lable at ScienceDirect

Journal of Cleaner Production 19 (2011) 129–137

Contents lists avai

Journal of Cleaner Production

journal homepage: www.elsevier .com/locate/ jc lepro

Life cycle assessment of hydrotreated vegetable oil from rape, oil palmand Jatropha

Rickard Arvidsson a,*, Sara Persson b, Morgan Froling b,c, Magdalena Svanstrom b

a Environmental Systems Analysis, Chalmers University of Technology, SE 41296 Goteborg, Swedenb Chemical Environmental Science, Chalmers University of Technology, SE 41296 Goteborg, Swedenc Ecotechnology and Environmental Science, Mid Sweden University, SE 83125 Ostersund, Sweden

a r t i c l e i n f o

Article history:Received 16 January 2009Received in revised form5 February 2010Accepted 5 February 2010Available online 12 February 2010

Keywords:LCAHydrotreatmentPalm oilRapeseed oilJatropha

* Corresponding author. Tel.: þ46 (0)31 772 21 61.E-mail address: rickard.arvidsson@chalmers.se (R.

0959-6526/$ – see front matter � 2010 Elsevier Ltd.doi:10.1016/j.jclepro.2010.02.008

a b s t r a c t

A life cycle assessment of hydrotreated vegetable oil (HVO) biofuel was performed. The study wascommissioned by Volvo Technology Corporation and Volvo Penta Corporation as part of an effort to gaina better understanding of the environmental impact of potential future biobased liquid fuels for cars andtrucks. The life cycle includes production of vegetable oil from rape, oil palm or Jatropha, transport of theoil to the production site, production of the HVO from the oil, and combustion of the HVO. The functionalunit of the study is 1 kWh energy out from the engine of a heavy-duty truck and the environmentalimpact categories that are considered are global warming potential (GWP), acidification potential (AP),eutrophication potential (EP) and embedded fossil production energy. System expansion was used totake into account byproducts from activities in the systems; this choice was made partly to make thisstudy comparable to results reported by other studies. The results show that HVO produced from palmoil combined with energy production from biogas produced from the palm oil mill effluent has thelowest environmental impact of the feedstocks investigated in this report. HVO has a significantly lowerlife cycle GWP than conventional diesel oil for all feedstocks investigated, and a GWP that is comparableto results for e.g. rape methyl ester reported in the literature. The results show that emissions from soilcaused by microbial activities and leakage are the largest contributors to most environmental impactcategories, which is supported also by other studies. Nitrous oxide emissions from soil account for morethan half of the GWP of HVO. Nitrogen oxides and ammonia emissions from soil cause almost all of thelife cycle EP of HVO and contribute significantly to the AP as well. The embedded fossil production energywas shown to be similar to results for e.g. rape methyl ester from other studies. A sensitivity analysisshows that variations in crop yield and in nitrous oxide emissions from microbial activities in soil cancause significant changes to the results.

� 2010 Elsevier Ltd. All rights reserved.

1. Introduction

Conventional crude oil derived fuels such as diesel oil have sincelong been pointed out as contributors to environmental impactssuch as global warming. Shrinking supplies of crude oil is anotherimportant driver for introduction of alternative fuels. However, theproduction and use of biofuels also have impacts on the environ-ment, and it is important that biofuels are evaluated for theirenvironmental impact throughout their whole life cycle in order toidentify fuels with the potential to be more sustainable in thisrespect. One type of biofuel that has been discussed in this contextis hydrotreated vegetable oil (HVO) fuel. This paper reports on a lifecycle assessment (LCA) study that compares three different HVOs.

Arvidsson).

All rights reserved.

The study was commissioned by Volvo Technology Corporation andVolvo Penta Corporation.

HVO is a paraffinic biobased liquid, with the chemical struc-ture CnH2nþ2, originating from vegetable oil (the process can alsobe applied to animal fat). The oil or fat is treated in a number ofprocesses, the most important being hydrogenation, in order tocreate a biobased liquid diesel fuel. During the hydrogenation,oxygen is removed from the triglyceride and converted intowater. Propane is formed as a byproduct and can be combustedand used for energy production. HVO can be used in conven-tional diesel engines, pure or blended with conventional diesel,due to its similar physical properties to diesel. The chemical andphysical properties of HVO are listed in Table 1 and compared tofatty acid methyl ester (FAME) and EN 590 as given in Rantanenet al. (2005). FAME is a biobased diesel substitute that similarlyto HVO is produced from oil vegetables and that has been

Table 1Chemical and physical properties of HVO, FAME and EN 590 diesel. HVO and FAMEare both made from vegetable oil, but have different chemical structures. EN 590diesel is a fossil fuel, and the chemical composition of HVO resembles that of EN 590diesel.

Property HVO FAME Diesel EN 590

Density at þ15 �C [kg/m3] 775–785 885 835Viscosity at þ40 �C [mm2/s] 2.9–3.5 4.5 3.5Cetane number 84–99 51 53Distillation 90 vol-% [�C] 295–300 355 350Cloud point [�C] �5 to �30 �5 �5Lower heating value [MJ/kg] 44 38 43Polyaromatic content [wt-%] 0 0 4Oxygen content [wt-%] 0 11 0Sulfur content [mg/kg] w0 <10 <10

R. Arvidsson et al. / Journal of Cleaner Production 19 (2011) 129–137130

investigated for its life cycle environmental performance previ-ously; see e.g. Bernesson (2004) and Edwards et al. (2007). EN590 is European standard diesel. The cetane number of HVO isvery high, the storage stability is good and HVO has no contentof sulfur, aromatics or ash. This gives benefits for the combustionprocess as well as for the catalytic aftertreatment processes.However, its density does not meet the European diesel fuelstandard EN 590, but up to 30% can be blended into diesel stillfulfilling the EN 590 norm.

Several companies, such as Universal Oil Products (UOP) andNeste Oil Corporation, are developing HVO fuels. The HVOproduced by Neste Oil has the trade name NExBTL (an acronym for‘‘next generation bio-to-liquid’’). The HVO produced by UOP has thetrade name Green Diesel. The purpose of this study is to assess thelife cycle environmental performance of HVOs from three differentvegetable feedstocks and to compare the environmental perfor-mance of the HVOs with the environmental performance of otherbiofuels. Results reported by Edwards et al. (2007) and Bernesson(2004) are used for comparisons.

2. Goal and scope

A life cycle assessment (LCA) of HVO from three differentfeedstocks was conducted. LCA has been used previously to assessthe environmental impact of several other biofuels; both diesel-like fuels and others, see for example publications by Leng et al.(2008), von Blottnitz and Curran (2007), Pleanjai et al. (2009),Bernesson (2004) and Edwards et al. (2007). Bernesson (2004)compared rape methyl ester from rapeseed oil and ethanol fromwheat, showing slightly better performance for ethanol regardingglobal warming potential, acidification and eutrophication, but theopposite for photochemical ozone creation potential and energyuse. Edwards et al. (2007) compared energy use and greenhousegas emissions for a number of biofuels (rape methyl ester, ethanol,biogas, dimethyl ether, etc.) and showed among other thingsa wide diversity in environmental performance. The low green-house gas emissions of used cooking oil methyl ester compared toconventional diesel was shown by Pleanjai et al. (2009), butimpact categories such as acidification and eutrophication was notstudied. Leng et al. (2008) showed that the difference in life cycleemissions between gasoline and E10 (consisting of 10 percent bio-ethanol and 90 percent gasoline) was very small regardinggreenhouse gases and that E10 resulted in higher acidic emissions.von Blottnitz and Curran (2007) conducted a review of many LCAstudies on biobased ethanol, showing for instance that ethanoloften contribute more to acidification than fossil fuels and thatethanol generally reduces greenhouse gas emissions compared tofossil fuels. Considering the results of these studies it should notbe taken for granted that biofuels have a lower environmental

impact than fossil fuels, and they show the importance ofincluding additional impact categories besides greenhouse gasemissions in LCA studies of biofuels. The goal of this LCA studywas to investigate the environmental impact of HVO combusted inheavy-duty trucks using pure HVO as fuel and with a geographicalfocus on Europe for consumption of the biofuel. The functionalunit used was 1 kWh out from the engine of a heavy-duty truck(kWhengine). The vegetable oil used for producing HVO can origi-nate from a number of different biological sources. In this study,rapeseed oil, palm oil and Jatropha oil were investigated. Rape-seed oil is a likely raw material if HVO is to be produced froma feedstock grown in Europe. Palm oil is currently the leastexpensive vegetable oil, and has the highest yield per ha of allvegetable oil feedstocks. It is also the feedstock that is currentlyused in HVO production by Neste Oil. Jatropha is a plant believedto be able to grow on wasteland and thus might be used for landreclamation. In this study, the rapeseed was considered to becultivated in Germany, the palm oil in Malaysia and the Jatrophain India. The hydrotreatment process is assumed to take place atNeste Oil’s production site in Porvoo, Finland, for all three feed-stocks. The HVO is considered to be used in truck engines incentral Germany. Regardless of which oil feedstock that is used forthe HVO production, the flow chart in Fig. 1 applies to the HVO lifecycle.

Electricity was modeled as the average of the electricity typesused in the different countries and not as marginal electricity (whichwould be the type of electricity that would be used to produce anextra MJ at peak load). In this study, all transports were consideredto use fossil diesel and not HVO or any other non fossil fuel (exceptfor in the actual consumption phase for HVO). Production capitalsuch as construction of buildings and transport vehicles, as well astransports of workers to facilities and electricity for offices weresystematically excluded. Large-scale production was assumed for allof the three different oil plants. Aspects related to preparation ofland for agricultural use are not included. This was partly due tomethodological difficulties, which are well described in Anton et al.(2007): there is a lack of definition regarding which parameters toconsider and what methodology that should be applied. Land usewas also excluded in order to facilitate comparison with results fromother LCA studies such as Edwards et al. (2007).

The impact categories assessed in the study are globalwarming potential (GWP, carbon dioxide equivalents), acidifica-tion potential (AP, sulfur dioxide equivalents), eutrophicationpotential (EP, nitrate equivalents) and embedded fossil produc-tion energy (MJ). The background to this choice of impact cate-gories is that greenhouse gas emissions and energy use are oftenconsidered the most important aspects in assessing biofuels, seee.g. Buchholz et al. (2008) or Edwards et al. (2007). Embeddedfossil production energy (i.e. MJ of fossil energy required for thefunctional unit of 1 kWh of HVO fuel) was also applied todescribe energy use in Luterbacher et al. (2009). Emissionscontributing to acidification and eutrophication is of interest forbiofuels; the contribution from the production chain was shownto have a significant impact on the total results e.g. in a study ofRME and ethanol by Bernesson (2004). Table 2 lists the emissionsconsidered in the inventory as well as their assumed contribu-tion to each impact category (Hauschild and Wenzel, 1998). Notethat the unit ‘‘tonne’’ in this study always refers to metric ton, i.e.1000 kg. Considering the variations in previous results of LCAstudies of biofuels, which are clearly demonstrated in the reviewof biobased ethanol studies by von Blottnitz and Curran (2007),the impact of some parameters on the results was studied ina sensitivity analysis. Sensitivity analyses are more and morecommon in LCA studies and were for instance applied in Lenget al. (2008) as well.

Production of vegetable oil

Transport

Production of HVO

Combustion

Transport

Fertilizers

Pesticides

Field work operations

Electricity

Hexane (rapeseed case only)

Diesel

Emissions from soil

Rape meal/palm cake/Jatropha seed cake

POME methane (palm oil scenario 1 only)

Process emissions

Emissions

Electricity

Hydrogen Emissions

Diesel Emissions

Emissions

Fig. 1. Process tree describing the HVO life cycle.

R. Arvidsson et al. / Journal of Cleaner Production 19 (2011) 129–137 131

3. Inventory

3.1. Production of vegetable oil

Data for the agricultural stage for rapeseed, oil palm and Jatro-pha originate mainly from Bernesson (2004), Edwards et al. (2007),Schmidt (2007), Yusoff and Hansen (2007), Prueksakorn andGheewala (2006).

Winter rapeseed, Brassica napus, is an oil plant with small darkseeds with an oil yield of 1045 kg per ha (Bernesson, 2004). Afterharvesting, the seed is dried and transported to the extractor, andthen the oil is pressed and then extracted with hexane as solvent.The byproduct rape meal is used as animal fodder and accountedfor by system expansion, replacing soymeal. Artificial fertilizers andpesticides are used during cultivation of rapeseed (Bernesson,2004; Schmidt, 2007).

The oil palm, Elaeis guineensis, is a perennial crop that gives anoil yield of 4220 kg per ha when both the palm oil and the palmkernel oil is included (Schmidt, 2007). The fresh fruit bunches areharvested, sterilized and stripped into fruits and empty fruitbunches. The fruits are then pressed, giving palm oil and a press

Table 2Emissions considered and their contributions to the different impact categories. Thelast two parameters, N and P are emissions to water and all other are emissions to air.

Name GWP100 years

[g CO2 eq./g]AP[g SO2 eq./g]

EP[g NO3

� eq./g]

Carbon dioxide CO2a 1

Methane CH4 25Nitrous oxide N2O 320Sulfur dioxide SO2 1Nitrogen oxides NOx 0.7 1.35Ammonia NH3 1.88 3.64Carbon monoxide CO 2Hydrocarbons HC 3Nitrogen N 4.43Phosphorous P 32.03

a Only fossil CO2 is considered.

cake. The press cake is burned in a boiler in order to produce steamand electricity, making the palm oil mill self-sufficient in energy(Schmidt, 2007; Yusoff and Hansen, 2007). The empty fruit bunchescan be burned for additional electricity production, but in mostcases they are landfilled (Schmidt, 2007), and therefore that is thescenario considered in this study. The palm kernels are pressedseparately in a palm kernel oil mill, which requires a small amountof electricity. The palm kernel press cake is used as animal fodder(Schmidt, 2007), which is accounted for by system expansion,replacing soy meal (Schmidt, 2007). Both palm oil and palm kerneloil are assumed to be utilized for producing HVO in this study.Artificial fertilizers and pesticides are used during the cultivation ofoil palms (Bernesson, 2004; Schmidt, 2007). The palm oil milleffluent (POME) is a residue from palm oil pressing, and consists ofboth water soluble parts of the palm fruit and materials that aresuspended in water, such as palm fibers. It could potentially giverise to large emissions of methane from decomposition of theorganic material (Schmidt, 2007; Yusoff and Hansen, 2007). ThePOME is often handled in either of two ways, uncontrolleddischarge into ponds or anaerobic digestion with gas collection.Both are assessed in this study, in two different scenarios; onescenario with the methane from the POME emitted to air (palmscenario 1), and one with the methane from the POME used forenergy production (palm scenario 2).

The oil rich fruits of Jatropha curcas Linnaeus can give an oil yieldof 1150 kg per ha (Prueksakorn and Gheewala, 2006). The fruits arepicked from the branches and collected and transported to theprocessing site where they are dried in sunlight to reduce the watercontent of the fruits. The dry fruits are placed in a cracking machineto remove the coats from the seeds. According to Prueksakorn andGheewala (2006) the coats are not further used although they couldpotentially be incinerated. The seeds are then pressed mechanicallywith a screw press to extract the oil, and finally the oil is purified byfiltering. The Jatropha fields are irrigated by diesel-driven irrigationpumps in the dry season according to Prueksakorn and Gheewala(2006). The Jatropha press cake cannot, in contrast to rape mealand palm kernel cake, be used as animal fodder, since Jatropha

R. Arvidsson et al. / Journal of Cleaner Production 19 (2011) 129–137132

contains substances that are toxic to animals (Gubitz et al., 1999;Francis et al., 2005). Instead, the cake can be used as a combinedfertilizer and biopesticide (Gubitz et al., 1999; Francis et al., 2005;Openshaw, 2000) that is how it is modeled in this study. Thus, noartificial fertilizers or pesticides are needed during the cultivationof Jatropha.

For all feedstocks, diesel is used to fuel all tractors.Emissions from microbial activities in the soil due to farming

operations, such as fertilizing, etc., were calculated according to themethod applied by Schmidt (2007) for all three oil plants. Emis-sions to air that were calculated are nitrous oxide (N2O), nitrogenoxides (NO and NO2) and ammonia (NH3) and emissions to waterare nitrate (recalculated as nitrogen (N) emissions) and phosphate(recalculated as phosphorus (P) emissions). As an example, Schmidt(2007) calculates the direct annual N2O emissions for a specific areausing an equation from IPCC (2000):

N2ODirect—N [hðFSN D FAM D FBN D FCRÞ$EF1

iD ðFOS$EF2Þ

½kg N2O—N�

FSN is the annual amount of synthetic nitrogen fertilizer appliedto the soil in kg N, FAM is the annual amount of nitrogen in animalmanure intentionally applied to the soil in kg N, FBN is the annualamount of nitrogen fixated by crops in kg N, FCR is the annualamount of nitrogen in crop residues returned to soils in kg N andFOS is the area of organic soil cultivated annually in ha. EF1 (inkg N2O–N per kg N input) and EF2 (in kg N2O–N per ha) are emis-sion factors. For EF1, a value of 1.25 is used, but variations in thisfactor were studied in the sensitivity analysis. For a more detaileddescription of the estimation of direct nitrous oxide emissions, andalso estimations of other emissions from soil, such as indirectemissions of nitrous oxide and emissions of ammonia and nitrogenoxides, see Schmidt (2007). The nursery stage is omitted because itis omitted in most references used (Schmidt, 2007; Yusoff andHansen, 2007; Prueksakorn and Gheewala, 2006).

3.2. Transport of vegetable oil

Different fuel emission standards (i.e. Euro I, II or IV) wereapplied for transport vehicles depending on what was consideredbeing most representative for the situation in different countries.Cultivation and pressing of the rapeseed is assumed to take place inGermany, the largest rape producer in the EU. The EU, in turn, is thelargest rape producer in the world (Mielke, 2005). The rapeseed oilis transported from the cultivation site to the port in Kiel bya heavy-duty truck with trailer running with a Euro IV dieselengine. The gross vehicle weight (GVW) is 40 tonne and the fuelconsumption is 4.9 l per 10 km. The transport distance is

Fig. 2. An example of how a triglyceride and hydrogen can be transformed into paraffinic cthen turned into a diesel-like fuel through isomerization.

approximately 300 km. At the port, the oil is reloaded and trans-ported by large cargo ships to Porvoo in Finland.

The palm oil is assumed to be cultivated and pressed inMalaysia, which is the largest palm oil producer in the world(Schmidt, 2007). It will then be transported by truck 100 km to PortKlang, one of the largest ports in Malaysia. The transport is con-ducted by a heavy-duty truck with an average fuel consumption of3.5 l per 10 km and a 26 tonnene GVW and running with a Euro Iengine.

Jatropha oil is assumed to originate from India, since the Indiangovernment has shown interest in Jatropha cultivation (Gubitzet al., 1999). The oil is transported by a heavy-duty truck with26 tonne GVW, running with a Euro II engine. The average fuelconsumption is 3.5 l per 10 km. The transport distance within Indiais 500 km from the center of the country to the port inVisakhapatnam.

At the ports in Germany, Malaysia and India, the freight isreloaded onto a large cargo ship and shipped to Porvoo in Finland,a distance of 1217 km, 16,720 km and 15,790 km, respectively. Fromthe port in Porvoo, the oils are transported by pipeline to the HVOproduction site.

The sea distances are all estimated from the MaritimeChainwebsite (http://www.maritimechain.com/port/port_distance.asp)and the transport emissions are calculated using the NTMCalc tool(http://www.ntm.a.se/ntmcalc/Default.asp). Land distances areestimated from maps.

3.3. Production of HVO

The HVO production involves several different activities, such asproduction of hydrogen, hydrogenation of the triglycerides (seeFig. 2), isomerization of the obtained paraffinic chains and combus-tion of the propane. Data used are provided by experts at Neste OilCorporation and based on the performance of a pilot plant and maynot correlate completely with full scale production. The hydrogenused in the reaction is produced in a steam reforming unit fromnatural gas. Inventory data for hydrogen production are obtainedfrom literature (Edwards et al., 2007; Marquevich et al., 2002).

3.4. Transport of HVO

From the production plant in Porvoo, the HVO is transported bypipeline to the port and from there by cargo ship 1217 km to Kiel. Thefuel is transported further (253 km, to an assumed location close toHannover) by a heavy-duty truck with trailer running with a Euro IVdiesel engine, a GVW of 40 tonne and a fuel consumption of 4.9 lper 10 km. Again, the sea distances are all received from theMaritimeChain website (http://www.maritimechain.com/port/port_

hains, propane, carbon dioxide and water by hydrogenation. The paraffinic chains are

R. Arvidsson et al. / Journal of Cleaner Production 19 (2011) 129–137 133

distance.asp) and the transport emissions are calculated using theNTMCalc tool (http://www.ntm.a.se/ntmcalc/Default.asp).

3.5. Combustion of HVO

The emissions from combustion of HVO are based on data fromtests in stationary truck engines in a test rig using the Europeanstationary cycle (Kleinschek, 2005). A 40 percent engine efficiencyis assumed, which is often considered to be the general perfor-mance for a truck diesel engine running on conventional diesel.

4. Results and discussion

4.1. Global warming and energy use

In Fig. 3, the first four bars show the contributions to GWP fromdifferent steps in the HVO life cycle for the four systems investi-gated: rape, palm (scenario 1 – POME methane emitted; scenario 2– POME methane used for energy production), and Jatropha. TheGWP is lowest for HVO from oil palm scenario 2, and highest forHVO from rape and Jatropha. Utilization of the POME methane forenergy production lowers the GWP significantly for palm oil basedHVO, mainly because of avoided emissions of methane, but doesnot change the ranking of the three feedstocks. HVO from oil palmand Jatropha have significantly larger contributions from transportscompared to HVO from rape, which is expected since the palm andJatropha oils are transported a much longer distance. There is noGWP contribution from the oil extraction phase for the two palm oilscenarios since the renewable palm kernel cake is used to fuel thepalm oil mill. For HVO from Jatropha, there is no contribution toGWP from fertilizers since the use of the Jatropha seed cake coversthe need of nutrients. The credit that HVO from rape receives fromthe replacement of soymeal by rapemeal lowers the GWP of HVOfrom rape significantly. GWP from field work operations aresignificantly higher for HVO from Jatropha compared to HVO frompalm and rape, because of the much higher energy use for fieldwork operations reported in Prueksakorn and Gheewala (2006) forJatropha compared with what is reported for rape and palm in

Fig. 3. GWP results from this study (first four bars). For comparison, results for RME accor(2004) are included (two last bars).

Bernesson (2004) and Schmidt (2007). This may be due to theirrigation of Jatropha fields reported in Prueksakorn and Gheewala(2006), but the GWP for different field work operations cannot bedistinguished in that report. However, it is clear from Fig. 3 thatemissions from soil give the largest contributions to GWP for HVO,regardless of feedstock (with N2O as the main contributor).

The HVO results in this study are compared to results for RME asdescribed in Edwards et al. (2007) and Bernesson (2004), repre-sented by the two last bars in Fig. 3. The impact from rapeseed oilproduction is expected to be very similar for RME and HVO fromrape. Fig. 3 shows that the largest difference between the studiedrape HVO and the two RME systems originates from assumptionsrelated to emissions from soil and fertilizers. It has to be remem-bered that nitrous oxide emissions from soil are difficult to estimate.The values from Schmidt (2007) used in the model of HVO from rapein this study are about twice as high per kg of oil as the values used inBernesson (2004) and Edwards et al. (2007) for RME. This indicatesthe need for more accurate and standardized but case sensitivemethods to estimate such emissions if production of biobased fuelsare to be compared. Edwards et al. (2007) has also given a muchsmaller credit for the rapemeal used as animal feed compared to inthis study, which uses data from Bernesson (2004). Note thatEdwards et al. (2007) model fuel use in cars. In order to comparewith the results from this study, the results were recalculated usingemission factors and engine efficiencies for trucks. The results fromEdwards et al. (2007) in Figs. 3 and 4 thus contain recalculated data.

Fig. 4 shows the embedded fossil production energy (i.e. MJ offossil energy used for 1 kWh of HVO fuel) for HVO from the threefeedstocks investigated in this study. The hydrotreatment processcontributes significantly for all feedstocks. In the case of HVO fromrapeseed, energy used for production of artificial fertilizers alsocontributes significantly, at the same time as the system expansionfor the use of rape meal as animal fodder, replacing soy meal,considerably decreases the total impact. More energy is requiredfor the transport of palm and Jatropha oil, since they are trans-ported a much longer distance. The high embedded fossil produc-tion energy of HVO from Jatropha may (as for GWP) be explained bythe irrigation needed using diesel-driven pumps as reported in

ding to Edwards et al. (2007) (case with glycerin used as chemical) and to Bernesson

Fig. 4. The embedded fossil production energy of HVO from rape, palm and Jatropha. For comparison, the result for RME according to Edwards et al. (2007) (case with glycerin usedas chemical) is included (last bar).

R. Arvidsson et al. / Journal of Cleaner Production 19 (2011) 129–137134

Prueksakorn and Gheewala (2006). The reported productionenergy for RME in Edwards et al. (2007) was recalculated from carsto trucks, as described earlier for GWP.

Fig. 5 compares the results from this study for HVO from rape,palm (scenarios 1 and 2) and Jatropha with GWP and total energyuse for some fuels studied in a screening involving several Euro-pean car manufacturers, reported by Edwards et al. (2007). Thefunctional unit used in Edwards et al. (2007) is 1 km for GWP and100 km for energy use, and the results from this study have

002-

051-

001-

05-

0

05

001

051

002

023003082062042022002

GW

P (g

C

O2

eq

/km

)

)mk001/JM(esuygrenE

Fig. 5. GWP and total energy use for HVO (the four black symbols) is compared to conventioSeveral production routes are listed for the biofuels.

therefore been recalculated into the same functional unit. Theresults from this study reported in Fig. 5 were also recalculated tocorrespond to the personal car used in Edwards et al. (2007), usingestimated combustion emissions from Rantanen et al. (2005).Edwards et al. (2007) use a special method to characterize energyuse for a biofuel, which includes so-called ‘‘biomass energy los-ses’’. This term accounts to some extent for inefficiencies in the useof biomass feedstocks in the production of biofuels. In order to beable to compare with the results in Edwards et al. (2007), energy

004083063043

leseidlanoitnevnoC

sadesunirecylG:EMRlacimehc

sadesunirecylG:EMRreddoflamina

morfleseidcitehtnySsaglarutan

morfleseidcitehtnySdoowdemraf

morfleseidcitehtnySrouqilkcalb/doowetsawdoowdemrafmorfEMD

etsawmorfEMDrouqilkcalb/doow

lapicinummorfsagoiBetsaw

erunamdiuqilmorfsagoiB

erunamyrdmorfsagoiB

epar,OVH

1oiranecsmlap,OVH

2oiranecsmlap,OVH

ahportaJ,OVH

nal diesel, RME, synthetic diesel, DME and biogas as reported by Edwards et al. (2007).

R. Arvidsson et al. / Journal of Cleaner Production 19 (2011) 129–137 135

contents in non-utilized biomass feedstock were therefore addedto the embedded fossil production energy and presented in Fig. 5.It should be noted that the ‘‘biomass energy losses’’, whenincluded, account for the majority of the energy use.

HVO has a comparatively low total energy use when compared tothe results from Edwards et al. (2007); the only biofuel in Fig. 5 thathas a lower total energy use is dimethyl ether (DME) produced fromblack liquor. The first generation biofuels, such as ethanol, RME andbiogas, have a higher total energy use (see Fig. 5). Regarding GWP,HVO has an average performance among the biofuels in Fig. 5 andabout one-third less compared to fossil diesel. The GWP of HVO isabout the same as that of RME from Edwards et al. (2007). However,the GWPs of biogas, synthetic diesel from wood or black liquor, andDME from farmed wood or black liquor according to Edwards et al.(2007) are much smaller than the GWP of HVO in this study.

4.2. Acidification

Results regarding acidification are shown in Fig. 6; the firstfour bars are for this study and the last bar is for Bernesson(2004). HVO from rape has the lowest AP (when the systemexpansion for rape meal is accounted for) and HVO fromJatropha has the highest AP. The low AP of rape is mainly dueto the combination of shorter transport distance and thesystem expansion for the use of rapemeal. The field workoperations contribute significantly to AP for HVO from Jatropha,probably due to diesel combustion required for irrigation, asearlier discussed for GWP. The nitrogen oxide and ammoniaemissions from soil are comparatively low for HVO from oilpalm because of the high oil yield of oil palm. The emissionsfrom soil are actually higher for oil palm per ha compared torape or Jatropha, but since the palm oil yield is about fourtimes higher, the emissions from soil per kWhengine are smaller.Nitrogen oxide emissions from combustion contribute signifi-cantly, which confirms the importance of aftertreatment to

Fig. 6. The AP of HVO from rape, palm and Jatropha in

reduce nitrogen oxide emissions in exhaust gases. The AP ofRME reported in Bernesson (2004) (the last bar in Fig. 6) issimilar to that of HVO from rape. The reported emissions fromsoil are smaller in Bernesson (2004) but the acidifying emis-sions from combustion are somewhat higher. AP is not includedin Edwards et al. (2007) and thus no comparison with thatstudy is performed regarding AP.

4.3. Eutrophication

The EP caused by HVO is described in Fig. 7. Emissions from soilclearly make up the largest contributor, regardless of feedstock.Nitrogen oxide emissions from combustion contribute to a less extent.No comparison with the other studies is shown regarding EP, sinceonly emissions to air are included in Bernesson (2004) and thusnitrate and phosphorus emissions, important contributors to EP, areomitted. Eutrophication is not assessed in Edwards et al. (2007).

4.4. Sensitivity analysis

In a sensitivity analysis, the impact of some key parameters onthe final results was investigated. The parameters vegetable oiltransport distance, crop yield and nitrous oxide emissions frommicrobial activities in soil were varied within a reasonable range.Assuming that the oil palm is cultivated somewhere in Africa (fromwhere it originates and is also cultivated today (Basiron, 2007)),instead of in Malaysia, the palm oil sea transport distance decreasesto about half. This would reduce the total GWP by 6 percent for HVOfrom palm, scenario 1, and by 9 percent for HVO from palm,scenario 2. Also, for both palm scenarios, it would reduce the AP by16 percent, EP by 4 percent and embedded fossil production energyby 13 percent. Jatropha oil is also cultivated and extracted in Africa(Gubitz et al., 1999; Openshaw, 2000), which would also reduce seatransports to about half. This lowers GWP by 6 percent, AP by 11percent, EP by 2 percent and embedded fossil production energy by

this study and RME according to Bernesson (2004).

Fig. 7. The EP of HVO from rape, palm and Jatropha.

R. Arvidsson et al. / Journal of Cleaner Production 19 (2011) 129–137136

9 percent for HVO from Jatropha. Since the transport distances inthe rape seed case is shorter, a change in transport distance with 50percent will affect the total results much less compared to the palmand Jatropha cases.

Regarding crop yields, Schmidt (2007) states that the palm fruityield varies �5 percent. This would give a change of 5 percent orless for all impact categories for HVO from palm oil. The rapeseedyield varies for different countries, fertilizing regimes and agricul-tural practices, and Schmidt (2007) applies a yield of 3.24 tonne perha to describe Danish conditions, which is substantially higher thanthe value used in this study. Applying the higher yield to this studyreduces GWP by 56 percent, AP by 48 percent, EP by 60 percent andembedded fossil production energy by 20 percent for HVO fromrapeseed oil. For Jatropha cultivation, the number of publishedstudies is rather limited. Seed yields between 0.5 and 12 tonne perha have been reported, due to differences in e.g. soil and rainfall(Francis et al., 2005; Openshaw, 2000). Assuming 0.5 tonne seedper ha (a tenfold decrease compared to the seed yield applied inthis study) would increase GWP by 770 percent, AP by 550 percent,EP by 880 percent and embedded fossil production energy by 470percent. Assuming 12 tonne seed per ha (a twofold increasecompared to the seed yield applied in this study) would decreaseGWP by 43 percent, AP by 30 percent, EP by 50 percent andembedded fossil production energy by 24 percent. Regardingnitrous oxide emissions from microbial activities in soil, there havelately been some indications in the scientific literature that theestimation method from IPCC might underestimate the nitrousoxide emissions by a factor of three or more (Crutzen et al., 2007;Destouni and Darracq, 2009). Thus the emission factor EF1, thepercent of applied nitrogen fertilizer that will end up as nitrousoxide, could increase from w1 percent to w3 percent. A change inthe emission factor of this magnitude would increase the GWP with30 percent for HVO from rapeseed, 70 percent for HVO from palmoil (scenario 1), 100 percent for HVO from palm oil (scenario 2) and80 percent for HVO from Jatropha oil.

From the sensitivity analysis it can be concluded that a variationof transport distances would change the results of this studymoderately, whereas variations in agricultural yield and nitrousoxide emissions connected to fertilizing and microbial activities insoil can cause large changes in the results. This further emphasizes

the earlier statement that there is a need for more precise and sitespecific methods for estimations of soil emissions from cultivationof biofuel feedstock. Also, this illustrates the need for inclusion ofnon-technical emissions in LCA, which has also been discussedelsewhere (Johansson et al., 2008).

5. Conclusion

The life cycle GWP, AP and embedded fossil production energyof HVO are similar in magnitude to those of RME published inprevious studies. HVO has about half the GWP compared toconventional diesel. The feedstock for HVO generating the lowestGWP, AP and EP is palm oil provided that methane is produced fromthe palm oil mill effluent and combusted for energy production.

Significant parts of the GWP of HVO originate from emissionsfrom soil during the agricultural phase, a result similar to what hasbeen reported in other studies (e.g. Edwards et al. (2007) andJohansson et al. (2008)). However, emissions from soil give an evenmore significant contribution to eutrophication, a larger contribu-tion than all other processes combined. Emissions from soil alsogive a large contribution to acidification. This illustrates the needfor inclusion of non-technical emissions in LCA. A sensitivity anal-ysis reveals that the results are highly sensitive to the parametersagricultural yield and nitrous oxide emissions related to fertilizingand microbial activities in soil.

Acknowledgements

We would like to thank Volvo Technology and Volvo Penta fortheir contribution to this study, and especially Peter Jozsa, PatrikKlintbom, Rolf Westlund, Jan Eismark and Lennart Cider. We alsothank Steven Gust from Neste Oil Corporation.

References

Anton, A., Castells, F., Montero, J.I., 2007. Land use indicators in life cycle assess-ment. Case study: the environmental impact of Mediterranean greenhouses.Journal of Cleaner Production 15 (5), 432–438.

Basiron, Y., 2007. Palm oil production through sustainable plantations. EuropeanJournal of Lipid Science and Technology 109 (4), 289–295.

R. Arvidsson et al. / Journal of Cleaner Production 19 (2011) 129–137 137

Bernesson, S., 2004. Life cycle assessment of rapeseed oil, rape methyl ester andethanol as fuels – a comparison between large- and small-scale production.Ph.D. Thesis at Swedish University of Agriculture Sciences, Agraria 497,Uppsala.

Buchholz, T., Luzadis, V.A., Volk, T.A., 2008. Sustainability criteria for bioenergysystems: results from an expert survey. Journal of Cleaner Production 17, S86–S98.

Crutzen, P.J., Mosier, A.R., Smith, K.A., Winiwarter, W., 2007. N2O release from agro-biofuel production negates global warming reduction by replacing fossil fuels.Atmospheric Chemistry and Physics Discussions 7 (4), 11191–11205.

Destouni, G., Darracq, A., 2009. Nutrient cycling and N2O emissions in a changingclimate: the subsurface water system role. Environmental Research Letters 4(035008) 7pp.

Edwards, R., Larive, J.-F., Mahieu, V., Rouveirolles, P., 2007. Well-to-wheel analysis offuture automotive fuels and powertrains in the European context, Version 2c.Report from Concawe, EUcar and Joint Research Centre of the EuropeanCommission, Brussels.

Francis, G., Edinger, R., Becker, K., 2005. A concept for simultaneous wastelandreclamation, fuel production, and socio-economic development in degradedareas in India: need, potential and perspectives of Jatropha plantations. NaturalResources Forum 29 (1), 12–24.

Gubitz, G.M., Mittelbach, M., Trabi, M., 1999. Exploitation of the tropical oil seedplant Jatropha curcas L. Bioresource Technology 67 (1), 73–82.

Hauschild, M., Wenzel, H., 1998. Environmental Assessment of Products, vol. 2.Chapman and Hall, London.

IPCC, 2000. Good Practice Guidance and Uncertainty Management in NationalGreenhouse Gas Inventories. Report from the Intergovernmental Panel onClimate Change of the United Nations Environmental Programme, Geneva.

Johansson, K., Perzon, M., Froling, M., Mossakowska, A., Svanstrom, M., 2008.Sewage sludge handling with phosphorus utilization – life cycle assessment offour alternatives. Journal of Cleaner Production 16 (1), 135–151.

Kleinschek, G., 2005. Emissions Tests with Synthetic Diesel Fuels (GTL & BTL) witha Modern Euro 4 (EGR) Engine. Report from Scania, Sodertalje, Sweden.

Leng, R., Wang, C., Zhang, C., Dai, D., Pu, G., 2008. Life cycle inventory and energyanalysis of cassava-based fuel ethanol in China. Journal of Cleaner Production 16(3), 374–384.

Luterbacher, J.S., Froling, M., Vogel, F., Marecchal, F., Tester, J.W., 2009. Hydro-thermal gasification of waste biomass: process design and life cycle asessment.Environmental Science & Technology 43 (5), 1578–1583.

Marquevich, M., Sonnemann, G.W., Castells, F., Montane, D., 2002. Life cycleinventory analysis of hydrogen production by the steam-reforming process:comparison between vegetable oils and fossil fuels as feedstock. GreenChemistry 4 (5), 414–423.

Mielke, T., 2005. Aussichten der Rapsproduktion in den nachsten 10 Jahren. Reportfrom ISTA Mielke GmbH (OIL WORLD), Hamburg.

Openshaw, K., 2000. A review of Jatropha curcas: an oil plant of unfulfilled promise.Biomass & Bioenergy 19 (1), 1–15.

Pleanjai, S., Gheewala, S.H., Garivait, S., 2009. Greenhouse gas emissions fromproduction and use of used cooking oil methyl ester as transport fuel inThailand. Journal of Cleaner Production 17 (9), 873–876.

Prueksakorn, K.; Gheewala, S.H., 2006. In Energy and Greenhouse Gas Implicationsof Biodiesel Production from Jatropha curcas L., 2nd Joint InternationalConference on ‘‘Sustainable Energy and Environment (SEE 2006)’’, 21–23November, Bangkok, Thailand.

Rantanen, L., Linnaila, R., Aakko, P., Harju, T., 2005. NExBTL – Biodiesel fuel of thesecond generation. Report by Neste Oil Corporation and Technical ResearchCentre of Finland, Espoo.

Schmidt, J.H., 2007. Life cycle assessment of rapeseed oil and palm oil. Ph.D. Thesisat Aalborg University, Aalborg.

von Blottnitz, H., Curran, M.A., 2007. A review of assessments conducted on bio-ethanol as a transportation fuel from a net energy, greenhouse gas, and envi-ronmental life cycle perspective. Journal of Cleaner Production 15 (7), 607–619.

Yusoff, S., Hansen, S.B., 2007. Feasibility study of performing an life cycle assess-ment on crude palm oil production in Malaysia. The International Journal of LifeCycle Assessment 12 (1), 50–58.