two-stage anaerobic digestion of biodegradable municipal solid waste using a rotating drum mesh...

6
Two-stage anaerobic digestion of biodegradable municipal solid waste using a rotating drum mesh filter bioreactor and anaerobic filter M. Walker * , C.J. Banks, S. Heaven Bioenergy and Organic Resources Research Group, School of Civil Engineering and the Environment, University of Southampton, University Road, Southampton SO17 1BJ, UK article info Article history: Received 7 January 2009 Received in revised form 23 March 2009 Accepted 24 March 2009 Available online 29 April 2009 Keywords: Anaerobic digestion Biodegradable municipal waste (BMW) Membrane Mesh Two-stage abstract A rotating drum mesh filter bioreactor (RDMFBR) with a 100 lm mesh coupled to an anaerobic filter was used for the anaerobic digestion of biodegradable municipal solid waste (BMW). Duplicate systems were operated for 72 days at an organic loading rate (OLR) of 7.5 gVS l 1 d 1 . Early in the experiment most of the methane was produced in the 2nd stage. This situation gradually reversed as methanogenesis became established in the 1st stage digester, which eventually produced 86–87% of the total system methane. The total methane production was 0.2 l g 1 VS added with 60–62% volatile solids destruction. No fouling was experienced during the experiment at a transmembrane flux rate of 3.5 l m 2 h 1 . The system proved to be robust and stably adjusted to a shock loading increase to 15 gVS l 1 d 1 , although this reduced the overall methane production to 0.15 l g 1 VS added . Ó 2009 Elsevier Ltd. All rights reserved. 1. Introduction There is increasing interest in anaerobic digestion (AD) as a means of stabilising biodegradable municipal waste (BMW) with the co-benefit of producing biogas as a renewable energy source (De Baere, 2006). In this work a novel high rate 1st stage rotating drum reactor was developed which incorporated a mesh filter allowing the solids and liquids retention time to be uncoupled. The purpose of this work was to explore the possibility of using a rotating drum mesh filter bioreactor (RDMFBR) in series with an anaerobic filter (AF) in a two stage configuration to provide a high rate, compact and robust AD process. The use of advanced membrane materials in membrane biore- actors (MBR) with pore sizes around 0.05–0.2 lm is growing rap- idly in full-scale applications, particularly in the wastewater treatment industry where they can show a number of advantages. These include the elimination of secondary sedimentation for bio- mass retention, and reduced energy consumption (Yang et al., 2006); and in some situations a decrease in sludge production (Bohdziewicz et al., 2008). Most commercial applications of MBR employ micro-porous membranes and are primarily used in the treatment of liquids or suspensions with very low solids such as domestic (Saddoud et al., 2007) and industrial wastewaters (Choo and Lee, 1996) and landfill leachate (Bohdziewicz et al., 2008). The use of membranes with larger pore sizes of 10–100 lm has also been reported for wastewater/sludge treatment, in what are described as mesh filter bioreactors (MFBRs). These units retain the flocculent biomass that provides the biological treatment capacity and produce an effluent similar to that of as a conven- tional activated sludge plant. This is not as high a quality as may be achieved by an MBR but the advantages are the high membrane fluxes that can be achieved, of up to 150 l m 2 h 1 at transmem- brane pressures (TMP) of 30–100 Pa, leading to reduced reactor footprint and cost (Kiso et al., 2005). High flux rates at low TMPs open up the potential for using MFBRs in the anaerobic digestion of solid materials, as a means of reducing the hydraulic retention time (HRT) relative to the solid retention time (SRT). This concept of uncoupling the solids and liq- uids retention time in the 1st stage of a two-stage system has been developed using high rate methanogenic reactors for recovery of biogas from the liquid phase. It has been used to provide greater process stability and improved substrate degradation rates for treating problematic feed materials, such as mixed abattoir wastes (Banks and Wang, 1999). Uncoupling of retention times also al- lowed increased process loadings compared to single stage reac- tors when degrading a mixture of paper and wood (Banks and Humphreys, 1998). Little work has been carried out on the poten- tial application of mesh filtration as the solids liquid separation system in the uncoupling process, except in the field of rumen inoculated anaerobic digestion where promising results were found in terms of biological performance (Gijzen et al. 1987; Dal- hoff et al., 2003). In previous work (Walker et al., 2008) a 1st stage digester with an integral nylon mesh filter cartridge was used as part of a two-stage AD process, and it was found that meshes with pore sizes in the range 30–100 lm gave good digestion 0960-8524/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.biortech.2009.03.066 * Corresponding author. E-mail address: [email protected] (M. Walker). Bioresource Technology 100 (2009) 4121–4126 Contents lists available at ScienceDirect Bioresource Technology journal homepage: www.elsevier.com/locate/biortech

Upload: m-walker

Post on 21-Jun-2016

218 views

Category:

Documents


3 download

TRANSCRIPT

Page 1: Two-stage anaerobic digestion of biodegradable municipal solid waste using a rotating drum mesh filter bioreactor and anaerobic filter

Bioresource Technology 100 (2009) 4121–4126

Contents lists available at ScienceDirect

Bioresource Technology

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

Two-stage anaerobic digestion of biodegradable municipal solid waste usinga rotating drum mesh filter bioreactor and anaerobic filter

M. Walker *, C.J. Banks, S. HeavenBioenergy and Organic Resources Research Group, School of Civil Engineering and the Environment, University of Southampton, University Road, Southampton SO17 1BJ, UK

a r t i c l e i n f o a b s t r a c t

Article history:Received 7 January 2009Received in revised form 23 March 2009Accepted 24 March 2009Available online 29 April 2009

Keywords:Anaerobic digestionBiodegradable municipal waste (BMW)MembraneMeshTwo-stage

0960-8524/$ - see front matter � 2009 Elsevier Ltd. Adoi:10.1016/j.biortech.2009.03.066

* Corresponding author.E-mail address: [email protected] (M. Wal

A rotating drum mesh filter bioreactor (RDMFBR) with a 100 lm mesh coupled to an anaerobic filter wasused for the anaerobic digestion of biodegradable municipal solid waste (BMW). Duplicate systems wereoperated for 72 days at an organic loading rate (OLR) of 7.5 gVS l�1 d�1. Early in the experiment most ofthe methane was produced in the 2nd stage. This situation gradually reversed as methanogenesis becameestablished in the 1st stage digester, which eventually produced 86–87% of the total system methane. Thetotal methane production was 0.2 l g�1 VSadded with 60–62% volatile solids destruction. No fouling wasexperienced during the experiment at a transmembrane flux rate of 3.5 l m�2 h�1. The system provedto be robust and stably adjusted to a shock loading increase to 15 gVS l�1 d�1, although this reducedthe overall methane production to 0.15 l g�1 VSadded.

� 2009 Elsevier Ltd. All rights reserved.

1. Introduction

There is increasing interest in anaerobic digestion (AD) as ameans of stabilising biodegradable municipal waste (BMW) withthe co-benefit of producing biogas as a renewable energy source(De Baere, 2006). In this work a novel high rate 1st stage rotatingdrum reactor was developed which incorporated a mesh filterallowing the solids and liquids retention time to be uncoupled.The purpose of this work was to explore the possibility of using arotating drum mesh filter bioreactor (RDMFBR) in series with ananaerobic filter (AF) in a two stage configuration to provide a highrate, compact and robust AD process.

The use of advanced membrane materials in membrane biore-actors (MBR) with pore sizes around 0.05–0.2 lm is growing rap-idly in full-scale applications, particularly in the wastewatertreatment industry where they can show a number of advantages.These include the elimination of secondary sedimentation for bio-mass retention, and reduced energy consumption (Yang et al.,2006); and in some situations a decrease in sludge production(Bohdziewicz et al., 2008). Most commercial applications of MBRemploy micro-porous membranes and are primarily used in thetreatment of liquids or suspensions with very low solids such asdomestic (Saddoud et al., 2007) and industrial wastewaters (Chooand Lee, 1996) and landfill leachate (Bohdziewicz et al., 2008).

The use of membranes with larger pore sizes of 10–100 lm hasalso been reported for wastewater/sludge treatment, in what are

ll rights reserved.

ker).

described as mesh filter bioreactors (MFBRs). These units retainthe flocculent biomass that provides the biological treatmentcapacity and produce an effluent similar to that of as a conven-tional activated sludge plant. This is not as high a quality as maybe achieved by an MBR but the advantages are the high membranefluxes that can be achieved, of up to 150 l m�2 h�1 at transmem-brane pressures (TMP) of 30–100 Pa, leading to reduced reactorfootprint and cost (Kiso et al., 2005).

High flux rates at low TMPs open up the potential for usingMFBRs in the anaerobic digestion of solid materials, as a meansof reducing the hydraulic retention time (HRT) relative to the solidretention time (SRT). This concept of uncoupling the solids and liq-uids retention time in the 1st stage of a two-stage system has beendeveloped using high rate methanogenic reactors for recovery ofbiogas from the liquid phase. It has been used to provide greaterprocess stability and improved substrate degradation rates fortreating problematic feed materials, such as mixed abattoir wastes(Banks and Wang, 1999). Uncoupling of retention times also al-lowed increased process loadings compared to single stage reac-tors when degrading a mixture of paper and wood (Banks andHumphreys, 1998). Little work has been carried out on the poten-tial application of mesh filtration as the solids liquid separationsystem in the uncoupling process, except in the field of rumeninoculated anaerobic digestion where promising results werefound in terms of biological performance (Gijzen et al. 1987; Dal-hoff et al., 2003). In previous work (Walker et al., 2008) a 1st stagedigester with an integral nylon mesh filter cartridge was used aspart of a two-stage AD process, and it was found that meshes withpore sizes in the range 30–100 lm gave good digestion

Page 2: Two-stage anaerobic digestion of biodegradable municipal solid waste using a rotating drum mesh filter bioreactor and anaerobic filter

sealed bearing

12V electric motor

gas outlet

pulley drive

sealed bearing

Rotating drum with mesh sections

effluent tube

influent tube (delivers to centre of drum)

outer casing (PVC)

temperature controlled in a water bath

Fig. 1. RDMFBR design (cross section).

4122 M. Walker et al. / Bioresource Technology 100 (2009) 4121–4126

performance when used with BMW as feedstock. The maximumorganic loading rate (OLR) that could be achieved in these stirreddigesters was 3.75 g VS l�1 d�1, due to difficulty in mechanicallystirring the digestate rather than flux limitation of the nylon meshfilter. The rotating drum mesh filter bioreactor (RDMFBR) de-scribed in the current paper was designed to overcome thislimitation.

2. Methods

A pair of two-stage digestion systems (S1 and S2) were operatedin parallel. Each system consisted of a 1st stage RDMFBR (volume1.5 l) and a 2nd stage AF reactor (volume 4 l). Each system wasoperated at mesophilic temperature and used a primary feedstockof simulated BMW which had the properties shown in Table 1. Thiswas prepared based on compositional analysis of waste collectedas part of a waste audit performed by the Resource Recovery For-um (RRF, 2001). The simulated BMW had a biochemical methanepotential (BMP) of 0.32 l g�1 VSadded (Walker et al., 2008). The li-quid effluent from the 1st stage passed forward to the AF whoseeffluent was in turn recirculated back to the RBMFBR.

2.1. System design

The RDMFBR consisted of two main parts: a rotating drum and agas tight outer casing, as shown in Figs. 1 and 2. The drum had a 2-lcapacity and was mounted within the outer casing in such a waythat it could rotate on its horizontal axis. The drum surface waspart covered with nylon mesh (100 lm pore size) supported onstainless steel mesh (�1 mm) sections of the same size, giving a to-tal filtration area of 360 cm2. Bars fixed to the inside the drum pro-vided a tumbling agitation to the digestate as the drum rotated. Apipe passing through the outer case entered the drum via a sealedbearing and allowed recirculated liquor from the AF to be deliveredto the inside of the drum. Filtrate permeated from the inner drumand was collected in the outer casing from where it drained at aconstant head level into a glass collection bottle. From there itwas pumped to the AF by a peristaltic pump via an effluent pipe.Biogas was collected in Tedlar bags (SKC Ltd. Dorset, UK) througha port in the lid of the outer casing. The casing was placed in awater bath to control the digester temperature at 37 �C. An exter-nally mounted motor was used to drive the internal drum at12 rpm.

The AF reactors had a working volume of 4 l, were filled with amixture of proprietary filtration media with specific surface area of240 m�1 (Flocor, Italy) and were operated in upflow mode at a HRTof 4 days. The reactor design is shown in Fig. 3. The reactors weremaintained at 37 �C by an external heating coil connected to athermostatically controlled circulating water heater.

2.2. Digester start up

45 g VS of BMW was added to the drum of each of the RDMFBRsand made up to a 1.5 l working volume using anaerobic digestersludge from a municipal wastewater treatment plant (Millbrook

Table 1Composition of simulated BMW.

Component

TS g TS g�1 wet 0.23 (0.01)VS g VS g�1 TS 0.85 (0.01)C % of VS 39.9 (1.7)N % of VS 1.31 (0.12)C:N – 30.5COD g COD g�1 VS 1.3 (0.1)BMP90 STP l g�1 VS 0.321 (0.027)

WWTW, Southampton, UK). The AFs had already operated over aperiod of years on a number of different substrates and did not re-quire inoculation, but had not previously been acclimated to BMWpermeate liquor.

2.3. Digester monitoring and operation

The system was operated at a 1st stage organic loading rate(OLR) of 7.5 g VS l�1 d�1 on the simulated BMW. Digestate was re-moved from the drum to maintain a constant working volume afteraddition of feed. This gave a SRT in the 1st stage of between 15 and20 days throughout the experiment. Each day 1 l of filtrate fromeach RDMFBR was fed to the AF reactor and the equivalent AFeffluent was recirculated back to the 1st stage reactor, giving a fluxthrough the mesh of 3.5 l m�2 h�1. The pair of systems operated inparallel in this way for a period of 72 days. During this time sam-ples were taken to measure gas production (daily) and composition(weekly); digestate TS and VS composition (3 �weekly); bothstage effluent VS composition (steady state), SCOD (variable fre-quency), alkalinity (steady state), VFAs (steady state) and pH(3 �weekly).

After 72 days changes were made to the operation of both of thesystems. In S1, the AF was removed and a direct liquor recycle wasused, in an attempt to see if stable digestion could be maintainedwithout the 2nd stage. In S2 the OLR was doubled to 15 g VS l�1 d�1

to assess the behaviour under higher loading conditions.

2.4. Residual specific methane potential digestate fibre/liquor mix

Samples taken from inside the drum on day 72 of the experi-mental run were placed in vials of working volume 15 ml in orderto measure the residual methane potential of the digestate. Eachheadspace purged was purged with nitrogen gas before sealing.The vials, without any additional inoculum, were incubated at35 �C in a shaking water bath for 60 days with Tedlar bags (SKCLtd., Dorset, UK) attached to collect the biogas produced.

2.5. Analytical techniques

pH was measured daily using a Jenway 3010 pH meter (Jenway,London, UK). TS and VS were measured by standard methods2540B, 2540E (APHA, 2005). SCOD was measured using a closedtube digestion method (Env.Agency, 2007). Gas volumes weremeasured a water displacement column containing acidified (pH2) tap water. Volumes were corrected for moisture content andconverted to standard temperature and pressure (273 K, 105 Pa).Composition of the gas produced by the reactors was measuredweekly using a gas analyser (Model GA 94A, Geotechnical Instru-ments, Leamington Spa, UK). Gas composition of the smaller gassamples produced in the residual methane test was made using a

Page 3: Two-stage anaerobic digestion of biodegradable municipal solid waste using a rotating drum mesh filter bioreactor and anaerobic filter

Fig. 2. RDMFBR photos with (left) and without (right) lid.

reactor outer casing (PVC)

filter medium

copper heating coils insulated casing

gas outlet

effluent tube

influent tube

Fig. 3. AF reactor design.

M. Walker et al. / Bioresource Technology 100 (2009) 4121–4126 4123

Varian CP 3800 gas chromatograph with a gas sampling loop usingargon as the carrier gas at a flow of 50 ml min�1. The chromato-graph was fitted with a Haysep C column and a molecular sieve13 � (80–100 mesh) operating at a temperature of 50 �C. Calibra-tion of both instruments was performed using a standard gas con-taining 35% CO2 and 65% CH4 (BOC, Guildford, UK).

Alkalinity was measured by titration with 0.25 N sulphuric acidto pH 4 and expressed as equivalent concentration of calcium car-bonate. Alkalinity ratio was defined as the ratio of the titre to pH5.7 and that between pH 5.7 and 4.3, which gives a good measureof the stability of the process (Ripley et al., 1986).

Volatile fatty acids were quantified in a Shimazdu 2010 gaschromatograph, using a flame ionization detector and a capillarycolumn type SGE BP 21 with helium as the carrier gas at a flowof 190.8 ml min�1, with a split ratio of 100 giving a flow rate of1.86 ml min�1 in the column and a 3.0 ml min�1 purge. The GCoven temperature was programmed to increase from 60 to210 �C in 15 min, with a final hold time of 3 min. The temperaturesof injector and detector were 200 and 250 �C, respectively. Sampleswere preserved by acidification in 10% formic acid and diluted with10% formic acid to give an appropriate concentration. Three stan-dard solutions containing 50, 250 and 500 mg l�1 of acetic, propi-onic, iso-butyric, n-butyric, iso-valeric, valeric, hexanoic andheptanoic acids were used to calibrate the instrument.

3. Results

The experimental run lasted for a total of 87 days during whichtime the RDMFBRs operated continuously without fouling. The

rotating drum was able to mix the high solids input material ap-plied at a high loading with little mechanical effort. A summaryof the main results is given in Table 2, which shows averages ofmeasurements taken between days 56 and 72. This period waschosen since at a SRT of 15–20 days most of the inoculum solidswould have been removed, and the data are likely to representthe longer term operating characteristics of the system.

The methane productions from each of the two systems andfrom the individual reactors of each pair are shown in Fig. 4. Overthe first 56 days of the trial S2 showed a lower methane productionthan S1. This was attributed to an erratic leakage of biogas from the1st stage of that system due to condensed moisture blocking thegas collection tube, which gave a back pressure sufficient to causeleakage through the sealed bearing of the outer case. The period of16 days after solution of this problem is taken as the stable opera-tional period and during this time methane production in the twosystems was very similar at 0.20 lg�1 VSadded with VS destructionsof 60–62%.

The pH of the 1st stage reactors dropped at the beginning of therun (Fig. 5) but showed a gradual recovery to reach a steady statepH value of 6.68–6.69. This pattern was reflected in the SCOD ofthe 1st stage effluent which showed an initial increase followedby a decrease.

From the methane production and SCOD measurements it waspossible to perform a COD balance around the 2nd stage AF reac-tors, and the results for this over the duration of the experimentalrun are given in Table 3. The results indicate that the total methaneproduction of the AF reactors was greater than the predicted theo-retical yield based on the SCOD values. The additional methanemust therefore have arisen from the solids present in the 1st stageeffluent. Taking this into account it can be seen that both S1 and S2gave very similar values of SMP of 0.24 and 0.23 l g�1 VSdestroyed.

The residual methane potential of the digestate taken from thereactors on day 72 of the run was measured as 0.112 and0.090 l g�1 VSadded after 60 days of incubation at 37 �C; the calcula-tions to derive this value are summarised in Table 4.

When the 1st stage digester was isolated from the AF after72 days of operation in S1 the SMP decreased to 0.15 (S.D. 0.02)l g�1 VSadded, representing a 25% loss of methane potential. Therewas, however, no evidence of process instability as pH and SCODremained stable over this period as can be seen in Fig. 5. Whenthe OLR applied to S2 was doubled to 15 g VS l�1 d�1 a decreasein SMP was observed to 0.16 (SD 0.02) l g�1 VSadded, but again withno signs of process instability.

4. Discussion

During the first 30 days of the trial, the 1st stage reactors wentthrough an acidic phase and produced an effluent with a COD of

Page 4: Two-stage anaerobic digestion of biodegradable municipal solid waste using a rotating drum mesh filter bioreactor and anaerobic filter

Table 2Summary of system performance – average process parameters between days 56 and 72.

System number S1 S2

Specific methane potential (combined 1st and 2nd stage) SMP STP l g�1 VS added 0.20 (0.04) 0.20 (0.04)Total methane produced in 1st stage % 87 (2) 88 (4)1st stage methane composition of biogas % 43 (2) 44 (1)2nd stage methane composition of biogas % 68 (5) 72 (1)SCOD concentration in 1st stage mg l�1 2101 (202) 2189 (282)VFA concentration in 1st stage mg COD l�1 15 (3) 23 (4)SCOD concentration in 2nd stage mg l�1 1710 (218) 1696 (108)VFA concentration in 2nd stage mg COD l�1 <10 <101st Stage pH 6.69 (0.05) 6.68 (0.07)2nd Stage pH 6.86 (0.05) 6.86 (0.05)TS destruction (1st stage) % 58 (4) 55 (3)VS destruction (1st stage) % 62 (4) 60 (4)2nd Stage total alkalinity mg CaCO3 l�1 3103 (164) 2779 (68)2nd Stage alkalinity ratio 0.35 (0.11) 0.32 (0.10)VS concentration in 1st stage effluent g l�1 4.51 (0.08) 4.62 (0.08)VS concentration in 2nd stage effluent g l�1 3.59 (0.01) 4.35 (0.02)

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0 20 40 60 80 100Time (Days)

Two-

stag

e SM

P (S

TP

l/gVS

adde

d)

S1 S2 S1 7-day average S2 7-day average

00.5

11.5

22.5

33.5

4

0 20 40 60 80 100Time (Days)

Met

hane

(STP

l)

RDMBR1 RBMBR2 AF1 AF2

Fig. 4. Systems and reactor individual methane production.

5.55.75.96.16.36.56.76.97.17.37.5

0 20 40 60 80 100Time (Days)

pH

RDMBR1 RDMBR2 AF1 AF2

0

1000

2000

3000

4000

5000

6000

0 20 40 60 80 100Time (Days)

CO

D (m

g/l)

RDMBR1 RDMBR2 AF1 AF2

Fig. 5. RDMFBR and AF reactor pH and effluent SCOD.

Table 3AF calculations for RBMBR trial (Period 0–72 days).

System number S1 S2

Total methane production STP l 58.1 53.2Total COD degraded (time based) g SCOD 107.3 109.7SMP of 2nd stage based on SCOD STP l g�1 SCOD degraded 0.54 0.48Theoretical SMP STP l g�1SCOD degraded 0.35 0.35Total above theoretical methane production STP l 20.4 14.4Average daily above theoretical methane production STP l day�1 0.28 0.20VS reduction in AF reactor g VS destroyed day�1 1.18 0.90SMP of 2nd stage based on VS STP l g�1 VS destroyed 0.24 0.23

4124 M. Walker et al. / Bioresource Technology 100 (2009) 4121–4126

4000–5000 mg l�1. During this time each AF reactor produced upto 1.5 l d�1 of methane. Gradually methanogenic conditions were

established in the 1st stage reactors and these began to contributea greater and greater proportion of the system methane and the AF

Page 5: Two-stage anaerobic digestion of biodegradable municipal solid waste using a rotating drum mesh filter bioreactor and anaerobic filter

Table 4Summary of residual methane potential results.

System number S1 S2

Residual SMP based on digestate VS STP l g�1 VSadded

0.122(0.011)

0.090(0.023)

Residual SMP based on input VS (assuming60% VS destruction)

STP l g�1 VSadded

0.074(0.006)

0.054(0.014)

Reactor SMP + Residual SMP based on inputVS

STP l g�1 VSadded

0.274(0.046)

0.254(0.054)

M. Walker et al. / Bioresource Technology 100 (2009) 4121–4126 4125

reactor methane production decreased to bellow 0.5 l d�1. At thesame time the effluents from the 1st stage were reduced instrength leading to decreasing methane production in the 2ndstage. This effect has been observed in other continuously-oper-ated two-stage systems in which effluent from the 2nd stage isrecycled to the 1st stage (Walker et al., 2007). It has also been re-ported in leach bed reactors where recirculation from a methano-genic reactor is known to provide the inoculum, nutrients andalkalinity that can lead to the establishment of methanogenic con-ditions (Chynoweth et al., 1992; Lai et al., 2001).

In both S1 and S2 once methanogenic conditions had beenestablished the concentration of VFA in the 1st stage effluentswas very low and other process parameters were very stable. Thesystem was robust and was able to withstand an initial load of7.5 g VS l�1 d�1 without an acclimatisation period. This is believedto be because the initial acid production in the 1st stage reactors isconverted rapidly to methane in the 2nd stage anaerobic filters andthe alkalinity is retained and recycled back to the 1st stage. With-out the 2nd stage being present initially it is likely that the loss ofalkalinity, the increase in acid production and the drop in pHwould lead to an unrecoverable digester failure.

The COD balance around the AF reactors (Table 3) showed agood agreement between methane production and apparent VSSremoval. There were some differences between S1 and S2 withthe filter of S1 showing a slightly higher methane production,but if the available SCOD was fully converted to methane and thesurplus methane production is attributed to VSS destruction bothsystems degraded similar amounts of VSS. This does not mean thatthe VSS was necessarily fully degraded in the AF reactors, but sim-ply that the proportion was similar in both systems. It is probablethat there was an accumulation of suspended solids inside the AFreactors, as the apparent VS destruction was higher than expected.This is based on the system methane production being 63% of theBMP while the BMP test itself found only 62% of the feed materialto be degraded. Solids accumulation in the AF reactors caused noproblems over the 86 days of the experimental run, but couldpotentially reach unmanageable levels in the longer term.

The methane production potential of the residual digestatewhen tested under batch conditions showed that 25–35% moremethane could be produced; this indicates that the full hydrolysisand acidification potential was not achieved in the RDMFBRs. Theresults shown in Table 4 suggests that when the methane produc-tion potential of the systems S1 and S2 are combined with the dig-estate residual methane potential the overall production would be0.25–0.27 l g�1 VSadded which is still lower than the 0.32 l g�1

VSadded as measured as the BMP of the feed.The increase in OLR to 15 g l�1 d�1 in S2 showed a reduction in

SMP from 0.20 to 0.15 l g�1 VSadded with no apparent process insta-bility. The loss of biogas productivity indicated that hydrolysis andacidification of the solid substrate was limiting at this loading. Re-moval of the 2nd stage filter in S1 showed a reduction in SMP from0.20 to 0.16 l g�1 VSadded but again no apparent instability. The 2ndstage is obviously important during the start up phase but oncemethanogenesis becomes established in the 1st stage the systemis less reliant on the 2nd stage for inoculation and/or removal of

SCOD. The 2nd stage at this point acts principally as a scavengerof any remaining SCOD and could be reduced in size to match thatrequirement. The responses to these two major process changesagain highlight the robustness of the system.

4.1. Filtration

The only TMP applied to the meshes in the drum reactors wasthe digestate hydrostatic head (4–5 cm) with some dynamic pres-sure head caused by the tumbling action on the digestate. Contin-uous filtration of the digestate was possible for the duration of thetrial at both loading rates of 7.5 and 15 g VS l�1 d�1. Because therewas no direct way to control the flux rate in the RDMFBRs, the fil-tration occurred at the natural rate under the rotating drum re-gime. The pumping speed was set to approximate the flow rateof the flux. The RDMFBRs had no mesh scouring mechanism, andsimply relied on the weight and structure of the digestate to allowtumbling under the rotation of the drum. For these reasons the fluxthrough the meshes on the RDMFBRs settled at around3.5 l m�2 h�1 compared to 44 l m�2 h�1 found in previous workwhere similar nylon meshes were used in a different configuration(Walker et al., 2008).

In practice, to further develop the RDMFBR system at a largerscale would require filtration at higher flux rate, achievable byintroducing pressure control and a backwash system to clean themesh. This would be necessary as the area of the drum only in-creases with an exponent of 2/3 relative to the volume, meaningthe useful mesh surface becomes smaller in comparison to the re-quired flux to give the appropriate HRT. For example a 500 m3

rotating drum digester would have a filtration surface area of147 m2 (assuming 2:1 aspect ratio of the cylinder and 50% filtra-tion surface area) and the flux required to give a 1.5 day HRT is94 l m�2 h�1. It may therefore be the case that the flux limitationrestricts the scale at which this type of reactor could operate; e.g.the previously achieved flux of 44 l m�2 h�1 would mean a maxi-mum reactor volume of 50 m3 (49.94) to maintain an HRT of1.5 days. In a reactor of this size an OLR of 7.5 g VS l�1 d�1 wouldlimit the total dry solids load applied to 436 kg TS d�1. A fullyinstrumented filtration system with backwashing and TMP controlmay, however, be able to sustain greater fluxes than were observedin the simple laboratory-scale equipment used. For example, in asimilar system Dalhoff et al. (2003) used a 1 lm polyethylene sub-merged membrane unit to filter digestate in which grass was thesubstrate, and optimisation of gas backwashing system allowedthe flux through the membrane to be increased from approxi-mately 20 to 120 l m�2 h�1.

4.2. Potential application of the RDMFBR system

The results from these experiments showed that a two-stageRDMFBR/AF system had good process stability at a high OLR. Themethane production was, however, only 63% of the BMP for thefeedstock. This could limit the potential for application of thetwo-stage system if the main source of income is from the methanegenerated. If gate fees for waste disposal are the main economicdriver then the system offers a very compact process with poten-tially low capital costs. A further alternative would be to add a3rd stage reactor to treat the residual solids, thus maximising bio-gas production. The system design would also have to considerprotection of the mesh surfaces, as these could be damaged by hea-vy and sharp materials. Removal of contaminants such as metal,glass and tough biodegradable materials such as wood would re-quire pre-treatment processes which would increase the capitalcost and overall process energy requirement. In this work the ny-lon mesh was surrounded on both sides by stainless steel with amuch larger mesh size, and a similar strategy could be employed

Page 6: Two-stage anaerobic digestion of biodegradable municipal solid waste using a rotating drum mesh filter bioreactor and anaerobic filter

4126 M. Walker et al. / Bioresource Technology 100 (2009) 4121–4126

at a larger scale. Pulping the feedstock material was also effectiveas this reduced the particle size of the organic fraction to a pointwhere damage to mesh surfaces was unlikely, and would allowthe densitometric removal of grit and other abrasive inorganicparticles.

The biological robustness of the two-stage system may be itsmost valuable characteristic as the process should lend itself tothe digestion of potentially problematic feed materials such asthose that undergo rapid breakdown e.g. fruit and vegetable wasteand food waste with a naturally low buffering capacity that re-stricts the loading which can be applied to conventional digestionsystems.

5. Conclusions

The RDMFBR was able to maintain continuous filtration over an86-day period and provide an effective means of uncoupling thesolids and liquid retention time in the 1st stage reactor, allowingthe system to operate at an OLR of 7.5 g VS l�1 d�1 when usingSBMW. The rotating drum provided a means of mixing the wasteat this high OLR which was not possible in a previously usedmechanically stirred digester design (Walker et al., 2008). The mix-ing energy in the drum system was also considerably reduced com-pared to other mixing systems. Sufficient data was obtained fromthe experimental runs to provide a mass balance of solids and toaccount for the methane production from both SCOD and VSdestruction. The specific methane yield was 0.20 l g�1 VSadded; thiswas lower than the BMP of the substrate but testing of the meth-ane potential of the process residual accounted for a further 25%of the BMP. Operating the RDMFBR as a single stage process anddoubling the OLR on the two stage process both showed a reduc-tion in methane yield but without signs of process instability.

Scale-up of the process may be limited by the TMP flux which,with this substrate and a 100 lm mesh, may be limited to around44 l m�2 h�1. This would restrict the reactor working volume to amaximum of around 50 m3. It may be possible to increase themembrane flux but even so other feed related factors may limitthe application to problematic feed materials or where a high or-ganic loading is required in order to achieve a compact design.

References

APHA, 2005. Standard Methods for the Examination of Water and Wastewater.American Technical Publishers.

Banks, C.J., Humphreys, P.N., 1998. The anaerobic treatment of a ligno-cellulosicsubstrate offering little natural pH buffering capacity. Water Science andTechnology 38 (4–5), 29–35.

Banks, C.J., Wang, Z., 1999. Development of a two phase anaerobic digester for thetreatment of mixed abattoir wastes. Water Science and Technology 40 (1), 69–76.

Bohdziewicz, J., Neczaj, E., et al., 2008. Landfill leachate treatment by means ofanaerobic membrane bioreactor. Desalination 221 (1–3), 559–565.

Choo, K.-H., Lee, C.-H., 1996. Membrane fouling mechanisms in the membrane-coupled anaerobic bioreactor. Water Research 30 (8), 1771–1780.

Chynoweth, D.P., Owens, J., et al., 1992. Sequential batch anaerobic composting ofthe organic fraction of municipal solid waste. Water Science and Technology 25(7), 327–339.

Dalhoff, R., Rababah, A., et al., 2003. Membrane separation to improve degradationof road side grass by rumen enhance solid incubation. Water Science andTechnology 48 (4), 163–168.

De Baere, L., 2006. Will anaerobic digestion of solid waste survive in the future.Water Science and Technology 53 (8), 187–194.

Env.Agency, 2007. UK Environment Agency – the determination of chemical oxygendemand in waters and effluents.

Gijzen, H.J., Lubberding, F.J., et al., 1987. Application of rumen microorganisms foran enhanced anaerobic digestion of solid organic waste materials. BiologicalWastes 22, 81–95.

Kiso, Y., Jung, Y.-J., et al., 2005. Coupling of sequencing batch reactor and meshfiltration: operational parameters and wastewater treatment performance.Water Research 39 (20), 4887–4898.

Lai, T.E., Nopharatana, A., et al., 2001. Cellulolytic activity in leachate during leach-bed anaerobic digestion of municipal solid waste. Bioresource Technology 80(3), 205–210.

Ripley, L.E., Boyle, W.C., et al., 1986. Improved alkalimetric monitoring for anaerobicdigestion of high-strength wastes. Journal of the Water Pollution ControlFederation 58 (5), 406–411.

RRF, 2001. Assessment of kerbside collection schemes of dry recylables, resourcerecovery forum (RRF).

Saddoud, A., Ellouze, M., et al., 2007. Anaerobic membrane bioreactor treatment ofdomestic wastewater in Tunisia. Desalination 207 (1–3), 205–215.

Walker, M., Banks, C. J., et al., 2007. Effects of pH control and methanogenic effluentrecycle on the two-stage anaerobic digestion of OFMSW. In: 11th IWA WorldCongress on Anaerobic Digestion, Brisbane, Australia, IWA.

Walker, M., Banks, C.J., et al., 2008. Development of a coarse membrane bioreactorfor two stage anaerobic digestion of biodegradable municipal solid waste. In:5th International Symposium of Anaerobic Digestion of Solid Waste and EnergyCrops, Hammamet, Tunisia, IWA.

Yang, W., Cicek, N., et al., 2006. State-of-the-art of membrane bioreactors:worldwide research and commercial applications in North America. Journal ofMembrane Science 270 (1–2), 201–211.