Thermophilic anaerobic digestion of pasteurised food wastes and dairy cattle manure in batch and large volume laboratory digesters: Focussing on mixing ratios

Thermophilic anaerobic digestion of pasteurised food wastes and dairy cattle manure in batch and large volume laboratory digesters: Focussing on mixing ratios

Renewable Energy 80 (2015) 432e440 Contents lists available at ScienceDirect Renewable Energy journal homepage: www.elsevier.com/locate/renene Ther...

2MB Sizes 0 Downloads 58 Views

Renewable Energy 80 (2015) 432e440

Contents lists available at ScienceDirect

Renewable Energy journal homepage: www.elsevier.com/locate/renene

Thermophilic anaerobic digestion of pasteurised food wastes and dairy cattle manure in batch and large volume laboratory digesters: Focussing on mixing ratios Ioannis S. Zarkadas a, *, Artemis S. Sofikiti a, Evangelos A. Voudrias b, Georgios A. Pilidis a a b

University of Ioannina, Department of Biological Applications and Technologies, Laboratory of Environmental Chemistry, 45110 Ioannina, Greece University of Thrace, Department of Environmental Engineering, Laboratory of Solid and Hazardous Waste Management, 67100 Xanthi, Greece

a r t i c l e i n f o

a b s t r a c t

Article history: Received 1 March 2014 Accepted 6 February 2015 Available online 5 March 2015

The potential of pasteurised food wastes in mixtures with cattle manure as feedstock for anaerobic digesters was assessed in batch and high volume laboratory digesters under thermophilic conditions. While food wastes is an attractive substrate for anaerobic digestion plants, their characteristics, especially the high nitrogen content, renders their treatment problematic. During this study, for the different mixtures of cattle manure and food wastes, methane yields of 281e385 m3CH4/tonVSadded have been achieved in organic loading rates of up to 6.85 kgVS/m3d with the TS levels of the influent reaching as high as 15.7%. However, as the OLR and TS levels of the influent stream increase, the specific methane production is adversely affected. Addition of 25% (w/w) food wastes to thermophilic digesters treating cattle manures can be considered safe and results in an improvement of the specific methane production by 86%, the volumetric methane production by 430% and the VS reduction by 35.2% compared to cattle manure monodigestion. © 2015 Elsevier Ltd. All rights reserved.

Keywords: Thermophilic anaerobic digestion Waste management Pasteurised food waste Cattle manure Biogas production Canteen wastes

1. Introduction Redirection of organic municipal wastes away from landfills is one of the challenges that waste managers face every year. Only in Greece more than 2.7 million tons of municipal organic wastes are generated annually [28]. Currently most of these are landfilled, resulting in wastage of a resourceful substrate, over exploitation and pollution of surface and ground waters, as well as in releases of greenhouse gases into the environment. The municipal organic wastes are mainly composed of cooked and uncooked food [2] while their characteristics present wide temporal and worldwide spatial variability, as cooking and nutrition habits vary among different human communities. Furthermore, the characteristics of food wastes (FW) such as high nitrogen [7], fat, moisture and low

Abbreviations: FW, food waste; CM, cattle manure; TOC, total organic carbon; VFA, volatile fatty acids; TS, total solids; VS, volatile solids; TAD, thermophilic anaerobic digestion; HRT, hydraulic retention time; OLR, organic loading rate; SRT, solids retention time; AD, anaerobic digestion process; s-CSTR, semi-continuous stirred tank reactors. * Corresponding author. Tel.: þ30 265108007; fax: þ30 2651007274. E-mail address: [email protected] (I.S. Zarkadas). http://dx.doi.org/10.1016/j.renene.2015.02.015 0960-1481/© 2015 Elsevier Ltd. All rights reserved.

pH (~5), may create significant problems when biological treatment methods are applied for the treatment of such wastes. The anaerobic digestion process (AD) provides a waste management option for FWs, while offers the opportunity for recovering marketable products both in the forms of biogas and slow release bio-fertilizers. AD is a waste treatment method which converts, with the utilization of microorganisms, the organic matter of wastes or wastewaters into biogas, which can be further exploited as a sustainable fuel for energy generation. As a result, fewer wastes are dumped into landfill sites, while at the same time the process can be used by local authorities to meet the waste redirection targets set by the European Community Landfill Directive (1999/31/EC). AD is commonly applied throughout the world as a wastewater treatment method for livestock manures with the methane recovery yields ranging between 12 and 13.9 m3/m3 of influent substrate [37]. However, in order to enhance economic viability of an anaerobic digestion system, yield of influent stream higher than 30 m3 biogas per ton waste must be achieved [15]. As a way to improve the bio-methane production of AD systems, different wastes-wastewaters can be mixed and treated together in codigestion schemes. This mixing of different substrates is not only desirable for improving methane recovery rates and

I.S. Zarkadas et al. / Renewable Energy 80 (2015) 432e440

reducing life cycle costs, but it also provides better organic load removal efficiencies as an effect of C/N ratio correction, pH balancing and improvement on the buffering capacity of the treatment systems [5]. A number of researchers have identified the FWs as an attractive substrate for AD due to their characteristics includes the high biodegradable organic content and the biomethane volume that can be recovered [22]. Nevertheless generalised inhibition phenomena of the process can be developed due to the high nitrogen availability of the wastes. It is well known that during the hydrolysis step of the AD process, proteins are converted into ammonia [9], which is highly toxic towards methanogens, even at concentrations lower than 1.7 g/L [19]. Moreover, high ammonia concentrations into the digester can result in high ammonia gas releases through the biogas. In case that this biogas is utilised as a fuel for electric energy production, it could lead to high NOx releases into the atmosphere, as well as corrosion to certain parts of the reciprocating engines widely employed as power plants for driving electric energy generators. In order to combat inhibition of the process and possible NOx releases into the environment, a “carrier” substrate must be selected and co-digested together with the FW. For proper application of AD this co-substrate should present the following characteristics: a) high C/N ratio, b) neutral to alkaline pH (taking into account that food wastes are acidic pH ~5), c) being widely available and cheap, d) containing low concentrations of lipids and oils, as these together with the high concentrations available in the food wastes could result in the flotation and washing of the microorganisms or inhibition of the process by long chain fatty acids accumulation [29], e) being liquid with low concentrations of TS and finally f) to provide anaerobic microorganisms into the digester, which will not only help the system to better utilise the organic load of the wastes, but it will also assist in the recovery of the system in the case of process failure. A wastewater which requires treatment and fulfils all of the above criteria is the dairy cattle manure (CM), while its abundance and continuous annual production ensures the required volumes essential to centralised AD plants. After the implementation of the EC Nitrates Directive (91/676/ EEC), livestock breeders are requested to handle the wastes derived by their operations in an environmental friendly manner. Nevertheless, most of the farmers either due to shortages of available land for field application of the manures or due to unavailability of treatment systems, fail to meet the targets of the directive, risking for heavy penalties. It is well documented that the application of untreated manures on land, despite the fact that is an environmental sound operation for improving soil fertility, might result in surface and ground water contamination both by nitrates and pathogens, with great consequences on the environmental quality and human health. While contamination incidents of potable water sources with pathogens due to mistreatment of manures are scarce in the developed world, only between 1990 and 2000 more than 100 people died and 500.000 were diagnosed with illnesses traced to the consumption of pathogenic contaminated water as a result of livestock manures mismanagement [17]. Thermophilic anaerobic digestion on the other hand, offers sanitization of the substrate in relation to E. coli and Salmonella spp. in solids retention times (SRT) higher than 16 days [23] and complete weed seeds mortality in hydraulic retention times (HRT) higher than 2 days [18]. The objectives of this work was to investigate the thermophilic anaerobic digestion of FW present in the municipal solid wastes together with CM, in large volume laboratory semi-continuous stirred tank reactors (s-CSTR) with special attention given to inhibitory phenomena and the organic loading of the systems. Furthermore this study aims on characterising the FWs generated

433

in the Mediterranean countries and their bio-methane potential under thermophilic conditions. While a number of researchers have examined the food wastes as candidate substrate for AD systems[11,12,15,25,35,37] most of them focused on the mesophilic digestion of such wastes and in low organic loading rates of less than 5.0 KgVSm3-d. To the knowledge of the authors this is the first time that FWs (up to 30% w/w addition to CM) are examined under thermophilic conditions in high OLRs (up to 6.85 KgVSm3-d) in pasteurised and undiluted form. 2. Material and methods 2.1. Substrates and inoculum FW used in the present study were collected from the student restaurant of the University of Ioannina, Greece. During sampling, the wastes collected by the cleaning crew after lunch were categorised, weighted and through coning and quartering a 20 kg representative sample was collected daily. The waste samples were then transferred to the laboratory where bones and foreign object were removed, then minced with a bench top disk cutter through a 3 mm screen and pasteurised in a laboratory oven according to the Animal By-Products Regulation (EC 1774/2002). A 12 L unpressurized stainless steel vessel with lid was used for pasteurisation. After cooling the wastes were freezed to 20  C. At the end of the sampling period the wastes were removed from the freezer, homogenized with the application of a ribbon blade mixer operating at 40 rpm for 6 min, separated in 10 kg batches and finally refrigerated until their utilisation. The composition of the FW used in the present study based on weight is presented in Fig. 1, while their physicochemical characteristics in Table 1. The FW were composed of meat (chicken, pork, beef and mixed meat products), rice, pasta, legumes, bread, salads (including olive oil, vinegar and lemon juice that applied to the salads during preparation), fish as well as sweets. Cattle manures were collected from a non-grazing intensive dairy farm located 10 km away from the University and from the underground manure collection tanks. Thermophilic anaerobic inoculum was acquired from 4 anaerobic digesters (R1-R4) of 50 L operating under steady state within the facilities of the University. In order to acclimate the inoculum to FWs, the digesters, prior to the initiation of the experiment, were

Fig. 1. The semi-continuous system experimental set-up.

434

I.S. Zarkadas et al. / Renewable Energy 80 (2015) 432e440

Table 1 Characterization of the substrates used in the present research and comparison with FW used by other researchers (*TKN concentrations for FW are presented as g/kg due to the consistency of the substrate). Waste composition

Present research (FW)

TS (%) VS/TS (%TS) TKN (g/l, g/kg*) TOC (%TS)

33.9 94.9 7.43 51.2

pH Lipids (%TS)

5.15 ± 0.14 17.9 ± 1.54

± ± ± ±

2.31 1.83 0.61* 2.10

Zhang & Jahng, (2012) [36] (FW)

Dai et al. (2012) (FW)

Zhang et al. (2007) (FW)

0.23 2.40 0.23 3.14

18.1 94.4 e e

21.2 ± 2.0 92.8 ± 2.2 e e

7.32 ± 0.13 4.94 ± 0.47

e 12.8

4.7 ± 0.7 e

30.9 85.3 3.16 46.78 (C) e e

Present research (CM) 6.73 84.7 1.40 40.0

± ± ± ±

sustained with a mixture of CM and FW 95-5% w/w, while pulses of 1.0 kg of FW (corresponding to an OLR of approximately 9 g/VS L) mixed with inoculum to facilitate pumping was added two times only to reactor 4 (R4) in order to assess the response of the microorganisms. Apart from some foaming observed in reactor R4 and the increase of the VFA levels up to 2.2 g/L, no further problems were observed and the volumetric biogas production returned back to the CM and FW 95-5% levels after 5e7 days. After this period and 25 days prior to the initiation of the co-digestion experiment the feeding for all 4 reactors was altered to CM mono-digestion. 2.1.1. Biogas measurement and analysis Biogas production measurements for the s-CSTR systems were conducted in a 4 channel electronic device operating according to the liquid displacement principle. This apparatus and its operation have been described in detail elsewhere [34]. Biogas production for the batch experiments was measured with a similar apparatus, as the s-CSTR systems metre, lacking the electrical and electronic parts. After each measurement the biogas was manually released into the atmosphere in order to reduce the pressure from inside the vials. All values were subsequently converted into standard temperature and pressure conditions. Biogas analysis was carried out every three and five days for batch and s-CSTR systems respectively. For the s-CSTR systems the sampling was taking place just before the second daily feeding of the reactors. The analyses were performed with a Shimadzu GC 2014 coupled to a thermal conductivity detector. For the separation of the gases a CARBOXEN 1000-60/80 column of 15 ft x 1/8 inch was used. 2.1.2. TC, VFA, TS-VS, total Kjeldahl nitrogen, ammonia, fats-oils and pH analyses A Shimatzu TOC-VCPH carbon analyzer coupled to a solid state combustion unit- SSM-5000A was used for TC and TOC analysis. For the VFA analysis the samples were centrifuged at 2761 Gunits for 10 min and the supernatant was analysed by a Shimadzu GC-17A Gas Chromatograph equipped with a flame ionization detector (FID). TKN was analysed with a Hach digestahl digestion apparatus (Hach method 8075). Ammonia analyses were conducted with the Salicylate method (Hach method 10031). Total solids, volatile solids and pH were analysed according to Standard Methods, [6]. Soxhlet extraction with petroleum ether (40e60  C) was used for quantification of fat and oil concentration. 2.1.3. Experimental set-up Two types of reactors, batch and s-CSTR, were used. In batch experiments, mixtures of FW with CM were digested in triplicate in 118 mL glass bottles. During the digestion one batch per experimental series was used sacrificially in order to extract liquid samples for VFA and pH analyses. The three different mixtures were prepared from the same waste-wastewater with the addition of FW

± ± ± ±

0.07 0.28 0.22 (N% TS) 1.15

Zhang et al. (2012) [35] (FW) 18.5 91.8 2.2 46.5

± ± ± ±

0.1 0.2 0.30 (N% TS) 1.50 (C)

5.2 ± 0.3 e

Cho et al. (2013) (FW)

Browne & Murphy, (2013) [10] (FW)

21.1 82.4 (TN) 13 e

29.4 95.3 3.53 (N% TS) 49.58 (C)

3.9 e

4.1 19

ranging between 10 and 30% by weight. The characteristics of the mixtures are presented in Table 2. The batch experiments were prepared as described by Angelidaki et al. (2009)[4] with a reaction time of 29 days. The bottles were filled with active inoculum (50 mL), 1 gVS from the waste-wastewater mixtures and distilled water to the final working volume of 70 mL in a VSinoc to VSsubst ratio of approximately 1.5:1. Subsequently nitrogen gas was fed within the liquid for 5 min in order to immediately expel all oxygen. Finally the vials were sealed with butyl rubber stoppers and aluminium clips and incubated at 55 ± 1  C. In the second phase of experiments the three waste mixtures treated in 4 identical 50L stainless steel s-CST Reactors (Fig. 2) with the OLR of the systems to be governed by the TS and VS content of the produced waste mixtures with no dilution to be employed. Heating was provided by a purpose built water-bath of 40 L, while feeding and removal of the effluent were controlled by 8 identical stainless steel-plastic 12 V DC macerator pumps. The digesters were operated with a HRT of 21 days similar to the 20 days HRT found as optimal for the mesophilic AD of similar substrates by El-Mashad and Zhang, (2010). Feeding was provided 7 days per week 2 times per day at 12 h intervals, while throughout mixing was ensured with the application of four geared 12 V DC motor agitators coupled to immersed propellers operating at 60 rpm for 3 min every 20 min. This high mixing frequency was chosen in order to minimize the possibility of foam or scum formation within the reactors, which was recently found to occur at OLRs higher than 4 kgVSm3-d for protein and lipid rich wastes [20]. In order to increase the SRT and as a consequence the VS reduction and to improve methane yields, the stirring of the reactors was stopped 45 min prior to the removal of the effluent. The increase of the SRT was achievable due to the design of the reactors where the effluent lines were installed to the mid depth of the systems, allowing in this way the removal of mainly liquids by permitting the settling and/or flotation of the solids under no stirring conditions. Due to the fact that R4 where 30% FW was added during the first two co-digestion HRTs show signs of overloading and reduced removal efficiencies, the feeding to this reactor was altered after day 58 with the reduction of FW addition to 25% by weight (Table 2).

Table 2 Characterization of the different mixtures of CM and FW. Mixture

FW (% w/w)

CM (% w/w)

TS (%)

1 (CM) 2 3 4 5

e 10 20 30 25

100 90 80 70 75

6.73 9.87 13.02 15.68 14.29

± ± ± ± ±

0.2 0.3 0.2 0.6 0.3

Average OLR VS (KgVS/m3-d)

pH

Feed based on VS (%) CM

FW

2.71 3.94 5.52 6.85 6.19

7.32 7.04 6.57 6.30 6.49

100 61.5 41.5 29.2 34.7

0 38.5 58.5 70.8 65.3

I.S. Zarkadas et al. / Renewable Energy 80 (2015) 432e440

435

Fig. 2. Composition of FW used in the present research (based on weight).

Table 1 compares characteristics of FW used in the present study and other studies available in the literature. FWs contain high concentration of TS (18.1e30.9%), VS (82.4e95.3% TS), fat (12.8e19% TS) and an acidic pH (3.9e5.2) with the concentration of carbon being close to 50% of the TS. The substrate used in the present study was within the reported values in the literature for FW except for TS and TOC for which it was presenting higher values by approximately 3% and at least 1.5% respectively. The high concentration of organic carbon and fats makes FW particularly attractive for anaerobic digestion systems, since large volumes of methane can be recovered during digestion. Besides, the low pH, although considered as an inhibitory factor for AD systems, in this case, together with the high nitrogen content of FW is advantageous and can provide stabilization of the systems by sustaining the pH lower than 8, while reducing in the same time the concentrations of free ammonia which is the main inhibitor of thermophilic digesters treating nitrogen-rich wastes. 3. Results and discussion 3.1. Batch digestion experiments During the batch digestion of the different mixtures (Fig. 3A) no problems related to inhibitory phenomena were observed and the biogas production initiated rapidly, with most of the biogas being produced within the first 10 days. A slight reduction of the pH up to 0.3 units was observed during the first 6 days of the experiment with a subsequent increase in the range of 7.9 and 8.2 units for all batches. The production in terms of mLCH4/gVSadded, (Fig. 3B) for the codigestion mixtures ranged between 272 and 408 mL with the highest yield produced by the mixture containing 30% FW (w/w) (Mix. 3). On the other hand CM mono-digestion produced 215 mLCH4/gVSadded, which is in agreement with the findings of Hartmann and Ahring (2005) [15] for the thermophilic digestion of dairy cattle manures with 200 mLCH4/gVSadded. The concentrations of volatile fatty acids increased during the first 5 days of the experiment, for the mixtures containing 20% and 30% FW to the

maximum observed concentration of 1.5 and 1.8 g/L respectively, with a subsequent reduction of their concentrations and their complete consumption before the end of the experiment for all batches. The maximum productivity of the different mixtures (Fig. 3C), calculated at 314 mLCH4/LR-d for the CM, peaked at 536 mLCH4/LRd for the mixture containing 20% FW and reduced to 478 mLCH4/LRd for the mixture containing 30% FW. This reduction of the daily productivity when 30% FW is added into the process may be attributed to the existence of large concentrations of fats and oils within the substrate. Degradation of fats is known to take place through b-oxidation [27], which is a slow process. This however, results in the production of biogas with greater content of methane compared to biogas produced through acetotrophic pathways. VS reduction for the different mixtures is shown in Fig. 3C. As illustrated, the removal efficiencies of VS are significantly improved with the addition of FW into the process from the 51.9 ± 2.4% achieved by the batches treating CM as single substrate, to 60.1 ± 3.1%, 67.4 ± 2.9% and 72.7 ± 3.3% for the by 10, 20 and 30% addition of FW to CM, respectively. Furthermore, the quality of the generated biogas was also improved with the addition of FW into the mixtures. The methane content of biogas was increased due to the addition of FW by 8.4, 15.1 and 22.3%, for mixtures 2, 3 and 4 respectively, when compared to the 53.9% of the control batches (CM mono-digestion). Increased CH4 concentration in the biogas has also been observed by Brown and Li (2013) [8]; when FW was used as co-substrate in batch digestion with yard wastes and which may be attributed to the high lipid and protein content of the FWs. In the literature a number of publications related to the production of CH4 from the co-digestion of livestock and food wastes in batch systems are available however these are related to mesophilic digestion. El-Mashad & Zhang (2010) [12] studied the co-digestion of such wastes in mixtures of 32:68 and 48:52 (food waste: livestock wastes) with an organic loading rate of 3 kgVS m3 and digestion time of 30 days, concluding that the specific CH4 production in the first mixture is 282 mL and in the second 311 mL with the methane content in the generated gases ranging between 62 and 59%. Ye et al. (2013) [33] studied the co-digestion of rice

436

I.S. Zarkadas et al. / Renewable Energy 80 (2015) 432e440

Fig. 3. Results of the batch systems: (A) Daily specific methane production and pH development versus time. (B) Cumulative methane production and Total VFA concentrations versus time (C) Productivity of the different substrate mixtures and the VS removal efficiencies for the different mixtures at 29 days digestion time.

straw, kitchen waste and pig manure in batch digesters and concluded that the maximum methane yield was 384 mL/gVSadded, in a mixture of 0.4:1.6:1, respectively, for the three substrates, accompanied by a VS reduction of 55.7%. Large scale inhibition phenomena were also identified in this research, when kitchen wastes addition was greater than 26%, with the levels of total VFAs reaching as high as 28 g/L in some of the batch systems.

In the present study, the specific methane production during the batch digestion was found to be about 6e44% higher compared to the ones obtained by other investigators. By taking into consideration that all the above described studies examined the substrates under mesophilic conditions, the increased production achieved by the present study can be justified by the higher temperature that these experiments took place (55  C), as it is well known that under

I.S. Zarkadas et al. / Renewable Energy 80 (2015) 432e440

thermophilic conditions the processes kinetics are improved, resulting in better utilization of the substrate. Several studies have examined and confirmed the improvements that the thermophilic digestion offers, mainly due to the increase in the hydrolysis rate, when compared to the mesophilic digestion [14,30]. Furthermore at thermophilic temperatures the growth rate of the methanogens increases by up to 3 times compared to mesophilic [16], something that allows better organic matter utilization in higher loading rates and lower retention times. On the other hand limited variations in the composition of the different substrates and the studied mixtures among the different studies, can easily affect the results and offer higher or lower methane yields. 3.2. CSTR digestion experiments 3.2.1. Daily methane production The operational conditions for the s-CSTR systems are presented in Table 2 while the results of the daily methane production in Fig. 4A. During the semi-continuous experiment, the operation of the first reactor (R1) was steady with the methane production ranging between 17 and 22 L/d with an average value of 19.6 L/d (i.e 0.56 LCH4/LR-d). For R2, where 10% of FW was introduced, a steady increase of the daily methane production during the first codigestion HRT was observed with the pick of the daily methane production achieved around date 50. For this reactor a 98% (i.e. þ19.1 LCH4-d) increase in methane production was observed, when compared to R1, offering on average 38.6 LCH4-d (i.e. 1.10 LCH4/LR-d). This production is in agreement with the values re n et al. (2013) [11] for the addition of 10% FW in ported by Castrillo TAD systems with 1.18 LCH4/LR-d. The reactors where higher concentration of FW 20:80 (R3) and 30:70 (R4) were introduced, showed a less stable operation during the startup period (Fig. 4A). The third reactor (R3) showed instability in its operation during the first 20 days of the co-digestion period, which may be attributed to difficulty in the acclimation of the microorganisms to the incoming substrate that contained high concentrations of total solids (13.02 ± 0.2%). The methane production for this reactor peaked between days 50 and 52 (very similar to that of R1) followed by a reduction in the daily volumetric production by 16.4% on day 64 and then stabilization of the production to levels above 69 LCH4d for the rest of the experiment. Nevertheless, R3 during the first 2 co-digestion HRTs offered the highest volumetric methane yield, producing on average 71.4 LCH4-d (i.e 2.04 LCH4/LR-d). The methane production for R4 during the co-digestion period was very erratic and showed strong fluctuations of up to 34.6% between days 18 and 20 while the process seemed to become stabilized after day 26 with fluctuations on the daily CH4 production of up to 23% still common. However, despite the higher OLR that this reactor was operating at, the daily volumetric methane production was very similar to that of R3 with an average value of 69.5 L (i.e. 1.98 LCH4/ LR-d), indicating that an inhibitor was present and responsible for the lower methane yields. This lower methane production was not accompanied by accumulation of volatile fatty acids and/or drop of the pH, something indicating ammonia inhibition. Thus, it was considered that responsible for the lower yields were either the lack of trace elements that are necessary for hydrolysis or acidogenesis as Qiang et al. (2012) [31] propose, or decreasing hydrolysis rate as an effect of the high TS levels of the incoming substrate (TS 15.7 ± 0.6%), something that has already been described by Abbassi-Guendouz et al. (2012) [1]; for substrates with TS levels greater than 10%. 3.2.2. Specific methane production, pH and VFA Comparing the results of the different reactors based on the production of mLCH4/gVSadded it becomes apparent that by adding

437

FW even as high as 30% leads to an increase of the specific methane production by 35.6 and 78.7% compared to the CM mono-digestion. The highest specific methane production was derived by reactor (R3) which processed 20% FW and 80% CM w/w with an average yield of 370 mLCH4/gVSadded (Fig. 4B). On the other hand, reactor (R4) offered a yield of 290 mLCH4/gVSadded, which is significantly reduced compared to the 408 achieved during the batch digestion of the same mixture. The pH levels (not shown) throughout the semi-continuous experiment, even for R4 which showed signs of reduced conversion efficiencies, stayed at relatively constant levels, ranging between 7.8 and 8.2 units. The concentrations of total volatile fatty acids (Fig. 4C) were at quite low levels throughout the experiment. The highest concentrations of VFAs 1.34 and 1.51 g/L respectively were observed during the 22nd day of the s-CSTR operation for reactors 3 and 4, followed by a sharp decrease to levels below 0.4 g/L for the rest of the experimental period for R3 and day 65 for R4. Removal of VS ranged between 53.3 ± 2.8% and 73.9 ± 2.1% for the different reactors with the maximum removal efficiency derived by R3, while R4 during the first assessed period achieved a VS reduction of 62.3 ± 4.7% with reducing removal efficiencies between the different HRTs, something that reflecting the inhibited state that it was operating at. 3.2.3. Ammonia Ammonia levels within the reactors at the end of the second codigestion HRT were found to vary between 0.71 ± 0.16 g/L for R1, 1.14 ± 0.31 for R2 and 2.11 ± 0.20 g/L for R3, while ammonia levels in R4 were 1.83 ± 0.22 g/L, indicating free ammonia nitrogen (FAN) concentrations of 0.24 ± 0.05, 0.31 ± 0.08, 0.58 ± 0.06 and 0.49 ± 0.06 g/L for the four reactors respectively. While these ammonia concentrations for R3 and R4 are well within inhibitory levels as it has been proposed by others [19], no inhibition of the methanogens was observed even in concentrations of 2.11 g/L something that is coming in agreement with the findings recently published by Wang et al. (2013) [32]; who concluded that TAN concentrations of 4.3 g/Kg might set the three groups of anaerobic microorganisms under stress, however they are still viable. Furthermore, the lower concentration of ammonia in R4 when compared to R3 is another sign that the lower specific and volumetric methane yields observed for this reactor were due to decreased hydrolysis rate, rather than trace element deficiencies or ammonia inhibition, as both the latter resulting to the accumulation of VFAs and acidification of the reactors [26,36]. 3.2.4. s-CSTR experiments for R4 after the second co-digestion HRT In order to tackle the inhibition problem observed in R4, its feeding was altered with the reduction of FW addition by 5% (Table 2, Mixture 5 on day 58) which reduced the TS levels of the influent stream to 14.3 ± 0.3% and the OLR to 6.19 gVS/LR-d. This alteration in the incoming mixture resulted to a slow but steady improvement on the daily methane production reaching as high as 90.8 L on day 96 (i.e. 2.59 LCH4/LR-d) with an average yield of 83.3 LCH4-d (i.e. 2.38 LCH4/LR-d). This indicates an improvement of approximately 20% compared to the volume of CH4 produced prior to the alteration of the feeding mixture. An increase in specific methane production from 290 mL to 385 mLCH4/gVSadded was observed, representing an improvement of 32.6%. Furthermore, after the alteration of the feeding mixture, R4 showed a steady operation with the daily methane production fluctuating only within 8% of the average daily production. A negative side effect of the alteration of the influent substrate was the increase of the VFA levels within this reactor. While during the first 65 days of the experiment the levels of VFAs in all reactors were very low ranging

438

I.S. Zarkadas et al. / Renewable Energy 80 (2015) 432e440

Fig. 4. Results of the s-CSTR systems: (A) Daily volumetric methane yields, (B) Specific methane production for the different reactors, (C) Total VFA concentrations over time.

between 0 and 0.4 g/L (not taking into account the high VFA concentrations observed during day 22 for R3 & R4), after day 65 an increase of VFAs levels within R4 was observed, that ranged between 0.3 and 0.9 g/L and composed mainly of acetic and butyric acids with the highest concentrations observed during day 132 of

the experiment. From the above increase of the VFA levels within R4, it can be concluded that the reason of the lower methane yields achieved during the first 57 days of the experiments by this reactor was due to inhibition of the hydrolysis and/or acidogenesis stages of the process, something that did not affect the overall

I.S. Zarkadas et al. / Renewable Energy 80 (2015) 432e440

performance of the methanogens. However, as the hydrolysis and acidogenesis recovered, after the alteration in the influent for R4, the methanogens started having difficulties to utilise the increased concentrations of intermediate products offered to them by the primary phases of the process. The above is also evident by the increased ammonia levels in this reactor compared with the first stage of the experiment (57 first days). After the decrease of the FW addition into the substrate the ammonia levels within the reactor increased from the 1.83 ± 0.22 g/L to 2.31 ± 0.19 g/L corresponding to a FAN concentration of 0.74 ± 0.06 g/L which is a significant increase, considering that TKN levels within the influent stream were reduced by approximately 8% with the alteration of the feeding mixture. On the other hand, the alteration of the feeding resulted into improved VS removal efficiencies of this reactor from the 62.3 ± 4.7% that was achieved until day 57e70.2 ± 2.0% and 72.1 ± 3.6% measured during days 97 and 121 respectively. 3.2.5. Comparison of batch and s-CSTR experiments Table 3 presents a comparison between the average specific methane yields obtained during digestion of the different mixtures in batch and s-CSTR systems. As presented in the table, the results obtained during the different phases of the experiment are very similar with the variation between batch and s-CSTR experiments being in the range of 5.5% (not taking into account the methane production from R4). These results are very encouraging, as it can be concluded that in the case that the maximum organic loading rate which can be accepted by the AD systems is known, the preliminary analysis of wastes or mixtures of wastes in batch systems can provide results with little deviations, compared to those obtained by analysing the substrates in continuous or semi-continuous systems where the cost and time requirements are much higher. 3.2.6. Most likely inhibitor The microorganisms responsible for the bioconversion of biomass to biogas may be inhibited by a plethora of substrates, substances and products which may enter the process from an external source or may be generated by the process. It is well known that in order the AD process to be efficient and effective, the products generated in different phases of the process must be removed by the next step in order to reach the final products which are the CH4 and the CO2. If any one of the stages, namely hydrolysis, acidogenesis and methanogenesis, is inhibited then the overall

439

process becomes less productive or may even collapse due to accumulation of intermediate products. In the present study the efficiency of the process was found to be inhibited by approximately 29% based on the specific methane production achieved between batch and semi-continuous digestion when the addition of FW reached 70.8% based on VS (30% w/w). An obvious inhibitor was not present as the two most common inhibitors, ammonia and lack of trace elements [26,36] result in accumulation of VFAs and in acidification of the reactors, something that was not observed in the present study. Obviously, the methanogenesis was not inhibited, as both the levels of VFAs as well as the pH was nearly steady throughout the period, where the low biogas production was observed, rendering inhibition of hydrolysis and acidogenesis as the most likely causes. However, there is a near complete absence of studies describing similar phenomena in anaerobic digesters, with the exception of Fujishima et al. (2000) [13]. It is postulated that the inhibition observed in the present study was caused either by mass transfer limitations within the reactor that affected the different stages of the process [1], or by inhibition of the hydrolysis, due to the high availability of starch in the substrate. However, the second possible cause requires first the simple sugar acetogens to be inhibited. If this is the case then the inhibition took place in two steps: firstly the acidogens were stressed by an inhibitor resulting in increased concentrations of available sugars in the reactor, followed by inhibition of the hydrolysis by simple sugars accumulation. A similar conclusion, with ammonia as the inhibitor, is presented by Fujishima et al. (2000) [13]; which is probably not the inhibitor encountered by the acetogens in the present study. This type of inhibition, due to simple sugar accumulation, has been presented many times in the past for hydrolytic reactors employed for the hydrolysis of cellulose or cellulose-rich substrates, when no products were removed from the process [3,24]. While the inhibition mechanism is not clear it is most likely that the reduced sugars act competitively to the enzymes excreted by the hydrolytic microorganisms [24], thus halting the formation of new products.

3.2.7. Comparison of thermophilic and mesophilic anaerobic digestion of manures and food waste Few studies related to the semi-continuous co-digestion of FW with manures are available in the literature. These studies are presented in Table 4 and compared on the basis of the specific methane production achieved.

Table 3 Comparison of the specific methane production and VS reduction between batch and semi-continuous digestion of the CM and FW mixtures. Mixture CM 100% (R1) Mixture 2 (90CM:10FW) Mixture 3 (80CM:10FW) Mixture 4 (70CM:30FW) Mixture 5 (75CM:25FW)

Specific production batches (mLCH4/gVSadded) (R2) (R3) (R4.1) (R4.2)

214.6 271.8 340.3 407.8 e

± ± ± ±

7.6 5.1 13.7 17.1

VS reduction (%) 51.9 60.1 67.4 72.7 e

± ± ± ±

2.4 3.1 2.9 3.3

Specific production CSTR (mLCH4/gVSadded)

VS reduction (%)

206.8 280.5 369.7 290.1 (inhibition) 384.7

53.3 64.1 73.9 62.3 72.1

± ± ± ± ±

2.8 2.6 2.1 4.7 3.6

Table 4 Comparison based on specific methane production and VS reduction between the different available studies. Type of substrate

Temperature

Production mLCH4/gVSadded

VS reduction

Loading rate

Reference

OfMSW & manures FW and CM FW and CM FW & swine manure FW and CM

Thermophilic Thermophilic Mesophilic Mesophilic Thermophilic

340 330 250 388 385

55% 83% 65% 75.6% 72.1%

3.3e4.0 kgVS/m3-d 6.92e6.99 kgCOD/m3-d 4.6 kgCOD/m3-d 4.36 kgVS/m3-d 6.19 kgVS/m3-d

Hartmann & Ahring (2005) [15] n et al. (2013) [11] Castillo Neves et al. (2009) [25] Zhang et al. (2011) [37] Current study

440

I.S. Zarkadas et al. / Renewable Energy 80 (2015) 432e440

Comparing the results of the present study and those reported in the literature, while taking into account the addition of FW for the different studies, it can be seen that when TAD is employed for the co-digestion of FW and CM, it results in higher methane yield and/or more robust process compared to the mesophilic digestion. The same can be seen by comparing the specific methane production achieved in the present study which is in agreement with the results previously presented by Zhang et al. (2011) [37] for the digestion of similar substrates under mesophilic conditions. However, the TAD used in the present study was able to provide a similar specific methane production (388e385 mLCH4/gVSadded) in a similar HRT (20e21-d) but with a higher OLR by 1.83 kgVSm3-d, resulting in a higher volumetric CH4 production by approximately 0.9 LCH4/LR-d. From the above it can be considered that thermophilic digestion is probably more suitable when high OLRs (>5KgVS/m3-d) are applied on anaerobic digesters utilized for the co-digestion of FW and animal manures, which is in agreement with the findings presented by Li et al. (2002) [21] for the high solid mono-digestion of lipid-rich FW. 4. Conclusion Thermophilic anaerobic digestion was shown to provide a tool for the treatment of mixtures of pasteurised food wastes and cattle manures without the requirement of dilution or alterations on the hydraulic retention times of the digesters. No large scale inhibition phenomena on TAD systems were observed, even when mixtures of FW and CM containing up to 65.3% FW (based on VS) or 25% w/w were added into the process. OLRs as high as 6.2 kgVS/m3-d with the influent TS levels up to 14.3% can be accepted by s-CSTR systems, when the HRT is sustained to at least 21 days, without jeopardizing the process . References [1] Abbassi-Guendouz A, Brockman D, Trably E, Dumas C, Delgenes JP, Steyer JP, et al. Total solid content drives high solid anaerobic digestion via mass transfer limitations. Bioresour Technol 2012;111:55e61. [2] Adhikari B, Barrington S, Martinez J, King S. Characterization of food waste and bulking agents for composting. Waste Manag 2008;28:795e804. [3] Andri c P, Meyer AS, Jensen PA, Dam-Johanser K. Effect and modeling of glucose inhibition and in situ glucose removal during enzymatic hydrolysis of pretreated wheat straw. Appl Biochem Biotechnol 2010;160:280e97. [4] Angelidaki I, Alves M, Bolzonella D, Borzacconi L, Campos LJ, Guwy JA, et al. Defining the biomethane potential (BMP) of solid organic wastes and energy crops: a proposed protocol for batch assays. Water Sci Technol 2009;59: 927e34. [5] Athanasoulia E, Melidis P, Aivasidis A. Co-digestion of sewage sludge and crude glycerol from biodiesel production. Renew Energy 2014;62:73e8. [6] APHA. In: Clesceri L, Greenberg A, Trussell R, editors. Standard methods for the examination of water and wastewater. 17th ed. Washington: American Public Health Association; 1989. [7] Banks C, Chesshire M, Heaven S, Arnold R. Anaerobic digestion of sourcesegregated domestic food waste: performance assessment by mass and energy balance. Bioresour Technol 2011;102:612e20. [8] Brown D, Li Y. Solid state anaerobic co-digestion of yard waste and food waste for biogas production. Bioresour Technol 2013;127:275e80. [9] Browne DJ, Allen E, Murphy DJ. Improving hydrolysis of food waste in a leach bed reactor. Waste Manag 2013;33:2470e7. [10] Browne DJ, Murphy DJ. Assessment of the resources associated with biomethane from food waste. Appl Energy 2013;104:170e7. n L, Maran ~o n E, Ferna ndez-Nava Y, Ormaechea P, Quiroga G. Ther[11] Castrillo mophilic co-digestion of cattle manure and food waste supplemented with crude glycerin in induced bed reactor (IBR). Bioresour Technol 2013;136: 73e7.

[12] El-Mashad H, Zhang R. Biogas production from co-digestion of dairy manure and food waste. Bioresour Technol 2010;101:4021e8. [13] Fujishima S, Miyahara T, Naike T. Effect of moisture content on anaerobic digestion of dewatered sludge: ammonia inhibition to carbohydrate removal and methane production. Water Sci Technol 2000;41:119e27. [14] Ge H, Jensen PD, Batstone DJ. Increased temperature in the thermophilic stage in temperature phased anaerobic digestion (TPAD) improves degradability of wastes activated sludge. J Hazard Mater 2011;187:355e61. [15] Hartmann H, Ahring BK. Anaerobic digestion of the organic fraction of municipal solid waste: influence of co-digestion with manure. Water Res 2005;39:1543e52. [16] Ho DP, Jensen PD, Batstone DJ. Methanosarcinaceae and acetate-oxidizing pathways dominate in high-rate thermophilic anaerobic digestion of wasteactivated sludge. Appl Environ Microbiol 2013;20:6491e500. [17] James E, Joyce M. Assessment and management of watershed microbial contaminants. Crit Rev Environ Sci Technol 2004;34:109e39. [18] Johansen A, Nielsen BH, Hansen MC, Andreasen C, Carlsgart J, HauggardNielsen H, et al. Survival of weed seeds and animal parasites as affected by anaerobic digestion at meso and thermophilic conditions. Waste Manag 2013;33:807e12. [19] Koster IW, Lettinga G. Influence of ammonium-nitrogen on the specific activity of pelletized methanogenic sludge. Agric Wastes 1984;9:205e16. [20] Kougias PG, Boe K, Angelidaki I. Effect of organic loading rate and feedstock composition on foaming in manure-based biogas digesters. Bioresour Technol 2013;144:1e7. [21] Li YY, Sasaki H, Yamashita K, Seki K, Kamigochi L. High-rate methane fermentation of lipid-rich food wastes by a high-solid codigestion process. Water Sci Technol 2002;45:143e50. [22] Lisboa SM, Lansing S. Characterizing food waste substrates for co-digestion through biochemical methane potential (BMP) experiments. Waste Manag 2013;33:2664e9. [23] Lloret E, Pastor L, Pradas P, Pascual JA. Semi full-scale thermophilic anaerobic digestion (TAnD) for advanced treatment of sewage sludge: stabilization process and pathogen reduction. Biochem Eng J 2013;233:42e50. [24] Miao Y, Chen YJ, Jiang X, Huang Z. Kinetic studies on the product inhibition of enzymatic lignocellulose hydrolysis. Appl Biochem Biotechnol 2012;167: 358e66. [25] Neves L, Oliveira R, Alves MM. Fate of LCFA in the co-digestion of cow manure, food waste and discontinuous addition of oil. Water Res 2009;43:5142e50. [26] Niu Q, Qiao W, Quiang H, Hojo T, Li Y-Y. Mesophilic methane fermentation of chicken manure at a wide range of ammonia concentration: stability, inhibition and recovery. Bioresour Technol 2013;137:358e67. [27] Noutsopoulos C, Mamais D, Antoniou K, Avramides C, Oikonomopoulos P, Fountoulakis I. Anaerobic co-digestion of grease sludge and sewage sludge: the effect of organic loading and grease sludge content. Bioresour Technol 2013;131:452e9. [28] Papachristou E, Hadjianghelou H, Darakas E, Alivanis K, Belou A, Ioannidou D, et al. Perspectives for integrated municipal solid waste management in Thessaloniki, Greece. Waste Manag 2009;29:1158e62. [29] Silvestre G, Illa J, Fernandez B, Bonmati A. Thermophilic anaerobic codigestion of sewage sludge with grease waste: effect of long chain fatty acids in the methane yield and its dewatering properties. Appl Energy 2014;117:87e94. [30] Song Y-C, Kwon S-J, Woo J-H. Mesophilic and thermophilic temperature cophase anaerobic digestion compared with single-stage mesophilic and thermophilic digestion of sewage sludge. Water Res 2004;38:1653e62. [31] Qiang H, Lang D-L, Li Y-Y. Highesolid mesophilic methane fermentation of food waste with an emphasis on Iron, cobalt and nickel requirements. Bioresour Technol 2012;103:21e7. [32] Wang Z, Xu F, Li Y. Effects of ammonia nitrogen concentration on solidstate anaerobic digestion of corn stover. Bioresour Technol 2013;144: 281e7. [33] Ye J, Li D, Sun Y, Wang G, Yuan Z, Zhen F, et al. Improved biogas production from rice straw by co-digestion with kitchen waste and pig manure. Waste Manag 2013;33:2653e8. [34] Zarkadas I, Pilidis G. Anaerobic co-digestion of table olive debittering and washing effluent, cattle manure and pig manure in batch and high volume laboratory anaerobic digesters: effect of temperature. Bioresour Technol 2011;102:4995e5003. [35] Zhang C, Xiao G, Peng L, Su H, Tan T. The anaerobic co-digestion of food waste and cattle manure. Bioresour Technol 2012;129:170e6. [36] Zhang L, Jahng D. Long-term anaerobic digestion of food waste stabilized by trace elements. Waste Manag 2012;32:1509e15. [37] Zhang L, Lee Y, Jahng D. Anaerobic co-digestion of food waste and piggery wastewater: focusing on the role of trace elements. Bioresour Technol 2011;102:5048e59.