Strategic uses for ancillary bioenergy in a carbon-neutral and fossil-free 2050 European energy system

SOLm (Sustainability and Organic Livestock model) is a bottom-up mass-flow model of the agricultural production and food sector originally having a focus on livestock and organic production but now, in its sixth version, covering the whole food system and also conventional production in similar detail (Muller et al 2017). It is by default calibrated with Food and Agriculture Organization Corporate Statistical Database (FAOSTAT) data and categories of crops and livestock at the national level. Thanks to its detailed categorisation and flexible model assumptions, we can estimate the ancillary biomass potential per crop/livestock and per activity (e.g. 114 types of primary crops residues and 16 types of nuts shells, as documented in appendix A). Therefore, we can extract the annual flow of residues and by-products not used for food or feeding. We run the model at national resolution (35 European countries). We update SOLm with crops-to-residue shares from the latest 2019 Intergovernmental Panel on Climate Change (IPCC) refinement ('Ratio of above-ground residue dry matter to harvested yield' in Volume 4, chapter 11) (Masson-Delmotte et al 2018). For a detailed description of SOLm, one may refer to its latest documentation (Müller et al 2020). For our assumptions and data processing of ancillary biomass potential, see appendix A for a detailed description.

We model the European energy system in 2050 with a national resolution, sector-coupled energy system model modified from the Sector-Coupled Euro-Calliope model, which we hereafter refer to as Euro-Calliope (Pickering et al 2022). Euro-Calliope is a representation of demand and available supply technologies in all energy-consuming sectors (household and commercial heat, passenger and freight transport, industrial process heat and feedstocks, and all other sectors, including agriculture) across 35 European countries. The model is designed to be linearly optimised in the Calliope energy modelling framework v0.6.8 (Pfenninger and Pickering 2018).

The original version of Euro-Calliope (Pickering et al 2022) models all biomass feedstocks as one energy carrier, which is compatible for all bioenergy conversion technologies. Instead, we add more detailed bioenergy feedstocks data from SOLm and pair every biomass feedstock with compatible bioenergy conversion technologies (figure 1). With the modified and more detailed AB module in Euro-Calliope, we name our model AB-Euro-Calliope (for the detailed AB costs and technologies data, please refer to appendix B; for codes and data files to reproduce all model runs, see our GitHub repository (Wu 2022) and (Bryn 2022)). In AB-Euro-Calliope, we run all scenarios with a two-hour resolution for a full year, and assume that the annual biomass potential can be used arbitrarily throughout the year (i.e. it can be stored and used when needed).

Zoom In Zoom Out Reset image size

Overall, we consider and model a self-sufficient pan-European energy system, including self-sufficient bioenergy supply (i.e. no energy imports or transmission from outside of Europe). We assume the national autarky of both synthetic fuels demand and biomass supply in 2050 Reference (table 1).

For the future ancillary biomass supply data, we model the agricultural sector using SOLm and adopt the forestry and municipal waste potential from JRC-EU-TIMES (Ruiz et al 2015). This is because there is in principle no food/feed/land conflicts when we source forestry residues or forestry by-products and municipal solid waste, so for those feedstocks, there is no need to re-estimate the potential with strict exclusion of food/feed/land conflicts. There are many estimates of future residue data available from the literature (table A6 in appendix A), but they either have aggregated agricultural feedstocks with different energy properties (i.e. not suitable for the same conversion technology) or do not completely rule out feed/food conflicts. Therefore, we use SOLm and its FAO 2050 Business-as-usual scenario (Muller et al 2017) to model the more detailed and stringent agricultural ancillary biomass potential, i.e. we first extract only the non-food/feed shares of (a) by-/co-products (16 types of shells and three types of fats and oil with high energy density), (b) crop residues (114 types), and (c) animal manure. Moreover, we leave enough residues (50%) and manure (25%) on fields to keep enough soil fertility and nutrients. For the detailed assumptions and calculation of ancillary biomass potential data, please refer to table A4 and appendix A.

Next, we briefly introduce the common assumptions across all scenarios (table 1), using our baseline and reference scenario, hereafter referred to as the 2050 Reference scenario. Overall, we consider and model a self-sufficient pan-European energy system, which assumption thus also applies to bioenergy supply (i.e. no energy imports or transmission from outside Europe). We assume the national autarky of both synthetic fuels demand and biomass supply in 2050 Reference (table 1). For non-bioenergy renewable supply options, we have solar (open-field and rooftop), wind (off-shore and on-shore), hydro (run-of-river and reservoir), biomass (six categories of ancillary biomass as in figure 1) and nuclear (only the capacity in operation or planned towards 2050) to produce carbon-neutral electricity, heat, or synthetic fuels. Solar and wind constitute the predominant supply sources, i.e. their capacity reaches 2.23 TW and 3.32 TW in the optimised 2050 Reference scenario. Meanwhile, hydro reservoir, biomass, and nuclear can provide comparably minor but flexible supply. The national difference is pronounced in terms of renewable energy capacity and supply structure (table D2 in appendix).

On the demand side, we adopt today's demand profiles by default—the same hourly demand profiles as in Euro-Calliope (Pickering et al 2022). This implies several key assumptions about demand-side decarbonisation (table 1). First, we assume the marine shipping and aviation cannot be electrified in 2050 and can only be decarbonised by synthetic fuels (diesel and kerosene). Second, for other transportation (road and rail, light, and heavy duty), heating, and electricity, both electrification and carbon-neutral fuels are allowed. Third, the industrial demand for methanol and methane feedstocks can only be met by synthetic fuels generated by biomass or hydrogen. Overall, we report the primary energy supply in PJ (section 3.1) and final energy consumption in TWh (sections 3.2–3.5) for differentiation.

Apart from the 2050 Reference scenario, we examine two sets of counterfactual and near-optimal scenarios (i.e. total system costs, or the optimisation objective, are no more than 10% above the optimal solution in 2050 Reference) to explore different strategic uses of AB. Table 2 specifies the difference among these scenarios, marked in bold and italic when they deviate from 2050 Reference.

For the first set (different utilisation cases), we explore the possible roles for AB when it can be utilised optimally, fully utilised (FullUtiAll), or not utilised at all (NoUti). This is realised by adding additional bioenergy utilisation constraints or infrastructure, where we change only one constraint per scenario. More specifically, first, for the FullUtiAgr and FullUtiAll scenarios, we force the 100% utilisation of agricultural (FullUtiAgr) or all kinds of ancillary biomass (FullUtiAll). Second, for GasStorage, we add existing underground methane storage facilities. This is based on the latest data on a national level, as documented in the Sector-coupled Euro-Calliope (Pickering et al 2022). Third, in the 'BioDistribution' scenario, we deploy an additional distributing network (trail/road) connecting neighbouring European countries for transporting liquid synthetic fuels (see the third type of costs in appendix B). Lastly, NoUti disallows all biomass supply or conversion technologies, which provides a counterfactual scenario where Europe realises carbon-neutral energy systems in 2050 without utilising any bioenergy.

The second set explores alternative strategic use cases of AB—(a) adding or removing controversial low-carbon energy sources (DedicatedBiomass: allowing additional supply of dedicated advanced biomass (miscanthus) with additional land use; NoNuclear: disallowing all nuclear capacity); (b) forcing all AB to be used for negative emissions via stationary BECCS (AllBECCS). For the combined scenarios (e.g. BioDistribution + FullUtiAll) we change multiple constraints by combing assumptions, as indicated by the scenario names.

Apart from modelling the energy flow (AB-Euro-Calliope) and mass flow (SOLm), we also conduct ex post analysis to compare three environmental metrics among scenarios—(a) land area used by the energy system (including onshore wind turbines and open-field PVs in all scenarios and the land used by miscanthus in DedicatedBiomass scenario); (b) agricultural nutrients lost through biomass incineration (via multiplying the Nitrogen and Phosphorus contents of biomass by their incinerated percent); (c) negative emissions potential of using all ancillary biomass at stationary BECCS plants (through their emission factors and CO₂ capture rate). For the detailed data source, assumptions, and calculation, please refer to appendix C in the supplementary material.

The total potential of AB reaches 185 PJ (sustainable) and 798 PJ (technical) in 2050 in Europe. To minimise the environmental impact, we use only the sustainable potential of AB for comparing feedstock-wise availability and modelling all scenarios ('potential' refers to sustainable potential hereafter). Feedstock-wise, agriculture is the predominant sector for providing AB across Europe (81 PJ), accounting for over 56% of the total potential. Forestry potential (48 PJ) is the second highest, followed by municipal solid waste (33 PJ). For national potential of sector-wise ancillary biomass, please refer to figure D1 in appendix.

Compared to recent estimations including dedicated bioenergy (the middle block in figure 2), the potential of AB estimated here is reasonably limited (i.e. 3–6 times lower) given its stringent prerequisite—no land-use or food/feed competition. The more detailed and stringent estimation of AB potential could better reveal its sustainable role in a fully renewable European energy system. The technical potential for AB is around 20% lower than that of residues from the high availability scenario of JRC-EU-TIMES. In figure 2, we display the lower bounds as sustainable potentials and higher bounds as technical potentials if applicable—for models without differentiating potentials, we use a line (e.g. price-induced market equilibrium system (PRIMES)). JRC (residue) and JRC are from the JRC-EU-TIMES model (Ruiz et al 2015). By non-dedicated bioenergy in JRC-EU-TIMES, we refer to the non-dedicated biomass from agricultural waste, manure, residues from landscape care, fuelwood residues, secondary forestry residues (woodchips and sawdust), municipal waste.; IMAGE (SSP2) is the bioenergy supply from the Shared Societal Pathway 2 scenario modelled by IMAGE (Huppmann et al 2019); for the PRIMES model, we use its input data of bioenergy potential in 2050 (European Commission. Directorate General for Energy et al 2016); the current supply and consumption of bioenergy is compiled from IEA World Energy Balances (International Energy Agency 2022).

Zoom In Zoom Out Reset image size

We specifically subtract the non-dedicated bioenergy from JRC-EU-TIMES for comparison (the second-left bar—'JRC (residue)' in figure 2), as it is the closest to AB with overlapping forestry and municipal datasets. We further break down and compare the different agricultural biomass potentials between our results and the JRC data by feedstocks (table A5 in appendix) and spatial distribution (figure 3). Generally, our total agricultural ancillary biomass potential (185 PJ) is similar to the non-dedicated biomass considered in JRC-EU-TIMES model (182 PJ). However, the spatial distribution varies, especially when we break it down into sub-categories. There are three major differences. First, we consider additional by-/co-products feedstocks that are not included in JRC-EU-TIMES (i.e. (3) & (4) in figure 3—nuts shells and animal fats). Their amount is minor, but they have high energy density with strategic decarbonisation uses (e.g. producing kerosene and diesel for the hard-to-decarbonise aviation and shipping sectors for which regular agricultural residues do not qualify). Second, we embrace a wider variety of crops residues (114 types) compared to JRC (11 types) leading to a doubled potential (81 PJ in this study compared to 44 PJ in JRC-EU-TIMES). Third, we use the sustainable potential of manure without changing nutrient balance, livestock system, and food/feed system (e.g. 25% of all manure is left on fields), while JRC allows all wet manure (pig and cattle) to be available.

Zoom In Zoom Out Reset image size

We first explore the possible roles for AB when it can be utilised optimally, fully utilised, or not utilised at all in our 2050 European energy system (figure 4). This section acts as the reference scenarios for comparing the following strategic use cases in sections 3.3–3.5. Hence, in this section we assume that (a) nuclear power is available, (b) dedicated biomass is disabled, and (c) no BECCS technologies are available to provide negative emissions.

Zoom In Zoom Out Reset image size

Consumption-wise, the average European utilisation of AB is low and reaches only 38% in the optimised 2050 Reference scenario (see the European and national utilisation rates in figure D1 in appendix), whereas 19 out of 35 European countries are below this rate. When adding bioenergy infrastructure (especially the continental distribution network of liquid synthetic fuels in BioDistribution), it can significantly boost the AB utilisation (by half) and alter sectoral uses (figures 4 and 5). Our optimisation results suggest different sectoral uses for bioenergy among scenarios (ancillary biomass only; solid colour bars in figure 4) compared to the current sectoral demand in Europe (dedicated biomass that is not optimised; the transparent bottom bar). In all scenarios, ancillary biomass is more attractive for decarbonising transport in 2050, instead of balancing variable renewable electricity supply or residential heating (the current major use of dedicated bioenergy in 2019).

Zoom In Zoom Out Reset image size

In figure 4, the current (2019) data refer to the total final consumption of biofuels and waste in Europe from the International Energy Agency (IEA) Energy Balances, which includes dedicated biomass and is not cost-optimised (International Energy Agency 2022). 'Others' refer to the agriculture, commercial, and other sectors. For the detailed sectoral data, refer to table E1 in appendix. For scenarios difference, there are two additional kinds of infrastructure and/or forced full utilisation of AB—(a) underground methane gas storage facilities (for enhancing flexibility in scenario GasStorage) and (b) a liquid biofuels distribution network connecting European countries using wheel loaders and trucks for 0.64 €/km/ton (for realising pan-European autarky of fuels in scenario BioDistribution); (c) FullUtiAgr where all agricultural ancillary biomass is forced to be used and (d) FullUtiAll where all ancillary biomass is forced to be used. (e) We also combine BioDistribution and FullUtiAll scenarios into the fifth scenario (BioDistribution + FullUtiAll) as they can both substantially change strategic uses. We then compare them to the 2050 Reference scenario, which does not include either of two facilities.

When fully utilising AB, the total system costs barely increase (less than 1% in FullUtiAgri and FullUtiAll, table E2). But it can further decarbonise industry sectors by partially replacing hydrogen-based synthetic fuels, thus freeing up land (6% land footprint reduction) and reducing renewable power curtailment. This synergy is more pronounced when we combine the constraints of full utilisation and distribution network (BioDistribution + FullUtiAll, figure 5 and table E2), which leads to the highest reduction of total system costs (−8%) and power curtailment (−11%).

However, even with additional distribution and/or forced full utilisation, the attractiveness of AB is still uncompetitive in producing synthetic fuels in most European countries, compared to hydrogen (figure E1 in appendix). To further investigate the extent to which AB is necessary, in the NoUti scenario, we disallow all bioenergy supply, conversion, or consumption. Compared to 2050 Reference, total system costs barely change (less than +0.6%) in NoUti—only the annual power curtailment slightly increases (+2%, see figure 5). However, compared to the highest utilisation case (BioDistribution+ FullUtiAll), NoUti requires a drastically different solar-to-wind ratio to balance the system—substantially more offshore wind and rooftop PV and less onshore wind capacity (table 3). Note that even when comparing the two extreme utilisation cases here (full and zero utilisation of AB), their total system cost difference is still not significant, i.e. within 10%.

NoUti can lead to additional environmental benefits of preserving agricultural nutrients (figure 6). When AB is not incinerated for energy purposes, it can prevent up to 22.6 kiloton of nitrogen loss annually (equal to 2% of EU consumption in 2019). Similarly, it could be available as feedstock for other non-energy purposes without land/food/feed competition—for example, bio-based industrial materials, papers, and textiles.

Zoom In Zoom Out Reset image size

When we add an abundant supply of dedicated biomass in the DedicatedBiomass scenario (via miscanthus with additional land use as described in appendix C), the total system cost does not change much (only 1% lower). This means that even the limited sustainable potential of ancillary biomass alone can contribute similarly to the energy system compared to the much larger dedicated biomass resource, while saving substantial areas of agricultural land (table 3). Although the attractiveness of bio-based diesel substantially increases by three times when adding dedicated biomass, we can achieve a similar attractiveness by fully utilising AB and by adding the distribution network (DedicatedBiomass vs BioDistribution + FullUtiAll in table 3). In other words, it is plausible that we replace land-intensive dedicated biomass with only limited but strictly sustainable AB for higher societal acceptance (via lower land use) and for higher attractiveness (for producing carbon-neutral fuels).

For a counterfactual nuclear-free European energy system (NoNuclear), we remove the enforcement of a minimum nuclear capacity in 2050. This inherently reduces the total system costs by removing the (assumed-to-be relatively high) capacity costs and operational costs of nuclear plants: the system is −5% cheaper than the 2050 Reference (table 3). Without the balancing capacity from nuclear plants, intermittent offshore wind power and hydrogen production (via the power-intensive electrolysis technology) are less attractive, which results in their drastically reduced capacity (−36% less offshore wind and −12% less electrolysis capacity, NoNuclear in table 3). In this case, ancillary biomass becomes a strategic resource for balancing intermittent power (−9%–13% power curtailment) and for producing synthetic fuels (replacing the reduced capacity of hydrogen-based synthetic fuels). For instance, bio-based diesel accounts for 38% of the total diesel demand (NoNuclear), which is 6 times higher than its share in 2050 Reference scenario with nuclear (first column in table 3). The strategic use of AB is more pronounced when we force the model to use them all in the nuclear-free scenario (NoNuclear + FullUtiAll)—bio-based methanol constitutes 24% of the total industrial demand, which leads to 13% lower land use as well as power curtailment at similar system costs (table 3).

In contrast, a biomass-free and nuclear-free energy system is lacking in both firm capacity (from nuclear) and dispatchable power (from biomass). The consequently higher power curtailment requires drastically more intermittent renewable energy, especially for the rooftop PV capacity (+93%). Moreover, the overall higher intermittent renewable capacity further increases total system costs (NoNuclear + NoUti). Hence, AB is especially critical in a nuclear-free and highly renewable European energy system as it can considerably reduce land uses and total system costs by balancing renewable intermittency.

As a final alternative, the entire available AB potential could also be used exclusively in stationary applications with the intention of providing negative emissions (over and above our assumption of an already carbon-neutral energy system). This could be achieved by enforcing that all ancillary biomass is used either at stationary power plants or Fischer–Tropsch diesel plants, both of which would be coupled with CCS. This AB potential for negative emissions can contribute around 253–623 Mtons CO₂ eq yr⁻¹, which equals to 8%–21% of 2019 EU carbon emissions. We describe in detail how to calculate the negative emissions in appendix C.

We find that it is equally feasible to use all ancillary biomass for additional negative emissions with (AllBECCS) or without nuclear (NoNuclear + AllBECCS) at similar total system costs (less than 1% above 2050 Reference). NoNuclear + AllBECCS, especially, can have the most synergies among all use cases (table 3)—additional negative emissions (goes beyond carbon-neutrality in the other scenarios), enhanced energy safety (no nuclear), and the highest land-use reduction (−13% to −18%).

Here we perform a sensitivity analysis to examine the robustness of our model and identify what may drive the total system cost reduction. We focus on the following uncertainties—inter-annual weather variability by varying the weather year modelled (figure 7) and the costs and efficiencies of future technologies related to or competing with bioenergy (figure 8). We carry out this sensitivity analysis using the 2050 Reference Scenario.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

First, energy system models may be sensitive to the inter-annual difference of various weather patterns (Zeyringer et al 2018). For a 2050-oriented energy modelling study, the future weather is uncertain, especially for estimating hourly PV and wind time series in the long term. This prediction also falls beyond the scope of our paper. However, we can examine historical weather years (from 2000 to 2018) to see how sensitive the model would be to observed variability. We check three aggregate results: AB utilisation, power curtailment, and total system cost. As shown in figures 7, 2018 is our reference weather used in all scenarios. Different weather-year runs comprise the corresponding time series data of solar photovoltaic, wind, and hydro hourly capacity factors, synthetic fuel demand profiles, and heating and transportation demand profiles. For the detailed data files of every weather year, please refer to our Data availability section.

Overall, these three global variables do not vary significantly between weather years (all within 10%), suggesting the robustness of our results to the choice of weather year. Total system costs and power curtailment are less sensitive to different weather years (changes <5%) compared to ancillary biomass utilisation (between 4% and 10%).

Second, we examine the future cost and efficiency uncertainty by changing the cost and efficiency of one technology category at a time (i.e. cost relaxations of ±30% in increments of 10%; efficiency uncertainty of ±15% in increments of 5%) and keep the rest unchanged (table F1 in appendix). These two uncertainty ranges (±30% and ±15%) are a reasonable estimate according to the Danish Energy Agency Technology Data for 2050 (Danish Energy Agency 2019). For technologies examined here, we look at bioenergy-related technologies (biomass supply, biofuels production and distribution, etc) and bioenergy-competing technologies (hydrogen production, synthetic fuels conversion, and storage) to identify under which circumstance it would render significantly cheaper total system costs. Overall, our results are not sensitive to technology cost relaxation or efficiency uncertainties, among which hydrogen technologies have the most perceivable effects on total system costs (around ±4%). Bioenergy-related technologies and storage options do not substantially alter total system costs in any case (less than 1%).