson, F., Tay, J.H., 2000. Prediction of the effect of recirculation on theeffluent quality of anaerobic filters by empirical models. Water Environ. Res. 72,217–224.

Zhang, W., Lang, Q., Wu, S., Li, W., Bah, H., Dong, R., 2014. Anaerobic digestioncharacteristics of pig manures depending on various growth stages and initialsubstrate concentrations in a scaled pig farm in Southern China. Bioresour.Technol. 156, 63–69.