Effect of Pasteurisation on Methane Yield from Food 2 Waste and other Substrates in Anaerobic Digestion 3

: The effect of pasteurisation and co ‐ pasteurisation on biochemical methane potential 9 values in anaerobic digestion (AD) was studied. Pasteurisation prior to digestion in a biogas plant 10 is a common hygienisation method for organic materials which contain or have been in contact with 11 animal by ‐ products. Tests were carried out on food waste, slaughterhouse waste, animal blood, 12 cattle slurry, potato waste, card packaging and the organic fraction of municipal solid waste 13 (OFMSW); pasteurisation at 70°C for 1 hour was applied. Pasteurisation increased the methane 14 yields of blood (+ 15%) and potato waste (+ 12%) only, which both had a low content of structural 15 carbohydrates (hemi ‐ cellulose, cellulose) but a particularly high content of either non ‐ structural 16 carbohydrates such as starch (potato waste) or proteins (blood). With food waste, card packaging 17 and cattle slurry, pasteurisation had no observable impact on the methane yield. Slaughterhouse 18 waste and OFMSW yielded less methane after pasteurisation in the experiments (but statistical 19 significance of the difference between pasteurised and unpasteurised slaughterhouse waste or 20 OFMSW was not confirmed in this work). It is concluded that pasteurisation can positively impact 21 the methane yield of some specific substrates, such as potato waste where the heat ‐ treatment may 22 induce gelatinisation with release of the starch molecules. For most substrates, however, 23 pasteurisation at 70°C is unlikely to increase the methane yield. It is unlikely to improve 24 biodegradability of lignified materials and it may reduce the methane yield from substrates which 25 contain high contents of volatile components. Furthermore, in this experimental study the obtained 26 methane yield was unaffected by whether the substrates were pasteurised individually and then co ‐ 27 digested or co ‐ pasteurised as a mixture before batch digestion. 28

Processes 2020, 8, x FOR PEER REVIEW 3 of 21 and observed very good performance up to a ratio of 25% food waste by wet weight; however, they 97 did not digest unpasteurised food waste, and also not food waste alone. Pagliaccia et al. [36] reported 98 a reduction of the methane yield obtained in mesophilic AD after thermal pre-treatment of food 99 waste, which occurred along with an increased initial hydrogen production in response to the 100 carbohydrate solubilisation; but the pre-treatment was conducted at 134°C and a pressure of 3.2 bar, 101 and thus might not occur for standard pasteurisation at 70°C.

102
The available studies about the impact of pasteurisation on AD are based on a few selected 103 organic materials and the findings are partially contradictory and difficult to interpret [1,10]. One 104 possible explanation for the variations among the findings is that methodologies applied, and 105 experimental methods used, were not the same [1]. Furthermore, in some studies the duration of the 106 experiments was too short to allow reliable conclusions about the ultimate methane yield from 107 pasteurised material [8]. Other studies applied co-digestion of pasteurised substrates but did not 108 study the performance of individual substrates.

109
The existing knowledge in this area is thus uncertain and incomplete, making it difficult for the 110 AD industry to assess the impact of pasteurisation in the development of biogas production 111 technology. While pasteurisation is required in many cases under the ABPR, it is often justified as 112 potentially increasing the methane yield, and thus contributing to a more favourable energy balance 113 of the AD process. The body of knowledge to support or criticise this argument, however, is

125
BMP testing was conducted on source-separated domestic food waste, slaughterhouse waste 126 (consisting of pig gut with flotation fat), animal blood, cattle slurry, potato waste, card packaging and 127 on the organic fraction of municipal solid waste (OFMSW) recovered in a mechanical-biological 128 treatment (MBT) plant. Potato waste, which is not an animal by-product but a vegetable waste, was 129 included in the study because in the UK it is a high-volume organic residue stream in the food sector, 130 and due to its low nitrogen content has been suggested as a suitable co-substrate to accompany in the 131 biogas plant the digestion of food waste or slaughterhouse waste (susceptible to AD inhibition due 132 to high nitrogen content) [37]. Card packaging was included because it can be collected together with    company site, there were no further process steps (e.g. dissolved air flotation) to remove fat from the 147 generated slaughterhouse wastewater stream, and thus in this study it was assumed that the 148 retrieved trap material is representative of separable fat occurring at slaughterhouses. Sampled pig 149 gut and recovered fat were mixed to represent the slaughterhouse waste. In the current study, the 150 proportion of mixed gut and fat used was 9:1 respectively on a VS (volatile solids) basis.

151
Animal blood: Sheep blood (20 kg) was obtained from an abattoir in Farnborough, Hampshire, 152 UK (operating company R.W. Newman and Partners).

153
Cattle slurry: A 20-kg sample of fresh material was obtained from a dairy farm (Parkers Farm,

154
Hampshire, UK). Using a tractor-mounted scraper, the slurry was secured from the milking area at 155 the farm immediately after the milking was done.

156
Potato waste: A 2-kg sample was provided by Forest Products Ltd, Dorset, UK. The potato waste 157 consisted of raw potato chip (before frying) rejected for the manufacturing of crisps and was 158 essentially a two-dimensional material (slice thickness of around 0.5 mm).

166
Organic fraction of municipal solid waste (OFMSW): 100 kg of mechanically-recovered OFMSW 167 were collected from Bursom Recycling Centre, Leicester, UK. This was the organic fraction remaining 168 after pre-processing of municipal solid waste to recover plastic, paper and card, glass and metal using 169 a combination of processes including a ball mill, magnetic separator, ballistic separator, and eddy 170 current separator. The mean particle size of the OFMSW was 6.0 mm, with most of the particles (>

184
The range of properties determined for the substrates is more extensive than typically available 185 in similar work, providing a very detailed picture of the materials and valuable data on their 186 constituents.

194
Processes 2020, 8, x FOR PEER REVIEW 5 of 21   The sample temperature was gradually raised to 72°C ± 2°C then maintained at this value for 1 hour.

209
Manual stirring was performed without breaking the parafilm cover. 228 slurry were also co-digested, i.e. the performance of the mixture of these two substrates was studied.

229
The mixture was tested both with the two components pasteurised separately and then mixed before

237
The inoculum was digested separately in four replicates as a control, allowing determination of 238 its residual methane production. In addition, a positive control was run in triplicate using a standard 239 reference material to ensure that the overall test procedure was capable of giving valid results. The 240 standard was a high purity cellulose powder fibrous in form and of medium particle size (Sigma-

241
Aldrich Company Ltd, UK, product no. C6288, CAS 9004-34-6, EC no. 232-674-9). The results for this 242 control, which are documented in Appendix A, confirmed that the test method was reliable.

287
Further characterisation was conducted on samples prepared by air drying to constant weight 288 then milling in a micro-hammer mill (Glen Creston Ltd, London, UK) to a particle size ≤ 0.5 mm.

319
Looking at the performance of source-separated domestic food waste, Figure 1a shows that 320 methane production from unpasteurised and pasteurised substrate was very similar throughout the 321 digestion experiment. Both materials showed rapid digestion after initiation of the experiment. The 322 methane production rate from unpasteurised food waste was slightly higher during day 2, but the

334
Unpasteurised and pasteurised card packaging both showed a one-day lag in methane 335 production at the early stage of the test, as can be seen in Figure 1c, and closely similar rates thereafter.

336
BMP values were 0.266 ± 0.010 STP m 3 kg -1 VS for unpasteurised and 0.267 ± 0.005 STP m 3 kg -1 VS for 337 pasteurised substrate respectively; this small difference in the measured methane yields of 338 unpasteurised and pasteurised material is statistically not significant (p = 0.884).

339
Pasteurised potato waste had a slightly higher rate of methane production than unpasteurised 340 during the first days of the test (Figure 1d), but by day 5 the cumulative productions were the same;  The methane production rate from unpasteurised slaughterhouse waste was higher than from 354 pasteurised slaughterhouse waste early in the test period ( Figure 1e)

368
Turning to the AD performance of sheep blood, Figure 1f shows that the unpasteurised substrate 369 initially had a slightly higher methane production rate compared to the pasteurised substrate; an 370 explanation for this may be the reduced specific surface area that was available for enzymic attack in 371 the pasteurised blood as a result of heat coagulation. From day 4 onward methane production from 372 test digesters with unpasteurised blood was lower than from the control (inoculum only), leading to 373 the decline in net specific cumulative methane production seen in Figure 1f. A similar decline was 374 seen from day 9 for pasteurised blood, but in both cases, these declines were subsequently reversed.

375
Methane production from the unpasteurised blood rose quickly from day 33. For pasteurised blood, 376 the degree of inhibition initially appeared to be less than in the digesters containing the unpasteurised 377 substrate, but methane generation from the pasteurised blood only began to recover from day 51. The

391
For mechanically-recovered OFMSW, the initial methane production rates from unpasteurised 392 and pasteurised material were closely similar, as can be seen in Figure 1g.

399
These results suggest that, for the materials that were tested in this study, pasteurisation had a

409
The comparison of the experimentally found methane yield (BMP values) and the theoretical 410 BMP value of a substrate (determined through calculation, using the biochemical composition, see

411
Section 2.4) shows the actual degree of exploitation of the theoretically available potential.

419
Methane values were also expressed on a wet weight basis (i.e. weight of fresh material) to take 420 into account the moisture content and inert fraction of the substrates (Table 1), and the results are 421 shown in Table 3. Of the materials tested, card packaging had the highest methane yield due to its 422 very low moisture (6%) and low inert fraction (only 16% of TS). Food waste and OFMSW had 423 comparable methane potentials. The very low methane yield of cattle slurry (less than 10% of that of 424 card packaging on a wet weight basis), makes it unattractive as a sole substrate for energy production;

425
and also confirms the suitability of high-solids, high-methane feedstocks such as food waste as co-

426
substrates, since these can give a significant boost to methane production [34].

431
Food waste and cattle slurry as co-substrates in AD, mixed at a ratio of 20:80% (VS basis), were 432 tested after both separate pasteurisation (i.e. food waste and cattle slurry pasteurised separately and 433 then mixed for digestion) and co-pasteurisation (i.e. food waste and cattle slurry mixed before 434 pasteurisation, then pasteurised as mixture and then digested). From Figure 2 it can be seen that the

508
For the substrates with higher lignin content (cattle slurry, OFMSW, card packaging), 509 pasteurisation did not increase the methane yield during testing. Card packaging yielded around 510 81% of its theoretical methane potential both with and without pasteurisation ( Table 2).

511
Unpasteurised cattle slurry yielded 68% of the theoretical potential, and OFMSW 91%; however, 512 pasteurisation did not improve the experimental methane yield for these substrates. This was also

540
The phenomenon outlined for OFMSW might also explain why, compared to the unpasteurised 541 material, methane production from pasteurised food waste and slaughterhouse waste was lower 542 during the first days of digestion (see Figure 1); the final methane yield was also observed to be lower 543 (Table 2)  temperature. More research is required to verify this explanation and to quantify this phenomenon.

554
The experimental BMP value of unpasteurised food waste reached 94% of its theoretical BMP 555 value (see Table 2), i.e. nearly the full theoretical potential was exploited. This suggests pre-treatment 556 of food waste to increase its specific methane yield is probably a waste of effort. Pasteurisation is still 557 required for hygienisation purposes, but it is not an effective strategy to increase methane yield.  (Table 2). Of all materials tested, blood had the highest content in protein 575 and the lowest content in carbohydrates (Table 1). The high nitrogen content is likely to have caused 576 ammonia inhibition during AD, but it is interesting to note that the pasteurised material was initially 577 less affected by inhibition then required longer to recover, and at the same time the final methane 578 yield was significantly increased. An increase in methane yield for pasteurised slaughterhouse waste 579 rich in blood was previously reported in the literature [18], but blood itself has not received much 580 attention so far. From this research it can be concluded that pasteurisation ultimately increased the 581 methane yield from blood, but it also slowed the recovery process after inhibition. It requires further 582 study to fully understand the nature of the different impacts observed for the processing of blood 583 and to explore whether the observed phenomena also occur in continuous AD operation.

584
It is evident that pasteurisation had a very differing impact on animal blood and slaughterhouse 585 waste, which in this study was composed of pig gut and flotation fat. The slaughterhouse waste was 586 rich in lipids and proteins, while the blood was very rich in proteins but very poor in carbohydrates 587 (Table 1). An increased methane yield was found only for blood, while pasteurisation altered AD 588 kinetics for both blood and slaughterhouse waste, but with very different patterns (Figure 1).

589
Digestion of pasteurised slaughterhouse proceeded more slowly than digestion of unpasteurised

649
Pasteurisation of food waste, cattle slurry and card packaging had no significant impact on 650 methane yield during anaerobic digestion. It is interesting to note that food waste yielded 93-94% of 651 its theoretical methane potential with and without pasteurisation, thus high exploitation of the biogas 652 potential of this material can be achieved regardless of whether thermal pre-treatment is applied or 653 not, and consequently pre-treatment is anyway not a promising approach to achieve more value from this substrate in AD (pasteurisation is still required for hygienisation purposes). For cattle slurry and 655 card packaging, the experimental BMP was remarkably lower than the theoretical value, but 656 pasteurisation was not effective to increase the methane yield. Co-digestion with food waste did not 657 improve methane yield from cattle slurry. Furthermore, it made no difference to the methane yield if 658 cattle slurry was pasteurised individually and then co-digested with food waste, or the substrates 659 were co-pasteurised as a mixture before batch digestion.

660
None of the substrates with a high content of lignified constituents (cattle slurry, card packaging,

661
OFMSW) benefited from pasteurisation with respect to the methane yield achieved. Methane 662 generation from cattle slurry and card packaging was not noticeably impacted by pre-pasteurisation.

663
With OFMSW, the pasteurised material in this study yielded less methane than the unpasteurised 664 substrate, i.e. pasteurisation had a negative impact on the produced methane quantity in the 665 experiments of this study, but statistical testing found the difference nonsignificant at the 95% 666 confidence level, and thus there is insufficient evidence to conclude that a lower methane yield is to 667 be expected for OFMSW due to pre-pasteurisation. A lower methane generation of the pasteurised 668 material was also observed with slaughterhouse waste in this work, but also here the difference 669 between unpasteurised and pasteurised material was not confirmed to be statistically significant. The 670 observations suggest that for substrates that contain easily degradable components and undergo 671 periods of storage or other steps in which microbial degradation can release volatile organic 672 compounds, the elevated pasteurisation temperature may cause a reduction of methane yield, but 673 more research is required to confirm this hypothesis and to quantify this effect.

674
Overall, this study shows that pasteurisation before AD results into higher methane yields 675 during AD only for some specific substrates such as potatoes and blood, while biogas production 676 from lignified biomass is not likely to be increased through pre-pasteurisation; there might also be a

684
Whilst the Animal By-products Regulations impose a requirement to pasteurise waste streams 685 which contain animal by-products (ABP) or have been in contact with such materials, the findings of 686 this study indicate that for most substrates pre-pasteurisation before feeding to a biogas plant is 687 unlikely to enhance the efficiency of the anaerobic digestion process itself. This study therefore shows 688 that pre-pasteurisation is not generally an effective strategy for the purpose of increasing the methane 689 generation of a biogas plant and for improving the energy balance of the AD facility.      for the reliability of the test method applied, as described in the technical report for this project [37], 705 and presented in the following. As can be seen in Figure A1, at the beginning of the test there was a 706 lag of approximately 3 days before methane production from cellulose commenced. This lag period 707 probably reflected the time needed to initiate hydrolysis of the complex macromolecular control 708 material. Methane generation was then rapid, amounting to a cumulative total of 0.361 ± 0.007 STP 709 m 3 kg -1 VS after the first 16 days, equivalent to 87.0% of the theoretical BMP of 0.415 STP m 3 kg -1 VS.

710
Methane generation continued after day 16, but at a considerably lower rate. On day 64, the methane

715
To obtain data on possible losses through dissolution, carbon dioxide production was also 716 recorded in this assay. In Figure A1 it can be seen that the trend in carbon dioxide production was