Scholarly article on topic 'Modelling biodiesel production within a regional context – A comparison with RED Benchmark'

Modelling biodiesel production within a regional context – A comparison with RED Benchmark Academic research paper on "Earth and related environmental sciences"

Share paper
Academic journal
Renewable Energy
{Biodiesel / LCA / Spatial / Regional / RED / N2O}

Abstract of research paper on Earth and related environmental sciences, author of scientific article — S. O'Keeffe, S. Majer, C. Drache, U. Franko, D. Thrän

Abstract Biodiesel is an important bridging biofuel for reducing greenhouse gases (GHG). In 2015, Germany introduced a new GHG based quota scheme for biofuels. However, the use of default GHG values for rapeseed cultivation could provide inaccurate for specific regions and locations. Therefore, the aim of this paper was to use RELCA (a REgional Life Cycle inventory Approach) to assess the regional and spatial variation of GHG emissions associated with biodiesel production in Central Germany and to compare these results with the default values of the Renewable Energy Directive (RED), as well as to identify potential mitigation options for biodiesel production. The RELCA simulations indicated GHG emissions of 31.9–39.83 CO2eq./MJ, with emission magnitude changing between biodiesel configurations due to their locations within the CG region. In comparison with typical RED values for biodiesel, the CG simulations showed 13–31% greater mitigation potential. The results also indicated that the configuration of biomass and conversion plant needs to be assessed to develop the most appropriate mitigation strategies. Current GHG mitigation strategies are limited to the energy sector, allowing leakages within the agricultural sector. Therefore, for more spatially targeting GHG accounting to be implemented, sustainability certification should be expanded to other biomass markets.

Academic research paper on topic "Modelling biodiesel production within a regional context – A comparison with RED Benchmark"

Accepted Manuscript

Modelling biodiesel production within a regional context - A comparison with RED Benchmark

S. O'Keeffe, S. Majer, C. Drache, U. Franko, D. Thràn

PII: S0960-1481(17)30110-6

DOI: 10.1016/j.renene.2017.02.024

Reference: RENE 8531

To appear in: Renewable Energy

Received Date: 12 May 2016 Revised Date: 30 January 2017 Accepted Date: 10 February 2017

Please cite this article as: O'Keeffe S, Majer S, Drache C, Franko U, Thràn D, Modelling biodiesel production within a regional context - A comparison with RED Benchmark, Renewable Energy (2017), doi: 10.1016/j.renene.2017.02.024.

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Modelling biodiesel production within a regional context- a comparison with RED Benchmark

Regionally d is tri buted Bio mass inventory

Regionally distributed Conversion technology inventory

Catchment delineated inventory

Regionär Foreground

StapS. Non-regions! modeling

RELCA (REgional life cylce inventory assessment Approach)

• Direct regional Emissions

p ö g - g I I

Indirect regional Emissions

GHG Emission profiles of Biodiesel catchments

Regional Hotspots

CG- _ _ L

123456789 10 RED ■ Cultivation eec Processing ep "transport


5 6 T S 9 M

Implementation of a regionally distributed modelling approach

Spatially and regionally resolved results for regional biodiesel catchments

Comparison with RED typical value for biodiesel production and assessment of mitigation options

2 Modelling biodiesel production within a regional context- a comparison with RED Benchmark

3 *S. O'Keeffe1, S. Majer2, C. Drache1,3, U. Franko4, D. Thrän1,2

4 1 Helmholtz Centre for Environmental Research (UFZ), Department of Bioenergy. Permoserstraße 15, 04318

5 Leipzig, Germany

6 2 Deutsches Biomasseforschungszentrum (DBFZ), Bioenergy Systems Department. Torgauer Straße 116, 04347

7 Leipzig, Germany

8 3 Deutsches Biomasseforschungszentrum (DBFZ), Biorefinery Department. Torgauer Straße 116, 04347 Leipzig,

9 Germany

10 4 Helmholtz Centre for Environmental Research (UFZ), Department of Soil Physics, Theodor-Lieser-Straße 4,

11 06120 Halle/Saale, Germany

13 *Email corresponding author:

15 Abstract

16 Biodiesel is an important bridging biofuel for reducing greenhouse gases (GHG). In 2015 Germany, introduced a

17 new GHG based quota scheme for biofuels. However, the use of default GHG values for rapeseed cultivation could

18 provide inaccurate for specific regions and locations. Therefore, the aim of this paper was to use RELCA (a

19 REgional Life Cycle inventory Approach) to assess the regional and spatial variation of GHG emissions associated

20 with biodiesel production in Central Germany and to identify potential mitigation options for biodiesel production,

21 as well as to compare these results with the default values of the Renewable Energy Directive (RED). The RELCA

22 simulations indicated GHG emissions of 31.9-39.83 CO2eq. /MJ, with emission magnitude changing between

23 biodiesel configurations due to their locations within the CG region. In comparison with typical RED values for

24 biodiesel, the CG simulations showed 13-31% greater mitigation potential. The results also indicated that the

25 configuration of biomass and conversion plant needs to be assessed to develop the most appropriate mitigation

26 strategies. Current GHG mitigation strategies are limited to the energy sector, allowing leakages within the

27 agricultural sector. Therefore, for more spatially targeting GHG accounting to be implemented, sustainability

28 certification should be expanded to other markets for biomass.

29 Key words: Biodiesel, LCA, spatial, regional, RED, N2O

30 1.Introduction

31 Biodiesel, in terms of production capacities and economical relevance, is one of the most important bridging

32 biofuels being promoted to wean society off fossil dependent mobility [1-3]. It is also particularly important for

33 Germany, with the second highest installed capacity of biodiesel production (approx. 4.4 Mio. t/a), in Europe [3].

34 The majority of which is derived from the conversion of rapeseed i.e. rapeseed methyl ester (RME) [4]. Such

1 biodiesel facilities are based on mature, relatively simple technologies, unlike advanced biofuels, which are still at

2 various stages of development and with relatively higher investment costs. Thus, it is foreseen that biodiesel is likely

3 to play an important role in the transportation sector at least until 2030 [5,6]. For this reason the environmental

4 sustainability of such biodiesel still needs to be ensured, particularly in terms of greenhouse gas (GHG) mitigation.

6 Indeed, one of the primary goals for using biodiesel instead of fossil diesel is the reduction of GHG emissions.

7 Going one step further to ensure reductions, Germany in 2015, introduced a new quota system for biofuels, changing

8 from energy and mass related quotas, as stipulated under the Renewable Energy Directive (RED) [7], towards a new

9 GHG based quota scheme. Under this scheme biofuels must now satisfy increasing requirements for GHG reduction

10 over the entire chain, from the field through to arrival at the biofuel production plant [8]. As a result, the competition

11 between different biofuel technologies and feedstocks is now based on the GHG-mitigation potential as the main

12 criteria for the success of a biofuel producer [8].

14 Under RED [7] a biodiesel producer can estimate the GHG-mitigation potential of their biofuel across the major

15 steps in the production chain; cultivation, conversion, transport by using: 1) the typical or default values outlined in

16 Annex V of the directive (a form of European average); 2) a combination of actual values with default values (e.g.

17 own plant data with default values for rapeseed production) or; 3) real empirical data collected across the whole

18 supply chain [7]. In general for cultivation of rapeseed, the use of default values are preferred, as this reduces the

19 amount of bureaucracy required to determine the GHG balances of the rapeseed produced [9]. However, rather than

20 using default values outlined in the RED, (29 g CO2eq./MJ), it was recommended instead for German biodiesel

21 producers to use emission values estimated for the different Federal states (i.e. NUTS2)1 [10,9]. In other words a

22 famer producing rapeseed in a particular Federal state would have an associated GHG value for their rapeseed (2323 25 g CO2 eq./MJ RME), as long as such cultivation complied with the good farming practices outlined by the

24 Common Agricultural Policy (CAP) [11,10]

26 With cultivation of rapeseed accounting to between 50-90% of the total GHG balance for biodiesel production, the

27 use of RED default values for ease of implementation, may provide inaccurate results for specific regions and

28 locations [5]. Indeed one such study for the Veneto region, in Italy, estimated values much higher than those

29 reported in RED for sunflower and rapeseed [12].The reason for such discrepancies relates to two major aspects.

30 The first is yield, which is dependent on specific geographical (e.g. soil, climate) and regional (e.g. management)

31 conditions [5]. The second relates to soil emissions, which Hennecke et al. [13] also identified as a blind spot within

32 the RED accounting system, as there was no obligation to include more spatially detailed accounting which could

33 capture the interaction between management practices and geographical conditions affecting such soil emissions, as

34 well as the soil emissions themselves.

1 NUTS - Nomenclature of territorial units for statistics used by the EU (

2 Life cycle thinking is generally employed in combination with RED calculations to estimate the GHG-mitigation

3 potential of biofuels [5]. Here too, Malfa, Freire [14] also identified in their review, that for many life cycle

4 assessments of biodiesel production, the spatial details to determine the major emissions from cultivated soils e.g.,

5 nitrous oxide ( N2O) and carbon dioxide (CO2), were still missing. Recently, regionally contextualised life cycle

6 concepts were promoted to include greater spatial details in the life cylce assessments of regional bioenergy

7 production [15]. A "within regional" context or scope was identified helping to determine the regional distribution

8 of emissions associated with bioenergy production within a regional foreground. For this context the "RELCA"

9 modelling approach was developed. The goal of which is to develop a regionally distributed life cycle inventory to

10 assess the potential regional and spatial variation in the environmental performance of bioenergy production within a

11 region [16]. Therefore, the aim of this paper is to: 1) determine the regional distribution of direct and indirect GHG

12 emissions associated with the production of biodiesel in the Central Germany region; 2) to explore how regionally

13 detailed life cycle approaches, such as RELCA, can be used to identify potential options for reduction and

14 improvement of the GHG emissions associated with the regional biodiesel production and 3) to compare such

15 detailed assessment results with the default values outlined in RED, in order to highlight the need for including

16 greater spatial details within the associated calculation methods.

17 2. Material and methods

18 2.1 RELCA modelling approach

19 A "within regional" life cycle scope was implemented [15], using the RELCA modelling approach [16] to develop a

20 "regionally distributed foreground" inventory (Figure 1). RELCA determines the regional distribution of GHG

21 emissions from the foreground activities, as well as GHG emissions from non- regional activities (indirect burdens).

22 The latter refers to the associated activities producing these flows and are assumed to be outside of the region, along

23 with their associated environmental burdens (i.e. released anywhere else but the region of focus and are therefore not

24 considered with a spatial orientation). RELCA combines conventional geographical modelling with conventional

25 life cycle software through the use of catchment delineation to assess the potential environmental implications of

26 bioenergy configurations (i.e. bioenergy plants and their biomass catchments).

28 The regional scope is defined as one scale lower than a country and denotes the foreground activities relating to the

29 bioenergy systems being assessed [15]. The regional conversion system, transesterification, investigated in this

30 paper refers to the combination of rapeseed with different biodiesel technologies (scales) used in the region to

31 produce a biodiesel product. The RELCA approach applied was retrospective and complied with the ISO LCA

32 standards [17], as well as GHG accounting method of the IPCC [18] and RED. Keeping in line with the RED

33 calculation method (supplementary material A1), an attributional life cycle accounting approach was implemented

34 and all GHG emissions were allocated based on energetic content.

35 (Insert Figure 1)

1 2.2 Geographical description of study region

2 The regional foreground is set to the eastern German region of Central Germany (CG), which consists of three

3 federal states, or "Bundesländer"; Saxony, Saxony-Anhalt and Thüringen [19]. For the CG region, there is a north

4 south divide with regard to climate, with northern areas having on (50 year) average relatively higher mean annual

5 temperatures (9-10°C) and lower mean annual rainfall (450-600mm), compared with the more mountainous regions

6 of the South's mean annual temperatures (6-7°C) and mean annual rainfall (600-1000 mm)[20,21]. The BÜK 1000

7 [22] is an official map of German soils, with a total of 72 soil types. In CG region approximately 44 soil types can

8 be found. The "Ackerzahl" values (Az), or agricultural production value ranges mostly from 31-60, with some areas

9 in the north central part of the region having a value as high as 90 [23].

10 2.3 Implementation of RELCA

11 2.3.1 Crop allocation modelling (CRAMod)

12 Approximately 3.5 million hectares are devoted to agricultural production in the CG region, of which over 400,000

13 hectares (ha) of arable land was devoted to rapeseed production in 2010 (approx. 16% of arable land) [19,24]. For

14 the base year (2010), rapeseed yields in the CG region ranged from 2.31 t/ ha to 4.41 t/ha (fresh matter) with an

15 average yield of 3.92 t/ ha, slightly higher than the 10 year average of 3.75 t /ha (2004-2013) [25]. The CRAMod

16 approach outlined in [16,26] was implemented here. The output geodataset provided the potential regional

17 distribution of rapeseed cultivated for the base year. For each 25 hectare rapeseed grid cell (500 x 500m2), important

18 regional geographical variables (e.g. climate, soil types, agricultural suitability) were also provided, in order to

19 model management and emissions associated with the regional cultivation of the rapeseed crop. This is outlined in

20 the next section (BioMod). Additionally, the output from this step provides the regional biomass availability

21 required for the CAMod step (section 2.3.4).

22 (Insert Figure 2)

23 2.3.2 Biomass inventory modelling (BioMod)-Direct regional flows

24 As part of the BioMod step the regional management practices were determined, as well as the direct emissions

25 associated with producing rapeseed within the CG region. It was assumed the rapeseed was sown in August and

26 harvested the following July and was also assumed to be in a rotation common to the region; Wheat-Rapeseed-

27 Wheat [27]. All flows relating to biomass cultivation until the point of harvesting were considered and all

28 calculations steps carried out for biomass cultivation were estimated for each rapeseed grid cell using MATLAB

29 2012b (The Math Works, Inc., Natick, Massachusetts, United States). For more detailed information on how

30 this was carried out please refer to [16].

31 (Insert Table 1)

1 Management input, flows

2 To estimate the amount of nitrogen fertiliser (N) applied per grid cell the "N-Basis-Sollwert" method (required N

3 rate) was used (eq. 1 & 2). Best farming practices were assumed for rapeseed production in the region [37-39], with

4 the recommended N rate dependent on yield [37] (supplementary material, A2).

NrateRec = 0.0286 x yield - 1.7143 (1) N applied = NrateRec - Nmin ± Addadj (2)

5 The average mineralized nitrogen in the soil (Nmin), estimated for the six "Bóden Klima" (soil-climate) regions [40]

6 found within CG were derived from various regional reports and datasets [37,41-43,39,27]. Additionally, if the

7 Akazahl value of a grid cell was less than 40, then the nitrogen fertiliser rate required (Nrat.eRec) was adjusted

8 (Addajj) by subtracting 10 kg N /ha. The other rapeseed management assumptions are outlined in Table 1.

9 (Insert Table 2)

10 Field operation input, flows - diesel demand

11 The tractability of the soil was also used to estimate potential diesel consumption and hence emissions from field

12 operations on a per grid cell basis. This was done using the online KTBL tool [28] as it provides fuel consumption

13 assumptions for three different soil types, which were assumed to be similar to those outlined by [45]; light. (<12%

14 clay), medium (13-25% clay) and heavy (>25% clay). Only the major field operations were considered (Table 2).

16 Nitrogen sourced emissions to air

17 Nitrous oxide emissions were estimated for each grid cell according to the German national guidelines outlined by

18 [46]. This required estimating emissions using a Tier 2 approach, shown in equation 3". The direct field emissions of

19 nitrous oxide (N20) induced due to fertiliser application was estimated using the emission factor derived by Brocks

20 et al. [44] (Table 3).

21 (Insert Table 3)

22 N2On = Y, [(¿"«appiisd X EFlBrock s IPCC NH3Fert * EF^hz) + (£"jv_jv0/ert * EF¡vo) (^N.min * EF^ jpcc )] (3)

En residues = CropDM x FracRenew x RAG x NAG x (1 - FracRemov) + (NBG x RG) (4)

23 The emissions resulting from rapeseed residues left on the field (ENresidues) were estimated using eq. 4 (adapted from

24 IPCC equation 11.6 [47]. The parameter assumptions are outlined in Table 4.

2 Modified IPCC equation forN20 emissions, using only those parameters relevant for the assumed conditions modelled in this study

1 (Insert Table 4)

3 The nitrous oxide emissions resulting from ammonia volatilisation (IPCC class of indirect N20 emissions) were

4 estimated using the emission factors outlined in Table 5. Nitric oxide emissions NO were estimated per grid cell

5 using the EF of 0.012 kg NO_N /kg N applied outlined by Stehfest, Bouwman [48] and according to [46] .

6 (Insert Table 5)

8 Emissions relating to soil organic carbon

9 No land transformations regarding the conversion of grassland into arable land were identified from the available

10 land use statistics for the CG region (Supplementary material A4). Therefore, no C02 emissions were assumed to be

11 released due to land use changes. However, changes in carbon fluxes from cropland remaining cropland were

12 accounted for [18]. Changes in soil С were considered using the approach outlined by Petersen et al. [49], to

13 account for the potential effect of soil carbon changes for one year on the atmospheric C02, independent of the

14 initial soil carbon level. It was estimated that approximately 10% of the С added to the soil will be sequestered in a

15 100-year perspective and this factor was implemented here as in previous studies [50,51]. Similar to Mogensen et al.

16 [51] we assumed a reference crop of 'wheat grown without manure input and with no straw removed' for our

17 regional biodiesel scenarios. The contribution to soil С from each crop was based on the carbon flux estimation

18 methods used in the Candy Carbon Balance model (CCB) [52,53] (Table 6 and equations 5-10). Assuming a steady

19 state, equation 10, was used to calculate the effect of cultivating both wheat (dCsoc wht) and rapeseed (dCS0Crj>s) on

20 soil carbon. The difference between both (dCSOc_y) is then used as an indicator of SOC change for the annual time

21 step occurring per grid cell, i.e. carbon is either lost or sequestered. The carbon balance (dCSOc_y) was then

22 multiplied by the Peterson [49] factor.

23 (Insert Table 6)

24 Straw:

StrawDM = CropFM x RAG (5)

Crep_str = StrawDM x Cstraw x SOMrep bp (6)

26 Roots:

Ninput = (CropFM x Ncoeffl x Nfact) + CropNTOT (7)

Cfom = CNrati„ x Ninput (8)

Crep_root = SOMrep rs x CF0M (9)

28 Total:

d CsOC crop ~ Crep Crep str Crep root (10)

dCsoc_y = ^^*soc_wht ~~ dCS0Crps (11)

3 Emissions from field operations

4 Emissions associated with field operations provided in Ecoinvent v2.0 were provided on a per kg diesel basis and

5 were converted to kg emission per ha worked for all field operations (Table 2). The emissions were then estimated

6 per ha and multiplied by 25 for each grid cell (i.e. 25ha) of rapeseed (Table A2).

8 2.3.3Conversion Plant modelling (CPMod Step) Technology inventory

9 Operational assumptions

10 The operational base year for biodiesel production was assumed to be from the point of harvest in autumn 2010,

11 through to autumn 2011. During this time period there were approximately 10 biodiesel producing plants in the CG

12 region3, producing less than one million tonnes of biodiesel, with oilseed rape as the main oilseed feedstock

13 [54,55]4. The production capacity for all biodiesel plants was assumed to be approximately 52% of the installed

14 capacity for all plants [56,57]. The locations for each biodiesel plant within the region were determined and their

15 coordinates generated using Google maps [16]. The biodiesel plants were seen to have some degree of spatial

16 clustering, with many of the small scale plants (installed capacity <10,000 t biodiesel/year) located in the southern

17 part of the region and the larger scaled plants (installed capacity >150,000 t biodiesel/year) located predominantly

18 towards the northern parts of the region (DBFZ).

19 (Insert Table 7)

21 Model plant concepts

22 The development of the plant concepts was carried out in collaboration with the DBFZ [61]. Through investigating

23 the biodiesel production in the region (literature, internet, contact with plant operators), it was determined that most

24 biodiesel plants were attached to an oil mill (i.e. no transport of oil between oil mill and transesterification plant).

25 Three biodiesel plant concepts were identified, differentiated by the different oil mill extraction technologies

26 employed (i.e. cold press, hot press and hexane extraction). Therefore, the oil output, as well as the energy and

3For the purpose of this study we assume the operational year starts from the harvested rapeseed in 2010 until the following years harvest. Therefore companies that operated through this year were used in this study

4 For reasons of data sensitivity we provide a map for rapeseed production only

1 auxiliary inputs differ between the three plant concepts. The mass and energy flows as well as important modelling

2 parameters are outlined in Table 7.

3 For the production of biodiesel, rapeseed oil is mixed with a catalyst (mostly potassium hydroxide, with sodium

4 methylate used in large scale plants) and methanol in order to obtain a 98% methyl ester (biodiesel) yield. After this

5 transesterification reaction the biodiesel and glycerine are separated and the biodiesel undergoes several purification

6 processes (e.g. to remove excess methanol) and a drying step resulting in biodiesel which complies with the standard

7 DIN EN 14214. The remaining glycerol rich stream (50%) which also contains a mixture of methanol, soaps and

8 catalyst under goes further process steps, depending on the scale of operation. It was assumed that for the smaller

9 and medium plants the glycerine stream is purified to a concentration of approx. 80% (technical grade glycerol),

10 whereas for the larger plant it was processed to pharmaceutical grade glycerine (>99%), as this was considered to be

11 more economically viable for the larger plant concepts. The rapeseed demand vector for each model biodiesel plant

12 was then determined by; 1) their installed capacities; 2) the assumed operational capacities (52%) and; 3) the

13 estimated conversion efficiency from rapeseed to biodiesel (Table 7).

15 2.3.4 Catchment allocation modelling (CAMod)

16 The purpose of the CAMod step is to combine the regionally distributed bioenergy technology inventory with the

17 regionally distributed biomass inventory as in Figure 1. This is done by assigning the biomass to the associated

18 conversion plants using the demand function determined in the CPMod step (i.e. tonnes of bioenergy crop required

19 for annual production capacities). In this way the spatial configuration of the biodiesel plants were estimated.

20 Oil mills

21 The production of biodiesel (RME) was not the only demand for rapeseed within the region. Therefore, the potential

22 demand coming from oil mills producing for another market (i.e. food), was also estimated from regional reports

23 [62-64] and web searches. It was assumed that for the base year around 14 oil mills were in operation, with the

24 CAMod simulations estimating approximately 25% of the regional rapeseeds being diverted to these plants, which

25 was similar to the German average reported [65].

26 Rapeseed transport

27 In order to estimate the foreground emissions associated with transporting the rapeseed from field to biodiesel plant

28 the lorry emissions outlined in Ecoinvent 2.0 [66] (CH: operation, lorry 20-28t, full, fleet average [Street])

29 (supplementary material, A6) were used. The harvested rapeseeds were transferred directly at field edge to the

30 transporting lorry. It was assumed that the transport was carried out by a logistic company, with the driving route for

31 the lorry calculated for one direction only, as it was also assumed the lorry had an unrelated job in the vicinity

32 before collection of the rapeseed. Therefore, only the transport of the rapeseed to the plant was estimated as part of

33 the catchment modelling.

1 Catchment modelling

2 Catchments were modelled according to [16] using MATLAB2012b (The MathWorks, Inc., Natick, Massachusetts,

3 United States) generated scripts. It was assumed that smaller plants were more likely to use "regionally" sourced

4 rapeseed. Therefore, the model assigned catchments to the smaller plants first. This assumption was also applied to

5 the oil mills (non- biodiesel) in CG, as many were determined as small scaled and decentralised. The catchments

6 grew in size until the demands of all the plants were satisfied in one simulation run. However, if a rapeseed grid cell

7 was closer to one oil mill/biodiesel plant than another, the rapeseed field was allocated to the closest, in order to

8 avoid catchment area overlap [67]. Both biomass and biodiesel plant inventories were combined for each biodiesel

9 configuration as a result of the CAMod step (Figure 1). Delineating the inventory for each biodiesel configuration

10 sets the boundary for aggregating the all relevant foreground flows (e.g. mass, energy and emissions).

12 2.3.5 Non Regional modelling (NoRIodM)

13 The regional boundary denotes the foreground activities relating to the bioenergy systems being assessed [16,15],

14 such foreground activities also require inputs from outside the region (e.g. fertiliser products, fossil fuels, grid

15 energy). Therefore, the purpose of the NoRiMod step in the inventory accounting is to link the indirect upstream

16 emissions generated outside the region, to the direct emissions produced within the regional boundary.

17 Auxiliaries

18 Fertiliser sales statistics for each Federal state (Saxony, Saxony-Anhalt, Thüringen) were used to determine the

19 percentage contribution of fertilizer products to their fertilizer mix [68] (Table A3). In order to simplify the fertilizer

20 mix, if a fertilizer product contributed less than 3% weight to the entire annual sales, it was excluded and its weight

21 was normalized across the remaining fertiliser products in the mix. This cut off was also cross checked with regard

22 the GHG contribution of each fertilizer type and no significant contribution was found by those that were excluded

23 in the mix. The fertiliser mix was then attributed to each constituent rapeseed grid cell found within the associated

24 Federal states using MATLAB2012b based scripts (see [16] for more details). Additionally, once a final list of plant

25 protection products were identified, the active ingredients of the crop protection products were then determined [33],

26 which could be linked to their upstream emissions (Table 1).

27 Biomass imports

28 For one of the larger scaled plants the output of the CAMod step resulted in a short fall of biomass available, due to

29 its position on the regional periphery. Therefore, to cover the biomass demand, the shortfall was assumed to be

30 imported into the region from Eastern Europe.. The input data provided to and used by the Biograce tool (i.e. data

31 from the JEC consortium (biofuel pathway RED method) [69]) was assumed here to represent the European average

32 and therefore, was used to develop the unit processes for the imported biomass (supplementary material, Table A8).

1 Estimating non-regional GHG emissions

2 At this step in the RELCA approach, flows relating to biomass cultivation (inputs and emissions) are aggregated per

3 catchment area and then divided by the amount of biomass used by the bioenergy plant (e.g. kg rapeseed used).

4 These catchment averages were then used as input flows for a rapeseed model developed in the life cycle software

5 GaBi 6.0, enabling the connection to be made with the upstream non -regional flows (e.g. fertiliser products). This

6 was then linked to the relevant regional biodiesel models (Table 7) modelled in GaBi 6.0 and coupled with

7 Ecoinvent 2.0 inventory [70]. The resulting output of the modelling step was the aggregated emissions per

8 functional unit of mega joule biodiesel produced. However, an additional step was also taken during the

9 development of the regionally distributed inventory in which characterised values for the indirect non-regional GHG

10 emissions were taken from Ecoinvent 2.0 (for more detailed explanation of this step please refer to [16]) and

11 allocated to all constitutent rapeseed grid cells. In this way RELCA was used to estimate both the direct and indirect

12 GHG emissions per functional unit of energy or per hectare supplied for the constituent grid cells. Furthermore

13 GHG emissions were allocated energetically using lower heating values (Table 7).

14 2.4 Functional units and allocation

15 The calculations of RED focus on the function of GHG mitigation based on one mega joule of energy produced.

16 Therefore, all flows and emissions associated with the production of biodiesel are usually normalized to a functional

17 unit of one mega joule of energy output. While this is useful for comparing across non-spatial assessments, with the

18 introduction of greater spatial details additional considerations need to be included to capture the spatial variation in

19 a meaningful way. Therefore, we present and discuss the results here also using, the GHG emission intensity per

20 function of land area input, required to meet the rapeseed demand of the biodiesel plant (CO2eq. / ha). The latter,

21 refers to the mean of the emission profile associated with a particular catchment. A catchment with a greater share of

22 grid cells with higher emissions, results in a higher emission intensity per function of land area input. Therefore,

23 both functional units are used in this paper to provide an overview of important spatial details, which can help with

24 the interpretation of the data. Furthermore, to remain comparable with the typical values outlined in the RED, all

25 results, unless otherwise stated, have undergone energetic allocation. However, non-allocated values where deemed

26 relevant, will also be shown, in order to support the interpretation of results. Additionally, we will focus the

27 discussion mainly on the cultivation of rapeseed, as this is where most of the variability is introduced.

29 2.5 Statistical analysis

30 MATLAB 2012b Statistics and Machine Learning Toolbox™ was used to generate description statistics, as well as

31 to determine the emission distributions, or emission profiles for each biodiesel catchment. Overall significance

32 between the base case and sensitivity analysis was tested using Kruskal Wallace (p<0.05) and a post hoc test

33 (multcompare) with a Bonferoni correction, to test for significance between different regional catchments, in order

34 to provide a better understanding of potential within regional trends. Further to this, a spatial statistical analysis was

1 carried out, Esri ArcGIS hotspot analysis (Getis Ord Gi ) with a spatially weighted matrix ( K nearest neighbour ,

2 n=8), to identify where in the region emissions from cultivation of Rapeseed were statistically higher ("hotspot") or

3 lower ("coldspot") than the overall regional average. Additional profiling of the underlying geographical parameters

4 (summary statistics) was then carried out in order to understand the potential reason for such significantly

5 higher/lower results.

6 2.6 Sensitivities Analysis - catchment and location

7 As the results are catchment delineated, it is important to test how sensitive the results are to catchment size and

8 location. Therefore a "uniform analysis" was carried out. This was done by keeping locations of the biodiesel plants

9 constant, and dividing the total biodiesel production capacity uniformly across all 10 plants, thus, assigning each

10 biodiesel plant the same production capacity and demand for rapeseed (supplementary material, Table A5). In this

11 way an alternative demand pattern for rapeseed across the region was established.

12 2.7 Scenario Analysis-testing one mitigation strategy

13 One proposal for mitigating GHG emissions from rapeseed production and hence biodiesel production has been to

14 reduce nitrogen fertiliser application resulting in potential yield loses [2,10]. However, the risk is that lowering

15 yields may not result in the anticipated reduction in emissions [71]. In order to assess the mitigation potential of

16 lowering fertiliser and hence potentially yields, two scenarios were tested, in which rapeseed was modelled with a

17 reduced yield of 10% (S10) and 25% (S25) from the base case, similar to yield reductions outlined by [10].

19 3. Results & Discussion

20 Using the "within regional" context has enabled the exploration of three major aspects with regards the regional

21 production of biodiesel: 1) the general regional variation in potential GHG emissions associated with the annual

22 production of biodiesel configurations5 within CG, as well as the percentage GHG produced within and outside the

23 region; 2) the spatial patterns and regional trends showing how emission magnitude changes between catchments

24 due their locations in CG, in a base case and under different scenarios and; 3) how regionally variability cannot be

25 captured with a simple regional average value, like those used in the RED calculation methods.

26 3.1 General overview - GHG emissions from biodiesel production in CG

27 3.1.1 Overall emission summary

28 The energetically allocated results of the RELCA simulations for the CG region indicate an overall regional average

29 of 36 g CO2eq. /MJ for producing biodiesel in CG (field-to-biofuel factory). For the smaller scale operations, which

5 Configurations refer to the combination of biodiesel technologies and rapeseed supplied to plant, catchments refer to the rapeseed supplied to the plants only, no conversion step is considered.

1 are sourced from catchments in the south west of the region, the emissions ranged from 32.35 - 39.83 g CO2eq. /MJ

2 (^=36.04).For the medium scale operations the range (no distinct spatial clustering) was 31.90 - 34.83 g CO2eq. /MJ

3 (^ =33.36) and for the large scale operations in the northern part of the region, the range was found to be 37.714 38.47 g CO2eq. /MJ (^ =38.02) (Figure 3).

5 For cultivation of associated rapeseed, the smaller operations in the south had a tendency towards higher cultivation

6 emissions per mega joule, than those of the medium or larger scale operations. The mean catchment emissions were

7 found to range for the small scale from 25.15 to 32.71 (^ =29.46) g CO2eq. /MJ; for the medium scale from 23.958 26.84 (^ =25.39) g CO2eq. /MJ and; for the larger scale from 25.5-26.35 g CO2eq. /MJ (^ =25.84). However, the 9 influence of production scale on the GHG emissions is not so clear (both for allocated and non-allocated results).

10 3.1.2 Contribution analysis -production steps

11 In general emissions associated with the conversion step contributed approximately between 19-32% of the total

12 emissions. The energy demand for conversion (thermal and electricity) ranged from 10% in the smaller conversion

13 plants to over 20% in the larger plants. Auxiliaries required by conversion plants (e.g. methanol, acids) contributed

14 between 9-10% of the total emissions. Unsurprisingly the cultivation of rapeseed was again found to contribute the

15 greatest share of emissions (67-81%) for the production of one mega joule biodiesel. Emissions from fertiliser

16 production (indirect) and field emissions (direct) contributed between 27-29 % and 35-41% of total GHG emissions

17 respectively, thus together contributing the greatest share of emissions for the regional biodiesel production (see

18 supplementary material, A10).

19 3.1.2 Contribution analysis -Regional vs Non-regional

20 Surprisingly, the results indicate that less than half of the total emissions come directly from the region (40-48 %).

21 The biggest shares of the GHG burdens (51-60%) were estimated to be generated outside of the region (i.e. indirect,

22 assumed to be produced and emitted somewhere else). However, when looking only at the cultivation of rapeseed,

23 the largest share of emissions comes from within the regional boundary (53-59%), with a lower share generated

24 outside of the region (41-46%) (supplementary material, A10).

25 (Insert Figure 3)

26 For the conversion step, no direct GHG emissions were identified, as all emissions were assumed to be generated

27 outside the region [72]. Naturally, the indirect GHG emissions will be higher for the larger conversion plants (~30%

28 of total), than for the smaller conversion plants (~21% of total). Therefore, the indirect burdens associated with the

29 conversion step will be concentrated towards the northern part of the region.

30 Emissions from soil, nitrous oxide, contributed the greatest share (78-83%) of direct regional emissions, with soil

31 carbon fluxes resulting in a release of CO2, contributing between 4-10% of the direct emissions. The use of

32 agricultural machinery (exhaust emissions), contributed approx. 11-16% of direct regional emissions. By far the

33 largest generator of total indirect emissions outside the region was associated with the production of nitrogen

1 fertilizer used for the cultivation of rapeseed (79-83% of indirect). This was followed by the other fertiliser products

2 applied (14-16%), diesel (2-3%) and pesticides (1-2%). The regional distribution of GHG emissions associated with

3 the cultivation step will be discussed further in the next section.

5 3.2 Exploring spatial patterns and trends

6 The use of the within regional modelling approach enables the exploration of the data in a more spatially detailed

7 way. It helps us not only understand what the regional emission averages are, or the regional distribution of

8 emissions, but it also helps to understand the potential underlying spatial trends and issues relevant for improved

9 regional biodiesel production and in particular rapeseed production.

10 3.2.1 Comparison of regional Biodiesel configurations and emission profiles

11 In the CG region, the greatest biodiesel energy output is concentrated in the northern part of the region, which was

12 also associated with higher ranging yields (3.9-4.4t/ha). The lowest biodiesel energy output is located in the more

13 southerly parts, which were also found to have relatively lower yield ranges (2.9-4.3t/ha). Therefore, as expected the

14 larger biodiesel operations in the northern areas of the region, with higher energy output and higher yields (i.e. lower

15 land area input per mega joule), perform in general better than the smaller biodiesel operations of the southern areas,

16 with lower energy output and lower yields (Table 8-non allocated). However, for energetically allocated values, this

17 trend alters slightly, with two of the smaller scaled catchments in the south central area of the region showing

18 relatively lower cultivation emissions, thus outperforming the larger scaled plants in the north. Both catchments are

19 located in areas which had relatively good yields (>3.5t/ha) and relatively lower emissions. Therefore, despite the

20 lower energy output of the smaller scaled plant, the lower emission intensities (per hectare) and more preferable

21 energetic allocation for biodiesel, results in overall lower mean emissions for these catchments. While allocation can

22 have a large effect on the outcome of an LCA, this is not the focus of this paper. The non-allocated values are

23 provided to show that energetic allocation improves the performance of the smaller scale plants and this must be

24 kept in mind when interpreting the results, further discussion on allocation is outside the scope of this paper.

25 However, what is shown here is that by using more detailed values, rather than average values, the potential range of

26 GHG emissions associated with the regional production of biodiesel, as well as potential outliers, can be identified.

27 (Insert Table 8)

28 Looking at the emissions profiles of the different catchments (Figure 4), it can be seen that the northern areas of the

29 region are at a relatively higher range than those found in the more southerly areas, with catchment 9 having a

30 significantly higher emission profile than all other catchments found in the region. Therefore, it is clear from these

31 emission profiles that the higher energetic output from these biodiesel catchments helped to counteract the higher

32 emissions profiles seen in Figure 4. To the south, the smaller scaled catchments had relatively lower emission

33 profiles. However, despite the relatively lower emission intensities, their relatively lower energy output results

1 combined with a greater LAi results in higher emissions per MJ biodiesel (we discuss this further in section 3.2.5).

2 Despite the counteracting effect of the catchment's energy output, to be able to introduce better GHG mitigation

3 strategies, we still need to understand and identify potential underlying geographical aspects for such emission

4 profiles, if we are to reduce them. Therefore, to identify catchment areas which had significantly higher and lower

5 emissions than those estimated for the rest of the region, an ArcGIS hotspot analysis (Getis Ord Gi ) was carried out

6 (next section).

7 (Insert Figure 4)

9 3.2.3. Hotspot analysis - regional emissions

10 Catchments to the north were found to have a greater share of grid cells with significantly higher total GHG

11 emissions (direct and indirect) for the production of rapeseed (Figure 5). The more southerly areas had a tendency

12 for lower emissions, more "cold spots". Profiling the geographical parameters with a frequency analysis indicated

13 that "hotspot" grid cells, relative to the other grid cells in the study region, were located in areas with a high

14 dominance of Loess soils, higher Ackerzahl values (Az), higher annual temperatures, and therefore, naturally higher

15 associated yields. They also had relatively higher Brocks emission factors (Table 9). These hotspots therefore, had

16 also higher associated soil N2O emissions (relating to higher yields /nitrogen fertiliser application) and a greater loss

17 of carbon out of the soil (due to higher wheat yields relative to rapeseed, therefore resulting in a lower ability to

18 replace soil carbon) and higher emissions associated with field operations (related to higher clay content in soil).

19 It is a well published issue that soil emissions contribute heavily to the GHG balance of biomass production [73-76].

20 The contribution of such emissions towards regional biodiesel production has also shown to be the case here too.

21 The results of the hotspots indicate that Loess soils with high Az values associated with higher yields were also

22 associated with higher emissions. Although, this is more than likely a function of the modelling assumptions i.e.

23 higher yields assumed a greater nitrogen fertiliser, in order to explore such results further, the dataset of [48] for

24 German sites only, was used to compare the emissions results found for Loess soils against the other observed soil

25 types in Germany. The results of this indicated slightly higher associated emissions for Loess soils, but not

26 (Insert Figure 5)

27 significantly higher (Supplementary material, A11). However, without empirical field trials to test for such a trend,

28 it is therefore, not possible to know if such soils, as shown here, are susceptible to more N2O losses in comparison to

29 other soils with similar climate and management conditions.

30 With regards to the direct N2O emissions estimated here, these were found to be between 2.68-12.61 N2O /ha/ a,

31 which are comparable to those found within the literature for Eastern Germany; 1.9-14.8 kg N2O ha-1 a-1 [77,73,44].

32 Although, within the ranges outlined in the literature, the mean values were slightly higher than those estimated for

33 the different Federal states [44,77]. The reason for this could be related to the relatively higher yields and rainfall

1 observed for 2010, ultimately resulting in a larger number of grid cells having greater assumed fertiliser application,

2 as well as a higher Brocks EF, which is dependent on rainfall level. This indicates the importance for not only

3 including more spatial details, but also for including more temporal details. However, while the use of Tier 3 IPCC

4 calculations or biophysical modelling of such cultivation system, may improve the results presented here [74,78,77],

5 as stated already, the ability to validate soil related N2O emissions is still an issue due to due to lack of spatially

6 distributed empirical studies. This is an area which requires much more research in order to support more spatially

7 detailed studies, such as the one presented here.

8 Additionally, for the soil carbon estimated in this study, a steady state was assumed and potential SOC build up or

9 management strategies prior to the focus time point of this study were not considered. Naturally, this is a limitation

10 of the modelling results, as when assessing complex systems, such soil emissions; it would be more beneficial to

11 include a greater temporal scope. However, as a first step the aim was to provide spatially distributed results for one

12 year and this is what has been done. The losses of carbon from soil were estimated here to be between 0.05-0.4 t C

13 ha-1 for this simulated annual time step, on the lower end of the literature (0.40-0.84 t C ha-1 yr-1) [14,79]. However,

14 with many LCA studies ignoring the carbon flux relating to SOC in their studies [80-83], the results presented here

15 show that the inclusion of such carbon fluxes, associated with rapeseed cultivation, are important for estimating the

16 associated overall GHG balances of biodiesel. Indeed only through accounting for such emissions can agricultural

17 practices be improved for cleaner production of biodiesel [79], as soil is a primary ingredient for the successful

18 cultivation of crops.

19 (Insert Table 9)

20 How soil is managed will determine the success of sustainable food, fodder and fuel production [84]. Therefore,

21 what the results of the study indicate is that including greater spatial details can help to distinguish where within a

22 region relatively higher soil emissions are potentially occurring. This can help to identify where management

23 practices may need to be adapted or where further more detailed assessments may need to be carried out.

24 3.2.4. Hotspot analysis -Direct and Indirect regional emissions

25 The direct GHG emissions match the overall spatial trend found for the total emissions (Figure 5),

26 however, for the indirect emissions the spatial distribution for "hot and cold" spots differs slightly to that

27 of the total emissions. The spatial distribution is not entirely yield dependent, but appears instead to be influenced

28 by the Federal state fertiliser mix, with an increase in the frequency of "hotspots" further

29 south(supplementary material A12). This was especially prevalent for the Federal state of Saxony which was

30 found to have a higher share of CAN in its mix, as CAN has the highest associated production emissions in

31 comparison to other fertiliser products [85]. Differences in spatial patterns can be seen in supplementary material

32 A12.

1 As indicated in section two, one of the biodiesel plants operating in the region was assumed to import approx. 75%

2 of its Rapeseed. With the assumption that it comes from Europe, the direct and indirect balance for this biodiesel

3 configuration will be slightly different from the other regional configurations, with 90% of its total GHG emissions

4 being estimated to occur outside of the region. Additionally, as the focus of the study was to determine the within

5 regional distribution of GHG emissions, we do not deal here with aspects or issues external to the regional

6 foreground focus.

8 3.2.5. Sensitivity Analysis -Uniform analysis

9 A comparison is drawn in Figure 6, between the mean emissions per mega joule of the base case (original) and the

10 uniform analysis (all plants having the same capacity). The factor of difference (FD) is also presented, which is the

11 ratio of change between the base and uniform analyses, with regards the change in energy output and change in land

12 area input required. A factor of one means that there is no change in the energy land ratio between the base and

13 uniform case, a factor greater than one means that in the uniform scenario the energy output per input of land

14 supplied has increased (supplementary material). Therefore, what we would expect to see for catchments with a

15 factor higher than one is a decrease in the emission per mega joule for the uniform analysis in comparison to the

16 base case (Figure 6). The associated emission profiles can be seen in the supplementary material (S12).Comparing

17 again across the different uniform catchments in Figure 7, despite all plants having the same rapeseed demands and

18 energy outputs, their emissions are still not equal.

19 (Insert Figure 6)

21 In Figure 7 it can be seen that, while catchment 2 and 6 have similarly low land input, catchment 6 does not perform

22 as well as catchment 2. This can be related to the higher emissions intensities per ha found in the regional location of

23 plant 6, therefore catchment 6 performs worse per mega joule than catchment 2. On the other hand although

24 catchment 4 has relatively low emission intensity per ha, this is counteracted by the greater land area input to meet

25 production capacity demands. This results in it having the greatest emissions per mega joule in comparison to the

26 rest.

27 What Figure 6 and 7 both show, is that while there is some sensitivity in relation to catchments size; it is difficult to

28 distinguish such sensitivity from the intrinsic factor of location. This is important to show, as although the final

29 GHG balances for the configurations are catchment delineated, the results are more highly dependent on location,

30 especially due to the relationship between emission intensity and land input i.e. rapeseed yields. (The trend also

31 applies to non-allocated values)..

32 (Insert Figure 7)

1 3.4 Scenario Analysis

2 The results from the scenario analysis investigating the influence of lower nitrogen fertiliser application and hence

3 lower yields, indicated that assessing either biomass or conversion plants alone is insufficient for an effective GHG

4 mitigation strategy. Figure 8 shows that reducing the emission intensity associated with rapeseed cultivation may in

5 fact result in increasing the overall emissions per mega joule biodiesel. The result of lowering yields, led to a greater

6 land area input to meet the required rapeseed demand of the conversion plants on a per mega joule biodiesel basis,

7 except for catchment one (too small to observe change) and catchment 8 (imported biomass). In general there is an

8 increase in emissions from the base case, ranging from 1.14-1.91 g CO2eq. /MJ for S10 to an increase of 3.64 to 5.71

9 g CO2eq. / MJ for S25. The regional trends also change slightly between the scenarios, with S10 still indicating a

10 tendency for the larger scaled plants to have the greater emissions per mega joule. However, in the S25 the

11 performance of the different biodiesel configurations shows no clear spatial trend.

12 In comparison to the base case, when comparing the scenarios on a per hectare basis, there is a reduction of

13 approximately 128-244 kg CO2eq. /ha for S10 to 308-397kg CO2eq./ ha for S25. The biggest reduction in GHG

14 losses relating naturally to the reduction in N2O emissions from soil and the indirect GHG emissions associated with

15 fertilizer production. However, it must be noted that while, the rapeseed yields are being reduced the yields of the

16 wheat comparator were kept constant (i.e. assumed not to change from base case). Therefore, an increase in soil CO2

17 loss was observed in the model output6, increasing from about 4-10% contribution to cultivation emissions in the

18 base case to 7-14% for S10 and 12-19% for S25. However, this doesn't have a strong influence on the end balances,

19 as the effect of nitrogen emissions in our model output is too large.

20 (Insert Figure 8)

21 These scenarios indicate that while taking one solution, such as reducing nitrogen fertilizer to improve GHG

22 balances may work on a per hectare basis, it does not appear to be so successful on a per mega joule basis (allocated

23 and non-allocated results). Additionally, what is also shown by the model output is that while N2O emissions from

24 soil decreases, consideration must also be given towards the management of the carbon in the system. Providing

25 advice in reducing nitrogen application only, is only advantageous if it is considered within a more comprehensive

26 management strategy. Therefore, the configuration of both biomass and conversion plants needs to be considered

27 and assessed in order to identify potential options for trade-offs between more intensive and extensive cultivation

28 options [2] and that such options need also to be reflected per mega joule biodiesel output.

6 This is because the reduction of rapeseed yields results in a loss of carbon being returned to the field, thus when this is compared with that of the previous wheat crop in the rotation (for which yields have remained at the original high yields determined for the base case), the carbon deficit is increased.

1 3.5 Comparison with RED

2 In general the GHG performance of the regional biodiesel configurations (i.e. biodiesel plant and catchment) for the

3 base case is distinctly better than that of its fossil diesel comparator, having a greater mitigation potential of between

4 51.9-61.94%. Comparing the results with that of the "typical value" outlined in the RED (Figure 4), showed that

5 biodiesel production in the CG region was found to have between 13-31% greater mitigation potential. Therefore,

6 the CG region is producing biodiesel within the GHG stipulated targets set out by the German Government, at least

7 until those at the end of 2017 (41.9 g CO2eq. /MJ).. Thus, highlighting that production location has a strong

8 influence on the magnitude of GHG emissions observed within and across the region per mega joule of biodiesel

9 produced. However, what is interesting is that for the conditions modelled in this study, along with the calculation

10 method implemented (energetic allocation) scale of biodiesel operation plays no distinctly obvious role in mitigation

11 potential. Naturally this picture could change if other means of allocation are introduced; however this is outside the

12 scope of this paper. For this study, rapeseed yield and management factors of biomass are the driving factors. Indeed

13 the variation in biomass yield due to regional and local variability (geo-climatic) is an important driver of the

14 localised biofuel production economics [86]

15 The GHG emissions for rapeseed cultivation within CG for the different biodiesel configurations, was found to

16 range from 23.95-32.71 g Co2eq. /MJ (^ = 27 g CO2 eq. /MJ), with most of the results being relatively higher than

17 the proposed NUTS 2 values for the same federal states at 23.7 g CO2 eq. /MJ (Saxony- Anhalt and Thüringen)

18 23.8g CO2 eq. /MJ (Saxony) [9,10]. Therefore, an additional aspect presented here, is that NUTS2 values might not

19 be sufficient enough to reflect effectively the spatial variability in the GHG emissions associated with a producer's

20 rapeseed production or indeed perhaps the temporal variability (the base year was wet, but with high yields).

21 Therefore, as previously identified, attempts at over simplifying GHG estimates for ease of implementation, may

22 provide incorrect results for specific locations [5].

23 (Insert Figure 9)

25 3.5.1 Issues with introducing greater spatial details -Outlook

26 The issue with introducing greater spatial detail within the RED calculations for the development of more within

27 regional based mitigation strategies may not be entirely attractive for farmers or indeed biofuel producers. Aside

28 from the increase in bureaucracy and work load, with more detailed calculations, farmers may find that their

29 rapeseed crop has been produced with a higher GHG value than stipulated by the RED. With biofuel producers

30 seeking to have the best mitigation values for their biofuel, this may result in some rapeseed farmers facing future

31 disadvantages in the biofuel's market. The farmer in turn may then sell his rapeseed to another market e.g. food,

32 feed fodder which do not need to conform to the strict cultivation emissions limits set by the RED (i.e. leakage

33 within the agricultural sector). Therefore, in addition to the leakage effects related to the iLUC debate, this begs the

34 question why not include all cultivated crops or indeed agricultural production systems in such GHG accounting

1 systems? It is a well discussed issue that agriculture is like the "sleeping beauty of EU Climate Policy" and to date

2 has largely been ignored [11]. Indeed many scientific authors have outlined the need for a greater cohesion between

3 the various EU polices and for more holistic GHG mitigation strategies. With several proposing different options for

4 integrating and expanding aspects of the RED GHG accounting to all cultivation systems, as part of the two pillars

5 of the Common Agricultural Policy (CAP) [87,88,11]. As agricultural policies are crucial for the development and

6 success of biofuels [89], designing agricultural landscapes which include biofuel production should have

7 sustainability criteria for all cultivated biomass, as this is necessary to successfully mitigate GHG emissions

8 effectively by avoiding leakage [84]. Therefore, only when such measures are taken could accounting approaches,

9 such as the one outlined in this paper, be used in a fair way to identify and develop more effective GHG mitigation

10 strategies for all parts of the regional agri-biofuel supply networks.

12 5. Conclusions

13 Accounting for the heterogeneous geographical characteristics found within a region, such as, soil, climate, yields

14 and biomass management, within a life cycle approach can provide much greater insight into the potential

15 underlying regional factors contributing to the GHG balance of biodiesel production and where across the region

16 they might be of concern. This study also showed that such modelling of life cycle thinking "within a regional"

17 context can help to identify:

18 1) the general regional variation and range of potential GHG emissions associated with the annual production of

19 biodiesel within CG, ranging from 31.9-39.83 g CO2eq. /MJ. It also helped to show how much of the GHG balance

20 for biodiesel production is produced directly within the region (40-48%) and how much is generated outside the

21 region (51-60%);

22 2) how spatial patterns and regional trends can influence the emission magnitude and mitigation potential between

23 the different biodiesel configurations due their locations within the region, ranging from 51.9-61.94% mitigation

24 potential in comparison to fossil diesel for CG. These results also indicated that the configuration of both biomass

25 and conversion plants need to be assessed together, in order to develop the most appropriate mitigation strategies, as

26 assessing either one or the other alone is not sufficient to indicate the potential emission burden.

27 3) how regional variability (23.95-32.71 g CO2eq. /MJ) cannot be captured with a simple regional average value

28 (23.7 g CO2 eq. /MJ), or default value (29 g CO2eq. /MJ) like what is being promoted and used in GHG calculation

29 methods, such as RED. However, as current GHG mitigation strategies are limited to the energy sector only;

30 allowing leakages within the agricultural sector, for more spatially targeting GHG accounting to work sustainability

31 certification should also be expanded to other markets for biomass.

33 Acknowledgements

34 This work was made possible by funding from the Helmholtz Association of German Research Centres within the

35 project funding "Biomass and Bioenergy Systems". The authors would like to thank our colleagues, Arne Gróngróft

36 and Marcel Klemm of the DBFZ (German Biomass Research Centre) for their support in the development of

1 conversion plant models and to our colleague Felix Whiting for assembling a crop management databank for CG.

2 We would also like to thank the German weather service (DWD) for the provision of meteorological data.

3 References

4 1. Brück TB, Bohnen FM (2013) The Biodiesel Dilemma: A Techno- and Socioeconomic Review of EU Legislative

5 Actions. JSM Biotechnology and Biomedical Engineering 1 (2):1009

6 2. Milazzo MF, Spina F, Vinci A, Espro C, Bart JCJ (2013) Brassica biodiesels: Past, present and future. Renewable

7 and Sustainable Energy Reviews 18:350-389. doi:

8 3. Naumann K, Oehmichen K, Zeymer M, Meisel K (2014) Monitoring Biokraftstoffsektor. DBFZ-Report Nr. 11, 2.

9 Auflage. Nelles, M. (Ed).DBFZ, Leipzig. .

10 4. UFOP (2015) Union zur Förderung von Oel- und Proteinpflanzen e. V. Biodiesel 2014/2015 Report on the current

11 situation and prospects - extract from the UFOP annual report. Available

12 at: Accessed Jan 2016.

13 5. Boldrin A, Astrup T (2015) GHG sustainability compliance of rapeseed-based biofuels produced in a Danish

14 multi-output biorefinery system. Biomass and Bioenergy 75:83-93.

15 doi:

16 6. Millinger M, Ponitk J, Arendt O, Thrän D (2016) Competitiveness of advanced and conventional biofuels: results

17 from least-cost modelling of biofuel competition in Germany. Submitted

18 7. Directive 2009/28/EC Directive 2009/28/EC of the European Parliament and of the council of 23 April 2009 on

19 the promotion of the use of energy from renewable sources and amending and subsequently repealing Directives

20 2001/77/EC and 2003/30/EC. OJL 140/16.

21 8. UFOP (2015) Germany is going ahead: The introduction of greenhouse gas quotas 2015 - new regulation

22 requirements and consequences for the biofuel industry. Available at:

23 of-legal-greenhouse-gas-reduction-requirements-from-2015-in-germany-possible-consequences-for-the-biodiesel-

24 sector/. Accessed September 2015.

25 9. OVID (2014) Verband der ölsaatenverarbeitenden Industrie in Deutschland e.V. Pursuant to the specification of

26 regional CO2-values when trading in rapeseed for the purpose of producing biodiesel. Available

27 at:

28 Werte_NUTS2.FINAL.pdf. Accessed October 2014.

29 10. UFOP (2013) Erneuerbare-Energien-Richtlinie - RED Biokraftstoffnachhaltigkeitsverordnung

30 Klimaschutzquote - NUTS2-THG-Werte. Available at:

31 Broschre_12082013.pdf. Accessed: Oct 2014.

32 11. Grosjean G, Fuss S, Koch N, Bodirsky BL, De Cara S, Acworth W (2016) Agriculture: Sleeping Beauty of EU

33 Climate Policy? Overcoming Barriers to Implementation. Available at SSRN: or


35 12. Buratti C, Barbanera M, Fantozzi F (2012) A comparison of the European renewable energy directive default

36 emission values with actual values from operating biodiesel facilities for sunflower, rape and soya oil seeds in Italy.

37 Biomass and Bioenergy 47:26-36. doi:

38 13. Hennecke AM, Faist M, Reinhardt J, Junquera V, Neeft J, Fehrenbach H (2013) Biofuel greenhouse gas

39 calculations under the European Renewable Energy Directive - A comparison of the BioGrace tool vs. the tool of

40 the Roundtable on Sustainable Biofuels. Applied Energy 102:55-62.

41 doi:

42 14. Malça J, Freire F (2011) Life-cycle studies of biodiesel in Europe: A review addressing the variability of results

43 and modeling issues. Renewable and Sustainable Energy Reviews 15 (1):338-351.

44 doi:

45 15. O'Keeffe S, Majer S, Bezama A, Thrän D (2016) When considering no man is an island—assessing bioenergy

46 systems in a regional and LCA context: a review. The International Journal of Life Cycle Assessment: 1 -

47 18.10.1007/s11367-11016-11057-11361. doi:10.1007/s11367-016-1057-1

48 16. O' Keeffe S, Wochele-Marx S, Thrän D (in press) RELCA: A REgional Life Cycle inventory for Assessing

49 bioenergy systems within a region. Energy Sustainability and Society

50 17. EC-JRC (2010) General Guide for Life Cycle Assessments -Detailed guidance document for Life Cycle

51 Assessment (LCA). ILCD handbook- International Reference Life Cycle Data System, European Union. Available

52 at:

53 Accessed Jan 2012.

1 18. IPCC (2006) IPCC Guidelines for National Greenhouse Gas Inventories.In: Agriculture, forestry and other land

2 use, vol. 4. Intergovernmental Panel of Climate Change (IPCC). Online at:

3 2006gl/index.html. Accessed:Jan2012.

4 19. German Federal Statistical Office (2011) Land- und Forstwirtschaft, Fischerei Bodenfläche nach Art der

5 tatsächlichen Nutzung.


7 PDF_2030510.pdf?_blob=publicationFile. Accessed 01/2012

8 20. DWD (2010) German Weather service (

9 21. O'Keeffe S, Wochele S, Thrän D Regional Bioenergy Inventory for the Central Germany Region. In:

10 Geldermann J, Schumann M (eds) First International Conference on Resource Efficiency in Interorganizational

11 Networks - ResEff 2013 -: November 13th-14th, 2013 Georg-August-Universität Göttingen, Papers, 2013.

12 Niedersächsische Staats- und Universitätsbibliothek,

13 22. BÜK 1000 (2007) Soil Map of Germany 1:1.000.000 (BÜK 1000) (2007) Source: Federal Institute for

14 Geosciences and Natural Resources.

15 23. Scheffer F, Schachtschabel P (2002) Lehrbuch der Bodenkunde. 15. Auflage, Spektrum Akademischer Verlag

16 GmbH, Heidelberg, 593 S.

17 24. German Federal Statistical Office (2013) Regionaldaten bank Deutschland.


19 25. Statistisches Bundesamt (2010) Landwirtschaftliche Bodennutzung und pflanzliche Erzeugung. Statistisches

20 Bundesamt, Wiesbaden.

21 26. Wochele S, Priess J, Thrän D, O'Keeffe S Crop allocation model "CRAM" - an approach for dealing with

22 biomass supply from arable land as part of a life cycle inventory. In: Hoffmann C, Baxter, D., Maniatis, K., Grassi,

23 A., Helm, P., (ed) EU BC&E Proceedings 2014, 2014. ETA-Florence Renewable Energies, Florence, p. 36 - 40,

24 27. Witing F. (Unpublished) Crop Managment Inventory for Central German Region.

25 28. KTBL (2012) Leistungs-Kostenrechnung Pflanzenbau. Available at:

26 Accessed: 11. 2012.

27 29. Nemecek T., Kägi T., Blaser S. (2007) Life Cycle Inventories of Agricultural Production Systems. Final report

28 ecoinvent v2.0 No.15. Swiss Centre for Life Cycle Inventories, Dübendorf, CH. .

29 30. Kellenberger D, Althaus HJ, Künniger T, Lehmann M, Jungbluth N, Thalmann P (2007) Life Cycle Inventories

30 of Building products. Data v2.0 Ecoinvent report No. 7.

31 31. Roßberg D, Gutsche V, Enzian S, Wick M (2002) NEPTUN 2000 - Erhebung von Daten zum tatsächlichen

32 Einsatz chemischer Pflanzenschutzmittel im Ackerbau Deutschlands. Berichte aus der BBA, H. 98, 27 S.


34 Journal für Kulturpflanzen, 65 (4):141-151. doi:10.5073/JFK.2013.04.02

35 33. BVL (2013) Bundesamt für Verbraucherschutz und Lebesmittlesicherheit online database on plant protection

36 products. Available at: Accessed 1.1.2014.

37 34. Farack M, Jentsch U, Günther K (2012) Landessortenversuche in Thüringen - Winterraps - Versuchsbericht

38 2011. TLL (Hrsg.). Jena. .

39 35. Sutter J (2010) Life cycle inventories of pesticides.Final report ecoinvent v2.2. Swiss Centre for Life Cycle

40 Inventories, St. Gallen, CH.

41 36. BayWa (2014) Kalkdünger. Available at:

42 Accessed: Feb 2014.

43 37. TLL (2007) Thüringer Ministerium für Landwirtschaft, Forsten, Umwelt und Naturschutz. Landwirtschaft und

44 Landschaftspflege in Thüringen.Düngung in Thüringen 2007 nach "Guter fachlicher Praxis,,. Naumburger Str. 98,

45 07743 Jena

46 38. LLG Landesanstalt für Landwirtschaft, Forsten und Gartenbau, Sachsen-Anhalt. Grundlagen der

47 Düngebedarfsermittlung für eine gute fachliche Praxis beim Düngen.

48 39. BEFU (2013) Düngebedarfsermittlung. Sächsisches Landesamt für Umwelt, Landwirtschaft und Geologie.

49 Available at: Accessed 02.2013.

50 40. Roßberg D., Michel V, Graf R, Neukampf R (2007) Definition von Boden-Klima-Räumen für die

51 Bundesrepublik Deutschland.Nachrichtenbl. Deut. Pflanzenschutzd., 59 (7), S. 155-161. ISSN 0027-7479. © Eugen

52 Ulmer KG, Stuttgart.

53 41. TLL (2010) Thüringer Ministerium für Landwirtschaft, Forsten, Umwelt und Naturschutz.Aktueller Rat zur

54 Nmin- und Smin-Situation Thüringer Böden im Frühjahr 2010.

55 42. LLFG (2010) Landesanstalt für Landwirtschaft, Forsten und Gartenbau, Sachsen-Anhalt. Nmin-Vergleichswerte

56 Frühjahr 2010. Available at:http://www.llfg.sachsen-

1 Accessed

2 02.2013.

3 43. LLFG (2010) Landesanstalt für Landwirtschaft, Forsten und Gartenbau, Sachsen-Anhalt. Nmin-Vergleichswerte

4 Frühjahr 2010 - 2. In formation der LLFG Sachsen-Anhalt

5 44. Brocks S, Jungkunst HF, Bareth G. (2014) A regionally disaggregated inventory of nitrous oxide emissions from

6 agricultural soils in Germany-A GIS based empirical approach. ERDKUNDE: 125-144

7 45. VDLUFA (2000) Verband Deutscher Landwirtschaftlicher Untersuchungs- und

8 Forschungsanstalten.Bestimmung des Kalkbedarfs von Acker- und Grünlandböden.

9 46. vTI (2009) Calculations of emission from German agriculture -National Emission Inventory Report (NIR) 2009

10 for 2007. Agriculture and Forestry Research

11 47. IPCC (2006) IPCC Guidelines for National Greenhouse Gas Inventories, prepared by the National Greenhouse

12 Gas Inventories Programme, edited by: Eggleston, H.S, Buendia, L., Miwa, K., Ngara, T. and Tanabe, K., Volume

13 4, Chapter 11, N2O emissions from managed soils, and CO2 emissions from lime and urea application, IGES,

14 Hayama, Japan, 2006.

15 48. Stehfest E, Bouwman L (2006) N2O and NO emission from agricultural fields and soils under natural

16 vegetation: summarizing available measurement data and modeling of global annual emissions. Nutr Cycl

17 Agroecosyst 74 (3):207-228. doi: 10.1007/s10705-006-9000-7

18 49. Petersen BM, Knudsen MT, Hermansen JE, Halberg N (2013) An approach to include soil carbon changes in life

19 cycle assessments. Journal of Cleaner Production 52 (0):217-224.

20 doi:

21 50. Knudsen MT, Meyer-Aurich A, Olesen JE, Chirinda N, Hermansen JE (2014) Carbon footprints of crops from

22 organic and conventional arable crop rotations - using a life cycle assessment approach. Journal of Cleaner

23 Production 64 (0):609-618. doi:

24 51. Mogensen L, Kristensen T, Nguyen TLT, Knudsen MT, Hermansen JE (2014) Method for calculating carbon

25 footprint of cattle feeds - including contribution from soil carbon changes and use of cattle manure. Journal of

26 Cleaner Production 73 (0):40-51. doi:

27 52. Franko U, Kolbe H, Thiel E, Ließ E (2011) Multi-site validation of a soil organic matter model for arable fields

28 based on generally available input data. Geoderma 166 (1): 119-134.

29 doi:

30 53. Franko U, Oelschlägel B, Schenk S (1995) Simulation of temperature-, water- and nitrogen dynamics using the

31 model CANDY. Ecological Modelling 81 (1-3):213-222. doi:10.1016/0304-3800(94)00172-e

32 54. VDB (2010) Verband der Deutschen Biokraftstoffindustrie e.V. (VDB). Factsheet_biodiesel.Available


34 148.html?file=tl_files/download/Daten_und_Fakten/factsheet_biodiesel.pdf. Accessed March 2014.

35 55. Braune M, Grasemann E, Gröngröft A, Klemm M, Oehmichen K, Zech K (2015) Die Biokraftstoffproduktion in

36 Deutschland - Stand der Technik und Optimierungsansätze (DBFZ-Report Nr. 22, in press). DBFZ, Leipzig. ISSN

37 2197-4632.

38 56. UFOP (2010) Union zur Förderung von Oel- und Proteinpflanzen e. V. Biodiesel 2009/2010 Report on the

39 Current Situation and Prospects - Abstract from the UFOP Annual Report. Available at:

40 Accessed: Feburary 2012.

41 57. UFOP (2013) Union zur Förderung von Oel- und Proteinpflanzen e. V. Biodiesel 2012/2013 Report on the

42 Current Situation and Prospects - Abstract from the UFOP Annual Report. Available at:

43 Accessed: Feburary 2012.

44 58. Thrän D., Pfeiffer D. ( 2015) (Eds.) Method Handbook - Material flow-oriented assessment of greenhouse gas

45 effects. In: Series of the funding programme „Biomass energy use", Vol. 04, Leipzig - ISSN online - 2364-897X.

46 59. Dones R., Bauer C., Bolliger R., Burger B., Faist Emmenegger M., Frischknecht R., Heck T., Jungbluth N.,

47 Röder A., M. T (2007) Life Cycle Inventories of Energy Systems: Results for Current Systems in Switzerland and

48 other UCTE Countries. ecoinvent report No. 5. Paul Scherrer Institut Villigen, Swiss Centre for Life Cycle

49 Inventories, Dübendorf, CH. .

50 60. Neeft J, te Buck S, Gerlagh T, Gagnepain B, Bacovsky D, Ludwiczek N, Lavelle P, Thonier G, Lechön Y, Lago

51 C, Herrera I., Georgakopoulos K, Komioti N, Fehrenbach H, Hennecke A, Parikka M, Kinning L., Wollin P. (2012)

52 BioGrace Publishable final report for Grant Agreement IEE/09/736 (

53 61. Ponitka P, Arendt O, Lenz V, Daniel-Gromke J, Stinner W, Ortwein A, Zeymer M, Gröngröft A, Müller-Langer

54 F, Klemm M, Braun J, Thrän D, O'Keeffe S, Millinger M (2015) Konversionspfade - zur energetischen

55 Biomassenutzung im 21. Jahrhundert. In: Thrän, D.; Ponitka, J.; Arendt, O. (Hrsg.): Focus on: Bioenergie-

56 Technologien. Leipzig 2015 - ISSN 2192-1156 (in press) ISSN: 2192-1156.

1 62. Haas R, Remmele E (2011) Status quo der dezebtralen Ölgewinnung-bundesweite Befragung. Im Auftrag der

2 Union zu Föderung von Oel-und Proteinpflanzen (UFOP). Berichte aus dem TFZ 26. Available:

3 . Accessed:

4 April 2013.

5 63. LfULG (2009) Qualitätssicherung dezentraler Ölmühlen. Schrichtenreiche des Landesamtes für Umwelt,

6 Landwirtschaft und Geologie. Heft 33/2009. Available at:

7 Accessed: Feb 2013.

8 64. FNR (2013) Ölmühlen. Daten bank Karten. Available at: Accessed:

9 Feb 2013.

10 65. BLE (2012) Bundesanstalt für Landwirtschft und Ernährung. Evaluations- und Erfahrungsbericht für das Jahr

11 2011.Biomassestrom-Nachhaltigkeitsverordnung Biokraftstoff-Nachhaltigkeitsverordnung. Available at:


13 11.pdf?_blob=publicationFile. Accessed: Aug 2014.

14 66. Spielmann M, Bauer C, Dones R, Tuchschmid M (2007) Transport services: Ecoinvent report no. 14. Swiss

15 Centre for Life Cycle Inventories, Dübendorf. Available at:

16 Accessed: Jan 2013.

17 67. Tobler WR (1970) A Computer Movie Simulating Urban Growth in the Detroit Region. Economic Geography

18 46 (ArticleType: research-article / Issue Title: Supplement: Proceedings. International Geographical Union.

19 Commission on Quantitative Methods / Full publication date: Jun., 1970 / Copyright © 1970 Clark University):234-

20 240. doi: 10.2307/143141

21 68. Bundesamt S (2011) Düngemittelversorgung Wirtschaftsjahr 2010/2011, Wiesbaden. Available at:


23 versorgungJ2040820117004.pdf?__blob=publicationFile. Accessed 05.2012.

24 69. Biograce (2013) List of input data provided by the JEC consortium (biofuel pathway RED method 14 Nov

25 2008.xls). Available at: Accessed: April 2013.

26 70. Frischknecht R, Jungbluth N, Althaus H-J, Doka G, Dones R, Heck T, Hellweg S, Hischier R, Nemecek T,

27 Rebitzer G, Spielmann M (2005) The ecoinvent Database: Overview and Methodological Framework (7 pp). The

28 International Journal of Life Cycle Assessment 10 (1):3-9. doi:10.1065/lca2004.10.181.1

29 71. Pahlmann I, Böttcher U, Sieling K, Kage H (2013) Possible impact of the Renewable Energy Directive on N

30 fertilization intensity and yield of winter oilseed rape in different cropping systems. Biomass and Bioenergy 57:16831 179. doi:

32 72. Muller-Langer F, Majer S, O'Keeffe S (2014) Benchmarking biofuels--a comparison of technical, economic and

33 environmental indicators. Energy, Sustainability and Society 4 (1):20

34 73. Butterbach-Bahl K, Kesik M, Miehle P, Papen H, Li C (2004) Quantifying the regional source strength of N-

35 trace gases across agricultural and forest ecosystems with process based models. Plant Soil 260 (1-2):311-329.

36 doi: 10.1023/B:PLSO.0000030186.81212.fb

37 74. Del Grosso SJ, Mosier AR, Parton WJ, Ojima DS (2005) DAYCENT model analysis of past and contemporary

38 soil N2O and net greenhouse gas flux for major crops in the USA. Soil and Tillage Research 83 (1):9-24.

39 doi:

40 75. Flessa H, Dörsch P, Beese F (1995) Seasonal variation of N2O and CH4 fluxes in differently managed arable

41 soils in southern Germany. Journal of Geophysical Research: Atmospheres 100 (D11):23115-23124.

42 doi: 10.1029/95jd02270

43 76. Leip A, Busto M, Winiwarter W (2011) Developing spatially stratified N2O emission factors for Europe.

44 Environmental Pollution 159 (11):3223-3232. doi:

45 77. Henseler M, Dechow R (2014) Simulation of regional nitrous oxide emissions from German agricultural mineral

46 soils: A linkage between an agro-economic model and an empirical emission model. Agricultural Systems 124

47 (0):70-82. doi:

48 78. Dufosse K, Gabrielle B, Drouet JL, Bessou C (2013) Using Agroecosystem Modeling to Improve the Estimates

49 of N2O Emissions in the Life-Cycle Assessment of Biofuels. Waste and Biomass Valorization 4 (3):593-606.

50 doi: 10.1007/s12649-012-9171-1

51 79. Reijnders L, Huijbregts MAJ (2008) Biogenic greenhouse gas emissions linked to the life cycles of biodiesel

52 derived from European rapeseed and Brazilian soybeans. Journal of Cleaner Production 16 (18): 1943-1948.

53 doi:

54 80. Cherubini F, Peters GP, Berntsen T, Str0Mman AH, Hertwich E (2011) CO2 emissions from biomass

55 combustion for bioenergy: atmospheric decay and contribution to global warming. GCB Bioenergy 3 (5):413-426.

56 doi: 10.1111/j. 1757-1707.2011.01102.x

1 81. Brandao M, Mila i Canals L, Clift R (2011) Soil organic carbon changes in the cultivation of energy crops:

2 Implications for GHG balances and soil quality for use in LCA. Biomass and Bioenergy 35 (6):2323-2336.

3 doi:

4 82. Mila i Canals L, Brandao M (2007) Chapter 7.2: life cycle impact assessment (LCIA) methodology for land use

5 impacts. In: Mila' i Canals L, editor. LCA methodology and modelling considerations for vegetable production and

6 consumption. CES Working Papers 02/07, Guildford.

7 83. Cherubini F (2010) GHG balances of bioenergy systems - Overview of key steps in the production chain and

8 methodological concerns. Renewable Energy 35 (7): 1565-1573. doi:10.1016/j.renene.2009.11.035

9 84. Franzluebbers AJ (2015) Farming strategies to fuel bioenergy demands and facilitate essential soil services.

10 Geoderma 259-260:251-258. doi:

11 85. Nemecek T, Erzinger S (2005) Modelling Representative Life Cycle Inventories for Swiss Arable Crops (9 pp).

12 The International Journal of Life Cycle Assessment 10 (1):68-76. doi:10.1065/lca2004.09.181.8

13 86. IEA (2009) Bioenergy e the impact of indirect land use change. In: IEA Bioenergy ExCo63 Workshop.

14 International Energy Agency (IEA Bioenergy), Rotterdam.

15 87. Zafeiriou E, Karelakis C (2016) Income volatility of energy crops: the case of rapeseed. Journal of Cleaner

16 Production 122:113-120. doi:

17 88. Bartolini F, Angelini L, Brunori G, Gava O (2015) Impacts of the CAP 2014-2020 on the Agroenergy Sector in

18 Tuscany, Italy. Energies 8 (2): 1058

19 89. Banse M, van Meijl H, Tabeau A, Woltjer G (2008) Will EU biofuel policies affect global agricultural markets?

20 European Review of Agricultural Economics 35 (2): 117-141. doi: 10.1093/erae/jbn023

23 Figure Captions

24 Figure 1. Outline of RELCA modelling steps adapted from [16]

25 Figure 2. Distribution of Rapeseed yields (t FM/ha) in CG region -output of CRAM modelling [26]

26 Figure 3. Contribution of each step (cultivation, conversion, transport-to biodiesel plant only) for the different biodiesel

27 configurations in CG (values along x axis refer to catchments,1-5 are small scale, 6-7 medium scale, 8-10 large scale)

28 Figure 4. Box plot illustrating the GHG emission profiles (totalled direct and indirect) for each biodiesel catchment

29 in CG. Emission profiles with same letter no significant difference different letters denote significant difference and

30 *** indicates significantly different from every other profile (P<0.001 at 5% level)). Box plot produced using

31 MATLAB 2012b (The Math Works, Inc., Natick, Massachusetts, United States). Light grey colour

32 indicates small scale, dark grey- medium scale, charcoal -large scale. For catchment 8 this refers only to

33 the regional rapeseed supplied to the biodiesel plant.

34 Figure 5. Results of Hotspot analysis for rapeseed cultivation assumed to be supplied to biodiesel plants within the

35 CG region. Map produced using Arc GIS® software by Esri. Due to data sensitivity plant locations not provided.

36 Blue areas denote statistically lower total GHG emissions and Red denotes statistically total higher GHG emissions.

37 Figure 6. Comparison of emissions per mega joule biodiesel for base case and uniform analysis, FD refers to factor

38 difference between base and uniform analysis for land used and energy output ratio and is along the secondary y

39 axis, FD=1 means no difference . Values along x axis refer to catchments.

40 Figure 7. Emissions per mega joule compared with hectares of rapeseed supplied for the different catchments, as

41 well as the mean emission intensity per hectare supplied (values along x axis refer to catchments).

1 Figure 8. Mean GHG emission intensity for the different catchment; a) kg CO2eq. per 100 ha; b) g CO2eq. per mega

2 joule biodiesel, for the base case and two scenarios S10 (10% yield reduction) and S25 (25% yield reduction)

3 (values along x axis refer to catchments).

4 Figure 9. The mitigation potential (%) for the different biodiesel catchments modelled for CG when compared to

5 fossil diesel and the emissions targets of the German government for 2013 (35% lower than fossil compactor 54.5g

6 Co2eq. /MJ), 2017 (50% lower, 41.9 g Co2eq. /MJ) (values along x axis refer to catchments).

Table 1. Crop management practices assumed for the CG region (All units are kg/ ha' a, unless otherwise stated). All management flows were allocated to each rapeseed grid cell._

Regional flows1 Quantities Source Non-regional product flows1 Source Ecoinvent modules2 Source


P 26-29 [27] See Table A3 - - -

K 29 [28] See Table A3 - - -

CaO 1000 [28] Lime [28] "CH: limestone, milled, packed, at plant" [29,30]

Crop protection4

Tebuconazol 0.2874 [31-33] Folicur [34] "Pesticide Unspecified". [35]

Metazachlor/Clomazone 0.75 Nimbus CS "Cyclic N compound"

Thiacloprid 0.072 Biscaya "Diazole compounds"

1. Regional flows and Non -regional product flows as defined by [15].

2. Ecoinvent modules -used to estimate the upstream GHG emissions.

3. Fertilisers - nutrient applied P= phosphorus provided by P2O5 in fertiliser; K= potassium provided by K2O in fertiliser; CaO = assuming it

takes 1.785 kg of CaCO3 to neutralise the same area as 1kg of CaO [36] .

4. Data on crop protection products and recommended dosages was gathered for the region, from various sources (including an unpublished survey-DBFZ). Once a final list of plant protection products were identified, the active ingredients of the crop protection products were determined [33]. The active ingredients, associated with the used fungicide, herbicide and pesticide products, were then cross checked with the national survey data of [32].

Table 2. The total weight of diesel required for field operations values taken from [28] (Conversion to mass kg/ha assuming a diesel density 0.832 kg/l)

Soil types1 Fertilising1 Ploughing Harrowing Sowing Crop protect2 Harvesting Liming Plough back Total Diesel

Light 1.66 12.33 3.00 3.59 2.37 19.75 0.51 7.51 50.72

Medium 1.66 18.51 4.19 3.88 2.37 19.75 0.51 9.69 60.57

Heavy 1.66 34.49 8.01 4.18 2.37 19.75 0.51 13.84 84.83

1. Refers to clay content; Light (clay< 15%), medium (clay 15-25%), heavy (clay >25 %).

2. Fertilising refers to total diesel demand for all fertiliser operations (N,P,K).

3. Crop protection-summed from application of herbicides, insecticides and fungicides.

4. Field size was assumed to be approx. 20 ha and average vicinity of farm to field was set to 2km.

Table 3. Emissions factor EF1Brock [44].A map with the distribution of Brock EF is provided in the supplementary material, section A3

Geo-climate categories Values 1

Redoximorphic soils2 1.02

Well-aerated & Cold3 4.29

Well-aerated & Warm-Dry4 1.21

Well aerated Warm-Wet5 1.64

1. Emission factors are for (%) of chemical nitrogen fertiliser applied.

2. Redoximporhic soils found in [22] ( associated soil numbers; 7,8,10,11,12,9,22,23, 24,28,43, 47, 48).

3. Areas which have > 100 days of frost per year.

4. Areas which have <100 days of frost and < 600 mm of precipitation.

5. Areas which have <100 days of frost and > 600 mm of precipitation

Table 4. Parameter assumptions for estimating N coming from crop residues (ENresidues).

Values Source

Crop DM1 0.91 g /Kg [37]

RAG2 1.7 [37]

NAG3 0.7 g/Kg [46]

NBG4 0.7 g/Kg [46]

RG5 0.59 [46]

1. DM= Dry matter

2. Ratio of above ground biomass - between seeds and stem (residues)

3. Nitrogen content of above ground biomass

4. Nitrogen of below ground residues

5. RG ration of below ground residues to harvested seeds FracRenew assumed to be 1 (arable land re-cropped annually). FracRenew was zero, as it was no residues were removed EFUpcc=1%_

Table 5. Mineral fertilisers NH3 emissions factors as a function of spring temperature (°C) taken from [46] used to estimate ENNH3Fert for each grid cell

Fertiliser type EF

Calcium ammonium nitrate 0.0008+0.0001.ts1

Anhydrous ammonium 2 0.0127+0.0012.ts

Urea 0.1067+0.0035.ts

Ammonium sulphate3 0.0107+0.0006.ts

Ammonium nitrate3 0.0080+0.0001.ts

1. Spring temperatures (ts) for the months March, April, May, which were found to be in the range of 4.6-8.5 °C

2. Assumed to be similar to Urea ammonium nitrate

3. The statistics referred to an N mixture which was assumed to be 50:50 Ammonium sulphate: Ammonium nitrate (supplementary material, A3).

Table 6. Carbon modelling parameters and assumptions from CCB model [52,53]

Wheat Rapeseed

CropNTot 1 10 6.6

Ncoeffl2 0.07 0.0984

Nfact3 2.7 4.47

RAG4 0.8 1.6

StrawDM5 0.86 0.86

C 6 straw 0.44 0.44

SOMrep_bp7 0.5 0.46

CNratio 50 30

SOMrep_rs9 0.55 0.4

1. CropNTot= N amount independent from yield (i. e. N amount in crop).

2. Ncoeff1 = linear coefficient describing the relation between Nresidues and Ncrop.

3. Nfact= factor relating the N amount in main product +by-product to the natural yield (main product); is used to calculate the total N in main and by-product in order to get the nitrogen amount in crop and root residues

4. RAG=Ratio between grain yield and amount of by-product (straw)

5. StrawDM = DM of by-product

6. Cstraw= C concentration in DM

7. SOMrep by = synthesis coefficient describing the efficiency for the replacement of SOM C from byproduct C

8. CNratio = C to N ratio in crop and root residues

9. SOMrep rs = synthesis coefficient describing the efficiency for the replacement of SOM C from the C of crop and root residues

10. Other acronyms from equations 5-11; StrawDM=straw dry matter; CropFM=crop fresh matter; Crep str=carbon supplied to soil by straw; CFOM= organic matter; Crep root=carbon replaced by root;

Table 7. Activity parameters for conversion plants in the regional foreground. All flows unless otherwise stated are tonnes per tonne biodiesel produced, values have been rounded up to the nearest decimal place._



Oil mill Biodiesel Plant

Oil mills


Electricity input (GJ/t)2 Thermal energy input (GJ/ t)3 Sodium hydroxide Phosphoric acid Hexane

Presscake (output)4 Biodiesel plants

Rapseed oil

Electricity input (GJ/t)2 Thermal energy input (GJ/t)3 Methanol

Potassium hydroxide Sodium hydroxide Sulphuric acid Hydrochloric acid Sodium methyloxide Biodiesel5 Glycerol4'5 FFA6 Fertiliser 6

Cold press Batch

2.97 0.994

1.01 0.088 1.030 0.13 0.01

0.01 0.01

0.13 0.02 0.02

Hot press Continuous

0.12 0.01 0.01

Hexane extraction Continuous


0.072 0.619 0.10

0.01 0.016 1

0.093 0.02

1. Rapeseeds were delivered to the oil mill, assumed to have approx. 9% moisture content and 42% oil content

2. Electricity mix for Germany 2010 was taken from [58]

3. Thermal energy refers to German natural gas mix, taken from [59]

4. Output in italics refers to by-products. Energetic allocation was carried out using the following lower heating values (LHV): Press cake was taken to be 18.7 MJ/kg, crude press oil 36MJ/kg and Biodiesel 37.2 MJ/kg Biodiesel [60,7].

5. For the small and medium plants, due to investment costs of upgrading it was assumed that they processed the glycerol by product to approx. 80% glycerol, however for the larger scale the glycerol produced was pharma glycerol 99% purity.

6. These by-products were not considered in the allocation. FFA = Free fatty acids can be sold to chemical companies for further processing. Fertiliser-derived from the potassium auxiliaries - has the possibility of being used locally.

Table 8. Comparison across the different biodiesel catchments

Catchment scale1 GHG Cultivation2 GHG Cultivation3 Emission Intensity4 Energetic Output 15 Energetic Output 26 LAi7 Yields8

m 23.9 44.84 2660 59.2 22.2 0.0169 3978

s 24.8 52.62 2788 52.5 18.8 0.0191 4226

l 25.5 44.33 2870 61.2 21.3 0.0164 3996

l 25.7 43.97 2985 63.2 21.2 X 0.0158 4126

m 26.8 50.29 2760 54.9 19.9 0.0182 3690

s 27.4 66.17 2853 43.0 15.1 0.0233 3440

s 27.6 58.55 2740 46.8 17.1 0.0214 3736

s 29.4 62.34 2773 44.2 15.9 0.0226 3550

s 32.3 68.43 2720 39.7 14.6 0.0252 3172

CG av. 27.1 54.62 2794 51.6 18.5 0.0199 3768

1. s=small scale; m=medium scale; l=large scale; CG av.=average for Central Germany.

2. Total GHG emissions from cultivation (CO2 eq. / MJ) -energetically allocated.

3. For comparison total GHG emissions from cultivation (CO2 eq. / MJ) -non-allocated results.

4. Emission Intensity of catchment (Kg CO2eq. / ha supplying)

5. Gross energy output per ha supplied (upstream energetic input have not been taken into account here) (GJ /ha)

6. Gross energetic output per associated GHG emission of cultivation (MJ/ kg CO2 eq.)

7. The amount of land area input (LAi) required per GJ of energy produced by the associated biodiesel plant in the configuration (ha /GJ)

8.Yields tFM/ ha (91% DM)

For ease of understanding' only 9 plants are shown ' the plant which processes mostly imported rapeseed was excluded Emission Intensity/energetic output 1 and 2 as well as LAi have not been calculated with energetically allocated results

Table 9. Summary statistics for profiled "Hot" and "cold" spots GHG emission analysis within the CG region. Results are mixture of frequency analysis and means, unless stated results are from a frequency analysis._

Base 1

Cold spots 1

Hot spots

Geographical parameters Ackerzahl value Clay (%)

Soil category 1 (% grid cells)2 Soil category 2 (% grid cells)2 Soil category 3 (% grid cells)2 Soil category 4 (% grid cells)2 Soil category 6 (% grid cells)2 Climate parameters (mean)3

Mean average daily temperature (°C)

Total annual percipitation (mm /a)

Spring temperature (°C)

No. Frost days (days)

Agronomic variables

Mean rapeseed yield (t Fm /ha)

Mean wheat yield (t Fm /ha )

Management parameters

Mean N fertiliser applied (kg /ha)

Mean Diesel consumption (kg /ha )

Emission factors 4

Brocks EF low (% grid cells)

Brocks EF medium (% grid cells)

Brocks EF high(% grid cells)

Estimated Emissions(mean) 5

Soil N2O ( kg CO2 eq. /ha )

CO2 loss from soil ( kg CO2 /ha )

Field operations GHGs ( kg CO2 eq. /ha )

GHG fertiliser production ( kg CO2 eq. /ha)

Total Direct GHG emissions (CO2 eq./ ha)

56.91 14.32 20.93 14.58 51.84 11.31 1.34

9.03 921 8.02 87

3.94 7.09

160 58.43

30.67% 69.24% 0.09%

1314 121 182 1003 2852

43.59 13.88 47.74 12.23 20.72 19.20 0.11

943 7.91

3.75 6.51

158 56.93

90.48% 9.52%

1010 102 177 990 2505

66.41 14.61 2.93 12.40 76.71 6.09 1.88

9.24 874 8.18

4.26 7.84

169 59.22

1.35% 98.05% 0.60%

1547 139 185 1048 3156

1. No of grid cells included in analysis: Base, n=8797; Cold spots, n=1839; Hot spots, n=1331.

2. Soil categories from [22] were identified as follows: 1) soils 8-17: Soils in Broad River values;2) soils 18-34: soils in undulating lowlands and hilly; 3) soils 35-48: soils in Loess areas; 4) soils 49-67: Mountain and hill soils from soil rocks their weather products and reposited material; 5) Alpine soils; 6) Anthrosols settlements and surface water.

3. Climate parameters provided by the [20].

4. Emission factors (see Table 3)

5. Emissions provided here were the largest contributors other smaller contributors are not included, hence why variables don't sum to totals.

Step1. Crop allocation modelling

Step2. Biomass inventory modelling

Spatially distributed geographical data

Step3. Conversion plant modelling

Regionally distributed Biomass inventory

Regionally distributed Conversion technology inventory

Step4. Catchment allocation modelling

Catchment delineated inventory

Regional Foreground

Step5. Non- regional


•v" " . V " i / 1 ' -

- i^-f. • • "It •1 ■ ~

• - - ' --v-r^ ;, . ,,

£ —\1 -'r '- .• ..• if ; : V

" thr.^-, ' Wiigk** <V - '-

"1 "if

r vw v. • --1 . r * .. ••


| 2.90 - 3. | 3.60 - 3. | 3.79 - 3. ] 3.93 - 4. I 4.14-4.

yields tFM/ha

: S? --

6 7 8 9 10 RED

■ Cultivation eec SS Processing ep ■ transport

Biodiesel Catchments

3 4 5 6 7 8 □ Base X Uniform ▲ Factor Diff

3 4 5 6 7 8

AC02eq per MJ + Ha supplying(OOO) » C02eq per ha (00)



8 9 10


123456789 ID

□ Base OS10 A S20

70% 60% 50% 40% 30% 20% 10% 0% -10% -20%

+ + + + + + + + + +

A A A A A ▲ ▲ A A A A A


1 1 1 2 1 3 1 1 4 + Fossil i 5 AT 2013 i 6 AT 2017 7 1 8 1 9 1 1 10


• Spatial variation in GHG emissions assessed for regional biodiesel production showed; o Mitigation potential dependent on production location within the region o Regional variability cannot be captured with a simple regional average value o Assessing biomass/conversion plant configurations needed for mitigation strategies