Bioresource Technology 101 (2010) 1153–1158

Contents lists available at ScienceDirect

Bioresource Technology

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

A methodology for optimising feed composition for anaerobic co-digestionof agro-industrial wastes

J.A. Álvarez *, L. Otero, J.M. LemaDepartment of Chemical Engineering, School of Engineering, University of Santiago de Compostela, Rúa Lope Gómez de Marzoa, 15782 Santiago de Compostela, Spain

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

Article history:Received 9 June 2009Received in revised form 17 September 2009Accepted 18 September 2009Available online 14 October 2009

Keywords:Anaerobic co-digestionOrganic wasteBatch assayLinear programming

0960-8524/$ - see front matter � 2009 Elsevier Ltd. Adoi:10.1016/j.biortech.2009.09.061

* Corresponding author. Tel.: +34 981563100x1601E-mail address: juanantonio.alvarez@usc.es (J.A. Á

An optimisation protocol for maximising methane production by anaerobic co-digestion of severalwastes was carried out. A linear programming method was utilised to set up different blends aimed atmaximising the total substrate biodegradation potential (L CH4/kg substrate) or the biokinetic potential(L CH4/kg substrate d). In order to validate the process, three agro-industrial wastes were considered: pigmanure, tuna fish waste and biodiesel waste, and the results obtained were validated by experimentalstudies in discontinuous assays.

The highest biodegradation potential (321 L CH4/kg COD) was reached with a mixture composed of 84%pig manure, 5% fish waste and 11% biodiesel waste, while the highest methane production rate(16.4 L CH4/kg COD d) was obtained by a mixture containing 88% pig manure, 4% fish waste and 8% bio-diesel waste.

Linear programming was proved to be a powerful, useful and easy-to-use tool to estimate methaneproduction in co-digestion units where different substrates can be fed.

� 2009 Elsevier Ltd. All rights reserved.

1. Introduction

Although anaerobic digestion of organic solid wastes is anestablished technology in Europe, with 120 full-scale plants treat-ing about 4 million tons per year, it represents on average, only27.5% of all of the biological waste treatment processes (De Baere,2006). The idea of co-digestion offers several possible ecological,technological and economical advantages, so it can improveorganic waste treatment through anaerobic digestion. Anaerobicco-digestion can increase CH4 production of manure digesters by50–200%, depending on the operating conditions and the co-sub-strates used (Amon et al., 2006; Callaghan et al., 1999; Ferreiraet al., 2007; Murto et al., 2004; Soldano et al., 2007). Currently,there is a increasing number of full-scale co-digestion plants treat-ing manure and industrial organic wastes, mainly in Denmark andGermany (Angelidaki and Ellegaard, 2003; Raven and Gregersen,2007; Weiland, 2000), and there is an increasing interest, mainlyin Europe, in using this technology for bioenergy production. Onthe other hand, it is well known that organic waste anaerobicdigestion produces a new semi-liquid waste: digestate, whichcan be used in agriculture after doing a stabilisation or compostprocess.

Co-digestion is defined as the anaerobic treatment of a mixtureof at least two different waste types with the aim of improving the

ll rights reserved.

6; fax: +34 981528050.lvarez).

efficiency of the anaerobic digestion process. Therefore, it is veryimportant to establish the best blend in order to maximise meth-ane production, avoid inhibition processes and make profitablebiogas plants.

The main issue for co-digestion process lies in balancing severalparameters in the co-substrate mixture: macro- and micronutri-ents, C:N ratio, pH, inhibitors/toxic compounds, biodegradable or-ganic matter and dry matter (Hartmann et al., 2003). Optimumvalues of C:N and COD:N ratios of 20 and 70, respectively, havebeen suggested for the stable performance of anaerobic digestion(Burton and Turner, 2003; Chen et al., 2008). However, lower val-ues of C:N ratios (between six and nine) have been reported assuitable for the anaerobic digestion of nitrogen-rich waste(Mshandete et al., 2004). Threshold limits of free ammonia and to-tal ammonia of 1.1 and 4 g N/L, respectively, in swine and cattlemanure digestion have been reported (Angelidaki and Ahring,1993; Chen et al., 2008; Hansen et al., 1998).

Alkalinity is necessary to avoid decreasing pH due to accumula-tion of volatile fatty acids when applying a high organic load.Anaerobic digesters work in a wide variety of alkalinity valuesdepending on the substrate to be degraded. These values rangefrom 2000 to 18000 mg CaCO3/L (Cuetos et al., 2008; Gelegeniset al., 2007a; Murto et al., 2004; Mshandete et al., 2004).

Moller et al. (2004) studied the specific methane productivity ofdifferent types of manure in batch tests. The specific methanepotentials measured were 148 ± 41, 356 ± 28 and 275 ± 36 L CH4/kg VS (volatile solids) for cattle, pig fattener and sow manure,

1154 J.A. Álvarez et al. / Bioresource Technology 101 (2010) 1153–1158

respectively. The manure specific methane potential has been im-proved by co-digestion with other substrates: sewage sludge (Mur-to et al., 2004), fruit and vegetable waste (Ferreira et al., 2007),energy crops (Lehtomäki et al., 2007), glycerine (Amon et al.,2006) and the organic fraction of municipal solid waste (Hartmannand Ahring, 2005).

Waste biomethanation potential depends on the concentrationof the three main organic components: proteins, lipids and carbo-hydrates, and a substrate characterisation is required to predictmethane production (Gelegenis et al., 2007a,b; Maya-Altamiraet al., 2008; Neves et al., 2008; Shanmugam and Horan, 2009).Although there is a sufficient methodology for determining bio-methanation potential, most of the approaches used to date arebased on experimental studies concerning the behaviour of differ-ent feedings with different properties of raw waste.

The aim of this work is to develop a methodology useful fordetermining the most adequate ratios of different co-substratesthat provide an optimised biodegradation potential or biokineticmethane potential.

For this propose, a linear programming optimisation methodbased on determining restrictions (minimum and maximum val-ues) on several characteristics of the mixture has been developed.In order to validate the methodology, three types of wastes withquite different characteristics (pig manure, fish waste and biodieselwaste) were considered as co-substrates.

2. Methods

2.1. Waste and inoculum origin and collection

Pig manure (PM) was taken from a sewer of a 150-pig fattenerand sow farm, which collects both faeces and urine. It was stored at4 �C until characterisation. PM samples were homogenised andsieved to 2 mm. Fish waste (FW) from a canning industry consistedof heads, tails, fish bones and viscera of tuna fish. FW was frozenfor storage until characterisation and triturated until homogenised.Biodiesel waste (BW) was sampled from a biodiesel factory. It con-tained mainly glycerine (glycerol) produced in the transesterifica-tion of triglycerides with methanol and sodium hydroxide togenerate biodiesel (methyl esters).

Granular biomass from a pilot hybrid reactor (UASB-FA) treat-ing wine waste and from an IC reactor treating brewery wastewa-ter was used as inoculum in the biodegradation assays andbiokinetic assays, respectively. During biodegradation assays,anaerobic hybrid reactor was washed and its biomass was lost,so a similar granular inoculum had to be used to carry out bioki-netic assays and the IC reactor biomass was selected. The specificmethanogenic activities of each inoculum were 0.15 and0.10 g CH4-COD/g SSV d, respectively.

2.2. Analytical methods

Standard methods (1995) were applied for pH measurementsand for determination of COD, TS, VS, TKN-N, NHþ4 –N and TA (totalalkalinity). Biogas composition (N2, CH4, CO2 and H2S percentage)was analysed by gas chromatography (HP, 5890 Series II) equippedwith a thermal conductivity detector (Molina et al., 2008). Volatilefatty acids (VFA) were determined by gas chromatography (HP,5890A) equipped with a flame ionisation detector (Molina et al.,2008). Samples were previously centrifuged (5 min, 3500 rpm),and the supernatant filtrated through 0.45-lm cellulose filters. To-tal lipid content was determined using a Standard Soxhlet method(Standard methods, 1995). Proteins were calculated from organicnitrogen composition. Carbohydrates were estimated as the

remaining fraction of VS or COD after proteins and lipids weredetermined.

2.3. Linear programming

Computational software used to solve linear programmingproblems is actually a strength technological alternative whichfacilitates feasibility studies. Specifically, ExcelTM Solver is an ade-quate tool of relatively easy initial programming and versatile pos-terior usage to apply for different problems solution. So, solvermethod from ExcelTM was chosen as linear programming tool in thiswork.

The reported method consists of maximising an objective func-tion, taking into account several restrictions that need to be ful-filled. Two different objective functions have been considered:the total substrate biodegradation potential (L CH4/kg wet weight(WW) of the substrate); taking into account substrate transforma-tion efficiency and the biokinetic potential (L CH4/kg WW d); tak-ing into account kinetic capacity of the anaerobic process. Thetheoretical biodegradation potential value was calculated fromthe COD content of the different wastes using a factor of350 L CH4/kg COD removed (Angelidaki and Sanders, 2004). Thetheoretical biokinetic potential was calculated by considering thefollowing values: 35 L CH4/kg lipid d; 42 L CH4/kg protein d and27 L CH4/kg carbohydrate d (Neves et al., 2008).

Minimum and maximum values of the following parameterswere considered as the restrictions on the system: COD/N ratio(calculated as COD/TKN-N); NHþ4 –N concentration, calculated byassuming a full protein digestion (0.124 g N/g protein) (Angelidakiand Sanders, 2004; Gelegenis et al., 2007a); lipid concentration; to-tal alkalinity; liquid fraction; COD/SO�2

4 ratio; chloride.As a result of each optimisation, which corresponds to a partic-

ular set of restrictions, the program gives the fraction of each sub-strate to be used in the blend to obtain the maximumbiodegradation and biokinetic potential.

The organic loading rate value was also considered as a restric-tion when the biokinetic potential was determined.

Restriction values of total alkalinity, liquid fraction, COD/SO�24

ratio and chloride were maintained constant in all cases. Their val-ues were the following: total alkalinity: between 3 and 20 g CaCO3;liquid fraction: between 85% and 100%; COD/SO�2

4 ratio: higherthan 15; chloride: lower than 3 g/L.

The following minor restrictions must be stipulated to developproper solver programming: the sum of the substrate fraction inthe blend must be equal to 100 and each fraction must be higherthan 0.

Summarising, the parts of linear programming method used inthis work are the following:

1. Input proposed restrictions.2. Calculate theoretical restrictions values according substrate

characterisation.3. Calculate fractions of each waste to set up the blend, which

have to fulfil proposed restrictions and maximised methaneproduction (total biodegradation or kinetic transformation).

2.4. Batch assay methodology

Batch assays were carried out in 500-mL glass flasks with coiledbutyl rubber stoppers. All tests were performed in triplicate assaysunder the following operating conditions: 35 �C, mixing at 120 rpmand 5 g VSS/L of inoculum. Substrate feed was composed by adding10 mL of the optimised blend, the substrate concentration varyingfrom 4 to 9 g TCOD/L. Control assays with only inoculum and withboth inoculum and PM were also performed.

J.A. Álvarez et al. / Bioresource Technology 101 (2010) 1153–1158 1155

Anaerobic conditions were maintained by using an anaerobicbasal medium composed of cysteine (0.5 g/L) and NaHCO3 (5 g/L),at a pH of 7.0–7.2. Before flushing the liquid and headspace withN2, 1.2 mL of Na2S (20 g/L) was added to each assay as a reducingagent (Molina et al., 2008). An initial liquid volume of 385 mL wasused in all assays. A pressure transducer was used to measure the

Table 1PM, FW and BW characterisation.

Parameter Pigmanure

Fishwaste

Biodieselwaste

Liquid fraction (%) 98.3 63.1 100pH 6.9 nd ndSoluble fraction conductivity (mS/

cm)29.5 140.4 45.5

Density (kg WWa/L) 1.0 1.1 1.0TS (g TS/kg WW) 17.3 369 0VS (g SV/kg WW) 11.7 270 0COD (g O2/kg WW) 28.9 409.6 1390Soluble COD (g O2/kg WW) 15.3 nd 1390TKN-N (g N/kg WW) 3.3 33.6 0.2NHþ4 –N (g N/kg WW) 3.1 0.7 0Total alkalinity (g CaCO3/L) 7.7 0.3 32Chloride (g/kg WW) 0.5 34.9 ndSulphate (g/kg WW) 0.04 0.7 ndVFA-COD (g VFA-COD/kg WW) 12.2 0 0Proteins (g prot.b/kg WW) 1.1 205.8 1.2Lipids (g lip.c/kg WW) 1.5 28 77.3Carbohydrates (g ch.d/kg WW) 9.2 36.2 921.5e

COD/N ratio 8.9 12.2 7315

a Wet weight.b Proteins.c Lipids.d Carbohydrates.e Carbohydrates were determined by the COD balance; nd: not determined.

Table 2Percentage of each waste on blends A, B and C, determined by linear programmingand theoretical biodegradation potential of each blend.

Blend Main inputrestrictionsa

Fulfilledrestrictionvalues

Wastepercentage(%WW b)

Theoreticalbiodegradationpotential c

(L CH4/kg WW)

A 50 < COD/N < 100 50.6 Manure: 91 530.2 < NHþ4 –N < 3.5 3.0 Fish waste: 00.5 < Lip < 8.3 8.3 Biodiesel waste: 9

B 50 < COD/N < 100 100 Manure: 82 950.2 < NHþ4 –N < 3.5 2.7 Fish waste: 00.5 < Lip < 15 15.0 Biodiesel waste: 18

C 50 < COD/N < 90 90 Manure: 74 1200.2 < NHþ4 –N < 3.5 3.5 Fish waste: 40.5 < Lip < 20 19.2 Biodiesel waste: 22

a NHþ4 –N and lip (lipids) in g/L.b Wet weight.c Determined by linear programming method based on total biodegradation

potential.

Table 3Experimental and theoretical specific methane production of mixtures A, B, C and of the c

Assay ExperimentalCH4-COD (g/L)

TheoreticalCH4-COD (g/L)

Experimental methane prod

(LCH4/kg WW) (LCH4

Blend A 2.9 3.9 37.6 249.0Blend B 0.5 7.1 6.8 25.0Blend C 0.8 8.9 11.3 33.1PM 4.7 7.0 6.7 230.4

a Obtained by linear programming.

pressure increase. The biogas was sampled regularly, and its com-position was determined by gas chromatography.

2.5. Calculations

Batch assay methane production was plotted as CH4-COD (g)against time (d). Firstly, moles of methane were calculated bythe ideal gas law:

CH4 moles ¼ PT � XCH4 � Vgas

R � ðT þ 273Þ ;

where PT is the total pressure measured by the transducer (mmHg);XCH4 is the methane molar fraction; Vgas is the headspace volume(mL); R is the ideal gas constant (62,364 mmHg mL/mol K); and Tis the assay temperature (�C). CH4-COD (g) can be calculated bymultiplying the moles of CH4 by 64 (g DQO/CH4 mol).

The experimental specific methane potential at standard tem-perature and pressure conditions was calculated by dividing theCH4 volume produced by the waste quantity at the beginning ofthe assay in wet weight and the COD bases (STP L CH4/kg WW orSTP L CH4/kg COD). The theoretical specific methane potentialwas calculated assuming 100% degradation of the waste or blendCOD.

3. Results and discussion

3.1. Waste characterisation

PM, FW and BW characteristics are indicated in Table 1. Basi-cally, PM contributes moisture to the mixture; FW contributesmainly nitrogen from proteins and carbohydrates in a minor pro-portion (mainly from fishbone fibre and cartilage) and lipids; final-ly BW contributes COD, lipids (from incomplete transesterification)and glycerol (as carbohydrates) to the blend.

3.2. Determination of total biodegradation potential

From the characterisation data of each waste (Table 1), threedifferent blends (A, B and C) with specific characteristics imposedby the restrictions were set up. Table 2 indicates the contributionsof each waste to blends (A, B and C) calculated by the linear pro-gramming method, and Table 3 the characteristics of blends andthe theoretical biodegradation potential of each one.

In blend A, the maximum lipid concentration restriction valueof 8.3 g/L limits the maximum methane production obtained.Therefore, by mixing PM and BW, a maximum total biodegradationwould be reached without adding FW, as BW and PM contributeenough to liquid fraction, alkalinity, COD, lipids and nitrogen tofulfil the restrictions imposed (Table 2).

Blend B was produced by increasing the lipid concentrationrestriction (Table 2), leading to a blend with a higher biodegrada-tion potential than in the case of blend A, as a consequence ofthe higher COD/N ratio. In this way, the BW percentage in blend in-creased to the maximum permitted by the upper limit of the lipid

ontrol batch.

uction Theoretical methane production Experimental/theoretical ratio (%)

/kg COD) (LCH4/kg WW) (LCH4/kg COD)

53a 350 71.195a 350 7.2

120a 350 9.510 350 65.8

Table 4Percentage of each waste on blends D, E and F, determined by linear programmingand theoretical biokinetic potential of each blend.

Blend Main inputrestrictionsa

Fulfilledrestrictionvalues

Wastepercentage(%WWb)

Theoreticalbiokineticpotentialc

(L CH4/kg WW�d)

D 20 < COD/N < 35 35.0 Manure: 88 3.00.2 < NHþ4 –N < 4 4.0 Fish waste: 40.5 < Lip < 10 8.6 Biodiesel

waste: 8

E 20 < COD/N < 45 45.0 Manure: 84 4.00.2 < NHþ4 –N < 4 4.0 Fish waste: 50.5 < Lip < 15 11.1 Biodiesel

waste: 11

F 20 < COD/N < 60 60.0 Manure: 79 5.40.2 < NHþ4 –N < 4 4.0 Fish waste: 50.5 < Lip < 15 14.9 Biodiesel

waste: 16

a NHþ4 –N and lip (lipids) in g/L.b Wet weight.

1156 J.A. Álvarez et al. / Bioresource Technology 101 (2010) 1153–1158

concentration restriction. Since BW is the waste with the highestCOD and lipid content, COD/N ratio also raised to the upper limit.Again, FW was not selected in blend B because protein additionwas not necessary to fulfil COD/N ratio and NHþ4 restrictions.

In the third mix (blend C), the lipid concentration restrictionwas further increased up to 20 g/L, and the COD/N ratio restrictionupper limit was decreased to determine blends with different COD/N ratios, thus resulting in a higher proportion of BW. Accordingly,FW fraction attained the highest value to fulfil the COD/Nrestriction.

As observed (Table 2), the three blends had different character-istics regarding the main parameters that influence the anaerobicco-digestion of organic substrates: COD/N ratio, lipid content andammonium concentration.

3.3. Batch assays of optimised mixtures based on total biodegradationpotential

Discontinuous assays were carried out to experimentally deter-mine the methane potential of the blends and to validate the cal-culations obtained by linear programming. Methane productionof the batch assays is shown in Fig. 1 and in Table 3 the theoreticaland experimental methane production are compared.

Blend A reached 71% of the estimated theoretical methane andproduced an increase in methane production, as compared to thePM assay, of 464% and 8% in wet weight and COD bases, respec-tively (Table 3).

Although, blends B and C were expected to have the highestmethane production based on the total biodegradation potential(95 and 120 L CH4/kg WW, respectively, Table 2), their experimen-tal methane production (6.8 and 11.3 L CH4/kg WW, respectively,Table 3) were noticeably lower than expected from theoreticalcalculations.

A possible reason for this could be the high lipid concentration(BW provided the main lipid content), which may cause an inhibi-tion by LCFA (long chain fatty acids). However, theoretical lipidconcentrations in all assays were much lower (0.2, 0.4 and 0.5 g/L in blends A, B and C) than that which could cause inhibition:2.3 g/L, a value obtained in a lipid-rich waste batch test by Cirneet al. (2007).

A more likely possibility of this behaviour is the lack of suffi-cient nutrient concentration in blends B and C due to the higherCOD/N ratio of 100 and 90, respectively.

3.4. Determination of biokinetic potential

The biodegradation kinetic potential is important to consider inthe preparation of co-digestion blends as it indicates the rate of theanaerobic process. Therefore, linear programming was also used to

0

1

2

3

4

5

0 5 10 15 20 25 30 35TIME (d)

CH

4-C

OD

(g/L

)

Pig controlBlend ABlend BBlend C

Fig. 1. Methane production of assays based on the total biodegradation potential.The methane production of the inoculum assay is already subtracted.

determine three optimised blends (D, E and F), in this case, maxi-mising the methane production rate (biokinetic potential). Table4 shows the blends determined by linear programming based onthe biokinetic potential, in which COD/N ratio was limited between20 and 60 to avoid a lack of nutrients in batch assays and because itwas considered an optimal range, taking into account the previoustest. In these experiments, FW was added in all blends to decreasethe COD/N ratio (between 20 and 60), meaning that blends needmore nitrogen provided by FW. Table 4 also provides the theoret-ical biokinetic potential (methane production rates) of each blend.

In these three new blends, COD/N ratio upper limit again con-strained the maximum methane production rate calculated by lin-ear programming. Additionally, as lipid concentrations wereincreased in blends D, E and F due to the increase in the BW frac-tion in the blends, FW fraction increased to provide enough nutri-ent content to the blend to fulfil the COD/N restriction (Table 4);besides, protein has the fastest biodegradation kinetics. At thesame time, nitrogen was limited by ammonium restriction, so that,the three blends had the highest permitted NHþ4 –N value (Table 4).

3.5. Batch assays of obtained mixtures based on biokinetic potential

Methane production of each batch assay is indicated in Fig. 2,while in Table 5 the comparison of theoretical and experimentalvalues is shown.

In contrast to the previous batch assays, no inhibition was ob-served in any assay and the three tests produced methane in accor-dance to their theoretical organic content. In this case, PM

0

1

2

3

4

5

6

0 5 10 15 20 25 30 35 40TIME (d)

CH

4-C

OD

(g/L

)

Pig controlBlend DBlend EBlend F

Fig. 2. Methane production of assays based on the biokinetic potential. Themethane production of the inoculum assay is subtracted.

c Determined by linear programming method based on biokinetic potential.

Table 5Experimental and theoretical specific methane production of mixtures D, E, F and of the control batch.

Assay ExperimentalCH4-COD (g/L)

TheoreticalCH4-COD (g/L)

Experimental methane production Theoretical methane production Experimental/theoretical ratio (%)

(LCH4/kg WW) (LCH4/kg COD) (LCH4/kg WW) (LCH4/kg COD)

Blend D 3.4 3.9 44.3 292.7 53 350 83.6Blend E 4.7 5.1 62.8 320.6 69 350 91.6Blend F 6.0 6.9 79.8 302.7 92 350 86.5PM 2.9 5.0 7.3 198.8 13 350 56.8

J.A. Álvarez et al. / Bioresource Technology 101 (2010) 1153–1158 1157

biodegradation was lower than in previous biodegradation tests,likely due to the manure samples being taken at least 3 or4 months before starting these batch assays, so the samples mayhave been partly biodegraded beforehand. The biodegradation ofthe three blends determined by linear programming based on thebiokinetic potential reached values higher than 80% (Table 5).

Blends D, E and F exhibited a significantly increased PM meth-ane yield, as can be observed in Table 5, while blend E achieved thehighest biodegradation percentage and the highest specific meth-ane production relating to the same organic matter content (kgCOD). Thus, the recommended fractions of FW and BW that shouldbe added to PM to achieve the maximum substrate biodegradationare between those of blend E and F, 5% and 11–16%, respectively.

Table 6 indicates the methane production rate of each assay car-ried out. The experimental methane production rates were com-pared to the theoretical ones. These values were calculated fromthe methane produced, the theoretical COD content, input sub-strate quantity and assay time before reaching a methane curveplateau; in this way, substrate limitations that can influence themethane production rate were avoided. The three blends exhibiteda significantly increased experimental specific methane productionrate, as compared to the PM control (Table 6).

Methane production rate in wet weight basis increased alongblends D, E and F (Table 6) as was predicted by the linear program-ming method based on the biokinetic potential (Table 4). This factwas due to the increased organic matter in blends D, E and F, ob-tained from adding more BW, with a higher COD value. Taking intoaccount the methane potential rate in COD basis (Table 6), blend Dreached the maximum rate with respect to the same amount of or-ganic matter (16.4 L CH4/kg COD d, Table 6), duplicating the PWmethane production rate (8.3 L CH4/kg COD d, Table 6).

The highest biodegradation potential was achieved with blend E(320.6 L CH4/kg COD, Table 5), while blend D exhibited the highestbiokinetic potential (16.4 L CH4/kg COD d, Table 6).

In a hypothetical biogas plant, the choice of blend E or blend Dwould depend on the aim of the plant. Blend E would provide amore stabilised digestate, while blend D would lead to a highermethane production rate. These aspects are obviously influencedby the hydraulic retention time (HRT) applied to the plant. For in-stance, an HRT of 30 d was proposed in linear programming perfor-mances based on biokinetic potential, and the OLR values

Table 6Experimental and theoretical specific methane production rates of mixtures D, E, F and of

Assay Experimental CH4-COD (g/L)a Experimental methane production rateb

(L CH4/kg WW d) (L CH4/kg COD d)

Blend D 2.9 2.5 16.4Blend E 3.4 2.9 14.4Blend F 3.9 3.4 12.8PM 1.8 0.3 8.3

a Experimental CH4-COD (g) produced at 15.6 d (time before reaching the methane cub Taking into account the experimental CH4-COD production at 15.6 d.c Taking into account the initial theoretical COD of each assay and 15.6 d of assay.

calculated by linear programming for blends D, E and F were 5,6.5 and 8.8 g COD/L d, respectively, which will have to be testedin continuous operation. Hence, more research in continuous oper-ation needs to be done to obtain further information about the to-tal biodegradation and methane production rates of mixtures ofthese substrates.

3.6. Effect of co-digestion of PM, FW and BW on methane potential

Taking into account the different blends developed by linearprogramming, either through total biodegradation or biokineticpotential, the best COD/N ratio for obtaining the maximum meth-ane production was determined to be between 45 and 60 (blends Eand F), which is similar to that reported in other studies (Burtonand Turner, 2003; Chen et al., 2008). However, these ratios canbe influenced by the inoculum nutrient demand and inoculumacclimation; thus, the inoculum used in the batch assays can causevariations in the substrate biodegradation.

Pig manure specific methane potential in VS basis obtained inthis work was between 570 and 620 L CH4/kg VS, which is almosttwice than that reported by Moller et al. (2004) (356 L CH4/kg S)and Ferreira et al. (2007) (375 L CH4/kg VS). Inoculum characteris-tics and substrate/inoculum ratios can influence the manure meth-ane potential. Cattle manure has a lower methane potential thanpig manure, as indicated by Callaghan et al. (1999) (300 L CH4/kg VS) and Moller et al. (2004) (148 L CH4/kg VS).

Finally, from results obtained with the different blends deter-mined by linear programming, it can be indicated that the increasein specific methane production obtained in blends A, D, E and F, incomparison to the specific methane production of PM, were 464%,508%, 763% and 996%, respectively.

Callaghan et al. (1999) achieved an increase in the methanepotential of 23.3% in VS basis with a mixture of 70% cattle man-ure, 20% fish offal and 10% inoculum. In another study, Ferreiraet al. (2007) achieved increases of 131–406% (VS basis) whenpig manure was blended with 5–30% fruit waste. An increase of182% (VS basis) in pig manure methane potential adding 6% ofglycerine was obtained by Amon et al. (2006). In this work, takinginto account the methane production in VS basis, increases be-tween 225% and 520% in the manure methane potential wereachieved.

the control batch.

Theoretical methane production ratec Experimental/theoretical ratio (%)

(L CH4/kg WW d) (L CH4/kg COD d)

3.4 22.4 73.24.4 22.4 66.15.9 22.4 56.90.8 22.4 37.0

rve plateau – see Fig. 2).

1158 J.A. Álvarez et al. / Bioresource Technology 101 (2010) 1153–1158

4. Conclusions

The following general conclusions can be drawn from this work:

– Linear programming is an useful and easy-to-use method toestimate methane production in co-digestion units where differ-ent substrates can be fed.

– Inoculum nutrient demand and inoculum acclimation can influ-ence the substrate biodegradation and methane potential. Inspite of using granular inoculum, inoculum of different sourcecaused variations in the substrate biodegradation.

– The recommended percentages of FW and BW that should beadded to PM to reach the maximum substrate biodegradationwould be 5% and 10%, respectively.

– Experiments in continuous operation are recommended toacquire broader information about the biodegradation andmethane production rates of the blends determined by linearprogramming.

Acknowledgements

This work was supported by project PSE-120000-2007-16/PRO-BIOGAS of the Ministry of Education and Science of Spain and theEnergy National Program.

References

Amon, T., Amon, B., Kryvoruchko, V., Bodiroza, V., Pötsch, E., Zollitsch, W., 2006.Optimising methane yield from anaerobic digestion of manure: effects of dairysystems and of glycerine supplementation. International Congress Series, vol.1293, pp. 217–220.

Angelidaki, I., Ahring, B., 1993. Thermophilic anaerobic digestion of livestock waste:effect of ammonia. Appl. Microbiol. Biotechnol. 38, 560–564.

Angelidaki, I., Ellegaard, L., 2003. Codigestion of manure and organic wastes incentralized biogas plants. Appl. Biochem. Biotechnol. 109, 95–105.

Angelidaki, I., Sanders, W., 2004. Assessment of the anaerobic biodegradability ofmacropollutants. Rev. Environ. Sci. Bio/Technol. 3, 117–129.

Burton, C., Turner, C., 2003. Manure Management. Silsoe Research Institute. pp.281–282.

Callaghan, F.J., Wase, D.A.J., Thayanithy, K., Forster, C.F., 1999. Co-digestion of wasteorganic solids: batch studies. Bioresour. Technol. 67, 117–122.

Chen, Y., Cheng, J.J., Creamer, K.S., 2008. Inhibition of anaerobic digestion process: areview. Bioresour. Technol. 99, 4044–4064.

Cirne, D.G., Paloumet, X.L., Björnsson, L., Alves, M.M., Mattiasson, B., 2007.Anaerobic digestion of lipid-rich waste—Effects of lipid concentration. Renew.Energy 32, 965–975.

Cuetos, M.J., Gómez, X., Otero, M., Morán, A., 2008. Anaerobic digestión of solidslaughterhouse waste (SHW) at laboratory scale: influence of co-digestion with

the organic fraction of municipal solid waste (OFMSW). J. Biochem. Eng. 40, 99–106.

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

Ferreira, L., Duarte, E., Silva, C., Malfeito, M., 2007. Fruit wastes bioconversion foranaerobic co-digestion with pig manure. Process development for the recyclingin decentralised farm scale plants. In: Proceedings of the InternationalConference Progress in Biogas. Stuttgart, Germany, pp. 135–140.

Gelegenis, J., Georgakakis, D., Angelidaki, I., Mavris, V., 2007a. Optimization ofbiogas production by co-digesting whey with diluted poultry manure. Renew.Energy 32, 2147–2160.

Gelegenis, J., Georgakakis, D., Angelidaki, I., Christopoulou, N., Goumenaki, M.,2007b. Optimization of biogas production from olive-oil mill wastewater, bycodigesting with diluted poultry-manure. Appl. Energy 84, 646–663.

Hansen, K.H., Angelidaki, I., Ahring, B.K., 1998. Anaerobic digestion of swinemanure: inhibition by ammonia. Water Res. 38, 5–12.

Hartmann, H., Ahring, B.K., 2005. Anaerobic digestion of the organic fraction ofmunicipal solid waste: influence of co-digestion with manure. Water Res. 39,1543–1552.

Hartmann, H., Angelidaki, I., Arhing, B.K., 2003. Co-digestion of the organic fractionof municipal waste with other waste types. In: Mata-Alvarez, J. (Ed.),Biomethanization of the Organic Fraction of Municipal Solid Wastes. IWAPublishing, UK.

Lehtomäki, A., Huttunen, S., Rintala, J.A., 2007. Laboratory investigations oncodigestion of energy crops and crop residues with cow manure for methaneproduction: effect of crop to manure ratio. Resour. Conserv. Recycl. 51, 591–609.

Maya-Altamira, L., Baun, A., Angelidaki, I., Schmidt, J.E., 2008. Influence ofwastewater characteristics on methane potential in food-processing industrywastewaters. Water Res. 42, 2195–2203.

Standard Methods for the Examination of Water and Wastewater 1995. 19th ed.American Public Health Association/American Water Works Association/WaterEnvironment Federation, Washington, DC, USA.

Molina, F., García, C., Roca, E., Lema, J.M., 2008. Characterization of anaerobicgranular sludge developed in UASB reactors that treat ethanol, carbohydratesand hydrolyzed protein based wastewaters. Water Sci. Technol. 57 (6), 837–842.

Moller, H.B., Sommer, S.G., Ahring, B.K., 2004. Methane productivity of manure,straw and solid fractions of manure. Biomass Bioenergy 26, 485–495.

Mshandete, A., Kivaisi, A., Rubindamayugi, M., Mattiasson, B., 2004. Anaerobic batchco-digestion of sisal pulp and fish wastes. Bioresour. Technol. 95, 19–24.

Murto, M., Björnsson, L., Mattiasson, B., 2004. Impact of food industrial waste onanaerobic co-digestion of sewage sludge and pig manure. J. Environ. Manage.70, 101–107.

Neves, L., Gonçalo, R., Oliveira, R., Alves, M.M., 2008. Influence of composition on thebiomethanation potential of restaurant waste at mesophilic temperatures.Waste Manage. 28, 965–972.

Raven, R.P.J.M., Gregersen, K.H., 2007. Biogas plants in Denmark: successes andsetbacks. Renew. Sustain. Energy Rev. 11, 116–132.

Shanmugam, P., Horan, N.J., 2009. Simple and rapid methods to evaluate methanepotential and biomass yield for a range of mixed solid wastes. Bioresour.Technol. 100, 471–474.

Soldano, M., Fabbri, C., Piccinini, S., 2007. Co-digestion plant in dairy cattle farm inEmilia Romagna region (Italy). In: Proceedings of the International ConferenceProgress in Biogas. Stuttgart, Germany, pp. 95–99.

Weiland, P., 2000. Anaerobic waste digestion in Germany – status and recentdevelopments. Biodegradation 11 (6), 415–421.