《生物质能工程》课程教学资源（阅读材料）4-Anaerobic digestion of source-segregated

Bioresource Technology 102 (2011)612-620 Contents lists available at ScienceDirect BIORESOURCE TECHNǒOGY Bioresource Technology ELSEVIER journal homepage:www.elsevier.com/locate/biortech Anaerobic digestion of source-segregated domestic food waste:Performance assessment by mass and energy balance Charles J.Banks2,Michael Chesshireb,Sonia Heaven,Rebecca Arnoldb School of Civil Engineering and the Environment,University of Southampton.Southampton S017 1BJ.UK BiogenGreenfinch,The Business Park,Coder Road,Ludlow SY8 1XE,UK ARTICLE INFO ABSTRACT Article history: An anaerobic digester receiving food waste collected mainly from domestic kitchens was monitored over Received 14 July 2010 a period of 426 days.During this time information was gathered on the waste input material.the biogas Received in revised form 1 August 2010 production,and the digestate characteristics.A mass balance accounted for over 90%of the material Accepted 2 August 2010 Available online 6 August 2010 entering the plant leaving as gaseous or digestate products.A comprehensive energy balance for the same period showed that for each tonne of input material the potential recoverable energy was 405 kWh.Bio- gas production in the digester was stable at 642 m3tonne-1VS added with a methane content of around Keywords: 62%.The nitrogen in the food waste input was on average 8.9 kg tonne-1.This led to a high ammonia con- Anaerobic digestion Food waste centration in the digester which may have been responsible for the accumulation of volatile fatty acids Energy that was also observed. Biogas 2010 Elsevier Ltd.All rights reserved. Mass balance 1.Introduction and industry in methods of processing source segregated house- hold food waste by the anaerobic digestion route. Many examples exist on the use of anaerobic digestion (AD)to There are,however.reasons why food waste has not been pop- treat the mechanically separated biodegradable fraction of munici- ular in the past as a single substrate,since digestion of this energy- pal waste.Both 'wet'and 'dry'anaerobic technologies have been rich material can lead to operational problems.The protein content used as part of mechanical-biological treatment (MBT)(Mata of food waste typically gives a high nitrogen content on hydrolysis, Alvarez,2003).There are also examples of the processing of mixed which leads to elevated concentrations of ammonia or ammonium source segregated biodegradable wastes such as kitchen and gar- ion in the digester.The distribution of the two species and their den wastes(Archer et al.,2005):but there are few reports of AD relative toxicity is pH dependent,with the more toxic form domi- plants operating entirely on source segregated household food nating at higher pH(Mata-Alvarez,2003).There is still uncertainty waste.Interest in this approach is growing within Europe due to concerning the concentration at which ammonia becomes inhibi- rising energy costs associated with the processing of wet waste. tory to methanogenesis,and this is reflected in the various limit the requirement to meet the diversion targets of the EU Landfill values given in recent literature.According to Mata-Alvarez directive(99/31/EC),and the need to comply with regulations for (2003),inhibition occurs at total ammonia concentrations of the disposal of animal by-products (EC 1774/2002).When AD is 1200 mgl-1 and above.Hartmann and Ahring (2005)showed used to process source segregated waste it not only produces bio- ammonia inhibition begins at free ammonia concentrations above gas,but also presents an opportunity to recover additional value 650 mg I'NH3-N,whereas Angelidaki et al.(2005)in a study of 18 from the waste material,in the form of a quality assured nutri- full-scale biogas plants in Denmark co-digesting manure and or- ent-rich fertiliser product that can applied to agricultural land used ganic waste only found decreases in efficiency when total ammo- in food production.If the waste is not source segregated and the nia was above 4000 mg NH3-NI-1.El Hadj et al.(2009)found organic fraction is recovered through a MBT plant,regulations in that methane generation in batch tests with a high-protein syn- many European countries do not permit the digestate product to thetic biowaste under mesophilic conditions fell by 50%at ammo- be used on land in this way (Stretton-Maycock and Merrington, nium ion concentrations of 3860 mg NH-N 1-1.Although 2009).Consequently,there is strong interest from government ammonia has been shown to create operational difficulties in anaerobic digesters,it is also recognised that populations can accli- mate,making it difficult to predict the exact concentration at Corresponding author.Tel.:+44(0)2380 594650:fax:+44 (0)2380 677519 which process instability or failure may occur (Fricke et al.. E-mail address:cjb@soton.ac.uk (C.J.Banks). 2007).It has been reported on a number of occasions that digestion 0960-8524/$-see front matter2010 Elsevier Ltd.All rights reserved. doi:10.1016j.biortech.2010.08.005

Anaerobic digestion of source-segregated domestic food waste: Performance assessment by mass and energy balance Charles J. Banks a, *, Michael Chesshire b , Sonia Heaven a , Rebecca Arnold b a School of Civil Engineering and the Environment, University of Southampton, Southampton SO17 1BJ, UK b BiogenGreenfinch, The Business Park, Coder Road, Ludlow SY8 1XE, UK article info Article history: Received 14 July 2010 Received in revised form 1 August 2010 Accepted 2 August 2010 Available online 6 August 2010 Keywords: Anaerobic digestion Food waste Energy Biogas Mass balance abstract An anaerobic digester receiving food waste collected mainly from domestic kitchens was monitored over a period of 426 days. During this time information was gathered on the waste input material, the biogas production, and the digestate characteristics. A mass balance accounted for over 90% of the material entering the plant leaving as gaseous or digestate products. A comprehensive energy balance for the same period showed that for each tonne of input material the potential recoverable energy was 405 kWh. Bio￾gas production in the digester was stable at 642 m3 tonne1 VS added with a methane content of around 62%. The nitrogen in the food waste input was on average 8.9 kg tonne1 . This led to a high ammonia con￾centration in the digester which may have been responsible for the accumulation of volatile fatty acids that was also observed. 2010 Elsevier Ltd. All rights reserved. 1. Introduction Many examples exist on the use of anaerobic digestion (AD) to treat the mechanically separated biodegradable fraction of munici￾pal waste. Both ‘wet’ and ‘dry’ anaerobic technologies have been used as part of mechanical–biological treatment (MBT) (Mata￾Alvarez, 2003). There are also examples of the processing of mixed source segregated biodegradable wastes such as kitchen and gar￾den wastes (Archer et al., 2005); but there are few reports of AD plants operating entirely on source segregated household food waste. Interest in this approach is growing within Europe due to rising energy costs associated with the processing of wet waste, the requirement to meet the diversion targets of the EU Landfill directive (99/31/EC), and the need to comply with regulations for the disposal of animal by-products (EC 1774/2002). When AD is used to process source segregated waste it not only produces bio￾gas, but also presents an opportunity to recover additional value from the waste material, in the form of a quality assured nutri￾ent-rich fertiliser product that can applied to agricultural land used in food production. If the waste is not source segregated and the organic fraction is recovered through a MBT plant, regulations in many European countries do not permit the digestate product to be used on land in this way (Stretton-Maycock and Merrington, 2009). Consequently, there is strong interest from government and industry in methods of processing source segregated house￾hold food waste by the anaerobic digestion route. There are, however, reasons why food waste has not been pop￾ular in the past as a single substrate, since digestion of this energy￾rich material can lead to operational problems. The protein content of food waste typically gives a high nitrogen content on hydrolysis, which leads to elevated concentrations of ammonia or ammonium ion in the digester. The distribution of the two species and their relative toxicity is pH dependent, with the more toxic form domi￾nating at higher pH (Mata-Alvarez, 2003). There is still uncertainty concerning the concentration at which ammonia becomes inhibi￾tory to methanogenesis, and this is reflected in the various limit values given in recent literature. According to Mata-Alvarez (2003), inhibition occurs at total ammonia concentrations of 1200 mg l1 and above. Hartmann and Ahring (2005) showed ammonia inhibition begins at free ammonia concentrations above 650 mg l1 NH3-N, whereas Angelidaki et al. (2005) in a study of 18 full-scale biogas plants in Denmark co-digesting manure and or￾ganic waste only found decreases in efficiency when total ammo￾nia was above 4000 mg NH3-N l1 . El Hadj et al. (2009) found that methane generation in batch tests with a high-protein syn￾thetic biowaste under mesophilic conditions fell by 50% at ammo￾nium ion concentrations of 3860 mg NHþ 4 -N l1 . Although ammonia has been shown to create operational difficulties in anaerobic digesters, it is also recognised that populations can accli￾mate, making it difficult to predict the exact concentration at which process instability or failure may occur (Fricke et al., 2007). It has been reported on a number of occasions that digestion 0960-8524/$ - see front matter 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.biortech.2010.08.005 * Corresponding author. Tel.: +44 (0)2380 594650; fax: +44 (0)2380 677519. E-mail address: cjb@soton.ac.uk (C.J. Banks). Bioresource Technology 102 (2011) 612–620 Contents lists available at ScienceDirect Bioresource Technology journal homepage: www.elsevier.com/locate/biortech

CJ.Banks et al/Bioresource Technology 102 (2011)612-620 613 at high ammonia concentrations can give stable biogas production and type of waste was recorded.Water usage was monitored by at alkaline pH over extended periods of time under continuous separate meters,one for the industrial process water and another loading conditions.In digesters treating food waste these condi- for staff facilities (e.g.toilets and washrooms). tions can also lead to operation at elevated levels of volatile fatty acids in the digestate (Banks et al.,2008:Neiva Correia et al., 2.2.2.Biogas sampling analysis and quantification 2008).Similar conditions have been reported in thermophilic cattle A biogas sample was taken daily from the gas holder feeding the slurry digesters(Neilsen and Angelidaki,2008),and in other nitro- CHP and analysed for methane and carbon dioxide content using a gen rich substrates such as slaughterhouse waste (Banks and GA2000 portable infrared gas analyser(Geotechnical instruments. Wang.1999;Wang and Banks 2003). Leamington Spa,UK).Biogas volumes were recorded on an indus- The current work presents the results of a mass and energy bal- trial gas flow meter,and readings were manually adjusted for ance over a 14-month period for a full-scale food waste digester water vapour content and expressed at standard temperature operating at high ammonia and VFA concentrations.During the and pressure(STP)of 273.15 K and 101.325 kPa. study period the digester was fed mainly on food waste collected from domestic properties mixed with small amounts of commer- 2.2.3.Waste input sampling and analysis cial food waste and municipal green waste.Since the study was Daily composite samples of the shredded feedstock were taken completed the plant has continued to operate successfully as a for analysis of the total solids (TS)and volatile solids(VS)content commercial facility processing food waste. according to Standard Method 2540 G (APHA,2005).Further com- posites were prepared from the daily composites over two-week 2.Methods periods for determination of Total Kjeldahl Nitrogen(N),phospho- rus(P),and potassium(K).Total Kjeldahl N was determined using a 2.1.Digestion plant Kjeltech block digestion and steam distillation unit according to the manufacturer's instructions (Foss Ltd.,Warrington,UK).Sam- The plant was commissioned in March 2006 and for the first ples for Potassium and Phosphorus were extracted using concen- 9 months of operation was fed on mixed kitchen and garden waste trated HNOs in a CEM Microwave Accelerated Reaction System collected from domestic properties.From January 2007 the feed for Extraction (MARSX)(CEM Corporation,North Carolina,USA). was gradually switched to source segregated food waste only. Potassium was quantified using a Varian Spectra AA-200 atomic The study period began on 1 June,2007(day 0).and data for the absorption spectrophotometer (Varian,Australia)according to mass and energy balances was collected for 426 days.During the the manufacturer's instructions.Phosphorus was measured spec- study 3936 tonnes of waste were processed of which 95.5%was trophotometrically by the ammonium molybdate method (ISO source-segregated domestic food waste,with the remainder con- 6878:2004) sisting of commercial food waste from restaurants and local busi- nesses (2.9%)including a small amount of whey,and grass 2.2.4.Digester and digestate sampling and analysis cuttings (1.6%).The food waste received at the plant was first Samples of digestate were taken on a regular basis for analysis. shredded in a rotary counter-shear shredder to reduce the particle Total and volatile solids were measured as above.Ammonia was size,then passed to a feed preparation vessel where it was mixed determined using a Kjeltech steam distillation unit according to with recirculated whole digestate and macerated to give a particle the manufacturer's instructions (Foss Ltd.,Warrington,UK).VFA size less than 12 mm.The feed to the digester was via a buffer stor- were quantified in a Shimazdu GC-2010 gas chromatograph,using age tank providing 3 days storage,to allow continuous feeding over a flame ionization detector and a capillary column type SGE BP-21 weekends and public holidays.The digester itself was a 900 m with helium as the carrier gas at a flow of 190.8 ml min-,with a tank that was completely mixed by continuous gas recirculation split ratio of 100 giving a flow rate of 1.86 ml min-in the column and maintained at 42C by external heat exchangers:the choice and a 3.0 ml min-purge.The GC oven temperature was pro- of temperature was based on the previous experience and prefer- grammed to increase from 60 to 210C in 15 min,with a final hold ence of the plant operator.The digestate was passed batch-wise time of 3 min.The temperatures of injector and detector were 200 to a pasteurisation tank(60 m)where it was heated to 70C for and 250 C,respectively.Samples were prepared by acidification in a minimum of 1 h.Pasteurised digestate was transferred to the dig- 2%formic acid.A standard solution containing acetic,propionic, estate storage tank(900 m3),where it was kept until being ex- iso-butyric,n-butyric,iso-valeric,valeric,hexanoic and heptanoic ported to local farms for use on agricultural land as either acids,at three dilutions giving individual acid concentrations of separated fibre,liquor or whole digestate.The biogas generated 50.250 and 500 mgI-1,respectively,was used for calibration. was used to produce electricity using a 195 kW MAN Combined Alkalinity was measured by titration using 0.25 N H2SO4 to end- Heat and Power(CHP)unit with an assumed electrical conversion points of 5.7 and 4.3(Ripley et al.,1986).Digestate pH was mea- efficiency of 32%at full load and a potential for 53%recovery of heat sured using a combination glass electrode and meter calibrated via the jacket and exhaust cooling water streams.Electricity pro- in buffers at pH 4,7 and 9. duced by the CHP and imports and exports to the grid were all me- tered.The power requirements of the plant were calculated from 3.Results and discussion (CHP generator meter grid import meter-grid export meter). Some of the heat produced by the CHP was fed back into the pro- cess.Temperatures in all tanks were recorded continuously using 3.1.Feedstock characteristics,organic loading rate and retention time a SCADA.More detailed descriptions of individual components of Fig.1 shows values for TS and VS throughout the study period the plant are given in Chesshire(2007)and Arnold et al.(2010). for the domestic food waste and the commercial food waste (not including whey)components of the feedstock.The average solids 2.2.Sampling,measurement and analysis content was similar for domestic food waste(TS 27.7%,VS 24.4%) and commercial food waste (TS 27.8%,VS 24.3%).As can be seen 2.2.1.Quantification of input waste and other materials in Fig.1a and b,there was some variation in the TS and VS content All vehicles delivering waste to the plant were weighed on a of individual samples of domestic food waste but no strong evi- weighbridge before and after discharging their load.The origin dence of seasonal variation and the VS:TS ratio remained fairly

at high ammonia concentrations can give stable biogas production at alkaline pH over extended periods of time under continuous loading conditions. In digesters treating food waste these condi￾tions can also lead to operation at elevated levels of volatile fatty acids in the digestate (Banks et al., 2008; Neiva Correia et al., 2008). Similar conditions have been reported in thermophilic cattle slurry digesters (Neilsen and Angelidaki, 2008), and in other nitro￾gen rich substrates such as slaughterhouse waste (Banks and Wang, 1999; Wang and Banks 2003). The current work presents the results of a mass and energy bal￾ance over a 14-month period for a full-scale food waste digester operating at high ammonia and VFA concentrations. During the study period the digester was fed mainly on food waste collected from domestic properties mixed with small amounts of commer￾cial food waste and municipal green waste. Since the study was completed the plant has continued to operate successfully as a commercial facility processing food waste. 2. Methods 2.1. Digestion plant The plant was commissioned in March 2006 and for the first 9 months of operation was fed on mixed kitchen and garden waste collected from domestic properties. From January 2007 the feed was gradually switched to source segregated food waste only. The study period began on 1 June, 2007 (day 0), and data for the mass and energy balances was collected for 426 days. During the study 3936 tonnes of waste were processed of which 95.5% was source-segregated domestic food waste, with the remainder con￾sisting of commercial food waste from restaurants and local busi￾nesses (2.9%) including a small amount of whey, and grass cuttings (1.6%). The food waste received at the plant was first shredded in a rotary counter-shear shredder to reduce the particle size, then passed to a feed preparation vessel where it was mixed with recirculated whole digestate and macerated to give a particle size less than 12 mm. The feed to the digester was via a buffer stor￾age tank providing 3 days storage, to allow continuous feeding over weekends and public holidays. The digester itself was a 900 m3 tank that was completely mixed by continuous gas recirculation and maintained at 42 C by external heat exchangers: the choice of temperature was based on the previous experience and prefer￾ence of the plant operator. The digestate was passed batch-wise to a pasteurisation tank (60 m3 ) where it was heated to 70 C for a minimum of 1 h. Pasteurised digestate was transferred to the dig￾estate storage tank (900 m3 ), where it was kept until being ex￾ported to local farms for use on agricultural land as either separated fibre, liquor or whole digestate. The biogas generated was used to produce electricity using a 195 kW MAN Combined Heat and Power (CHP) unit with an assumed electrical conversion efficiency of 32% at full load and a potential for 53% recovery of heat via the jacket and exhaust cooling water streams. Electricity pro￾duced by the CHP and imports and exports to the grid were all me￾tered. The power requirements of the plant were calculated from (CHP generator meter + grid import meter grid export meter). Some of the heat produced by the CHP was fed back into the pro￾cess. Temperatures in all tanks were recorded continuously using a SCADA. More detailed descriptions of individual components of the plant are given in Chesshire (2007) and Arnold et al. (2010). 2.2. Sampling, measurement and analysis 2.2.1. Quantification of input waste and other materials All vehicles delivering waste to the plant were weighed on a weighbridge before and after discharging their load. The origin and type of waste was recorded. Water usage was monitored by separate meters, one for the industrial process water and another for staff facilities (e.g. toilets and washrooms). 2.2.2. Biogas sampling analysis and quantification A biogas sample was taken daily from the gas holder feeding the CHP and analysed for methane and carbon dioxide content using a GA2000 portable infrared gas analyser (Geotechnical instruments, Leamington Spa, UK). Biogas volumes were recorded on an indus￾trial gas flow meter, and readings were manually adjusted for water vapour content and expressed at standard temperature and pressure (STP) of 273.15 K and 101.325 kPa. 2.2.3. Waste input sampling and analysis Daily composite samples of the shredded feedstock were taken for analysis of the total solids (TS) and volatile solids (VS) content according to Standard Method 2540 G (APHA, 2005). Further com￾posites were prepared from the daily composites over two-week periods for determination of Total Kjeldahl Nitrogen (N), phospho￾rus (P), and potassium (K). Total Kjeldahl N was determined using a Kjeltech block digestion and steam distillation unit according to the manufacturer’s instructions (Foss Ltd., Warrington, UK). Sam￾ples for Potassium and Phosphorus were extracted using concen￾trated HNO3 in a CEM Microwave Accelerated Reaction System for Extraction (MARSX) (CEM Corporation, North Carolina, USA). Potassium was quantified using a Varian Spectra AA-200 atomic absorption spectrophotometer (Varian, Australia) according to the manufacturer’s instructions. Phosphorus was measured spec￾trophotometrically by the ammonium molybdate method (ISO 6878: 2004). 2.2.4. Digester and digestate sampling and analysis Samples of digestate were taken on a regular basis for analysis. Total and volatile solids were measured as above. Ammonia was determined using a Kjeltech steam distillation unit according to the manufacturer’s instructions (Foss Ltd., Warrington, UK). VFA were quantified in a Shimazdu GC-2010 gas chromatograph, using a flame ionization detector and a capillary column type SGE BP-21 with helium as the carrier gas at a flow of 190.8 ml min1 , with a split ratio of 100 giving a flow rate of 1.86 ml min1 in the column and a 3.0 ml min1 purge. The GC oven temperature was pro￾grammed to increase from 60 to 210 C in 15 min, with a final hold time of 3 min. The temperatures of injector and detector were 200 and 250 C, respectively. Samples were prepared by acidification in 2% formic acid. A standard solution containing acetic, propionic, iso-butyric, n-butyric, iso-valeric, valeric, hexanoic and heptanoic acids, at three dilutions giving individual acid concentrations of 50, 250 and 500 mg l1 , respectively, was used for calibration. Alkalinity was measured by titration using 0.25 N H2SO4 to end￾points of 5.7 and 4.3 (Ripley et al., 1986). Digestate pH was mea￾sured using a combination glass electrode and meter calibrated in buffers at pH 4, 7 and 9. 3. Results and discussion 3.1. Feedstock characteristics, organic loading rate and retention time Fig. 1 shows values for TS and VS throughout the study period for the domestic food waste and the commercial food waste (not including whey) components of the feedstock. The average solids content was similar for domestic food waste (TS 27.7%, VS 24.4%) and commercial food waste (TS 27.8%, VS 24.3%). As can be seen in Fig. 1a and b, there was some variation in the TS and VS content of individual samples of domestic food waste but no strong evi￾dence of seasonal variation and the VS:TS ratio remained fairly C.J. Banks et al. / Bioresource Technology 102 (2011) 612–620 613

614 CJ.Banks et aL/Bioresource Technology 102 (2011)612-620 0.5 a 0.4 0.3 ·TS VS -21-day TS 。 21-day VS 0.0 0 100 200 300 400 Day 0.5 0.5 b C 0.4 0.4 03 0.3 ⊙0.2 0.2 y=0.796x+0.023 y=0.884x-0.002 0.1 R2=0.881 0.1 R2=0.970 0.0 0.0 0.0 0.1 0.2 0.3 0.4 0.5 .0 0.1 0.2 0.3 0.4 0.5 Domestic food waste TS.gg Commercial food waste TS.gg Fig.1.TS and VS content of domestic (a and b)and commercial(c)food waste during the study period (points show average value for triplicate determinations:lines show rolling 21-day averages). constant.Fig.1c shows the TS and VS values for commercial food The average nutrient content of the domestic food waste during waste:these spanned a greater range than for the domestic food the study period was 8.9,1.9 and 3.3 kg tonne-on a wet weight waste,reflecting greater differences in moisture content,but again (WW)basis for Total Kjeldahl Nitrogen (TKN),Phosphorus (P) the ratio of VS:TS was consistent. and Potassium (K).respectively,while the equivalent values for 12 10 a 8 64 2 0 10 b y6IeM JeM 9Uuo]6 86 A 2 0 122138 171182 245258 259/272 2744288 289303 305318 319350 366381 Days ▣Nitrogen(TKN)■Phosphorus(P)▣Potassium(K Fig.2.Variability in nutrient content of domestic food waste composite samples(a)and of digestate samples(b)during the study period

constant. Fig. 1c shows the TS and VS values for commercial food waste: these spanned a greater range than for the domestic food waste, reflecting greater differences in moisture content, but again the ratio of VS:TS was consistent. The average nutrient content of the domestic food waste during the study period was 8.9, 1.9 and 3.3 kg tonne1 on a wet weight (WW) basis for Total Kjeldahl Nitrogen (TKN), Phosphorus (P) and Potassium (K), respectively, while the equivalent values for Fig. 1. TS and VS content of domestic (a and b) and commercial (c) food waste during the study period (points show average value for triplicate determinations; lines show rolling 21-day averages). Fig. 2. Variability in nutrient content of domestic food waste composite samples (a) and of digestate samples (b) during the study period. 614 C.J. Banks et al. / Bioresource Technology 102 (2011) 612–620

CJ.Banks et al./Bioresource Technology 102 (2011)612-620 615 0.10 a 0.08 pue 0.06 o 以。 00 ◆TS 0.04 ▣VS 4品 0.02 0.00+ 0 100 200 300 400 Days 0.30 ◆ 0.06 b ) 0.20 9 0.04 0.02 ejql 0.10 y=0.630x+0.001 y=0.615x+0.033 R2=0.925 R2=0.620 0.00+ 0.00+ 0.00 0.02 0.04 0.06 0.08 0.10 0.00 0.10 0.20 0.30 0.40 Digestate TS.gg Fibre TS.gg Fig.3.TS and VS content of digestate (a and b)and fibre(c)during the study period(points show average value for triplicate determinations:note different scales for digestate and fibre). commercial food waste were 8.7,1.8 and 3.4 kg tonne-1 WW. than those for feedstock,apart from one high value for nitrogen Fig.2a shows the variability in fortnightly composite samples of in day 215-227.A nutrient mass balance taking into account water domestic food waste.The variations between consecutive samples additions showed outputs equal to 86.1%,32.8%and 96.4%of the may reflect the fact that only a small amount of material is ulti- input values of TKN,P and K,respectively.The lower recovery of mately used in laboratory analysis,and however much effort is TKN and particularly of P may indicate losses by precipitation made to prepare representative composites,subsamples may show e.g.of struvite(NH4MgPO6H2O)within the digester system. slight non-homogeneity due to unavoidable scale factors. The average organic loading rate during the study period was 2.5 kg Vs m-3 day-1 based on the nominal digester volume of 3.3.Biogas output and variability 900 m',or 2.7 kg VS m day-'based on the average volume of di- gester contents.The maximum and minimum weekly average Table 1 shows the total biogas production and the proportion of loadings based on actual volume were 3.46 and 0.91 kg VS m methane and carbon dioxide based on daily measurements.The day-,respectively,with the minimum corresponding to a Christ- specific biogas and methane yields are given on both a wet weight mas closure period.The average hydraulic retention time was and a VS basis,and show the food waste has a high methane poten- 80 days,based on the nominal digester volume divided by the tial in comparison to typical municipal residual waste streams.The mass input on a wet weight basis.More detailed information on high moisture content means,however,that the biogas production day-to-day variations in feedstock quantities is given in Arnold etal.(2010). Table 1 Gas production parameters during mass and energy balance period. 3.2.Digestate characteristics Item Unit Value 黑 Methane m3STP 385,488 62.6 Values for TS and VS content of digestate and fibre throughout Carbon dioxide mSTP 229984 37.4 the study period are shown in Fig.3.The average solids content Biogas m3 STP 615,472 100.0 was TS 4.5%,VS 2.9%for the digestate and TS 23.8%,VS 17.9%for Food waste input kg ww 3936.504 the fibre.As can be seen in Fig.3.there was some variation in kg VS 959.209 m3tonne-1wW 156 the TS and VS content of individual samples of digestate but the Specific biogas yield m3 tonne-1 VS 642 VS:TS ratio remained fairly constant. Specific methane yield m3tonne-1ww 98 The average nutrient content of the digestate during the study mtonne-1Vs 402 period was 5.6.0.4 and 2.3 kg tonne-1 WW for TKN.P and K. Volumetric biogas yield m m-3 reactor 1.59 respectively.Fig.2b shows the variability between fortnightly Volumetric methane yielda m3m-3 reactor 1.00 composite samples.As expected the values were more consistent a Based on volume of digester only

commercial food waste were 8.7, 1.8 and 3.4 kg tonne1 WW. Fig. 2a shows the variability in fortnightly composite samples of domestic food waste. The variations between consecutive samples may reflect the fact that only a small amount of material is ulti￾mately used in laboratory analysis, and however much effort is made to prepare representative composites, subsamples may show slight non-homogeneity due to unavoidable scale factors. The average organic loading rate during the study period was 2.5 kg VS m3 day1 based on the nominal digester volume of 900 m3 , or 2.7 kg VS m3 day1 based on the average volume of di￾gester contents. The maximum and minimum weekly average loadings based on actual volume were 3.46 and 0.91 kg VS m3 day1 , respectively, with the minimum corresponding to a Christ￾mas closure period. The average hydraulic retention time was 80 days, based on the nominal digester volume divided by the mass input on a wet weight basis. More detailed information on day-to-day variations in feedstock quantities is given in Arnold et al. (2010). 3.2. Digestate characteristics Values for TS and VS content of digestate and fibre throughout the study period are shown in Fig. 3. The average solids content was TS 4.5%, VS 2.9% for the digestate and TS 23.8%, VS 17.9% for the fibre. As can be seen in Fig. 3, there was some variation in the TS and VS content of individual samples of digestate but the VS:TS ratio remained fairly constant. The average nutrient content of the digestate during the study period was 5.6, 0.4 and 2.3 kg tonne1 WW for TKN, P and K, respectively. Fig. 2b shows the variability between fortnightly composite samples. As expected the values were more consistent than those for feedstock, apart from one high value for nitrogen in day 215–227. A nutrient mass balance taking into account water additions showed outputs equal to 86.1%, 32.8% and 96.4% of the input values of TKN, P and K, respectively. The lower recovery of TKN and particularly of P may indicate losses by precipitation e.g. of struvite (NH4MgPO46H2O) within the digester system. 3.3. Biogas output and variability Table 1 shows the total biogas production and the proportion of methane and carbon dioxide based on daily measurements. The specific biogas and methane yields are given on both a wet weight and a VS basis, and show the food waste has a high methane poten￾tial in comparison to typical municipal residual waste streams. The high moisture content means, however, that the biogas production Fig. 3. TS and VS content of digestate (a and b) and fibre (c) during the study period (points show average value for triplicate determinations; note different scales for digestate and fibre). Table 1 Gas production parameters during mass and energy balance period. Item Unit Value % Methane m3 STP 385,488 62.6 Carbon dioxide m3 STP 229,984 37.4 Biogas m3 STP 615,472 100.0 Food waste input kg WW 3936,504 – kg VS 959,209 – Specific biogas yield m3 tonne1 WW 156 – m3 tonne1 VS 642 – Specific methane yield m3 tonne1 WW 98 – m3 tonne1 VS 402 – Volumetric biogas yielda m3 m3 reactor 1.59 – Volumetric methane yielda m3 m3 reactor 1.00 – a Based on volume of digester only. C.J. Banks et al. / Bioresource Technology 102 (2011) 612–620 615

616 CJ.Banks et al /Bioresource Technology 102(2011)612-620 15000 10000 -biogas ◆-CH4 ◆-C02 5000 100 200 300 400 Day 6 b 75 CM 65 -Average CH4 55 45 100 200 300 400 Day Fig.4.Weekly gas production(a)and daily methane percentage in biogas(b)during the study period. per tonne of imported material is similar to typical values reported The reasons for this change are not clear but the ammonia concen- for municipal solid waste(MSW).Volumetric gas production is cal- tration in the digesters had been increasing steadily and reached culated based on the volume of the digester only. around 5000 mg I-1 at this time.Subsequent work in laboratory- Variability in the biogas production and composition is shown scale digesters has suggested that high ammonia concentrations by reference to the weekly values for methane,carbon dioxide may cause a shift in the biochemical pathways leading to methane and biogas in Fig.4a.Total biogas production over a one-week per- formation (Banks and Zhang,2010).A slight decrease in biogas iod varied from a minimum 6364 m3 to a maximum of 13,438 m3 methane concentration can be seen before day 342 followed by although some of the peaks and troughs can be explained by differ- recovery.Total VFA concentrations continued to increase and ap- ences in the incoming load (e.g.suspension of some deliveries dur- proached 15.000 mgl-1 by the end of the monitoring period.of ing the Christmas-New Year period in 2007).Fig.4b shows the which propionic acid made up 11,500 mgI-.Despite the high variability in methane concentration based on daily readings,com- VFA values the specific and volumetric biogas yields remained pared to the calculated average for the whole study period. unaffected (Fig.6). 3.4.Digestion parameters 3.5.Overall mass balance Digestion parameters are reported from day 0 corresponding to The mass balance around the plant was calculated in two ways: the start of the mass and energy balance study,although measure- by wet weight (Table 2)and on a VS basis (Table 3).In the wet ment of VFA and ammonia only began some time after this. weight balance water additions from both the process and facilities The average digester pH in the study period was 8.13 with val- supplies were included as inputs.Methane and carbon dioxide vol- ues remaining mainly between 8.0 and 8.25(Fig.5a).From day 252 umes were corrected to STP and it was assumed that the spot val- to day 304,however,the pH rose to 8.64,then fell sharply to a min- ues for methane concentration are representative of a 24-h period. imum value of 7.24 by day 342.This fall appears to have been a re- Weights of digestate,fibre and rejects were taken from weigh- sult of a shift in alkalinity,with an increase in intermediate bridge data for materials leaving site.Stored materials are based alkalinity(IA).a fall in partial alkalinity (PA)and a rise in the IA/ on tank volumes and estimated quantities of fibre in the digestate PA ratio to 2.74(Fig.5b).Prior to this the IA/PA ratio was around hall.Weight data on all wastes generated by the operation (includ- 0.4 indicating stable operation (Ripley et al.,1986). ing canteen wastes and litter as well as feedstock contamination) The major factor affecting the intermediate alkalinity is the con was only collected from April 2008,and therefore underestimates centration of undissociated VFA.This fell between day 250-300. the total weight of material leaving the plant by this route.Con- with a decrease in the propionic acid concentration,followed by tamination of the feedstock itself,assessed by hand sorting of sam- a rapid increase after day 300 in both acetic and propionic acid ples(not reported here).was minimal.Evaporative water losses and a slower rise in the concentration of butyric etc.(Fig.5c). from the gas mixing system due to supersaturation followed by

per tonne of imported material is similar to typical values reported for municipal solid waste (MSW). Volumetric gas production is cal￾culated based on the volume of the digester only. Variability in the biogas production and composition is shown by reference to the weekly values for methane, carbon dioxide and biogas in Fig. 4a. Total biogas production over a one-week per￾iod varied from a minimum 6364 m3 to a maximum of 13,438 m3 , although some of the peaks and troughs can be explained by differ￾ences in the incoming load (e.g. suspension of some deliveries dur￾ing the Christmas–New Year period in 2007). Fig. 4b shows the variability in methane concentration based on daily readings, com￾pared to the calculated average for the whole study period. 3.4. Digestion parameters Digestion parameters are reported from day 0 corresponding to the start of the mass and energy balance study, although measure￾ment of VFA and ammonia only began some time after this. The average digester pH in the study period was 8.13 with val￾ues remaining mainly between 8.0 and 8.25 (Fig. 5a). From day 252 to day 304, however, the pH rose to 8.64, then fell sharply to a min￾imum value of 7.24 by day 342. This fall appears to have been a re￾sult of a shift in alkalinity, with an increase in intermediate alkalinity (IA), a fall in partial alkalinity (PA) and a rise in the IA/ PA ratio to 2.74 (Fig. 5b). Prior to this the IA/PA ratio was around 0.4 indicating stable operation (Ripley et al., 1986). The major factor affecting the intermediate alkalinity is the con￾centration of undissociated VFA. This fell between day 250–300, with a decrease in the propionic acid concentration, followed by a rapid increase after day 300 in both acetic and propionic acid and a slower rise in the concentration of butyric etc. (Fig. 5c). The reasons for this change are not clear but the ammonia concen￾tration in the digesters had been increasing steadily and reached around 5000 mg l1 at this time. Subsequent work in laboratory￾scale digesters has suggested that high ammonia concentrations may cause a shift in the biochemical pathways leading to methane formation (Banks and Zhang, 2010). A slight decrease in biogas methane concentration can be seen before day 342 followed by recovery. Total VFA concentrations continued to increase and ap￾proached 15,000 mg l1 by the end of the monitoring period, of which propionic acid made up 11,500 mg l1 . Despite the high VFA values the specific and volumetric biogas yields remained unaffected (Fig. 6). 3.5. Overall mass balance The mass balance around the plant was calculated in two ways: by wet weight (Table 2) and on a VS basis (Table 3). In the wet weight balance water additions from both the process and facilities supplies were included as inputs. Methane and carbon dioxide vol￾umes were corrected to STP and it was assumed that the spot val￾ues for methane concentration are representative of a 24-h period. Weights of digestate, fibre and rejects were taken from weigh￾bridge data for materials leaving site. Stored materials are based on tank volumes and estimated quantities of fibre in the digestate hall. Weight data on all wastes generated by the operation (includ￾ing canteen wastes and litter as well as feedstock contamination) was only collected from April 2008, and therefore underestimates the total weight of material leaving the plant by this route. Con￾tamination of the feedstock itself, assessed by hand sorting of sam￾ples (not reported here), was minimal. Evaporative water losses from the gas mixing system due to supersaturation followed by Fig. 4. Weekly gas production (a) and daily methane percentage in biogas (b) during the study period. 616 C.J. Banks et al. / Bioresource Technology 102 (2011) 612–620