Challenges in Quantifying Greenhouse Gas Impacts of Waste-Based Biofuels in EU and US Biofuel Policies: Case Study of Butanol and Ethanol Production from Municipal Solid Waste

Conversion of wastes to biofuels is a promising route to provide renewable low-4 carbon fuels, based on a low- or negative-cost feedstock whose use can avoid 5 negative environmental impacts of conventional waste treatment. However, current 6 policies that employ LCA as a quantitative measure are not adequate for assessing 7 this type of fuel, given their cross-sector interactions and multiple potential product/service streams (energy, fuels, materials, waste treatment service). We employ a case study of butanol and ethanol production from mixed municipal solid 10 waste to demonstrate the challenges in using life cycle assessment to appropriately 11 inform decision-makers. from gCO eq./MJ biofuel (under US policies that employ system expansion approach), gCO 2 eq./MJ biofuel gCO 2 eq./MJ biofuel (under initial and current EU policies employ energy-based gasoline emissions

4 1 Introduction 28 Liquid biofuels can play a key role in the decarbonisation of the transport sector, and 29 have been studied extensively with life cycle assessment (LCA) tools to quantify their 30 net contribution to addressing greenhouse gas (GHG) emissions associated with 31 conventional, fossil fuels. LCA methodologies have been developed as a quantitative 32 element of transport fuel policies globally, wherein they are used to determine a fuel's 33 eligibility (US Energy Independence and Security Act; EU Renewable Energy 34 Directive) or to calculate its contribution to reducing emissions related to fuel use (e.g.,

47
LCA plays a central and quantitative role in global policies aimed at reducing GHG 48 emissions of transport fuels. In the EU, the Fuel Quality Directive regulates a minimum 49 of 6% reduction of the life cycle GHG intensity of transport fuels by 2020 compared to 50 2010 level, which can be achieved through the use of biofuels as one means. 1 In order 51 to be considered as renewable biofuels, life cycle GHG emissions must be at least 52 50% lower than from the fossil fuel they replace and 60% for newer installations from 53 January 2018. Similar thresholds are present in US policy: the Energy Independence 54 and Security Act (EISA) requires biofuels to achieve a life cycle GHG reduction 55 threshold as compared to a 2005 petroleum baseline for different types of biofuels 56 (e.g., 60% reduction for cellulosic biofuel, 50% reduction for advanced biofuel from 8 87 waste treatment processes, such as landfilling, are also credited to the biofuel product 88 (e.g., 15 ). With credits from co-products considered, biofuels can in some cases be 89 attributed with negative emissions: credits from co-products exceed the total 90 emissions associated with producing and using the fuel (e.g., 16,17  The overall environmental performance of converting the organic content of MSW to 148 biofuels and concurrently avoiding current waste treatment practices is evaluated.
149 Given the wide range of potential products/co-products (energy outputs; recovered 150 metals/glass/plastics) with diverse materials and energy market applications (see

236
The US EISA and California LCFS employ a system expansion approach, wherein 237 co-products are assumed to displace production elsewhere, with associated avoided 238 impacts credited to the primary biofuel product. We assume direct displacement for 239 co-product electricity (avoiding average UK grid generation), hydrogen (avoiding 240 production from fossil fuel sources), and acetone (avoiding primary production). Scrap 19 241 materials require further processing before they displace alternative production in a 242 market; these downstream processes to convert scrap to saleable materials are 243 included in the model. Plastic waste, of average composition 31 is input to a mechanical 244 recycling process to recover, per 1000kg input, 236 kg PET, 63 kg PP, 122 kg PE, and 245 1 kg PVC. 32 Recovered materials are assumed to displace primary production.
246 Unrecyclable materials (films, wastes and residues, 580kg) are disposed of by 247 incineration (71%) and landfill (29%). 27 For metals recycling, we use inventory data 248 from Gabi and Ecoinvent database. 28,33 Glass is assumed to be recovered to replace 257 All other co-products are considered with energy allocation.

261
Two additional allocation methods are considered that are able to account for non-262 energy co-products. Mass allocation distributes the GHG emissions associated with 263 main products and co-products based on their respective mass. A two-stage mass 264 allocation is employed: first, upstream processes and waste pre-treatment are 265 allocated between the biofibre and non-biogenic content on a mass basis (see Figure   266 S5 in the SI). Second, we allocate a share of biofuel production impacts to co-product 267 acetone and hydrogen (electricity and heat have no allocation as they have no mass).
268 Finally, economic allocation apportions impacts between co-products on the basis of 269 their financial value. We conduct the allocation considering the overall production 21 270 outputs, as intermediate product (biofibre) does not have a financial value (see Table   271 1).
272 Table 1 Partitioning ratio for mass, energy, and economic allocation.  293 gCO 2 eq./MJ biofuel ) (see Figure S7). It is noted in this case study, the relatively low sugar 294 yield by hydrolysis and correspondingly low biofuel yield results in larger quantities of 295 residual biomass available for co-product electricity production than with conventional 296 feedstocks. Major GHG emissions sources arise from the manufacture of enzymes 297 (187 gCO 2 eq./MJ biofuel ), included in the total biorefinery emissions indicated in Figure   298 2a (also see Table S6). Other process inputs (pH control; fermentation nutrients; 299 microorganism) have smaller impacts, totalling 20.87 g CO 2 eq/MJ biofuel . Treatment of 300 residual waste from autoclave has a large GHG emission of 141.01 g CO 2 eq/MJ biofuel . 301 Collection and transport accounts for about 4% while fuel distribution and use 302 accounts for less than 1% of the total PED and GHG emissions (see Figure S7 in the 24 303 SI). On balance, with substantially negative GHG emissions, the MSW-derived 304 biofuels would by far surpass the eligibility requirements for the US EISA policy.

305
GHG emissions are substantially higher when allocation is used to evaluate the 306 MSW-derived biofuels. The initial RED I policy employs energy allocation between 307 products, with the exception of co-product electricity: excess electricity is evaluated by 308 system expansion, and the credit from displacing generation elsewhere is allocated 309 between the biorefinery products. No impacts are allocated to the recovered metal and 310 glass co-products, as these material do not have an energy content. RED I results in 311 GHG emissions of 86 gCO 2 eq./MJ biofuel , achieving only a minor reduction of 9% relative 312 to gasoline and therefore would not qualify as an eligible biofuel under the policy.
313 Enzyme production 29 represents approximately 85% of net emissions allocated to 314 biofuel production. The higher net GHG emissions, relative to the system expansion 315 approach, result from the exclusion of avoided waste treatment and the apportioning 316 of the co-product electricity credit between biofuels and other products: of the total 102 317 gCO 2 eq/MJ biofuel credit, only 31 gCO 2 eq. is credited to the biofuel product. Thus, 25 318 although the production of biofuels from MSW would achieve significant overall GHG 319 reductions when all products are considered, this pathway would not be eligible under 320 the original RED policy.

321
In contrast, under the revised RED II policy, the MSW-derived biofuels would be 322 eligible, with overall GHG emissions of 23 gCO 2 eq/MJ biofuel , a reduction of 75%. With 323 exergy allocation applied to the co-product electricity and heat, a large share of 324 biorefinery emissions (86%) are applied to these outputs; correspondingly, fewer 325 emissions are attributed to the biofuel product. Excess electricity is attributed with 326 GHG emissions of 86 gCO 2 eq/MJ, which represents a 12% reduction compared to UK 327 grid electricity mix 35 (see Table S5 in the SI). Enzyme production still contributes the 328 largest share of GHG emissions attributed to the biofuel outputs (68%).

339
Mass and economic allocation are considered as alternatives to system expansion 340 and energy allocation approaches, as these allow allocation to non-energy products 341 (recovered metal, glass) (see Figure 2a and Table S6)

459
A key question facing the analysis of waste-derived fuels is how avoided waste 460 treatment should be included within LCA calculations. Avoided waste treatment is 461 excluded in EU policy, but this approach ignores the "co-service" of waste treatment 462 provided by biofuel production and thus overestimates the impacts of waste-based 463 fuels. In contrast, system expansion gives full credit to biofuels for waste diversion, 464 despite this being but one product of the biorefinery system, and ignoring any other 465 changes occurring in the waste treatment sector, including those in response to policy 466 drivers to limit or reduce waste to landfill (and increasingly, to incineration). In future, 467 multiple viable opportunities may exist to utilise MSW, and therefore the role of a single 468 use in avoiding conventional waste treatment would be questionable. Sector 488 Figures S7-9 shows the environmental efficiency of waste to biofuel, comparison with 489 other waste treatment routes and sensitivity analysis results. Table S1 summarises 490 the outputs of the autoclave and biorefinery process. Table S2 is an overview of 491 current biofuel regulations in the EU and US.