Waste Management 52 (2016) 286–294

Contents lists available at ScienceDirect

Waste Management

journal homepage: www.elsevier .com/ locate/wasman

Anaerobic co-digestion of commercial food waste and dairy manure:Characterizing biochemical parameters and synergistic effects

http://dx.doi.org/10.1016/j.wasman.2016.03.0460956-053X/� 2016 Elsevier Ltd. All rights reserved.

⇑ Corresponding author.E-mail address: tatasp@rit.edu (T.A. Trabold).

Jacqueline H. Ebner a, Rodrigo A. Labatut b, Jeffrey S. Lodge c, Anahita A. Williamson a,d,Thomas A. Trabold a,⇑aGolisano Institute for Sustainability, Rochester Institute of Technology, Rochester, NY 14623, United StatesbDepartamento de Ingeniería Hidráulica y Ambiental, Pontificia Universidad Católica de Chile, Av. Vicuña Mackenna 4860, 7820436 Santiago, Chilec Thomas H. Gosnell School of Life Sciences, Rochester Institute of Technology, Rochester, NY 14623, United StatesdNew York State Pollution Prevention Institute, Rochester Institute of Technology, Rochester, NY 14623, United States

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

Article history:Received 19 September 2015Revised 22 March 2016Accepted 24 March 2016Available online 15 April 2016

Keywords:Biochemical methane potential (BMP)Anaerobic co-digestion (AcoD)BiowasteCommercial food waste

Anaerobic digestion of commercial food waste is a promising alternative to landfilling commercial foodwaste. This study characterized 11 types of commercial food wastes and 12 co-digestion blends. Bio-methane potential, biodegradable fraction, and apparent first-order hydrolysis rate coefficients werereported based upon biochemical methane potential (BMP) assays. Food waste bio-methane potentialsranged from 165 to 496 mL CH4/g VS. Substrates high in lipids or readily degradable carbohydratesshowed the highest methane production. Average bio-methane potential observed for co-digested sub-strates was �5% to +20% that of the bio-methane potential of the individual substrates weighted by VScontent. Apparent hydrolysis rate coefficients ranged from 0.19 d�1 to 0.65 d�1. Co-digested substratesshowed an accelerated apparent hydrolysis rate relative to the weighted average of individual substraterates. These results provide a database of key bio-digestion parameters to advance modeling and utiliza-tion of commercial food waste in anaerobic digestion.

� 2016 Elsevier Ltd. All rights reserved.

1. Introduction

Anaerobic digestion (AD) has been promoted for its ability togenerate clean renewable energy, treat waste and recycle nutri-ents. Early adoption of AD in the U.S. has primarily occurred onconcentrated animal feeding operations (CAFOs) where it also pro-vides odor reduction and increased manure management flexibil-ity. However, single substrate digestion (mono-digestion) ofmanure can result in low biogas yield due to low organic loadand high N concentrations of manure may lead to inhibition andprocess instability. Combining feedstock substrates or anaerobicco-digestion (AcoD) can increase organic loading and improve per-formance relative to mono-digestion by diluting toxic or inhibitorycompounds and providing macro or micro nutrients (Khalid et al.,2011; Mata-Alvarez et al., 2011). In addition, AcoD of manure andfood waste can improve project economics through additional rev-enue in the form of ‘‘tipping fees” for the imported food waste.Thus recent years have witnessed a trend toward AcoD with 98of the 260 farm-based biogas plants in the U.S. now co-digesting

additional feedstocks (USEPA, 2015). With this trend has comethe need to develop methods that could improve the performanceas well as the efficiency of this process, including analysis of co-digestion substrates to exploit their complementary characteristicsand the use of mathematical models simulating the AcoD process,as recognized by Esposito et al. (2012).

Currently, industrial food processing wastes and agriculturalwastes are the predominate co-digestion feedstocks (USEPA,2015). However, increasing regulation of organic disposal in land-fills is driving interest in AcoD among solid waste generators(Massachusetts, 2013). These landfill bans or mandates often tar-get commercial establishments that landfill large quantities of foodwaste. Commercial food waste is mainly composed of retail foodwaste and food service waste. Supermarkets are a large source ofretail food waste consisting of rotting produce, damaged packagedgoods or otherwise unmarketable product. Food service wasteconsists of scraps generated during food preparation as well aspost-consumer plate waste and un-served food. While a portionof commercial food waste may be reduced or diverted to feed thehungry, some commercial food waste is unavoidable. In fact, over40% of the food produced in the U.S. ends up in a landfill withoutreaching a table, from which 19% originates from the retail-levelfood supply (Gunders, 2012).

J.H. Ebner et al. /Waste Management 52 (2016) 286–294 287

Commercial food waste generated from different operationswithin an establishment or at different types of establishmentscan be categorized and often source separated. These waste prod-ucts can become valuable resources for renewable energy produc-tion when anaerobically digested or co-digested. While AcoD hasreceived increasing attention in the literature, most studies havefocused on the organic fraction of municipal solid waste (OFMSW),industrial wastes or agricultural wastes as co-digestion feedstocks(Mata-Alvarez et al., 2011). This study has collected a representa-tive array of commercial organic waste substrates to analyze asfeedstock for AD. The objectives of this research were threefold:(1) provide data on representative commercial food waste compo-sition; (2) provide key biodegradability parameters, namelybio-methane potential, degradation extent and hydrolysis ratecoefficients; and (3) assess potential synergistic or antagonisticeffects when these complex substrates are co-digested.

2. Methods

2.1. Substrate description

2.1.1. Source-separated commercial food wastesSamples of retail food waste were obtained from the food bank

for the Finger Lakes region of New York (Foodlink, Inc.) where non-distributable food was source-separated into several retail wastecategories: fruit and vegetable waste (FVW), stale baked goods(BG), damaged canned goods (CG), non-distributable yogurts andfrozen desserts (YFD), salad mix waste (SM), and dried goods,which were further separated into sweet (SDG) and unsweetened(UDG) dried goods (Table 1). Furthermore, the following kitchenwaste samples were obtained from the source-separated wastecollection bins of the Grace Watson dining hall (GWDH) at theRochester Institute of Technology: kitchen food preparation waste(PREP), spent coffee grounds and filter paper (COF), and post-consumer waste (POST) and soiled napkins (SN) from the returnedtrays after meals (see Ebner et al., 2014). Approximately 20 kg ofeach of the 11 substrate samples were collected. The substrateswere first manually mixed, and then ground, using a VitaMix�

blender (1825 Professional Series 750) to reduce particle size toless than 2 mm and produce a homogenous slurry or powder mate-rial. Samples were stored at 4 �C until prepared (approximately5 days) and then immediately frozen until used again.

2.1.2. Food sector co-digestion blendsSelected source-separated food wastes were combined to model

the potential waste streams of three commercial food sectors: (1)Cafe (CAFE) – combined BG and COF in a 60:40 proportion (% freshweight (%w/w)); (2) food service waste (SERVICE) – combinedPOST and PREP in a 80:20 proportion (% w/w) (see Ebner et al.,2014); and (3) retail (RETAIL) – combined FVW (57%), SDG (7%),BG (21%), CG (8%), YFD (7%) (% w/w) to replicate the reported com-position of the food bank waste.

2.1.3. Manure-food waste co-digestion blendsDairy manure slurry (M) was co-digested with food wastes and

sector blends in a 70:30 ratio (%w/w). This ratio was based upondata reported on New York State’s largest manure-based anaerobicco-digestion facility (Ebner et al., 2015). The dairy manure slurrywas obtained from the receiving pit of a dairy farm equipped witha scrape manure collection system. The 24 substrates evaluated aresummarized in Table 1.

2.1.4. Substrate characterizationTotal solids (TS) dry matter and Volatile solids (VS) were deter-

mined according to the APHA Standard Methods 2540B and 2540E,

which involves a gravimetric moisture determination at 105 �C,followed by an ashing (ignition) of the dried sample at 550 �C(APHA, 1998).

Crude protein was calculated from nitrogen (N) measurementusing a heated block digestion with copper catalyst, followed bysteam distillation into a boric acid solution per modified AOACMethod 984.13 (AOAC, 2012a). The sample was digested in sulfuricacid using copper sulfate as a catalyst. This converts bound nitro-gen into ammonia, which was distilled and titrated with standardacid. A 6.25 conversion coefficient was used to calculate proteinconcentration from measured total Kjeldahl nitrogen (TKN).

Crude fat was measured via solvent extraction per modifiedAOAC 991.36 (AOAC, 2012b). Soluble fat-based materials areextracted from dried test samples via a two-step submersion treat-ment with hexane solvent. The crude fat content was determinedby measuring weight after drying the hexane extracts.

Crude carbohydrates were calculated as the mass-balance dif-ference of the crude fat, protein, moisture (calculated from totalsolids measurement) and ash determinations. This is a generalizedapproach for certain types of foods and biosolids. An example ofthis approach can be found in AOAC Method 986.25, where thegeneral formula is presented as ‘‘Carbohydrate = total solids �(proteins + fat + ash)” (AOAC, 2012c).

2.2. Biochemical methane potential assay

Biochemical methane potential (BMP) assays were conductedbased upon the original protocol described by Owen et al. (1979)and revised by others (Angelidaki and Sanders, 2004; Angelidakiet al., 2009; ASTM, 2008; Hansen et al., 2004). A total of 149 assayswere prepared and conducted in 6 different phases. Microcrys-talline, 20-lm, cellulose (SigmaCell type 20) was used as positivecontrol samples across each phase. Inoculum was harvested fromthe post solid separated, effluent, from a full-scale, complete mixanaerobic digester operated at mesophilic temperatures that co-digested dairy manure with assorted food wastes (i.e.,whey, greasetrap waste, and fruit and vegetable processing waste). Inoculumwas pre-incubated at 37 �C for five days to minimize gas produc-tion from un-digested biomass. Samples were prepared to achievean inoculum to substrate ratio (ISR) of 2 (gVS inoculum: gVS sub-strate added) to prevent biomass limiting kinetics (Jensen et al.,2011). Total solids content was less than 3% in all prepared sam-ples. Basic nutrient requirements for anaerobic microorganismswere provided by the dairy manure-based inoculum (Gustafson,2000; Labatut et al., 2011). No additional external nutrients/traceelements were added in order to evaluate the synergistic effectsof co-digestion in providing these requirements. Measurementsof pH for each sample prior to the start of the test ranged from6.9 to 7.6. (Measurement at the end of the test ranged from 7.2to 7.9.) Samples were flushed with N2 to create an anaerobic envi-ronment and incubated at 37� (±1 C) with mixing at 10 s per min-ute. BMP vessels were 0.5 L with working volumes ranging from300 to 400 mL. Bio-methane production was measured continu-ously using the AMPTS II (Bioprocess Control). The efficiency ofthe CO2 fixing system was periodically verified by measuring CO2

and CH4 concentrations before and after entering the system usinggas chromatography (TCD with helium carrier gas and HaysepQpacked column). Bio-methane production of substrates, blanksand controls were adjusted to standard temperature and pressure(STP) conditions (i.e., 0 �C, 1 atm). The BMP assay was conductedfor 33 days, after which bio-methane production for all sampleshad reached a plateau. Blank samples containing only inoculum,were run in triplicate for each phase. Substrate bio-methane pro-duction was obtained by subtracting background methane produc-tion observed in the blanks.

Table 1Description and sources of substrates evaluated.

Source separated commercial food waste

Substrate Description Source

Baked goods (BG) Stale bagels, muffins and donuts. FoodlinkCanned goods (CG) Damaged cans of crushed tomatoes, diced tomatoes, green beans, beets, chicken broccoli soup, cream of chicken

soup, cheese pot pie soup, baked beans, papaya, pineapple chunks, tuna fish and mandarin oranges in damagedplastic cups

Foodlink

Coffee grounds (COF) Spent coffee grounds (medium roast) and coffee filter paper GWDHSweet dry goods

(SDG)Assorted breakfast cereals (Cocoa O’s�, Cap’n Crunch�, Shredded Wheat�, Lucky Charms�, Chex�, FrostedFlakes�) and dry goods (quick oats, pasta, Cliff� cereal bar)

Foodlink

Unsweetened drygoods (UDG)

Assorted grains (rice, oatmeal, bread crumbs, cream of wheat) Foodlink

Fruit and vegetablewaste (FVW)

Approximately 50% rotting bagged lettuce and 50% rotting whole or prepared fruit or vegetables (pineapple, melon,strawberries, grapes, tomatoes, oranges and blackberries.)

Foodlink

Napkins (SN) Soiled paper napkins GWDHPost-consumer (POST) Pieces of pizza crusts, French fries, mashed potatoes/gravy, ham scraps, home fries, chicken fingers,

salad/dressing/grated cheese, pancakesGWDH

Kitchen Preparationwaste (PREP)

Approximately 90% assorted melon rinds and seeds with balance consisting of rotting tomato, celery scraps, olives,kiwi peels, strawberry tops, carrot peelings and coffee grounds

GWDH

Salad mix (SM) Rotting lettuce and bagged lettuce mixes FoodlinkYogurt and frozen

desserts (YFD)Greek yogurt (chocolate), Low-fat ice cream (blueberry), sorbet (mango), frozen greek yogurt (black cherry) Foodlink

Food sector blends

Substrate Contents Composition (% w/w)

Food Service blend(SERVICE)

Post-consumer (POST) plate waste combined with kitchen preparation waste (PREP) POST:PREP (80:20)

Café blend (CAFÉ) Baked goods (BG) combined with coffee/filter paper (COF) BG:COF (60:40)Retail blend (RETAIL) Combination of the fruit and veg waste (FVW), sweet dry goods (SDG), baked goods (BG), canned goods (CG) and

yogurt and frozen desserts (YFD)FVW:SDG:BG:CG:YFD(57:8:21:7:7)

Dairy manure co-digestion blend description

Substrate Contents Composition (% w/w)

BG:M Baked goods (BG) and dairy manure (M) BG:M (30:70)CAFE:M Café mix (CAFÉ) and manure (M) CAFÉ:M (22:78)a

CG:M Canned goods (CG) and dairy manure (M) CG:M (30:70)FVW:M Fruit and vegetable waste:manure (FVW:M) FVW:M (30:70)POST:M Post-consumer:manure (POST:M) POST:M (30:70)PREP:M Kitchen Prep waste (PREP) and dairy manure (M) PREP:M (30:70)RETAIL:M Retail blend (RETAIL) and dairy manure (M) RETAIL:M (30:70)SDG:M Sweet dry goods (SDG) and dairy manure (M) SDG:M (30:70)UDG:M Unsweetened dry goods (UDG) and manure (M) UDG:M (30:70)

a Sample preparation error resulted in a non-standard co-digestion ratio for this sample.

288 J.H. Ebner et al. /Waste Management 52 (2016) 286–294

2.2.1. Bio-methane potential, BoSubstrate bio-methane production was normalized by VS to

report observed bio-methane potential (Bo). In addition to Bo, thestandard specific methane yield reporting on a basis of VS added(mL CH4/g VS), bio-methane potential was also reported basedupon fresh mass of substrate digested (Lo) (m3 CH4/t).

2.2.2. Theoretical methane potential Bu and extent of degradation, fdTheoretical methane potential (Bu) was calculated based upon

the organic fraction composition (OFC) as described by Raposoet al. (2011) and Nielfa et al. (2015) as follows:

Bu ¼ 415 �%Carbohydratesþ 496 �%Proteinsþ 1014 �%Lipid

ð1Þ

fd can be calculated by the ratio of Bo to Bu, as follows (Raposo et al.,2011):

f d ¼Bo

Buð2Þ

where fd is the extent of degradation or substrate biodegradablefraction (decimal), and Bo and Bu correspond to the observed andtheoretical bio-methane potential on a VS basis (mL CH4/g VSadded).

2.2.3. Hydrolysis rate coefficientHydrolysis is the rate-limiting step during the anaerobic diges-

tion of particulate materials (Eastman and Ferguson, 1981). Thusfor complex feedstock, parameters obtained from BMP tests shouldbe directly applicable to characterize biodegradability in modelssuch as the ADM1, i.e., the extent of degradation, fd and the appar-ent first order hydrolysis rate coefficient, kh (Batstone et al., 2002).

The rate of hydrolysis of the biodegradable fraction ofsubstrates was assumed to be first order and equivalent to thedifference between the observed bio-methane potential, Bo, andthe bio-methane production, B, at any given time, t.

dSdt

¼ �khðBo � BÞ ¼ �khðf dBu � BÞ ð3Þ

where S is the biodegradable substrate and t is time (d). The extentof degradation (fd) and apparent hydrolysis rate coefficient (kh)where estimated using the secant method of Aquasim 2.1 g thatsimultaneously fits these two parameters from the BMP data(Gustafson, 2000; Reichert, 2014).

2.3. Co-digestion performance index (CPI) and co-digestion rate index(CRI)

AcoD can result in increased bio-methane production when theorganic load of the combined substrate is higher than that of the

J.H. Ebner et al. /Waste Management 52 (2016) 286–294 289

original substrate. However, the combination of substrates can alsoresult in synergistic effects. Synergistic effects may arise from dilu-tion of inhibitory intermediaries, addition of valuable nutrientsthat result in increased bio-degradability, and/or a change in themicrobiome that results in an enhanced metabolism. Labatutet al. (2011) suggested comparing the bio-methane potential of aco-digested substrate with the weighted sum of the single sub-strate bio-methane potentials as a measure of synergistic or antag-onistic interactions. A co-digestion performance index (CPI) wascalculated as the ratio of the bio-methane potential of the co-digestion blend (Boi;n ) to the weighted average (Boi;n) based uponVS content (%VS) of the individual substrate bio-methane poten-tials (Bo;i):

CPIi;n ¼ Bi;n

Boi;n

¼ Bi;nPn

i %VSiBo;ið4Þ

where substrates i through n are co-digested such thatPn

i %VSi ¼ 1.Thus, a CPI > 1 indicates a synergistic effect of co-digestion andCPI < 1 indicates an antagonistic affect.

Similarly, a co-digestion rate index (CRI) was calculated to com-pare the apparent hydrolysis rate coefficient for co-digested sub-strates, khi;n with the rate obtained from a predictive curveresulting from adding the methane production curves of the indi-vidual substrates. The rate coefficient for the sum of two cumula-tive exponential decay curves could not be determinedmathematically. Therefore a simulation was used to determinean appropriate relationship between the rate coefficients of indi-vidual curves and that of the curve resulting from summing them;see Supplementary Material S.2 for details. One thousand sub-strates were simulated with parameters (kh and Bo) within therange of the data. The best estimate of the combined hydrolysis

rate coefficient was obtained by the geometric mean ðGkhi;n Þ ofthe individual substrate hydrolysis rate coefficients. Thus, the co-digestion rate index was calculated as the measured rate coeffi-

cient ðkhi;n Þ over the predicted rate coefficient ðGkhi;n Þ:

CRIi;n ¼ khi;nGkhi;n

¼ khi;nPni expðð%VSi � Bo;iÞ � lnðkh;iÞÞ=

Pni ð%VSi � Bo;iÞ

ð5Þ

where substrates i through n are co-digested such thatPn

i %VSi ¼ 1.The maximum bio-methane production for each constituent is thebio-methane potential of the substrate (Bo;i) weighted by the %VSof the substrate in the blend. A CRI > 1 indicates that co-digestionhad an accelerating effect on apparent hydrolysis rate and aCRI < 1 indicates that co-digestion had a slowing effect.

Table 2Substrate characterization.

Substrates %TS/FM %VS/TS TVS (%VS/FM)

Baked goods (BG) 91.6% 97.9% 88.9%Canned goods (CG) 10.5% 90.7% 9.6%Coffee grounds (COF) 29.3% 99.3% 29.1%Fruit and vegetable waste (FVW) 7.7% 93.3% 7.1%Post consumer (POST) 46.6% 97.1% 45.2%Preparation waste (PREP) 14.3% 100.0% 14.3%Salad mix (SM) 3.8% 90.6% 3.4%Soiled napkins (SN) 91.1% 100.0% 91.1%Sweet dry goods (SDG) 92.7% 95.0% 88.0%Unsweetened dry goods (UDG) 92.4% 97.8% 90.4%Yogurt and frozen deserts (YFD) 30.9% 97.9% 30.3%

Manure (M) 10.2% 83.6% 8.5%

NA = not measured.All samples were measured in triplicate.

a Rounding error may lead to nutrients not summing to 100% total solids.

3. Results and discussion

3.1. Substrate characterization

Characterization of the waste categories is shown in Table 2.Although all samples would be disposed of as solid wastes, severalsamples (canned goods (CG), spent coffee grounds (COF), fruit andvegetable waste (FVW), salad mix (SM) and kitchen prep waste(PREP)) had solid content <30%. All food wastes showed VS/TSratios over 90% (vs. 83.6% for manure). Measured carbohydratecontent ranged from 61% to 85% of TS. Protein constituted 10–20% of TS for most samples (with SM showing a higher contentand PREP waste a lower content). Post-consumer waste (POST)and stale baked goods (BG) contained the highest lipid content.

3.2. BMP test results

Key bio-methane kinetic parameters are summarized in theSupplementary Material Table S.1. Bio-methane potential of thecellulose controls across all phases showed good agreement withexpected results measuring 353 (r = 44) mL CH4/g VS (n = 15)and fd of 85%. The average apparent first-order hydrolysis rate coef-ficient for cellulose of 0.32 d�1 showed good agreement withJensen et al. (2011) who reported a kh based uponmethane produc-tion of 0.36 d�1 at an ISR of 2.

3.2.1. Bio-methane potentialDairy manure resulted in a Bo of 238 ± 19 mL CH4/g VS (n = 12),

which compares remarkably well with previously reported results(Labatut et al., 2011; Hoffmann et al., 2008; El-Mashad and Zhang,2010) (Fig. 1) Food service waste (SERVICE) resulted in 496 mLCH4/g VS which was the highest Bo observed; manure co-digested with kitchen prep waste (PREP:M) resulted in the lowestBo (165 mL CH4/g VS (r = 19)) (Fig. 1). All food waste substratesshowed higher Bo than dairy manure when digested alone. Sub-strates with high lipid content, such as POST and BG, resulted inhigher Bo. Raw fruits and vegetables (FVW) resulted in higher aver-age Bo than processed fruits and vegetables (CG) (although this wasattributed to the substrate composition as both substrates werenearly completely bio-degraded (Fig. 2)). Both fruit and vegetablesubstrates produced more methane than the purely vegetable,salad mix (SM) substrate. However, only SM and CG showed a sta-tistical difference based upon a pairwise student t-test at p < 0.05.This was attributed in part to the large variability observed in theFVW results. Several other substrates presented high variabilitynotably, FVW:M, UDG and POST waste which reported relative

Composition of solids (TS)a

% ash/TS % crude fat/TS % crude protein/TS % carbohydrate/TS

3% 11% 10% 76%9% 2% 15% 74%1% 4% 17% 79%7% 0% 10% 83%3% 21% 17% 59%0% 3% 15% 82%11% 2% 23% 65%NA NA NA NA5% 2% 11% 82%2% 1% 12% 85%2% 5% 14% 79%

16% 1% 14% 69%

Fig. 1. Standard bio-methane yield (Bo) for the substrates tested (mL CH4/g VS) shown in red with axis below graph. Methane yield per unit mass (Lo) (m3 CH4/tFW) shown inblue with axis above graph. Substrates were tested in triplicate (n = 3) unless otherwise noted. Error bars represent one standard deviation (r).

290 J.H. Ebner et al. /Waste Management 52 (2016) 286–294

standard deviations (RSD) of 30%, 27% and 19% respectively (where(RSD = r/l)). Potential sources of variability include substrate non-homogeneity and clumping of pulverized samples (UDG).

Both of the dried goods samples performed similarly (pairwisestudent t-test p > 0.05). SDG, demonstrated slightly higher degrad-ability than UDGwhich was unexpected as it contained higher con-centrations of glucose, fructose and crude lipids (Table S.1). Bothdried goods substrates demonstrated lower bio-methane poten-tials than the fruit and vegetable substrates, although statisticaldifference was only shown with the dried goods samples and CG(p < 0.05).

Results were shown to be reasonable when compared to similarsubstrates found in the literature. Gunaseelan (2004) tested 24fruit and vegetable wastes collected in South India and found sub-stantial differences among the varieties of FVW and even amongdifferent parts of the plant with methane yields ranging from

180 to 732 mL CH4/g VS. Cabbai et al. (2013) analyzed samples col-lected from Italian canteens, supermarkets, restaurants, fruit/veg-etable markets and bakery shops. Their supermarket and marketwaste contained only fruits and vegetables and ranged from 99to 363 mL CH4/g VS. This was lower than the results for FVW inthe current study, however, the composition of the wastes differed.The results reported by Cabbai et al. (2013) for bakery wasteshowed good agreement with the BG sample although the pastriesand fillings comprising the Italian bakery waste reported a higherlipid content. The bio-methane potential of the Italian food servicewastes were higher (571–675 mL CH4/g VS) than the SERVICE(496 mL CH4/g VS) and POST (483 mL CH4/g VS) samples in thisstudy which is again attributed to temporal and regional variation.Menardo and Balsari (2012) tested several waste substrates fromthe European retail market. This included a dairy waste substrateconsisting of waste milk, yogurt and cheese, which reported a

J.H. Ebner et al. /Waste Management 52 (2016) 286–294 291

bio-methane production higher (545 mL CH4/g VS) than YFD in thisstudy (454 mL CH4/g VS) which was attributed to the lower fatcontent of the U.S. dairy products. Menardo and Balsari’s resultsfor stale bread were consistent with the UDG sample in this study.There are limited reports of anaerobic digestion of coffee produc-tion waste and variation in substrate characteristics (i.e., TS, %lipid) can be observed in these studies (Dinsdale et al., 1996;Qiao et al., 2013). However, Neves et al. (2006) tested severalblends containing coffee and coffee substitutes and reported bio-

Fig. 2. Comparison of observed bio-methane potential (Bo) to theoretical bio-methaneexperimental data and estimated error of the theoretical calculation of 3% based upon mshown as a percentage in Table 1 in Supporting Materials.)

methane production consistent with the coffee ground/filter papersample (COF) in this study.

Results for bio-methane production were also expressed on afresh weight (FW) basis (Fig. 1). This illustrates the large effect thatmoisture content can have on substrate methane potential per unitmass. While the %VS/TS ranged from 90% to 100% for the commer-cial food waste substrates, the large variation in solids contentresulted in TVS ranging from 3.4% to 90.1% of fresh weight. Thishad a large effect on bio-methane yield per unit mass (Lo).

yield (Bu) calculated based upon OFC. Error bars indicate standard deviation of theethod error estimation. (The ratio of Bo/Bu is the extent of degradation (fd) and is

Table 3Apparent hydrolysis rate constants (kh) based upon substrate BMP testing.

Substrate Apparent hydrolysis ratecoefficientkh (d�1)a

Cellulose (C) (n = 15) 0.32 (0.032)Manure (M) (n = 12) 0.19 (0.111)Baked goods (BG) 0.26 (0.007)Canned goods (CG) 0.32 (NA)b

Coffee/filter paper (COF) (n = 6) 0.14 (0.009)Fruit and Veg Waste (FVW) 0.34 (0.010)Soiled napkins (N) (n = 2) NAb

Post consumer (POST) (n = 6) 0.27 (0.016)Kitchen prep waste (PREP) (n = 9) 0.48 (0.027)Sweet dry goods (SDG) 0.20 (0.003)Salad mix (SM) 0.64 (0.049)Unsweetened dry goods (UDG) (n = 6) 0.47 (0.033)Yogurt/Frozen desserts (YFD) 0.45 (0.059)Cafe blend (CAFE) (n = 6) 0.38 (0.011)Food service blend (SERVICE) 0.28 (0.015)Retail blend (RETAIL) (n = 9) 0.42 (NA)b

Baked goods:manure (BG:M) 0.27 (NA)b

Canned goods:manure (CG:M) (n = 6) 0.27 (0.007)Fruit/Veg waste:manure (FVW:M) 0.19 (0.005)Kitchen Prep:manure (PREP:M) 0.35 (0.014)Post consumer:manaure (POST:M) NAb

Retail blend:manure (RETAIL:M) (n = 6) 0.44 (0.019)Sweet dry goods:manure (SDG:M) 0.25 (0.011)Food service blend:manure (SERVICE:M) 0.30 (0.011)Unsweetened dry goods:manure (UDG:M)

(n = 8)0.41 (0.023)

a Standard error of non-linear parameter estimation shown in parenthesis.b Not available due to the nature of the data and the parameter estimation

software.

292 J.H. Ebner et al. /Waste Management 52 (2016) 286–294

Substrates with high solids content (baked goods, soiled napkinsand dry goods) resulted in Lo that were an order of magnitudehigher than those of substrates with higher moisture content.

3.2.2. Theoretical methane yield (Bu) and extent of biodegradation (fd)The extent of bio-degradation was calculated via Eq. (1) and

compares the observed bio-methane potential (Bo) to the theoret-ical bio-methane potential (Bu) (Fig. 2). Several substrates showedan extent of bio-degradation (fd) > 95%. These substrates wereobserved to be rich in readily hydrolysable carbohydrates (decayedFVW and processed CG) and fats (YFD, BG, CAFE and RETAIL). Thelowest conversion efficiency was observed in manure (54%), which

Fig. 3. Co-digestion performance index (CPI) of co-digestion substrates. CPI > 1 indicatesmanure; indicates food waste co-digestion blends. Error bars indicate standard devia

was attributed to a higher content of lignin or other recalcitrantcarbon than food wastes. Kitchen preparation waste (PREP) alsoresulted in low bio-degradability (56%); this could be due to thelignin content in the seeds and rinds of the preparation waste,nutrient deficiencies or inhibitory compounds.

Buswell’s equation (Buswell and Neave, 1930), which is thebasis of the methane potentials used for the organic fractions (pro-tein, carbohydrate, lipid) in the OFC method, is based upon a bal-anced redox equation where the substrate (and water) iscompletely converted to CH4 and CO2, therefore Bu should alwaysbe greater than the observed Bo due to cellular synthesis andincomplete digestion. Raposo et al. (2011) estimated the organicmatter consumed in microbial biomass to be near 15% for referencecarbohydrate and proteinaceous substrates, but cite literatureranging from 3% to 15%. In this study, some degradation extentsfor individual assays were observed near or >95% (CG, RETAIL,FVW, POST). The theoretical bio-methane yield (Bu) is subject touncertainty due to sample heterogeneity and the application ofthe OFCmethod of estimation applied. Heterogeneity in the samplemay have resulted in a difference between the sample character-ized (and in turn the calculated theoretical bio-methane yield(Bu)) and the tested substrate (resulting in observed bio-methanepotential (Bo)). This is supported by the large variability observedin the individual sample results. Furthermore, calculation of Bubased upon the generalized formulas for organic composition (car-bohydrate, protein and lipid content) is an approximation. In par-ticular, calculating a single form carbohydrate content as theremainder of nutrients does not account for more complex nutri-ents or structures (including lignocellulosic compounds).

3.2.3. Apparent hydrolysis rate coefficient (kh)Apparent first-order hydrolysis rate coefficients ranged from

kh = 0.14 (0.01) d�1 for coffee and filter paper (COF) to kh = 0.64(0.05) d�1 for salad mix (SM) (Table 3). The apparent first-orderhydrolysis rate for cellulose (0.32 d�1) agreed well with Jensenet al. (2011) whom reported first-order hydrolysis rate constants(kh) for ISR > 2 measured by cellulose solubilisation (0.36 d�1)and by methane production (0.41 d�1). It is noted that a lag phasewas observed for cellulose which impacted the standard error(0.032) for the fit of the first-order decay model. Other substrates(YFD, SM and UDG) also showed a high standard error (0.059,0.049, 0.033 respectively) indicating a poor fit to the first-order

synergistic effect, CP < 1 indicates antagonistic effect. Indicates co-digestion withtion.

Fig. 4. Co-digestion rate index (CRI) of co-digestion substrates. indicates substrates co-digestion with manure; indicates food waste co-digestion blends. (Standard errorassociated with estimating hydrolysis rate coefficients could not be used to estimate statistical significance.)

J.H. Ebner et al. /Waste Management 52 (2016) 286–294 293

decay model however none of the (non control) substrates exhib-ited a lag phase.

Generalizable conclusions regarding apparent first-orderhydrolysis rate coefficients are difficult to draw based upon sub-strate characteristics (Table 3). FVW and CG showed similar ratecoefficients of 0.32 d�1 and 0.34 d�1 respectively. However thesalad mix (SM) and the kitchen preparation waste (PREP) showedsignificantly faster degradation profiles. While SM had a higherprotein concentration suggesting an improved C:N ratio, PREPhad a lower protein concentration, yet both showed higher hydrol-ysis coefficients than FVW and CF. Interestingly, despite similarcompositions, unsweetened dried goods (UDG) resulted in a higherapparent hydrolysis coefficient than sweetened dry goods (SDG),suggesting that other factors beside composition, play a role indigestion kinetics.

3.2.4. Co-digestion parameters3.2.4.1. Co-digestion performance index (CPI). Co-digestion perfor-mance indices (CPI) ranged from 0.68 for PREP:M to 1.21 forUDG:M (Fig. 3) Nine of the 13 co-digested samples indicated a syn-ergistic effect, based upon mean Bo values, while 4 indicated anantagonistic affect. However, all but three of the samples did notshow an effect that was statistically different from the weightedaverage of the individual substrates (CPI = 1). Food service blend:manure (SERVICE:M) and canned goods:manure (CG:M) showeda statistically significant synergistic effect. This is presumed to bedue to synergistic mechanisms such as the buffering of volatilesolids in AcoD between manures and C-rich wastes as describedby Mata-Alvarez et al. (2011). The reason for the highly antagonis-tic effect observed for PREP:M was not evident. Near neutral pH atthe end of the assay did not indicate a build up of VFA and the highapparent hydrolysis rate coefficient observed for PREP appeared tobe moderated by the addition of manure resulting in a reduction inkh for the PREP:M mixture. Toxic or inhibitory compounds in thePREP waste are suspected although a review of the literature didnot reveal any insight; thus further characterization and testingis suggested. Nutrient deficiency is also a potential cause toconsider.

3.2.4.2. Co-digestion rate index (CRI). The range of apparent hydrol-ysis rates for co-digested substrates ranged from 0.19 d�1 for FVW:M to 0.44 d�1 for RETAIL:M (Table 3). Apparent hydrolysis ratecoefficient for the co-digested substrates was higher than the geo-

metric weighted average of the individual substrate coefficients for10 of the 12 co-digested substrates (Fig. 4). Only FVW:M and UDG:M resulted in co-digestion rate indices below 1 (0.80 and 0.95respectively). The co-digestion indices were for RETAIL (1.68) andCAFE blends (1.59) were the highest.

These results are in agreement with the observations of Astalset al. (2014) who reported a general improvement in processkinetics without a significant change in biodegradability whencomparing varying co-digestion mixtures of pure and slaughter-house carbohydrates, protein and lipids. They attributed theirresults to mitigation of inhibitory compounds, particularly dilu-tion of fat concentration and mitigation of long-chain fatty acids(LCFA) inhibition. The high CRI’s observed in this study for theRETAIL (1.68) and RETAIL:M blends (1.16) may be attributed tothis effect as lipid rich baked goods (BG) were a constituent ofboth RETAIL and RETAIL:M blends. However, other high lipidcontent substrates did not exhibit such a significant kineticsynergism (i.e., BG:M and SERVICE). It is worth noting that BGand POST, although high in lipid content for commercial foodwastes (11% TS and 19% TS respectively) have significantly lowercontent than the pure lipids or olive oil used in the Astals et al.(2014) study thereby resulting in less LCFA-related inhibition tomitigate. Another possible cause for the strong synergismsobserved in RETAIL and RETAIL:M may be the supply of nutri-ents or trace elements from the co-substrates. Whereas, additionof a nutrient medium as cited in the BMP protocol referenced byAstals et al. may have masked this type of synergy. As a furtherexample, combining BG with COF, both of which had higher lipidcontent resulted in a higher apparent hydrolysis rate coefficient(in the CAFÉ blend) than either of the individual substrates (kh,BG,COF = 0.38 vs. kh,BG = 0.26 and kh,COF = 0.14) and a co-digestionratio index of 1.59. The significant synergism observed may bedue to dilution of another inhibitory compounds such as theunidentified inhibition observed in digesting coffee grounds byLane (1983). Thus, the use of actual food waste substrates, alongwith information on their micro- and macro-nutrients is impor-tant to uncovering possible causes of synergism (or antagonism)observed in co-digestion mixtures.

4. Conclusions

Bio-methane potential was a result of substrate composition aswell as biodegradability. Substrates with high fat content resulted

294 J.H. Ebner et al. /Waste Management 52 (2016) 286–294

in higher bio-methane production. Substrates rich in readilyhydrolysable carbohydrates and fats showed high bio-degradability. Co-digestion resulted in bio-methane productionclose to that of the weighted average of the individual substrates(�5%/+20% on average). Co-digestion apparent hydrolysis ratesshowed an increase in 10 of 12 substrates which was attributedto dilution of inhibitory effects and improved nutrient balancesas substrate complexity increased. Macro-nutrient compositionalone was not sufficient to explain synergistic impacts pointingto other factors such as provision of micro-nutrients, build up/dilu-tion of inhibitory compounds.

Acknowledgements

We gratefully acknowledge Fritz Ebner for his contribution inperforming the simulations to justify the use of geometric meanas a basis for co-digestion hydrolysis rate (kh,i,n) parameter estima-tion. Funding provided by the New York State Pollution PreventionInstitute (NYSP2I) through a grant from the NYS Department ofEnvironmental Conservation, which also provided a GraduateResearch Assistantship for J.H. Ebner. Any opinions, findings, con-clusions or recommendations expressed are those of the author(s) and do not necessarily reflect the views of the Department ofEnvironmental Conservation.

Appendix A. Supplementary material

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.wasman.2016.03.046.

References

Angelidaki, I., Sanders, W., 2004. Assessment of the anaerobic biodegradability ofmacropollutants. Rev. Environ. Sci. Biotechnol. 3 (2), 117–129.

Angelidaki, I., Alves, M., Bolzonella, D., Borzacconi, L., Campos, J.L., Guwy, A.J.,Kalyuzhnyi, S., Jenicek, P., van Lier, J.B., 2009. Defining the biomethane potential(BMP) of solid organic wastes and energy crops: a proposed protocol for batchassays. Water Sci. Technol. 59 (5), 927–934.

AOAC, Official Methods of Analysis of AOAC International, 2012a. 19th ed., AOACInternational, Gaithersburg, MD, USA (984.13).

AOAC, Official Methods of Analysis of AOAC International, 2012b. 19th ed., AOACInternational, Gaithersburg, MD, USA (991.36).

AOAC, Official Methods of Analysis of AOAC International, 2012c. 19th ed., AOACInternational, Gaithersburg, MD, USA (986.25).

APHA, AWWA, WPCF, 1998. Standard Methods for the Examination of Water.Wastewater and Sludge, Washington, D.C.

Astals, S., Batstone, D.J., Mata-Alvarez, J., Jensen, P.D., 2014. Identification ofsynergistic impacts during anaerobic co-digestion of organic wastes. Bioresour.Technol. 169, 421–427.

ASTM Standard E2170-01, 2008. Standard Test Method for Determining AnaerobicBiodegradation Potential of Organic Chemicals under Methanogenic Conditions.ASTM International, West Conshohocken, PA.

Batstone, D.J., Keller, J., Angelidaki, I., Kalyuzhnyi, S.V., Pavlostathis, S.G., Rozzi, A.,Vavilin, V.A., 2002. The IWA Anaerobic Digestion Model No 1 (ADM 1). WaterSci. Technol. 45 (10), 65–73.

Buswell, E.G., Neave, S.L., 1930. Laboratory Studies of Sludge Digestion. IllinoisDivision of State Water Survey. Bulletin No. 30. 1930, Urbana, Illinois.

Cabbai, V., Ballico, M., Aneggi, E., Goi, D., 2013. BMP tests of source selected OFMSWto evaluate anaerobic codigestion with sewage sludge. Waste Manage. 33 (7),1626–1632.

Dinsdale, R.M., Hawkes, F.R., Hawkes, D.L., 1996. The mesophilic and thermophilicanaerobic digestion of coffee waste containing coffee grounds. Water Res. 30(2), 371–377.

Eastman, J.A., Ferguson, J.F., 1981. Solubilization of particulate organic carbonduring the acid phase of anaerobic digestion. Journal (Water Pollution ControlFederation), 352–366.

Ebner, J., Win, S.S., Hedge, S., Vadney, S., Williamson, A., Trabold, T., 2014.Estimating the biogas potential from colleges and universities. In:Proceedings of the ASME 2014 8th International Conference on EnergySustainability & 12th Fuel Cell Science, Engineering and TechnologyConference, ESFuelCell2014, June 30–July 2, Boston, MA, USA.

Ebner, J., Labatut, R., Rankin, M., Pronto, J., Gooch, C., Williamson, A., Trabold, T.,2015. Lifecycle greenhouse gas analysis of an anaerobic co-digestion facilityprocessing dairy manure and industrial food waste. Environ. Sci. Technol. 49,11199–11208.

El-Mashad, H.M., Zhang, R., 2010. Biogas production from co-digestion of dairymanure and food waste. Bioresour. Technol. 101, 4021–4028.

Esposito, G., Frunzo, L., Giordano, A., Liotta, F., Panico, A., Pirozzi, F., 2012. Anaerobicco-digestion of organic wastes. Rev. Environ. Sci. Bio/Technol. 11 (4), 325–341.

Gunaseelan, V.N., 2004. Biochemical methane potential of fruits and vegetable solidwaste feedstocks. Biomass Bioenergy 26 (4), 389–399.

Gunders, D., 2012. Wasted: How America is Losing Up to 40 Percent of Its Food fromFarm to Fork to Landfill. Natural Resources Defense Council Issue Paper. 26pp.

Gustafson, G.M., 2000. Partitioning of nutrient and trace elements in feed amongmilk, feces and urine by lactating dairy cows. Acta Agric. Scand. Sect. A – Anim.Sci. 50, 111–120.

Hansen, T.L., Schmidt, J.E., Angelidaki, I., Marca, E., Jansen, J.C., Mosbæk, H.,Christensen, T.H., 2004. Method for determination of methane potentials ofsolid organic waste measurement of methane potentials of solid organic waste.Waste Manage. 24 (4), 393–400.

Hoffmann, R.A., Garcia, M.L., Veskivar, M., Karim, K., Al-Dahhan, M.H., Angenent, L.T., 2008. Effect of shear on performance and microbial ecology of continuouslystirred anaerobic digesters treating animal manure. Biotechnol. Bioeng. 100, 38.

Jensen, P.D., Ge, H., Batstone, D.J., 2011. Assessing the role of biochemical methanepotential tests in determining anaerobic degradability rate and extent. WaterSci. Technol. 64 (4), 880.

Khalid, A., Arshad, M., Anjum, M., Mahmood, T., Dawson, L., 2011. The anaerobicdigestion of solid organic waste. Waste Manage. 31 (8), 1737–1744.

Labatut, Rodrigo A., Angenent, Largus T., Scott, Norman R., 2011. Biochemicalmethane potential and biodegradability of complex organic substrates.Bioresour. Technol. 102 (3), 2255–2264.

Lane, A.G., 1983. Anaerobic digestion of spent coffee grounds. Biomass 3 (4), 247–268.

Massachusetts Formally Proposes Commercial Food Waste Ban, July 12, 2013.Biocycle <http://www.biocycle.net/2013/07/12/massachusetts-formally-proposes-commercial-food-waste-ban/>.

Mata-Alvarez, J., Dosta, J., Macé, S., Astals, S., 2011. Codigestion of solid wastes: areview of its uses and perspectives including modeling. Crit. Rev. Biotechnol. 31(2), 99–111.

Menardo, S., Balsari, P., 2012. An analysis of the energy potential of anaerobicdigestion of agricultural by-products and organic waste. BioEnergy Res. 5 (3),759–767.

Neves, L., Oliveira, R., Alves, M.M., 2006. Anaerobic co-digestion of coffee waste andsewage sludge. Waste Manage. 26 (2), 176–181.

Nielfa, A., Cano, R., Fdz-Polanco, M., 2015. Theoretical methane productiongenerated by the co-digestion of organic fraction municipal solid waste andbiological sludge. Biotechnol. Rep. 5, 14–21.

Owen, W., Stuckey, D., Healy Jr., J., Young, McCarty, P., 1979. Bioassay for monitoringbiochemical methane potential and anaerobic toxicity. Water Res. 13, 485–492.

Qiao, W., Takayanagi, K., Shofie, M., Niu, Q., Yu, H.Q., Li, Y.Y., 2013. Thermophilicanaerobic digestion of coffee grounds with and without waste activated sludgeas co-substrate using a submerged AnMBR: system amendments andmembrane performance. Bioresour. Technol. 150, 249–258.

Raposo, F., Fernández-Cegrí, V., De la Rubia, M.A., Borja, R., Béline, F., Cavinato, C., DeWilde, V., 2011. Biochemical methane potential (BMP) of solid organicsubstrates: evaluation of anaerobic biodegradability using data from aninternational interlaboratory study. J. Chem. Technol. Biotechnol. 86 (8),1088–1098.

Reichert, P., 2014. AQUASIM a tool for simulation and data analysis of aquaticsystems. Water Sci. Technol. 30 (2), 21–30.

USEPA. AgSTAR Database of Livestock Digesters, May 2015 <http://www2.epa.gov/agstar/livestock-anaerobic-digester-database>.