- Email: [email protected]
Accounting for co-products in energy use, greenhouse gas emission savings and land use of biodiesel production from vegetable oils
G Model
ARTICLE IN PRESS
INDCRO-8542; No. of Pages 8
Industrial Crops and Products xxx (2015) xxx–xxx
Contents lists available at ScienceDirect
Industrial Crops and Products journal homepage: www.elsevier.com/locate/indcrop
Accounting for co-products in energy use, greenhouse gas emission savings and land use of biodiesel production from vegetable oils W.J. Corré a,∗ , J.G. Conijn a , K.P.H. Meesters b , H.L. Bos b a b
Plant Research International, Wageningen University and Research Centre, P.O. Box 616, 6700 AP Wageningen, The Netherlands Food and Biobased Research, Wageningen University and Research Centre, P.O. Box 17, 6700 AA Wageningen, The Netherlands
a r t i c l e
i n f o
Article history: Received 30 July 2015 Received in revised form 23 October 2015 Accepted 24 November 2015 Available online xxx Keywords: Biodiesel Oil crops LCA GHG emission Energy use Land use
a b s t r a c t Accounting for co-products of vegetable oil production is essential in reviewing the sustainability of biodiesel production, especially since oil crops produce valuable protein-rich co-products in different quantities and qualities. Two accounting methods, allocation on the basis of energy content and system expansion, are compared. Significant differences in results exist between the methods where system expansion is to be preferred because it can take actual use of co-products into account. Results are very sensitive to the choices made in system expansion. Differences can be large, especially between a system expansion where primarily the use of co-products of the oil crops is taken into account and an expansion that also includes direct oil exchange of the vegetable oil used for biodiesel for the marginal oil in the market. © 2015 Published by Elsevier B.V.
1. Introduction With the purpose of reducing greenhouse gas (GHG) emission and to ensure a more sustainable energy supply, the EC established targets for the production of renewable energy in the EU for the year 2020 of 20% of the total energy use and 10% of the energy use in the transport sector (EC, 2009). Currently, renewable energy use in the transport sector relies largely on biofuels produced from agricultural crops, a production of which the sustainability is heavily debated. The EC Renewable Energy Directive (RED, EC, 2009) gives criteria for the minimum required GHG emission reduction of biofuels in order to be considered as renewable energy and complying with the target of renewable energy production. A realistic calculation of the GHG emission reduction requires that the production of co-products used for other purposes than energy production is taken into account. The RED (EC, 2009) prescribes allocation on the basis of energy content as accounting method for co-products in the reporting of the Member States on the performance of biofuel production regarding greenhouse gas (GHG) emission reduction. Allocation is a simple technique of dividing, e.g., environmental burdening or costs over main product and co-products. This division is performed in a ratio based on the ratios of mass, energy
∗ Corresponding author. E-mail address: [email protected] (W.J. Corré).
content or economic value of the different products. Use of this method does not comply with the principles of Life Cycle Assessment (LCA). LCA is an established technique, described in ISO 14040 and 14044 (ECS, 2006a,b), in which the division of environmental burdening takes the real use of co-products into account as much as possible. For this division, LCA prescribes the use of system expansion (substitution method) whenever possible. In system expansion, the products that can be replaced by the co-products of the process described are included in the system, and thus in the LCA analysis. This is acknowledged in the RED, where article 23–4 reads: “In reporting on greenhouse gas emission saving from the use of biofuels, the Commission shall use the values reported by Member States and shall evaluate whether and how the estimate would change if co-products were accounted for using the substitution approach ”. Nevertheless, allocation, being easier to apply, received much more attention in accounting for co-products than system expansion, possibly stimulated by the idea that both methods produce results that are generally comparable (EC, 2009). However, appreciable differences in GHG emission reduction between the methods have been reported, in favor of both allocation and system expansion (Hoefnagels et al., 2010; Wang et al., 2011) and although system expansion for agricultural production can easily lead to complex system descriptions, it is in principle always possible to use system expansion (Weidema, 2001). In this paper we present both system expansion as well as allocation. Moreover, we include energy use and land use as important sustainability issues in our analysis.
http://dx.doi.org/10.1016/j.indcrop.2015.11.062 0926-6690/© 2015 Published by Elsevier B.V.
Please cite this article in press as: Corré, W.J., et al., Accounting for co-products in energy use, greenhouse gas emission savings and land use of biodiesel production from vegetable oils. Ind. Crops Prod. (2015), http://dx.doi.org/10.1016/j.indcrop.2015.11.062
G Model INDCRO-8542; No. of Pages 8
ARTICLE IN PRESS
2
W.J. Corré et al. / Industrial Crops and Products xxx (2015) xxx–xxx
Energy use is important since it determines the efficiency of the biodiesel produced and land use is an important issue because productive arable land is scarce worldwide and oil crops for biodiesel do compete with crops for food production. Global production of biodiesel from vegetable oil is almost completely based on three oil crops: oil palm, soybean, and rapeseed. In Europe vegetable oil from these three crops makes up over 95% of biodiesel production (Conijn et al., 2009). All three crops have a protein-rich meal as co-product but in very different quantities and qualities. Hence, an increase in biodiesel production from vegetable oil will increase total vegetable oil production, in turn resulting in an increased meal production of which the quantity and quality depends on the oil crop involved. However, in a study focusing on the effects of biodiesel production all other activities are supposed to be left unchanged and the increase in meal production has to be compensated for by a decreased production of other proteinrich feed components. The most logical choice is replacement of a marginal product, i.e., a product that is supposed to function as buffer in the market and the production of which responds strongly to a change in demand. This will involve one of the major oil crops, soybean, since this is an oil crop as well as a protein crop and is in fact currently regarded as the world’s marginal protein source (Schmidt and Weidema, 2008). A decrease in soybean production will result in a decrease in (soy) oil production which will be compensated for by an increased production of another oil crop, most likely oil palm, since palm oil is currently regarded as the marginal vegetable oil (Schmidt and Weidema, 2008). This again results in increased meal production and – in the end – establishment of a new equilibrium between oil crops in which more oil production goes together with the production of an amount of feed ingredients from which a comparable amount and quality of feed can be produced as before. However, the three oil crop meals differ in protein content and digestibility and can therefore not simply substitute each other (Lywood et al., 2009). A qualitative equivalent replacement is only possible when a combination is made with an energy-rich feed ingredient like wheat (Lywood et al., 2009) or barley (Dalgaard et al., 2008). Another difference between the RED prescriptions and the substitution method regarding biodiesel production is found in the accounting for glycerine, a co-product of the hydrolysis of vegetable oil. According to the RED no allocation to glycerine should be made because of its current low value. In system expansion all co-products should be taken into account, and when no current product can be replaced waste disposal is eventually part of its life cycle. A problem in the use of vegetable oil not addressed in system expansion is the exchangeability of the different vegetable oils: system expansion only accounts for the use of co-products. However, since different vegetable oils are equally suitable for major applications (e.g., cooking) an increased demand for one oil could be compensated for by an increased production of another. This means that the use of rapeseed or soy oil for biodiesel production could be expected to cause an increase of the use of palm oil, currently the world’s marginal vegetable oil, for other purposes. The two methods of accounting for co-products for the production of biodiesel from palm oil, soy oil and rapeseed oil for the impact categories GHG emission, energy use and land use will be quantitatively compared in this paper. A variant of system expansion, where the vegetable oil used for biodiesel production can be directly replaced by an increased production of the marginal vegetable oil, will be used as well. In the calculations primarily palm oil is assumed to be the marginal vegetable oil; results of an alternative calculation with rapeseed oil as marginal vegetable oil are also presented.
Table 1 Outputs of crop production systems relevant for biodiesel production in the EU (from: BioGrace (2011) and Corré and Conijn (2015)). Product output in kg ha−1 Yield seeds Fresh fruit bunches Moisture content Yield (dry matter) Oil content in dry matter Crude vegetable oil Biodiesel Glycerine (pure) Meal content in dry matter Meal (dry matter) Moisture content in meal Crude protein content in meal (dry matter) a b c
Oil palm – 19,000c 34% 12,540 34.4% 4314 4008 401 3.75% 470 18% 16.5%
Soybean 2800 – 15% 2380 22% 524 487 49 78% 1860 18.3% 48%
Rapeseed b
3550 – 10% 3195 45.5% 1454 1351 135 54.5% 1740 17% 34%
Wheata 7200b – 13% 6264 – – – – – – – –
Wheat is selected as feed compound to be replaced by oil crop meals. Values refer to varieties of both crops sown in late autumn. Average yield level over the life cycle of oil palms.
2. Materials and methods 2.1. Crop production systems The analysis is based on representative production systems, relevant to biodiesel production in Europe, which reached just over 10 Mt in 2013 (EBB, 2015). Palm oil is produced in South-East Asia, soybean in Brazil, and rapeseed and wheat in Western Europe. Yield levels and associated (co-) products are presented in Table 1. All products are transported to and processed in the Netherlands. Palm oil is produced in perennial plantations. Annual production reaches its maximum five years after planting and a life cycle of approximately 25 years provides optimal total production. Fresh fruit bunches of oil palm are harvested and transported to a palm oil mill where the fruits are separated from the bunches. The empty fruit bunches are composted and recycled to the oil palm fields. The fruits are separated from the kernels and oil is extracted from both fruits and kernels. The remainders of the fruits are separated into a solid (fibers) and a liquid (palm oil mill effluent, POME) fraction. The solid fraction is, together with the kernel shells, used for energy supply to the palm oil mill, which is supposed to follow best practices and to use no external energy (BioGrace, 2011). POME is treated as waste water avoiding methane emission. The meal of the kernels (Palm Kernel Expeller, PKE) is used in animal feed. The crude oil, mostly fruit oil and kernel oil mixed, is transported to a harbor, refined, and shipped. Soybean is an annual crop, in Brazil partly grown in a double cropping system with maize in winter and soybean in summer and partly grown in a single cropping system. Since only a minor proportion of the soybeans in Brazil is grown in a double cropping system (Babcock and Carriquiry, 2010) and information on the effects of other crops in system expansion is lacking, this complication in land use is not included in the analysis; this means that the results in this paper are limited to soybean grown in a single cropping system. After harvest soybeans are transported to a harbor and mostly shipped as whole beans. Oil is extracted and refined in importing countries and the remaining meal is used in animal feed. Rapeseed and wheat are annual crops, in Europe grown in rotation with other arable crops in single cropping systems. After harvest, rapeseed is transported to a processing plant where the oil is extracted and refined; the remaining meal is used in animal feed. After harvest wheat is transported to locations where it is processed into animal feed. Yield levels (Table 1) of rapeseed and wheat were calculated as the average data for production in the period 2007–2009 in Germany and France, the main West European rapeseed- and wheat-producing countries (Eurostat, 2015).
Please cite this article in press as: Corré, W.J., et al., Accounting for co-products in energy use, greenhouse gas emission savings and land use of biodiesel production from vegetable oils. Ind. Crops Prod. (2015), http://dx.doi.org/10.1016/j.indcrop.2015.11.062
G Model INDCRO-8542; No. of Pages 8
ARTICLE IN PRESS W.J. Corré et al. / Industrial Crops and Products xxx (2015) xxx–xxx
3
Biodiesel is produced from crude vegetable oil by transesterification with methanol, yielding fatty acid methyl esters (FAME) and the co-product glycerine. 2.2. Use of co-products from biodiesel production The world’s three dominating biodiesel crops also produce a protein-rich meal; details on quantity and protein content of the meals are presented in Table 1. Due to the large difference in protein content between the meals, each type of meal to be replaced requires combination of a different quantity of energy rich wheat with soy meal in order to reach feed with a comparable protein to energy ratio that can be used as replacement. Furthermore, differences in digestibility result in different quantities of the various combinations to be used in order to reach the same feeding value. Data for the replacement ratios of oil palm meal and rapeseed meal relative to soy meal were calculated by Lywood et al. (2009) for the average use in the EU compound feed market. Due to animalspecific digestibility characteristics, these ratios differ between ruminants, pigs and poultry. The average ratio is the weighted average for use in feed for ruminants, pigs and poultry. As weight factor Lywood et al. (2009) used the total production of compound feed. Recalculation of the data for fresh matter from Lywood et al. (2009), using the moisture contents from Table 1, results in replacement ratios for dry matter of 1 kg of oil palm meal for 0.127 kg of soy meal plus 0.643 kg of wheat, and in 1 kg of rapeseed meal for 0.596 kg of soy meal plus 0.152 kg of wheat. Besides meal, oil palm produces empty fruit bunches, fibers, effluent and kernel shells. These co-products are used in the system as compost or energy source, or discarded as waste. Furthermore, wood (oil palm) and straw (soybean and rapeseed) are other coproducts of oil crops. These co-products are suitable for energy production but currently this application is hardly developed and will be left out of the analysis. Usually, these co-products are left on the land after harvest. Another co-product of biodiesel production from vegetable oil is glycerine. The quantity is equal for all oil types: 0.14 kg crude glycerine (containing appr. 15% moisture and 15% waste) comprising 0.10 kg pure glycerine per kg biodiesel (Table 1). Purified glycerine can replace valuable synthetic glycerine but since the increased biodiesel production has saturated the market for high-value applications, part of the glycerine is currently used in low-value applications like energy production. In our analysis, crude glycerine is assumed to be used for energy production by producing biogas through anaerobic co-digestion with other organic materials, such as animal manure. Advantage of this way of using glycerine is that crude glycerine can be used and no costly purification is necessary. 2.3. Accounting for co-products Allocation according to the RED results in allocation ratios of oil to meal of 95/5 for oil palm, 37/63 for soybean, and 64/36 for rapeseed, based on energy contents of 36 MJ kg−1 for oil and 17 MJ kg−1 for meal. The allocation ratios between biodiesel and glycerine are calculated according to the same principle; the RED, however, does not allow allocation to glycerine and calculations are made with and without allocation to glycerine. Allocation to glycerine involves an allocation ratio of biodiesel to glycerine of 96/4 (energy content of pure glycerine is 16.5 MJ kg−1 ). A precondition of allocation is that, e.g., the energy use of all activities up to and including the separation of the product and co-product has to be allocated over the product and co-product and that the energy use of activities concerning the product after separation has to be allocated completely to the product. In the case of oil production, energy use of agriculture, transport and oil extraction are allocated over oil and meal,
Fig. 1. Interactions of increased vegetable oil demand with crop production.
energy use of transesterification of oil are allocated over biodiesel and (if applicable) glycerine. System expansion according to LCA principles is based on the expected use of marginally produced co-products: use of oil crop meals as feed components, replacing comparable feed components produced from other crops, and use of crude glycerine for energy production through anaerobic digestion, replacing natural gas. Depending on the digestible protein and energy contents of the different meals, different replacement ratios are used (see Section 2.2). In principle, the produced meal replaces the marginal protein compound first: soy meal. In order to compensate for a possibly lower protein-to-energy ratio of the meal, wheat (a commonly used energy compound in feed) can be replaced as well as soy meal. Replaced soy meal causes a decreased production of soy oil, to be compensated for by increased production of the marginal oil: palm oil. This increased oil production in turn causes increased meal production, resulting in a feed-back loop in the calculations as is illustrated in Fig. 1. However, this feed-back is of minor importance, as shown in Table 2, since the oil production of 1 ha soybean can be replaced by 0.12 ha oil palm while 0.12 ha oil palm produces only appr. 60 kg palm kernel expeller whereas 1 ha of soybean produces almost 1900 kg soy meal. Replacements based on production levels (Table 1) and replacement ratios, including feed-back, are presented in Table 2. Biogas produced from glycerine is supposed to replace natural gas in a 1:1 energy ratio without upgrading the gas to natural gas quality. Palm oil is functioning as the global marginal vegetable oil since around the year 2000, before that rapeseed oil was considered as the marginal oil for a long time (Schmidt and Weidema, 2008). A further increase of vegetable oil production could quite well hamper the current fast increase of the oil palm production area and a return to rapeseed oil as the marginal oil seems possible. Against this background another calculation with rapeseed oil as marginal oil has been performed. Parameters for this system expansion are presented in Table 2. Comparison with palm oil as marginal oil shows an increased feed-back due to the higher meal production of rapeseed compared with oil palm, resulting in higher values of total wheat and soy meal replaced. On the assumption that increased oil production always leads to increased production of the marginal oil by direct exchange for this oil, the use of soy oil and rapeseed oil for the production of
Please cite this article in press as: Corré, W.J., et al., Accounting for co-products in energy use, greenhouse gas emission savings and land use of biodiesel production from vegetable oils. Ind. Crops Prod. (2015), http://dx.doi.org/10.1016/j.indcrop.2015.11.062
G Model INDCRO-8542; No. of Pages 8
ARTICLE IN PRESS
4
W.J. Corré et al. / Industrial Crops and Products xxx (2015) xxx–xxx
Table 2 Parameters of the system expansion method (on dry matter basis) with palm oil and rapeseed oil as marginal oils. Oil palm ha kg
Soybean kg ha
Rapeseed kg Ha
Oil produced Meal produced
4314 470
1 1
524 1860
1 1
1454 1740
1 1
Wheat replaced Soy meal replaced Soy oil replaced
302 60 17
0.048 0.032 0.032
0 1860 524
0 1 1
264 1037 292
0.042 0.558 0.558
Palm oil as marginal oil Palm oil produced Palm meal produced
17 2
0.004 0.004
524 57
0.121 0.121
292 32
0.068 0.068
Wheat replaced total Soy meal replaced total Additional palm oil produced
303 60 17
0.048 0.032 0.004
37 1867 526
0.006 1.004 0.122
284 1041 293
0.045 0.600 0.068
Rapeseed oil as marginal oil Rapeseed oil produced Rapeseed meal produced
17 20
0.012 0.012
524 627
0.360 0.360
292 349
0.201 0.201
Wheat replaced total Soy meal replaced total Additional rapeseed oil produced
318 75 21
0.051 0.040 0.015
119 2328 656
0.019 1.251 0.451
330 1297 365
0.053 0.698 0.251
Table 3 Inputs of crop production and transport distances for products. Input in kg ha−1
Oil palma
Soybeana
Rapeseedb
Wheatb
Dieseld Fertiliser N Compost N Fertiliser P2 O5 Fertiliser K2 O Pesticides Seeds N in crop residues
47.8 128 12 144 200 8.4 0 12
48.6 8 0 66 62 2.7 76 37.5
63 152c 0 44 77 2 5 69
65 123c 0 60 105 5 165 84
Transport distance in km Fresh fruit bunches Truck Seeds Truck Ship Train Truck Crude oil Ship
Oil palma 20
Soybeana
Rapeseedb
Wheatb
1100 600 10,000
50 600
50 600
a b c d
150 10,000
From: BioGrace (2011). From: Conijn et al. (2011). From: FAO (2002). Specific density of diesel is 0.83 kg−l .
biodiesel will in the current market lead to an increased production of palm oil to be used for other purposes. In this situation, only the calculations for palm oil remain relevant when system expansion is applied. Comparably, when rapeseed oil is assumed to be the marginal oil, direct exchange of oils would mean that increased use of soy oil or palm oil would lead to an increased production of rapeseed oil and the only relevant calculation in system expansion would be the calculation for rapeseed oil with rapeseed oil as the marginal oil.
Table 4 Energy use and GHG emission of inputs in biodiesel production from vegetable oil (from Corré and Conijn (2015) and Omni Tech (2010) (transesterification)).
Cultivation
Energy use MJ kg−1
GHG emission kg CO2 -eq kg−1
Diesel Fertiliser N (World/Europe) Compost N Fertiliser P2 O5 (World/Europe) Fertiliser K2 O Pesticides Seeds -Rapeseed -Soybean -Wheat N in crop residues
62.3a 49.2/40.3 0 4.7/3.4 5.6/4.6 263 6.2 12.6 2.3 0
4.68a 9.30b /11.45b 6.63b 0.32/0.22 0.38/0.31 18.4 0.61 1.26 0.25 5.70b
Processing Oil extraction Palm Soybean Rapeseed Transesterification
GJ t−1 biodiesel
kg CO2 -eq t−1 biodiesel
0c 6.29 3.16 5.13
0c 440 220 256
Transport Truck Train Ship, ocean
MJ t−1 km−1 1.08 0.76 0.12
kg CO2 -eq t−1 km−1 0.080 0.048 0.0095
a Including a 27% surcharge for lubricants and manufacturing and maintenance of machinery. b Including emissions from production and direct and indirect field emissions of N2 O after IPCC (2006). c Energy use in palm oil extraction is based on renewable energy from crop residues.
Table 5 Energy use for biodiesel production from agricultural crops. Oil palm
Soybean
Rapeseed
Wheat
Energy use in GJ ha Cultivation Transport of fruits/seeds Oil extraction Transport of oil Oil trans-esterification
13.31 0.41 0 5.67 20.53
5.77 7.96 3.06 – 2.50
11.15 1.81 4.27 – 6.92
11.40 3.67
15.07
−1
Total
39.92
19.29
24.15
Accounting for meal (GJ ha−1 ) Allocation Total after allocation Credits in a system expansion Total after system expansion
0.67 39.25 1.18 38.74
11.13 8.16 14.58 4.71
6.63 17.52 9.43 14.72
Energy use in GJ t−1 biodiesel Total After allocation After system expansion
9.96 9.79 9.67
39.6 16.8 9.67
17.9 13.0 10.9
emissions are based on scenario studies where system expansion is applied (e.g., Laborde, 2011), these data are not suitable for comparison of methods. The importance of LUC-related GHG emissions will be further addressed in Section 4. 3. Results
2.4. Other parameters Crop production inputs and transport distances are presented in Table 3 and energy use and GHG emission from crop production inputs, processing and transport are presented in Table 4. Although land use chance (LUC), direct and indirect, is an inevitable result of increased vegetable oil production, this is not included in our calculations of GHG emission. The RED acknowledges the importance of LUC effects on GHG emission and the need to include these emissions in reporting the GHG reduction of the production and use of biofuels but since data of LUC-related GHG
Results of the calculations are presented in Tables 5 (energy use), 6 (GHG emission) and 7 (land use) for the three oil crops and for wheat, insofar as necessary for the system expansion. The end results, expressed per tonne of biodiesel produced, for the three oil crops are also shown in Fig. 2, including the results from the calculation with rapeseed as marginal oil crop instead of oil palm. Glycerine has not been taken into account in the calculations presented in Tables 5–7 as stipulated in the RED. For comparison, the effects of including glycerine in the analysis are presented separately in Section 3.5.
Please cite this article in press as: Corré, W.J., et al., Accounting for co-products in energy use, greenhouse gas emission savings and land use of biodiesel production from vegetable oils. Ind. Crops Prod. (2015), http://dx.doi.org/10.1016/j.indcrop.2015.11.062
G Model INDCRO-8542; No. of Pages 8
ARTICLE IN PRESS W.J. Corré et al. / Industrial Crops and Products xxx (2015) xxx–xxx
5
Fig. 2. Energy use, GHG emission, and land use in biodiesel production from vegetable oils. With different methods of accounting for co-products; allocation: allocation on the basis of energy content; syst. exp. P = M: system expansion with palm oil as marginal oil; syst. exp. R = M: system expansion with rapeseed oil as marginal oil; oil exchange: increased oil demand is filled by increased palm oil production.
3.1. Energy use Large differences in energy use per hectare exist between the crops (Table 5). Energy use in cultivation is relatively low for soybean, due to the very low use of N fertilizer. Total transport energy use is high in oil palm and soybean due to the long transport distances from the crop production sites to Europe. Energy use in oil extraction is zero in oil palm due to the use of crop residues for energy. Differences in energy use in refining and transesterification reflect the differences in oil yield. Large differences between crops also appear when total energy use is expressed per tonne biodiesel (Table 5), with a factor of almost 4 between the highest (soybean) and the lowest (oil palm)
value. However, after accounting for the meal these differences decrease considerably: after allocation the factor declines to less than 2 and after system expansion differences are almost absent between the highest and lowest values. It illustrates the effects of taking meal into account and shows that the system expansion approach in our study gives higher credits to meal in comparison with the allocation method based on energy contents. 3.2. GHG emission GHG emission results (Table 6) strongly reflect the results of energy use, except for the emission from cultivation, where N2 O emission from N inputs causes a higher GHG emission to energy
Please cite this article in press as: Corré, W.J., et al., Accounting for co-products in energy use, greenhouse gas emission savings and land use of biodiesel production from vegetable oils. Ind. Crops Prod. (2015), http://dx.doi.org/10.1016/j.indcrop.2015.11.062
G Model INDCRO-8542; No. of Pages 8
ARTICLE IN PRESS
6
W.J. Corré et al. / Industrial Crops and Products xxx (2015) xxx–xxx
Table 6 GHG emission for biodiesel production from agricultural crops. Oil palm
3.4. Oil exchange
Soybean
Rapeseed
Wheat
GHG emission in tonne CO2 -eq ha Cultivation 1.84 Transport of fruits/seeds 0.03 Oil extraction 0 0.44 Transport of oil Oil transesterification 1.03
0.71 0.59 0.21 0 0.12
2.50 0.12 0.30 0 0.35
2.37 0.24
3.34
2.61
−1
1.64
3.26
Accounting for meal (tonne CO2 -eq ha−1 ) 0.09 Allocation Total after allocation 3.25 Credits in a system expansion 0.16 3.18 Total after system expansion
1.00 0.65 1.25 0.39
1.08 2.18 0.87 2.39
GHG emission in tonne CO2 -eq t−1 biodiesel 0.84 Total 0.81 After allocation 0.79 After system expansion
3.37 1.35 0.79
2.42 1.61 1.77
Total
3.5. Glycerine
Table 7 Land use for biodiesel production from agricultural crops. Oil Palm
Soybean
Rapeseed
Wheat
1
1
1
1
0.05 0.95 0.076 0.924
0.63 0.37 0.887 0.112
0.36 0.64 0.577 0.423
Land use in ha year t−1 biodiesel 0.250 Total After allocation 0.237 0.230 After system expansion
2.053 0.692 0.230
0.740 0.466 0.310
Gross land use in ha year−1 Accounting for meal (ha ha−1 ) Allocation Total after allocation Credits in a system expansion Total after system expansion
As explained in Section 2.3, when assuming that an increased demand for oil to be used for production of biodiesel is always compensated for by an increased production of palm oil, only the calculation for palm oil following system expansion is relevant. Energy use, GHG emission and land use, expressed per tonne of biodiesel produced, are necessarily equal, irrespective of the oil actually used for biodiesel production (Fig. 2). In fact, system expansion for soybean results in a direct exchange of soy oil with palm oil, due to the fact that the co-product of soy oil (soy meal) is also the marginal protein source and therefore replaces itself in the system expansion (see Table 2).
use ratio. This effect is largest in rapeseed which has the highest N input per hectare. In soybean the effect is small because soy bean requires a low N input. For all crops, GHG emission from cultivation is the single highest emission value in the total GHG emission per ha. Expressed per tonne biodiesel, total GHG emission is highest for soybean due to its low oil yield. After allocation, diverting part of the emission to the meal, rapeseed has the highest emission per tonne (twice as high as the lowest value of oil palm), which is due to its high N input and associated N2 O emission relative to the oil yield. After system expansion the value for rapeseed is more than two times as high as the value for oil palm and soybean. Both allocation and system expansion can cause a large decline in the GHG emission per tonne biodiesel in soybean and rapeseed, in allocation strictly depending on the oil-to-meal ratio and in system expansion also depending on what products are replaced by the meal produced. 3.3. Land use Total land use per tonne of biodiesel (Table 7) is very different, reflecting the different oil yields per ha. This difference decreases by taking the production of meal into account, but only partly. After allocation, land use for soybean biodiesel is three times as high as land use for palm oil biodiesel, with the value for rapeseed in between. After system expansion, the value for rapeseed is again highest, 35% higher than the value for oil palm and soybean. Land use is expressed in hectare year t−1 , following the yields which are expressed in tonne ha−1 year−1 . For soybean this is disputable because a minor part of the soybean is grown in a double cropping system and in this case the year cannot be solely assigned to the soybean crop. Possible consequences of applying a double cropping system are addressed in Section 4.
Because the RED prescribes no allocation to glycerine, glycerine was also left out of the system expansion in the results above. However, a proper system expansion should include glycerine and therefore another calculation, including accounting for glycerine by allocation and system expansion, was made. In allocation on the basis of energy content, 4.25% of energy use, GHG emission and land use of biodiesel production allocated to biodiesel in the results of Tables 5–7 should be allocated to glycerine. For system expansion, per tonne of biodiesel a net energy production of 1.13 GJ (=appr. 10%) and a net GHG emission reduction of 0.067 t CO2 -eq (=appr. 3% for rapeseed and appr. 7% for oil palm and soybean) result from replacing natural gas by biogas from anaerobic co-digestion with animal manure (based on data from Corré and Conijn, 2015). Land use is not considered in system expansion since the glycerine does not replace any agricultural (co-) product. This results in a reduction of the energy use of appr. 10% for all crops and a greenhouse gas emission reduction of appr. 3% for rapeseed and appr. 7% for oil palm soybean. Due to the small proportion of glycerine relative to the oil yield, the effects of including accounting for glycerine are small but depending on crop, impact category and accounting method, the difference between inclusion and non-inclusion of glycerine varies between 0 and appr. 10%. 3.6. Rapeseed as marginal oil crop The results of the calculations per tonne biodiesel, both with rapeseed oil and palm oil as marginal oil, are presented in Table 8 and are also illustrated in Fig. 2. Comparable to the identical results for palm oil and soy oil in system expansion when palm oil is assumed to be the marginal oil, the results for rapeseed oil and soy oil are identical when assuming rapeseed oil as the marginal oil and the results for soybean comply with an exchange of soy oil for rapeseed oil. For oil palm, producing only a minor quantity of meal, the difference between palm oil and rapeseed oil being the marginal oil is very small. For rapeseed the effects are larger with increased energy use of 10%, GHG emission of 15%, and land use of 20%. The calculations in this study show that the choice of marginal oil really makes a difference in the results, especially in GHG emission for soy oil. The choice of marginal oil also has significant effects when considering direct oil exchange. Direct exchange for rapeseed oil increases energy use by 20%, land use by 60% and GHG emission by 160% compared to direct exchange for palm oil. 4. Discussion As shown in Fig. 2, different methods of accounting for coproducts yield different results. This implies that the choice of accounting method really does matter and that the method of system expansion is therefore indeed to be preferred since it takes account of the assumed real use of co-products. However, a
Please cite this article in press as: Corré, W.J., et al., Accounting for co-products in energy use, greenhouse gas emission savings and land use of biodiesel production from vegetable oils. Ind. Crops Prod. (2015), http://dx.doi.org/10.1016/j.indcrop.2015.11.062
G Model INDCRO-8542; No. of Pages 8
ARTICLE IN PRESS W.J. Corré et al. / Industrial Crops and Products xxx (2015) xxx–xxx
7
Table 8 Comparison of energy use, GHG emission and land use for biodiesel production from agricultural crops using system expansion with oil palm or rapeseed as marginal oil crop.
Energy use in GJ t−1 biodiesel GHG emission in tonne CO2 -eq t−1 biodiesel Land use in ha year t−1 biodiesel
Marginal oil: rapeseed oil Oil palm
Soy-bean
9.7 0.79 0.23
11.8 2.07 0.37
standard system expansion starting from the use of the co-products is not sufficient. The possibility of direct exchange of different oils should be included since different vegetable oils are equally suitable for major purposes. In this case the results change and in principle only the results from a system expansion of the marginal oil crop are relevant. The very different results from calculations with either palm oil or rapeseed oil as marginal oil and the effects of direct oil exchange illustrate that the choices to be made in the description of a system expansion are essential for the outcome and should be made explicit and need to be justified. In the current market, the choices for palm oil as marginal oil, soy meal as marginal protein source, and the possibility of exchanging different oils to be used for other purposes are most realistic (Schmidt and Weidema, 2008). The fact that in system expansion (without oil exchange) for soybean the results are identical to oil palm when palm oil is assumed to be the marginal oil and identical to rapeseed when rapeseed oil is assumed to be the marginal oil, is caused by the fact that soybean is not only an oil crop but also the world’s marginal protein crop. This results in a decrease of soy meal production in reaction to increased soy oil production and an increased production of the marginal oil to be used for other purposes. In fact, soy meal replaces itself and the complete increase in oil production is filled in by palm oil (or rapeseed oil), exactly as in the case of direct oil exchange. This analysis assumes that a future increase of vegetable oil production will presumably be filled in by palm oil. This is valid if the demand for vegetable protein will not change. However, this demand is certainly projected to increase in the future with an increased soybean production as a result. Nevertheless, this analysis only compares situations with and without increased use of vegetable oil for the production of bio-based materials. This increased use will indeed preferably be filled in by increased production of the marginal oil, independent of the actual shares of different oil crops in the total oil production, which is ultimately determined by the total demand for oil and for protein. Therefore, the analysis is not dependent on the level of protein demand and remains valid as long as palm oil functions as the world’s marginal oil and soybean functions as the world’s marginal protein source. As is shown in the calculations, results of applying system expansion are very sensitive to the choices made in the analysis. Regarding the choice for a marginal oil, it is good to be aware that although an increased demand will lead to increased production of predominantly one (the marginal) oil, it is quite well possible that such an increased supply will partly also be realized by an increased production of other oil crops. This indicates that the choice for one (marginal) product to be replaced is a simplification of the real world. The importance of assumptions concerning the use of co-products is illustrated by comparing Hoefnagels et al. (2010) with our results. In the analysis of Hoefnagels et al. (2010) rapeseed meal replaces soy meal without regarding the different protein-to-energy ratio, which results in the GHG emission from biodiesel production from rapeseed being circa 50% lower after system expansion compared with allocation on energy basis, while in our analysis the difference in GHG emission between allocation and system expansion is negligible. Furthermore, in the analysis of Hoefnagels et al. (2010) soy meal replaces wheat, again without
Rape-seed
Marginal oil: palm oil Oil palm
Soy-bean
Rape-seed
11.8 2.07 0.37
9.7 0.79 0.23
9.7 0.79 0.23
10.9 1.77 0.31
taking the difference in protein-to-energy ratio into account. This results in the GHG emission from biodiesel production from soybean being almost twice as high in system expansion as in allocation on energy basis, while in our analysis it is appreciably lower in system expansion. This is due to the choice of soy meal to replace the marginal protein instead of a low-protein commodity like wheat. Another example of the importance of the choices made in a system analysis is the very low GHG emission from biodiesel production from soybean after system expansion in Wang et al. (2011). This was probably caused by replacing glycerine produced from petrol by glycerine from biodiesel production. Replacing the production of petrol glycerine results in a large GHG emission reduction of 1.39 t CO2 -eq t−1 biodiesel (Reinhardt and Jungk, 2001) but due to saturation of the glycerine market it is not realistic to assume that increased glycerine production still can replace petrol glycerine. The RED (EC, 2009) prescribes no allocation to glycerine and for a better comparison, the results of the system expansion in Tables 5–7 were derived without taking glycerine into account. Since glycerine is produced in a constant ratio with biodiesel and the proportion of glycerine in the production is small, results of including glycerine in the analysis differ up to a maximum of only 10%. However, despite the differences being small, they give a good illustration of the fundamental differences of the methods used. In system expansion the same amount of glycerine has the same effects for different oils, while the effects are different for different oils when allocation is used. Furthermore, in system expansion the effect can be different in different impact categories, while in allocation the effect is necessarily the same in all impact categories. An example of the last effect is land use: in system expansion glycerine replaces fossil energy and hence no land use is accounted to it, while in allocation – independent of the use of glycerine – 4.25% of the land use is accounted to it. For the calculation of land use all yields are considered to be expressed on a per ha per year basis. This is correct for crops that are grown in a single cropping system, like rapeseed and oil palm. However, part of the soybeans in Brazil are grown in combination with other crops (notably corn) in a double cropping system, so both crops are harvested within one year from the same field. In this case the cropping area should be assigned to the oil crop only for part of the year. Since no analysis of the effects of a decreased soybean production on the double cropping system is available, the effects of double cropping could not be included in our analysis and it was assumed that only the area of soybean in a single cropping system would change. As shown in Table 2, for situations with palm oil as the marginal oil, the effect of double cropping can only be of importance when rapeseed oil is used for biodiesel production and soy meal is replaced by rapeseed meal. In this case the area of soybean shows a major change: 0.6 ha per hectare rapeseed, while this change is only 0.032 ha per hectare oil palm when palm oil is used for biodiesel production, and only 0.004 ha per hectare soybean when soy oil is used for biodiesel production. Theoretically, the average land use attributed to the production of soybean could be lowered by 10% considering the current 20% of soybean produced in double cropping systems in Brazil when soy bean is assumed to be produced during half the growing season (Babcock and Carriquiry, 2010). However, the yield in double cropping
Please cite this article in press as: Corré, W.J., et al., Accounting for co-products in energy use, greenhouse gas emission savings and land use of biodiesel production from vegetable oils. Ind. Crops Prod. (2015), http://dx.doi.org/10.1016/j.indcrop.2015.11.062
G Model INDCRO-8542; No. of Pages 8
ARTICLE IN PRESS
8
W.J. Corré et al. / Industrial Crops and Products xxx (2015) xxx–xxx
systems is lower than in single cropping systems due to the necessity of growing early ripening varieties (Babcock and Carriquiry, 2010). Considering a (theoretical maximum of) 10% decrease of the land use for soybean production (i.e., the production of 2800 kg in Table 1 is realized on 0.9 ha year−1 ), land use for biodiesel produced from rapeseed would increase from 0.31 (Table 7) to 0.35 ha year t−1 biodiesel (+13%). With direct oil exchange, however, double cropping of soybean has no effect on land use since the use of soy oil will not lead to an increased soybean production. Using allocation in this case would directly lead to a 10% lower land use for the production of soy oil. GHG emissions from (direct and indirect) land use change (LUC) are not included in this analysis. Data for individual ecosystem types and global regions are available but cannot be simply linked to increased land use for individual crops (Searchinger et al., 2008) and therefore LUC-related emissions were left out of the comparison of methods. Nevertheless, these emissions are too important to ignore. Figures for future scenarios in which system expansion is already applied are available (Bauen et al., 2010; Bowyer, 2010; Edwards et al., 2010; Laborde, 2011), where the option of exchange of crops is considered as well. In Laborde (2011), e.g., the use of this option resulted in very comparable emissions from LUC for the three oils: 2.0 t CO2 -eq t−1 biodiesel for palm and rapeseed oil and 2.1 t CO2 -eq t−1 biodiesel for soy oil. These figures are very sensitive to the assumed proportion of the expansion of oil palm production taking place on peat soils. In a similar way, Bauen et al. (2010) calculated LUC-related emissions of appr. 0.5–3.7 t CO2 -eq t−1 biodiesel for oil palm. Despite the large uncertainty, both in actual amount and location of land use change and in the GHG emissions related to these changes, the data show that estimated emissions from LUC are appreciable and can therefore not be ignored. It even seems probable that with LUC-related emissions taken into account for current agriculture, no GHG emission reduction at all can be reached from the use of biodiesel, irrespective of the oil used (Bowyer, 2010; Delzeit et al., 2011). From our analysis, a GHG emission of 0.73 t CO2 -eq t−1 biodiesel, as found in system expansion and direct oil exchange with palm oil as the marginal oil, including the use of glycerine, seems the most realistic value. Combined with an ILUC-related emission of 2.0 t CO2 -eq t−1 a total emission of 2.7 t CO2 -eq t−1 biodiesel is found. This is only slightly lower than the emission from an equivalent amount of fossil diesel, which is 3.1 t CO2 -eq assuming an energy content of 37 MJ kg−1 for biodiesel and an emission of 83.8 g CO2 -eq MJ−1 for fossil diesel (RED; EC, 2009), and it is certainly much too high to be considered sustainable. 5. Conclusions The two different accounting methods used can lead to different results with respect to energy use, GHG emission, and land use of biodiesel production. Values are generally higher when allocation is used but the differences vary between the three oil crops and between the three sustainability indicators analyzed. The choices concerning the use of co-products in system expansion have a large effect on the results. Besides the use of coproducts, direct exchangeability of different vegetable oils needs to be considered. Land Use Change effects could not be integrated in our comparison of methods but their potential effect on GHG emission is very large and should be taken into account.
References Babcock, B.A., Carriquiry, M., 2010. An exploration of certain aspects of CARB’s approach to modelling indirect land use from expanded biodiesel production. In: Staff Report 10-SR 105. Center for Agricultural and Rural Development, Iowa State University. Bauen, A., Chudziak, C., Vad, K., Watson, P., 2010. A causal descriptive approach to modelling the GHG emissions associated with the indirect land use impacts of biofuels. E4tech Report. BioGrace, 2011. Harmonised calculations of biofuel greenhouse gas emissions in Europe. Version 4. Accessible at: www.biograce.net (accessed 29.09.2015). Bowyer, C., 2010. Anticipated Indirect Land Use Change associated with expanded use of biofuels and bioliquids in the EU—An analysis of the national renewable energy action plans. Institute for European Environmental Policy. Accessible at: www.ieep.eu (accessed 29.09.2015). Conijn, J.G., Corré, W.J., de Ruijter, F.J., Rutgers, B., 2011. Economic and environmental performance of oilseed cropping systems for biodiesel production. In: Report 418. Plant Research International, Wageningen, the Netherlands. Corré, W.J., Conijn, J.G., 2015. Parameterisation of E-CROP. In: Report. Plant Research International, Wageningen, the Netherlands (in preparation). Dalgaard, R., Schmidt, J., Halberg, N., Christensen, P., Thrane, M., Pengue, W.A., LCA, Case, S., 2008. LCA of soybean meal. Int. J. Life Cycle Assess. 13, 240–254. Delzeit, R., Klepper, G., Lange, M., 2011. Review of IFPRI Study Assessing the Land Use Change Consequences of European Biofuel Policies and Its Uncertainties. Kiel Institute for the World Economy, Kiel, Germany. EBB, 2015. Statistics of the European biodiesel Board. Accessible at: http://www. ebb-eu.org/stats.php (accessed 29.09.2015). EC, 2009. Directive 2009/28/EC of the European Parliament and of the Council on the promotion of the use of energy from renewable sources. Official Journal of the European Union L 140. European Commission, Brussels, Belgium. ECS, 2006a. Environmental management–life cycle assessment–principles and framework. In: European Standard ISO 14040. European Committee for Standardisation, Brussels, Belgium. ECS, 2006b. Environmental management–life cycle assessment–requirements and guidelines. In: European Standard ISO 14044. European Committee for Standardisation, Brussels, Belgium. Edwards, R., Mulligan, D., Marelli, L., 2010. Indirect Land Use Change from increased biofuels demand. Comparison of models and results for marginal biofuels production from different feedstocks. JRC Report Brussels, Belgium. Eurostat, 2015. Statistical data on agricultural production. Accessible at: http://ec. europa.eu/eurostat/data/database (accessed 29.09.2015). FAO, 2002. Fertiliser use by crop. In: Joint Report FAO, IFA, IFDC, PPI, IPI, Fifth Edition. FAO, Rome, Italy. Hoefnagels, R., Smeets, E., Faaij, A., 2010. Greenhouse gas footprints of different biofuel production systems. Renew. Sust. Energy Rev. 14, 1661–1694. IPCC, 2006. IPCC Guidelines for national greenhouse gas inventories, Volume 4, Agriculture, forestry and other land use. Chapter 11: N2 O emissions from managed soils, and CO2 emissions from lime and urea application. Accessible at: http://www.ipcc-nggip.iges.or.jp/public/2006gl/vol4html (accessed 29.09.2015). Laborde, D., 2011. The Land Use Change consequences of European biofuel policies. International Food Policy Research Institute. Accessible at: http://trade.ec. europa.eu/doclib/docs/2011/october/tradoc 148289pdf (accessed 29.09.2015). Lywood, W., Pinkney, J., Cockerill, S., 2009. Impact of protein concentrate coproducts on net land requirement for European biofuel production. GCB Bioenergy 1, 346–359. Omni Tech International, 2010. Life Cycle Impact of Soybean Production and Soy Industrial Products. http://www.biodiesel.org/reports/20100201 gen-422.pdf (accessed on 29.09.2015). Reinhardt, G., Jungk, R., 2001. Pros and Cons of RME Compared to Conventional Diesel Fuel. Institut für Energie- und Umweltforschung, Heidelberg, Germany. Schmidt, J.H., Weidema, B.P., 2008. Shift in the marginal supply of vegetable oil. Int. J. Life Cycle Assess. 13, 235–239. Searchinger, T., Heimlich, R., Houghton, R.A., Dong, F., Elobeid, A., Fabiosa, J., Tokgoz, S., Hayes, D., Yu, T.H., 2008. Use of U.S. croplands for biofuels increases greenhouse gases through emissions from Land-Use Change. Science 319: 1238. Supporting Online Information. Accessible at: www.sciencemag.org/cgi/ content/full/1151861/DC1 (accessed 29.09.2015). Wang, M., Huo, H., Arora, S., 2011. Methods of dealing with co-products of biofuels in life-cycle analysis and consequent results within the U.S. context. Energy Policy 39, 5726–5736. Weidema, B.P., 2001. Avoiding co-product allocation in life cycle assessment. J. Ind. Ecol. 4–3, 11–33.
Acknowledgement This study was funded by the Dutch Ministry of EZ (Economic Affairs) as part of the policy support program Biobased Economy, BO-12.05-002.
Please cite this article in press as: Corré, W.J., et al., Accounting for co-products in energy use, greenhouse gas emission savings and land use of biodiesel production from vegetable oils. Ind. Crops Prod. (2015), http://dx.doi.org/10.1016/j.indcrop.2015.11.062
ARTICLE IN PRESS
INDCRO-8542; No. of Pages 8
Industrial Crops and Products xxx (2015) xxx–xxx
Contents lists available at ScienceDirect
Industrial Crops and Products journal homepage: www.elsevier.com/locate/indcrop
Accounting for co-products in energy use, greenhouse gas emission savings and land use of biodiesel production from vegetable oils W.J. Corré a,∗ , J.G. Conijn a , K.P.H. Meesters b , H.L. Bos b a b
Plant Research International, Wageningen University and Research Centre, P.O. Box 616, 6700 AP Wageningen, The Netherlands Food and Biobased Research, Wageningen University and Research Centre, P.O. Box 17, 6700 AA Wageningen, The Netherlands
a r t i c l e
i n f o
Article history: Received 30 July 2015 Received in revised form 23 October 2015 Accepted 24 November 2015 Available online xxx Keywords: Biodiesel Oil crops LCA GHG emission Energy use Land use
a b s t r a c t Accounting for co-products of vegetable oil production is essential in reviewing the sustainability of biodiesel production, especially since oil crops produce valuable protein-rich co-products in different quantities and qualities. Two accounting methods, allocation on the basis of energy content and system expansion, are compared. Significant differences in results exist between the methods where system expansion is to be preferred because it can take actual use of co-products into account. Results are very sensitive to the choices made in system expansion. Differences can be large, especially between a system expansion where primarily the use of co-products of the oil crops is taken into account and an expansion that also includes direct oil exchange of the vegetable oil used for biodiesel for the marginal oil in the market. © 2015 Published by Elsevier B.V.
1. Introduction With the purpose of reducing greenhouse gas (GHG) emission and to ensure a more sustainable energy supply, the EC established targets for the production of renewable energy in the EU for the year 2020 of 20% of the total energy use and 10% of the energy use in the transport sector (EC, 2009). Currently, renewable energy use in the transport sector relies largely on biofuels produced from agricultural crops, a production of which the sustainability is heavily debated. The EC Renewable Energy Directive (RED, EC, 2009) gives criteria for the minimum required GHG emission reduction of biofuels in order to be considered as renewable energy and complying with the target of renewable energy production. A realistic calculation of the GHG emission reduction requires that the production of co-products used for other purposes than energy production is taken into account. The RED (EC, 2009) prescribes allocation on the basis of energy content as accounting method for co-products in the reporting of the Member States on the performance of biofuel production regarding greenhouse gas (GHG) emission reduction. Allocation is a simple technique of dividing, e.g., environmental burdening or costs over main product and co-products. This division is performed in a ratio based on the ratios of mass, energy
∗ Corresponding author. E-mail address: [email protected] (W.J. Corré).
content or economic value of the different products. Use of this method does not comply with the principles of Life Cycle Assessment (LCA). LCA is an established technique, described in ISO 14040 and 14044 (ECS, 2006a,b), in which the division of environmental burdening takes the real use of co-products into account as much as possible. For this division, LCA prescribes the use of system expansion (substitution method) whenever possible. In system expansion, the products that can be replaced by the co-products of the process described are included in the system, and thus in the LCA analysis. This is acknowledged in the RED, where article 23–4 reads: “In reporting on greenhouse gas emission saving from the use of biofuels, the Commission shall use the values reported by Member States and shall evaluate whether and how the estimate would change if co-products were accounted for using the substitution approach ”. Nevertheless, allocation, being easier to apply, received much more attention in accounting for co-products than system expansion, possibly stimulated by the idea that both methods produce results that are generally comparable (EC, 2009). However, appreciable differences in GHG emission reduction between the methods have been reported, in favor of both allocation and system expansion (Hoefnagels et al., 2010; Wang et al., 2011) and although system expansion for agricultural production can easily lead to complex system descriptions, it is in principle always possible to use system expansion (Weidema, 2001). In this paper we present both system expansion as well as allocation. Moreover, we include energy use and land use as important sustainability issues in our analysis.
http://dx.doi.org/10.1016/j.indcrop.2015.11.062 0926-6690/© 2015 Published by Elsevier B.V.
Please cite this article in press as: Corré, W.J., et al., Accounting for co-products in energy use, greenhouse gas emission savings and land use of biodiesel production from vegetable oils. Ind. Crops Prod. (2015), http://dx.doi.org/10.1016/j.indcrop.2015.11.062
G Model INDCRO-8542; No. of Pages 8
ARTICLE IN PRESS
2
W.J. Corré et al. / Industrial Crops and Products xxx (2015) xxx–xxx
Energy use is important since it determines the efficiency of the biodiesel produced and land use is an important issue because productive arable land is scarce worldwide and oil crops for biodiesel do compete with crops for food production. Global production of biodiesel from vegetable oil is almost completely based on three oil crops: oil palm, soybean, and rapeseed. In Europe vegetable oil from these three crops makes up over 95% of biodiesel production (Conijn et al., 2009). All three crops have a protein-rich meal as co-product but in very different quantities and qualities. Hence, an increase in biodiesel production from vegetable oil will increase total vegetable oil production, in turn resulting in an increased meal production of which the quantity and quality depends on the oil crop involved. However, in a study focusing on the effects of biodiesel production all other activities are supposed to be left unchanged and the increase in meal production has to be compensated for by a decreased production of other proteinrich feed components. The most logical choice is replacement of a marginal product, i.e., a product that is supposed to function as buffer in the market and the production of which responds strongly to a change in demand. This will involve one of the major oil crops, soybean, since this is an oil crop as well as a protein crop and is in fact currently regarded as the world’s marginal protein source (Schmidt and Weidema, 2008). A decrease in soybean production will result in a decrease in (soy) oil production which will be compensated for by an increased production of another oil crop, most likely oil palm, since palm oil is currently regarded as the marginal vegetable oil (Schmidt and Weidema, 2008). This again results in increased meal production and – in the end – establishment of a new equilibrium between oil crops in which more oil production goes together with the production of an amount of feed ingredients from which a comparable amount and quality of feed can be produced as before. However, the three oil crop meals differ in protein content and digestibility and can therefore not simply substitute each other (Lywood et al., 2009). A qualitative equivalent replacement is only possible when a combination is made with an energy-rich feed ingredient like wheat (Lywood et al., 2009) or barley (Dalgaard et al., 2008). Another difference between the RED prescriptions and the substitution method regarding biodiesel production is found in the accounting for glycerine, a co-product of the hydrolysis of vegetable oil. According to the RED no allocation to glycerine should be made because of its current low value. In system expansion all co-products should be taken into account, and when no current product can be replaced waste disposal is eventually part of its life cycle. A problem in the use of vegetable oil not addressed in system expansion is the exchangeability of the different vegetable oils: system expansion only accounts for the use of co-products. However, since different vegetable oils are equally suitable for major applications (e.g., cooking) an increased demand for one oil could be compensated for by an increased production of another. This means that the use of rapeseed or soy oil for biodiesel production could be expected to cause an increase of the use of palm oil, currently the world’s marginal vegetable oil, for other purposes. The two methods of accounting for co-products for the production of biodiesel from palm oil, soy oil and rapeseed oil for the impact categories GHG emission, energy use and land use will be quantitatively compared in this paper. A variant of system expansion, where the vegetable oil used for biodiesel production can be directly replaced by an increased production of the marginal vegetable oil, will be used as well. In the calculations primarily palm oil is assumed to be the marginal vegetable oil; results of an alternative calculation with rapeseed oil as marginal vegetable oil are also presented.
Table 1 Outputs of crop production systems relevant for biodiesel production in the EU (from: BioGrace (2011) and Corré and Conijn (2015)). Product output in kg ha−1 Yield seeds Fresh fruit bunches Moisture content Yield (dry matter) Oil content in dry matter Crude vegetable oil Biodiesel Glycerine (pure) Meal content in dry matter Meal (dry matter) Moisture content in meal Crude protein content in meal (dry matter) a b c
Oil palm – 19,000c 34% 12,540 34.4% 4314 4008 401 3.75% 470 18% 16.5%
Soybean 2800 – 15% 2380 22% 524 487 49 78% 1860 18.3% 48%
Rapeseed b
3550 – 10% 3195 45.5% 1454 1351 135 54.5% 1740 17% 34%
Wheata 7200b – 13% 6264 – – – – – – – –
Wheat is selected as feed compound to be replaced by oil crop meals. Values refer to varieties of both crops sown in late autumn. Average yield level over the life cycle of oil palms.
2. Materials and methods 2.1. Crop production systems The analysis is based on representative production systems, relevant to biodiesel production in Europe, which reached just over 10 Mt in 2013 (EBB, 2015). Palm oil is produced in South-East Asia, soybean in Brazil, and rapeseed and wheat in Western Europe. Yield levels and associated (co-) products are presented in Table 1. All products are transported to and processed in the Netherlands. Palm oil is produced in perennial plantations. Annual production reaches its maximum five years after planting and a life cycle of approximately 25 years provides optimal total production. Fresh fruit bunches of oil palm are harvested and transported to a palm oil mill where the fruits are separated from the bunches. The empty fruit bunches are composted and recycled to the oil palm fields. The fruits are separated from the kernels and oil is extracted from both fruits and kernels. The remainders of the fruits are separated into a solid (fibers) and a liquid (palm oil mill effluent, POME) fraction. The solid fraction is, together with the kernel shells, used for energy supply to the palm oil mill, which is supposed to follow best practices and to use no external energy (BioGrace, 2011). POME is treated as waste water avoiding methane emission. The meal of the kernels (Palm Kernel Expeller, PKE) is used in animal feed. The crude oil, mostly fruit oil and kernel oil mixed, is transported to a harbor, refined, and shipped. Soybean is an annual crop, in Brazil partly grown in a double cropping system with maize in winter and soybean in summer and partly grown in a single cropping system. Since only a minor proportion of the soybeans in Brazil is grown in a double cropping system (Babcock and Carriquiry, 2010) and information on the effects of other crops in system expansion is lacking, this complication in land use is not included in the analysis; this means that the results in this paper are limited to soybean grown in a single cropping system. After harvest soybeans are transported to a harbor and mostly shipped as whole beans. Oil is extracted and refined in importing countries and the remaining meal is used in animal feed. Rapeseed and wheat are annual crops, in Europe grown in rotation with other arable crops in single cropping systems. After harvest, rapeseed is transported to a processing plant where the oil is extracted and refined; the remaining meal is used in animal feed. After harvest wheat is transported to locations where it is processed into animal feed. Yield levels (Table 1) of rapeseed and wheat were calculated as the average data for production in the period 2007–2009 in Germany and France, the main West European rapeseed- and wheat-producing countries (Eurostat, 2015).
Please cite this article in press as: Corré, W.J., et al., Accounting for co-products in energy use, greenhouse gas emission savings and land use of biodiesel production from vegetable oils. Ind. Crops Prod. (2015), http://dx.doi.org/10.1016/j.indcrop.2015.11.062
G Model INDCRO-8542; No. of Pages 8
ARTICLE IN PRESS W.J. Corré et al. / Industrial Crops and Products xxx (2015) xxx–xxx
3
Biodiesel is produced from crude vegetable oil by transesterification with methanol, yielding fatty acid methyl esters (FAME) and the co-product glycerine. 2.2. Use of co-products from biodiesel production The world’s three dominating biodiesel crops also produce a protein-rich meal; details on quantity and protein content of the meals are presented in Table 1. Due to the large difference in protein content between the meals, each type of meal to be replaced requires combination of a different quantity of energy rich wheat with soy meal in order to reach feed with a comparable protein to energy ratio that can be used as replacement. Furthermore, differences in digestibility result in different quantities of the various combinations to be used in order to reach the same feeding value. Data for the replacement ratios of oil palm meal and rapeseed meal relative to soy meal were calculated by Lywood et al. (2009) for the average use in the EU compound feed market. Due to animalspecific digestibility characteristics, these ratios differ between ruminants, pigs and poultry. The average ratio is the weighted average for use in feed for ruminants, pigs and poultry. As weight factor Lywood et al. (2009) used the total production of compound feed. Recalculation of the data for fresh matter from Lywood et al. (2009), using the moisture contents from Table 1, results in replacement ratios for dry matter of 1 kg of oil palm meal for 0.127 kg of soy meal plus 0.643 kg of wheat, and in 1 kg of rapeseed meal for 0.596 kg of soy meal plus 0.152 kg of wheat. Besides meal, oil palm produces empty fruit bunches, fibers, effluent and kernel shells. These co-products are used in the system as compost or energy source, or discarded as waste. Furthermore, wood (oil palm) and straw (soybean and rapeseed) are other coproducts of oil crops. These co-products are suitable for energy production but currently this application is hardly developed and will be left out of the analysis. Usually, these co-products are left on the land after harvest. Another co-product of biodiesel production from vegetable oil is glycerine. The quantity is equal for all oil types: 0.14 kg crude glycerine (containing appr. 15% moisture and 15% waste) comprising 0.10 kg pure glycerine per kg biodiesel (Table 1). Purified glycerine can replace valuable synthetic glycerine but since the increased biodiesel production has saturated the market for high-value applications, part of the glycerine is currently used in low-value applications like energy production. In our analysis, crude glycerine is assumed to be used for energy production by producing biogas through anaerobic co-digestion with other organic materials, such as animal manure. Advantage of this way of using glycerine is that crude glycerine can be used and no costly purification is necessary. 2.3. Accounting for co-products Allocation according to the RED results in allocation ratios of oil to meal of 95/5 for oil palm, 37/63 for soybean, and 64/36 for rapeseed, based on energy contents of 36 MJ kg−1 for oil and 17 MJ kg−1 for meal. The allocation ratios between biodiesel and glycerine are calculated according to the same principle; the RED, however, does not allow allocation to glycerine and calculations are made with and without allocation to glycerine. Allocation to glycerine involves an allocation ratio of biodiesel to glycerine of 96/4 (energy content of pure glycerine is 16.5 MJ kg−1 ). A precondition of allocation is that, e.g., the energy use of all activities up to and including the separation of the product and co-product has to be allocated over the product and co-product and that the energy use of activities concerning the product after separation has to be allocated completely to the product. In the case of oil production, energy use of agriculture, transport and oil extraction are allocated over oil and meal,
Fig. 1. Interactions of increased vegetable oil demand with crop production.
energy use of transesterification of oil are allocated over biodiesel and (if applicable) glycerine. System expansion according to LCA principles is based on the expected use of marginally produced co-products: use of oil crop meals as feed components, replacing comparable feed components produced from other crops, and use of crude glycerine for energy production through anaerobic digestion, replacing natural gas. Depending on the digestible protein and energy contents of the different meals, different replacement ratios are used (see Section 2.2). In principle, the produced meal replaces the marginal protein compound first: soy meal. In order to compensate for a possibly lower protein-to-energy ratio of the meal, wheat (a commonly used energy compound in feed) can be replaced as well as soy meal. Replaced soy meal causes a decreased production of soy oil, to be compensated for by increased production of the marginal oil: palm oil. This increased oil production in turn causes increased meal production, resulting in a feed-back loop in the calculations as is illustrated in Fig. 1. However, this feed-back is of minor importance, as shown in Table 2, since the oil production of 1 ha soybean can be replaced by 0.12 ha oil palm while 0.12 ha oil palm produces only appr. 60 kg palm kernel expeller whereas 1 ha of soybean produces almost 1900 kg soy meal. Replacements based on production levels (Table 1) and replacement ratios, including feed-back, are presented in Table 2. Biogas produced from glycerine is supposed to replace natural gas in a 1:1 energy ratio without upgrading the gas to natural gas quality. Palm oil is functioning as the global marginal vegetable oil since around the year 2000, before that rapeseed oil was considered as the marginal oil for a long time (Schmidt and Weidema, 2008). A further increase of vegetable oil production could quite well hamper the current fast increase of the oil palm production area and a return to rapeseed oil as the marginal oil seems possible. Against this background another calculation with rapeseed oil as marginal oil has been performed. Parameters for this system expansion are presented in Table 2. Comparison with palm oil as marginal oil shows an increased feed-back due to the higher meal production of rapeseed compared with oil palm, resulting in higher values of total wheat and soy meal replaced. On the assumption that increased oil production always leads to increased production of the marginal oil by direct exchange for this oil, the use of soy oil and rapeseed oil for the production of
Please cite this article in press as: Corré, W.J., et al., Accounting for co-products in energy use, greenhouse gas emission savings and land use of biodiesel production from vegetable oils. Ind. Crops Prod. (2015), http://dx.doi.org/10.1016/j.indcrop.2015.11.062
G Model INDCRO-8542; No. of Pages 8
ARTICLE IN PRESS
4
W.J. Corré et al. / Industrial Crops and Products xxx (2015) xxx–xxx
Table 2 Parameters of the system expansion method (on dry matter basis) with palm oil and rapeseed oil as marginal oils. Oil palm ha kg
Soybean kg ha
Rapeseed kg Ha
Oil produced Meal produced
4314 470
1 1
524 1860
1 1
1454 1740
1 1
Wheat replaced Soy meal replaced Soy oil replaced
302 60 17
0.048 0.032 0.032
0 1860 524
0 1 1
264 1037 292
0.042 0.558 0.558
Palm oil as marginal oil Palm oil produced Palm meal produced
17 2
0.004 0.004
524 57
0.121 0.121
292 32
0.068 0.068
Wheat replaced total Soy meal replaced total Additional palm oil produced
303 60 17
0.048 0.032 0.004
37 1867 526
0.006 1.004 0.122
284 1041 293
0.045 0.600 0.068
Rapeseed oil as marginal oil Rapeseed oil produced Rapeseed meal produced
17 20
0.012 0.012
524 627
0.360 0.360
292 349
0.201 0.201
Wheat replaced total Soy meal replaced total Additional rapeseed oil produced
318 75 21
0.051 0.040 0.015
119 2328 656
0.019 1.251 0.451
330 1297 365
0.053 0.698 0.251
Table 3 Inputs of crop production and transport distances for products. Input in kg ha−1
Oil palma
Soybeana
Rapeseedb
Wheatb
Dieseld Fertiliser N Compost N Fertiliser P2 O5 Fertiliser K2 O Pesticides Seeds N in crop residues
47.8 128 12 144 200 8.4 0 12
48.6 8 0 66 62 2.7 76 37.5
63 152c 0 44 77 2 5 69
65 123c 0 60 105 5 165 84
Transport distance in km Fresh fruit bunches Truck Seeds Truck Ship Train Truck Crude oil Ship
Oil palma 20
Soybeana
Rapeseedb
Wheatb
1100 600 10,000
50 600
50 600
a b c d
150 10,000
From: BioGrace (2011). From: Conijn et al. (2011). From: FAO (2002). Specific density of diesel is 0.83 kg−l .
biodiesel will in the current market lead to an increased production of palm oil to be used for other purposes. In this situation, only the calculations for palm oil remain relevant when system expansion is applied. Comparably, when rapeseed oil is assumed to be the marginal oil, direct exchange of oils would mean that increased use of soy oil or palm oil would lead to an increased production of rapeseed oil and the only relevant calculation in system expansion would be the calculation for rapeseed oil with rapeseed oil as the marginal oil.
Table 4 Energy use and GHG emission of inputs in biodiesel production from vegetable oil (from Corré and Conijn (2015) and Omni Tech (2010) (transesterification)).
Cultivation
Energy use MJ kg−1
GHG emission kg CO2 -eq kg−1
Diesel Fertiliser N (World/Europe) Compost N Fertiliser P2 O5 (World/Europe) Fertiliser K2 O Pesticides Seeds -Rapeseed -Soybean -Wheat N in crop residues
62.3a 49.2/40.3 0 4.7/3.4 5.6/4.6 263 6.2 12.6 2.3 0
4.68a 9.30b /11.45b 6.63b 0.32/0.22 0.38/0.31 18.4 0.61 1.26 0.25 5.70b
Processing Oil extraction Palm Soybean Rapeseed Transesterification
GJ t−1 biodiesel
kg CO2 -eq t−1 biodiesel
0c 6.29 3.16 5.13
0c 440 220 256
Transport Truck Train Ship, ocean
MJ t−1 km−1 1.08 0.76 0.12
kg CO2 -eq t−1 km−1 0.080 0.048 0.0095
a Including a 27% surcharge for lubricants and manufacturing and maintenance of machinery. b Including emissions from production and direct and indirect field emissions of N2 O after IPCC (2006). c Energy use in palm oil extraction is based on renewable energy from crop residues.
Table 5 Energy use for biodiesel production from agricultural crops. Oil palm
Soybean
Rapeseed
Wheat
Energy use in GJ ha Cultivation Transport of fruits/seeds Oil extraction Transport of oil Oil trans-esterification
13.31 0.41 0 5.67 20.53
5.77 7.96 3.06 – 2.50
11.15 1.81 4.27 – 6.92
11.40 3.67
15.07
−1
Total
39.92
19.29
24.15
Accounting for meal (GJ ha−1 ) Allocation Total after allocation Credits in a system expansion Total after system expansion
0.67 39.25 1.18 38.74
11.13 8.16 14.58 4.71
6.63 17.52 9.43 14.72
Energy use in GJ t−1 biodiesel Total After allocation After system expansion
9.96 9.79 9.67
39.6 16.8 9.67
17.9 13.0 10.9
emissions are based on scenario studies where system expansion is applied (e.g., Laborde, 2011), these data are not suitable for comparison of methods. The importance of LUC-related GHG emissions will be further addressed in Section 4. 3. Results
2.4. Other parameters Crop production inputs and transport distances are presented in Table 3 and energy use and GHG emission from crop production inputs, processing and transport are presented in Table 4. Although land use chance (LUC), direct and indirect, is an inevitable result of increased vegetable oil production, this is not included in our calculations of GHG emission. The RED acknowledges the importance of LUC effects on GHG emission and the need to include these emissions in reporting the GHG reduction of the production and use of biofuels but since data of LUC-related GHG
Results of the calculations are presented in Tables 5 (energy use), 6 (GHG emission) and 7 (land use) for the three oil crops and for wheat, insofar as necessary for the system expansion. The end results, expressed per tonne of biodiesel produced, for the three oil crops are also shown in Fig. 2, including the results from the calculation with rapeseed as marginal oil crop instead of oil palm. Glycerine has not been taken into account in the calculations presented in Tables 5–7 as stipulated in the RED. For comparison, the effects of including glycerine in the analysis are presented separately in Section 3.5.
Please cite this article in press as: Corré, W.J., et al., Accounting for co-products in energy use, greenhouse gas emission savings and land use of biodiesel production from vegetable oils. Ind. Crops Prod. (2015), http://dx.doi.org/10.1016/j.indcrop.2015.11.062
G Model INDCRO-8542; No. of Pages 8
ARTICLE IN PRESS W.J. Corré et al. / Industrial Crops and Products xxx (2015) xxx–xxx
5
Fig. 2. Energy use, GHG emission, and land use in biodiesel production from vegetable oils. With different methods of accounting for co-products; allocation: allocation on the basis of energy content; syst. exp. P = M: system expansion with palm oil as marginal oil; syst. exp. R = M: system expansion with rapeseed oil as marginal oil; oil exchange: increased oil demand is filled by increased palm oil production.
3.1. Energy use Large differences in energy use per hectare exist between the crops (Table 5). Energy use in cultivation is relatively low for soybean, due to the very low use of N fertilizer. Total transport energy use is high in oil palm and soybean due to the long transport distances from the crop production sites to Europe. Energy use in oil extraction is zero in oil palm due to the use of crop residues for energy. Differences in energy use in refining and transesterification reflect the differences in oil yield. Large differences between crops also appear when total energy use is expressed per tonne biodiesel (Table 5), with a factor of almost 4 between the highest (soybean) and the lowest (oil palm)
value. However, after accounting for the meal these differences decrease considerably: after allocation the factor declines to less than 2 and after system expansion differences are almost absent between the highest and lowest values. It illustrates the effects of taking meal into account and shows that the system expansion approach in our study gives higher credits to meal in comparison with the allocation method based on energy contents. 3.2. GHG emission GHG emission results (Table 6) strongly reflect the results of energy use, except for the emission from cultivation, where N2 O emission from N inputs causes a higher GHG emission to energy
Please cite this article in press as: Corré, W.J., et al., Accounting for co-products in energy use, greenhouse gas emission savings and land use of biodiesel production from vegetable oils. Ind. Crops Prod. (2015), http://dx.doi.org/10.1016/j.indcrop.2015.11.062
G Model INDCRO-8542; No. of Pages 8
ARTICLE IN PRESS
6
W.J. Corré et al. / Industrial Crops and Products xxx (2015) xxx–xxx
Table 6 GHG emission for biodiesel production from agricultural crops. Oil palm
3.4. Oil exchange
Soybean
Rapeseed
Wheat
GHG emission in tonne CO2 -eq ha Cultivation 1.84 Transport of fruits/seeds 0.03 Oil extraction 0 0.44 Transport of oil Oil transesterification 1.03
0.71 0.59 0.21 0 0.12
2.50 0.12 0.30 0 0.35
2.37 0.24
3.34
2.61
−1
1.64
3.26
Accounting for meal (tonne CO2 -eq ha−1 ) 0.09 Allocation Total after allocation 3.25 Credits in a system expansion 0.16 3.18 Total after system expansion
1.00 0.65 1.25 0.39
1.08 2.18 0.87 2.39
GHG emission in tonne CO2 -eq t−1 biodiesel 0.84 Total 0.81 After allocation 0.79 After system expansion
3.37 1.35 0.79
2.42 1.61 1.77
Total
3.5. Glycerine
Table 7 Land use for biodiesel production from agricultural crops. Oil Palm
Soybean
Rapeseed
Wheat
1
1
1
1
0.05 0.95 0.076 0.924
0.63 0.37 0.887 0.112
0.36 0.64 0.577 0.423
Land use in ha year t−1 biodiesel 0.250 Total After allocation 0.237 0.230 After system expansion
2.053 0.692 0.230
0.740 0.466 0.310
Gross land use in ha year−1 Accounting for meal (ha ha−1 ) Allocation Total after allocation Credits in a system expansion Total after system expansion
As explained in Section 2.3, when assuming that an increased demand for oil to be used for production of biodiesel is always compensated for by an increased production of palm oil, only the calculation for palm oil following system expansion is relevant. Energy use, GHG emission and land use, expressed per tonne of biodiesel produced, are necessarily equal, irrespective of the oil actually used for biodiesel production (Fig. 2). In fact, system expansion for soybean results in a direct exchange of soy oil with palm oil, due to the fact that the co-product of soy oil (soy meal) is also the marginal protein source and therefore replaces itself in the system expansion (see Table 2).
use ratio. This effect is largest in rapeseed which has the highest N input per hectare. In soybean the effect is small because soy bean requires a low N input. For all crops, GHG emission from cultivation is the single highest emission value in the total GHG emission per ha. Expressed per tonne biodiesel, total GHG emission is highest for soybean due to its low oil yield. After allocation, diverting part of the emission to the meal, rapeseed has the highest emission per tonne (twice as high as the lowest value of oil palm), which is due to its high N input and associated N2 O emission relative to the oil yield. After system expansion the value for rapeseed is more than two times as high as the value for oil palm and soybean. Both allocation and system expansion can cause a large decline in the GHG emission per tonne biodiesel in soybean and rapeseed, in allocation strictly depending on the oil-to-meal ratio and in system expansion also depending on what products are replaced by the meal produced. 3.3. Land use Total land use per tonne of biodiesel (Table 7) is very different, reflecting the different oil yields per ha. This difference decreases by taking the production of meal into account, but only partly. After allocation, land use for soybean biodiesel is three times as high as land use for palm oil biodiesel, with the value for rapeseed in between. After system expansion, the value for rapeseed is again highest, 35% higher than the value for oil palm and soybean. Land use is expressed in hectare year t−1 , following the yields which are expressed in tonne ha−1 year−1 . For soybean this is disputable because a minor part of the soybean is grown in a double cropping system and in this case the year cannot be solely assigned to the soybean crop. Possible consequences of applying a double cropping system are addressed in Section 4.
Because the RED prescribes no allocation to glycerine, glycerine was also left out of the system expansion in the results above. However, a proper system expansion should include glycerine and therefore another calculation, including accounting for glycerine by allocation and system expansion, was made. In allocation on the basis of energy content, 4.25% of energy use, GHG emission and land use of biodiesel production allocated to biodiesel in the results of Tables 5–7 should be allocated to glycerine. For system expansion, per tonne of biodiesel a net energy production of 1.13 GJ (=appr. 10%) and a net GHG emission reduction of 0.067 t CO2 -eq (=appr. 3% for rapeseed and appr. 7% for oil palm and soybean) result from replacing natural gas by biogas from anaerobic co-digestion with animal manure (based on data from Corré and Conijn, 2015). Land use is not considered in system expansion since the glycerine does not replace any agricultural (co-) product. This results in a reduction of the energy use of appr. 10% for all crops and a greenhouse gas emission reduction of appr. 3% for rapeseed and appr. 7% for oil palm soybean. Due to the small proportion of glycerine relative to the oil yield, the effects of including accounting for glycerine are small but depending on crop, impact category and accounting method, the difference between inclusion and non-inclusion of glycerine varies between 0 and appr. 10%. 3.6. Rapeseed as marginal oil crop The results of the calculations per tonne biodiesel, both with rapeseed oil and palm oil as marginal oil, are presented in Table 8 and are also illustrated in Fig. 2. Comparable to the identical results for palm oil and soy oil in system expansion when palm oil is assumed to be the marginal oil, the results for rapeseed oil and soy oil are identical when assuming rapeseed oil as the marginal oil and the results for soybean comply with an exchange of soy oil for rapeseed oil. For oil palm, producing only a minor quantity of meal, the difference between palm oil and rapeseed oil being the marginal oil is very small. For rapeseed the effects are larger with increased energy use of 10%, GHG emission of 15%, and land use of 20%. The calculations in this study show that the choice of marginal oil really makes a difference in the results, especially in GHG emission for soy oil. The choice of marginal oil also has significant effects when considering direct oil exchange. Direct exchange for rapeseed oil increases energy use by 20%, land use by 60% and GHG emission by 160% compared to direct exchange for palm oil. 4. Discussion As shown in Fig. 2, different methods of accounting for coproducts yield different results. This implies that the choice of accounting method really does matter and that the method of system expansion is therefore indeed to be preferred since it takes account of the assumed real use of co-products. However, a
Please cite this article in press as: Corré, W.J., et al., Accounting for co-products in energy use, greenhouse gas emission savings and land use of biodiesel production from vegetable oils. Ind. Crops Prod. (2015), http://dx.doi.org/10.1016/j.indcrop.2015.11.062
G Model INDCRO-8542; No. of Pages 8
ARTICLE IN PRESS W.J. Corré et al. / Industrial Crops and Products xxx (2015) xxx–xxx
7
Table 8 Comparison of energy use, GHG emission and land use for biodiesel production from agricultural crops using system expansion with oil palm or rapeseed as marginal oil crop.
Energy use in GJ t−1 biodiesel GHG emission in tonne CO2 -eq t−1 biodiesel Land use in ha year t−1 biodiesel
Marginal oil: rapeseed oil Oil palm
Soy-bean
9.7 0.79 0.23
11.8 2.07 0.37
standard system expansion starting from the use of the co-products is not sufficient. The possibility of direct exchange of different oils should be included since different vegetable oils are equally suitable for major purposes. In this case the results change and in principle only the results from a system expansion of the marginal oil crop are relevant. The very different results from calculations with either palm oil or rapeseed oil as marginal oil and the effects of direct oil exchange illustrate that the choices to be made in the description of a system expansion are essential for the outcome and should be made explicit and need to be justified. In the current market, the choices for palm oil as marginal oil, soy meal as marginal protein source, and the possibility of exchanging different oils to be used for other purposes are most realistic (Schmidt and Weidema, 2008). The fact that in system expansion (without oil exchange) for soybean the results are identical to oil palm when palm oil is assumed to be the marginal oil and identical to rapeseed when rapeseed oil is assumed to be the marginal oil, is caused by the fact that soybean is not only an oil crop but also the world’s marginal protein crop. This results in a decrease of soy meal production in reaction to increased soy oil production and an increased production of the marginal oil to be used for other purposes. In fact, soy meal replaces itself and the complete increase in oil production is filled in by palm oil (or rapeseed oil), exactly as in the case of direct oil exchange. This analysis assumes that a future increase of vegetable oil production will presumably be filled in by palm oil. This is valid if the demand for vegetable protein will not change. However, this demand is certainly projected to increase in the future with an increased soybean production as a result. Nevertheless, this analysis only compares situations with and without increased use of vegetable oil for the production of bio-based materials. This increased use will indeed preferably be filled in by increased production of the marginal oil, independent of the actual shares of different oil crops in the total oil production, which is ultimately determined by the total demand for oil and for protein. Therefore, the analysis is not dependent on the level of protein demand and remains valid as long as palm oil functions as the world’s marginal oil and soybean functions as the world’s marginal protein source. As is shown in the calculations, results of applying system expansion are very sensitive to the choices made in the analysis. Regarding the choice for a marginal oil, it is good to be aware that although an increased demand will lead to increased production of predominantly one (the marginal) oil, it is quite well possible that such an increased supply will partly also be realized by an increased production of other oil crops. This indicates that the choice for one (marginal) product to be replaced is a simplification of the real world. The importance of assumptions concerning the use of co-products is illustrated by comparing Hoefnagels et al. (2010) with our results. In the analysis of Hoefnagels et al. (2010) rapeseed meal replaces soy meal without regarding the different protein-to-energy ratio, which results in the GHG emission from biodiesel production from rapeseed being circa 50% lower after system expansion compared with allocation on energy basis, while in our analysis the difference in GHG emission between allocation and system expansion is negligible. Furthermore, in the analysis of Hoefnagels et al. (2010) soy meal replaces wheat, again without
Rape-seed
Marginal oil: palm oil Oil palm
Soy-bean
Rape-seed
11.8 2.07 0.37
9.7 0.79 0.23
9.7 0.79 0.23
10.9 1.77 0.31
taking the difference in protein-to-energy ratio into account. This results in the GHG emission from biodiesel production from soybean being almost twice as high in system expansion as in allocation on energy basis, while in our analysis it is appreciably lower in system expansion. This is due to the choice of soy meal to replace the marginal protein instead of a low-protein commodity like wheat. Another example of the importance of the choices made in a system analysis is the very low GHG emission from biodiesel production from soybean after system expansion in Wang et al. (2011). This was probably caused by replacing glycerine produced from petrol by glycerine from biodiesel production. Replacing the production of petrol glycerine results in a large GHG emission reduction of 1.39 t CO2 -eq t−1 biodiesel (Reinhardt and Jungk, 2001) but due to saturation of the glycerine market it is not realistic to assume that increased glycerine production still can replace petrol glycerine. The RED (EC, 2009) prescribes no allocation to glycerine and for a better comparison, the results of the system expansion in Tables 5–7 were derived without taking glycerine into account. Since glycerine is produced in a constant ratio with biodiesel and the proportion of glycerine in the production is small, results of including glycerine in the analysis differ up to a maximum of only 10%. However, despite the differences being small, they give a good illustration of the fundamental differences of the methods used. In system expansion the same amount of glycerine has the same effects for different oils, while the effects are different for different oils when allocation is used. Furthermore, in system expansion the effect can be different in different impact categories, while in allocation the effect is necessarily the same in all impact categories. An example of the last effect is land use: in system expansion glycerine replaces fossil energy and hence no land use is accounted to it, while in allocation – independent of the use of glycerine – 4.25% of the land use is accounted to it. For the calculation of land use all yields are considered to be expressed on a per ha per year basis. This is correct for crops that are grown in a single cropping system, like rapeseed and oil palm. However, part of the soybeans in Brazil are grown in combination with other crops (notably corn) in a double cropping system, so both crops are harvested within one year from the same field. In this case the cropping area should be assigned to the oil crop only for part of the year. Since no analysis of the effects of a decreased soybean production on the double cropping system is available, the effects of double cropping could not be included in our analysis and it was assumed that only the area of soybean in a single cropping system would change. As shown in Table 2, for situations with palm oil as the marginal oil, the effect of double cropping can only be of importance when rapeseed oil is used for biodiesel production and soy meal is replaced by rapeseed meal. In this case the area of soybean shows a major change: 0.6 ha per hectare rapeseed, while this change is only 0.032 ha per hectare oil palm when palm oil is used for biodiesel production, and only 0.004 ha per hectare soybean when soy oil is used for biodiesel production. Theoretically, the average land use attributed to the production of soybean could be lowered by 10% considering the current 20% of soybean produced in double cropping systems in Brazil when soy bean is assumed to be produced during half the growing season (Babcock and Carriquiry, 2010). However, the yield in double cropping
Please cite this article in press as: Corré, W.J., et al., Accounting for co-products in energy use, greenhouse gas emission savings and land use of biodiesel production from vegetable oils. Ind. Crops Prod. (2015), http://dx.doi.org/10.1016/j.indcrop.2015.11.062
G Model INDCRO-8542; No. of Pages 8
ARTICLE IN PRESS
8
W.J. Corré et al. / Industrial Crops and Products xxx (2015) xxx–xxx
systems is lower than in single cropping systems due to the necessity of growing early ripening varieties (Babcock and Carriquiry, 2010). Considering a (theoretical maximum of) 10% decrease of the land use for soybean production (i.e., the production of 2800 kg in Table 1 is realized on 0.9 ha year−1 ), land use for biodiesel produced from rapeseed would increase from 0.31 (Table 7) to 0.35 ha year t−1 biodiesel (+13%). With direct oil exchange, however, double cropping of soybean has no effect on land use since the use of soy oil will not lead to an increased soybean production. Using allocation in this case would directly lead to a 10% lower land use for the production of soy oil. GHG emissions from (direct and indirect) land use change (LUC) are not included in this analysis. Data for individual ecosystem types and global regions are available but cannot be simply linked to increased land use for individual crops (Searchinger et al., 2008) and therefore LUC-related emissions were left out of the comparison of methods. Nevertheless, these emissions are too important to ignore. Figures for future scenarios in which system expansion is already applied are available (Bauen et al., 2010; Bowyer, 2010; Edwards et al., 2010; Laborde, 2011), where the option of exchange of crops is considered as well. In Laborde (2011), e.g., the use of this option resulted in very comparable emissions from LUC for the three oils: 2.0 t CO2 -eq t−1 biodiesel for palm and rapeseed oil and 2.1 t CO2 -eq t−1 biodiesel for soy oil. These figures are very sensitive to the assumed proportion of the expansion of oil palm production taking place on peat soils. In a similar way, Bauen et al. (2010) calculated LUC-related emissions of appr. 0.5–3.7 t CO2 -eq t−1 biodiesel for oil palm. Despite the large uncertainty, both in actual amount and location of land use change and in the GHG emissions related to these changes, the data show that estimated emissions from LUC are appreciable and can therefore not be ignored. It even seems probable that with LUC-related emissions taken into account for current agriculture, no GHG emission reduction at all can be reached from the use of biodiesel, irrespective of the oil used (Bowyer, 2010; Delzeit et al., 2011). From our analysis, a GHG emission of 0.73 t CO2 -eq t−1 biodiesel, as found in system expansion and direct oil exchange with palm oil as the marginal oil, including the use of glycerine, seems the most realistic value. Combined with an ILUC-related emission of 2.0 t CO2 -eq t−1 a total emission of 2.7 t CO2 -eq t−1 biodiesel is found. This is only slightly lower than the emission from an equivalent amount of fossil diesel, which is 3.1 t CO2 -eq assuming an energy content of 37 MJ kg−1 for biodiesel and an emission of 83.8 g CO2 -eq MJ−1 for fossil diesel (RED; EC, 2009), and it is certainly much too high to be considered sustainable. 5. Conclusions The two different accounting methods used can lead to different results with respect to energy use, GHG emission, and land use of biodiesel production. Values are generally higher when allocation is used but the differences vary between the three oil crops and between the three sustainability indicators analyzed. The choices concerning the use of co-products in system expansion have a large effect on the results. Besides the use of coproducts, direct exchangeability of different vegetable oils needs to be considered. Land Use Change effects could not be integrated in our comparison of methods but their potential effect on GHG emission is very large and should be taken into account.
References Babcock, B.A., Carriquiry, M., 2010. An exploration of certain aspects of CARB’s approach to modelling indirect land use from expanded biodiesel production. In: Staff Report 10-SR 105. Center for Agricultural and Rural Development, Iowa State University. Bauen, A., Chudziak, C., Vad, K., Watson, P., 2010. A causal descriptive approach to modelling the GHG emissions associated with the indirect land use impacts of biofuels. E4tech Report. BioGrace, 2011. Harmonised calculations of biofuel greenhouse gas emissions in Europe. Version 4. Accessible at: www.biograce.net (accessed 29.09.2015). Bowyer, C., 2010. Anticipated Indirect Land Use Change associated with expanded use of biofuels and bioliquids in the EU—An analysis of the national renewable energy action plans. Institute for European Environmental Policy. Accessible at: www.ieep.eu (accessed 29.09.2015). Conijn, J.G., Corré, W.J., de Ruijter, F.J., Rutgers, B., 2011. Economic and environmental performance of oilseed cropping systems for biodiesel production. In: Report 418. Plant Research International, Wageningen, the Netherlands. Corré, W.J., Conijn, J.G., 2015. Parameterisation of E-CROP. In: Report. Plant Research International, Wageningen, the Netherlands (in preparation). Dalgaard, R., Schmidt, J., Halberg, N., Christensen, P., Thrane, M., Pengue, W.A., LCA, Case, S., 2008. LCA of soybean meal. Int. J. Life Cycle Assess. 13, 240–254. Delzeit, R., Klepper, G., Lange, M., 2011. Review of IFPRI Study Assessing the Land Use Change Consequences of European Biofuel Policies and Its Uncertainties. Kiel Institute for the World Economy, Kiel, Germany. EBB, 2015. Statistics of the European biodiesel Board. Accessible at: http://www. ebb-eu.org/stats.php (accessed 29.09.2015). EC, 2009. Directive 2009/28/EC of the European Parliament and of the Council on the promotion of the use of energy from renewable sources. Official Journal of the European Union L 140. European Commission, Brussels, Belgium. ECS, 2006a. Environmental management–life cycle assessment–principles and framework. In: European Standard ISO 14040. European Committee for Standardisation, Brussels, Belgium. ECS, 2006b. Environmental management–life cycle assessment–requirements and guidelines. In: European Standard ISO 14044. European Committee for Standardisation, Brussels, Belgium. Edwards, R., Mulligan, D., Marelli, L., 2010. Indirect Land Use Change from increased biofuels demand. Comparison of models and results for marginal biofuels production from different feedstocks. JRC Report Brussels, Belgium. Eurostat, 2015. Statistical data on agricultural production. Accessible at: http://ec. europa.eu/eurostat/data/database (accessed 29.09.2015). FAO, 2002. Fertiliser use by crop. In: Joint Report FAO, IFA, IFDC, PPI, IPI, Fifth Edition. FAO, Rome, Italy. Hoefnagels, R., Smeets, E., Faaij, A., 2010. Greenhouse gas footprints of different biofuel production systems. Renew. Sust. Energy Rev. 14, 1661–1694. IPCC, 2006. IPCC Guidelines for national greenhouse gas inventories, Volume 4, Agriculture, forestry and other land use. Chapter 11: N2 O emissions from managed soils, and CO2 emissions from lime and urea application. Accessible at: http://www.ipcc-nggip.iges.or.jp/public/2006gl/vol4html (accessed 29.09.2015). Laborde, D., 2011. The Land Use Change consequences of European biofuel policies. International Food Policy Research Institute. Accessible at: http://trade.ec. europa.eu/doclib/docs/2011/october/tradoc 148289pdf (accessed 29.09.2015). Lywood, W., Pinkney, J., Cockerill, S., 2009. Impact of protein concentrate coproducts on net land requirement for European biofuel production. GCB Bioenergy 1, 346–359. Omni Tech International, 2010. Life Cycle Impact of Soybean Production and Soy Industrial Products. http://www.biodiesel.org/reports/20100201 gen-422.pdf (accessed on 29.09.2015). Reinhardt, G., Jungk, R., 2001. Pros and Cons of RME Compared to Conventional Diesel Fuel. Institut für Energie- und Umweltforschung, Heidelberg, Germany. Schmidt, J.H., Weidema, B.P., 2008. Shift in the marginal supply of vegetable oil. Int. J. Life Cycle Assess. 13, 235–239. Searchinger, T., Heimlich, R., Houghton, R.A., Dong, F., Elobeid, A., Fabiosa, J., Tokgoz, S., Hayes, D., Yu, T.H., 2008. Use of U.S. croplands for biofuels increases greenhouse gases through emissions from Land-Use Change. Science 319: 1238. Supporting Online Information. Accessible at: www.sciencemag.org/cgi/ content/full/1151861/DC1 (accessed 29.09.2015). Wang, M., Huo, H., Arora, S., 2011. Methods of dealing with co-products of biofuels in life-cycle analysis and consequent results within the U.S. context. Energy Policy 39, 5726–5736. Weidema, B.P., 2001. Avoiding co-product allocation in life cycle assessment. J. Ind. Ecol. 4–3, 11–33.
Acknowledgement This study was funded by the Dutch Ministry of EZ (Economic Affairs) as part of the policy support program Biobased Economy, BO-12.05-002.
Please cite this article in press as: Corré, W.J., et al., Accounting for co-products in energy use, greenhouse gas emission savings and land use of biodiesel production from vegetable oils. Ind. Crops Prod. (2015), http://dx.doi.org/10.1016/j.indcrop.2015.11.062