Land use changes, greenhouse gas emissions and fossil fuel substitution of biofuels compared to...

b i om a s s an d b i o e n e r g y 3 5 ( 2 0 1 1 ) 4 0 6 0e4 0 7 4

Avai lab le at www.sc iencedi rect .com

ht tp : / /www.e lsev ier . com/ loca te /b iombioe

Land use changes, greenhouse gas emissions and fossil fuelsubstitution of biofuels compared to bioelectricity productionfor electric cars in Austria

Johannes Schmidt*, Viktoria Gass, Erwin Schmid

Institute for Sustainable Economic Development, University of Natural Resources and Life Sciences, Vienna, Austria

a r t i c l e i n f o

Article history:

Received 22 March 2011

Received in revised form

27 June 2011

Accepted 1 July 2011

Available online 23 July 2011

Keywords:

Biofuels

Bioelectric mobility

2020 goals

Spatially explicit optimization

* Corresponding author. Tel.: þ43 1 47654 35E-mail address: johannes.schmidt@boku.

0961-9534/$ e see front matter ª 2011 Elsevdoi:10.1016/j.biombioe.2011.07.007

a b s t r a c t

Bioenergy is one way of achieving the indicative target of 10% renewable energy in the

transportation sector outlined in the EU Directive 2009/28/EC. This article assesses the

consequences of increasing the use of bioenergy for road transportation on land use,

greenhouse gas (GHG) emissions, and fossil fuel substitution. Different technologies,

including first and second generation fuels and electric cars fuelled by bioelectricity are

assessed in relation to existing bioenergy uses for heat and power production. The article

applies a spatially explicit energy system model that is coupled with a land use optimi-

zation model to allow assessing impacts of increased biomass utilization for energy

production on land use in agriculture and forest wood harvests. Uncertainty is explicitly

assessed with Monte-Carlo simulations of model parameters. Results indicate that electric

mobility could save GHG emissions without causing a significant increase in domestic land

use for energy crop production. Costs of electric cars are still prohibitive. Second genera-

tion biofuels are more effective in producing fuels than first generation ethanol. However,

competition with power and heat production from ligno-cellulosic feedstock causes an

increase in GHG emissions when introducing second generation fuels in comparison to

a baseline scenario.

ª 2011 Elsevier Ltd. All rights reserved.

1. Introduction biodiesel and0.60TWhof ethanol in 2008 [2].A further increase

The directive 2009/28/EC requires all EU member states to

guarantee a share of 10% of renewable fuels in the trans-

portation sector by 2020. The target can be attained by various

measures, including an increase in the share of biofuels and an

increase in the utilization of renewably produced electricity in

the transportation sector. In Austria, bioenergy is traditionally

important providing around 8% of the primary energy demand

in 2006, mainly for heating purposes [1]. Other uses of bio-

energy developed in recent years include biofuel and power

production. Austria has compliedwith the 5.75% indicative EU

biofuel target since late 2008 and used around 4.00 TWh of

94; fax: þ43 1 47654 3692.ac.at (J. Schmidt).ier Ltd. All rights reserve

of the supply of biofuelswill be difficult to achieve, particularly

if only domestic biomass supply is processed. However, new

technologies are emerging that aim to increase biofuel

productivity and diversify feedstock supply. Second genera-

tion biofuels that may use ligno-cellulosic feedstock for fuel

production are regarded as a sustainable alternative to first

generation biofuels which aremainly produced from food and

feed crops [3],[4]. A technological alternative to the Internal

Combustion Engine (ICE) is the Battery Electric Vehicle (BEV).

Technical and economic barriers currently prevent the large

scale introduction of electric cars. However, future potentials

are considered to be significant [5],[6].

d.

mailto:[email protected]

http://www.sciencedirect.com

http://www.elsevier.com/locate/biombioe

http://dx.doi.org/10.1016/j.biombioe.2011.07.007



b i om a s s a n d b i o e n e r g y 3 5 ( 2 0 1 1 ) 4 0 6 0e4 0 7 4 4061

Since the large scale introduction of biofuels in the US and

the EU, an extensive discussion has evolved about direct and

indirect landuse changesandgreenhousegas (GHG) emissions

[3],[7], as well as the increasing competition between food and

fuels [8],[9]. Assessments of the effect of increasing bioenergy

utilization in the transportation sector should therefore

explicitly address the issue of land use change. Another

important issuewhenassessing bioenergy policies is the effect

of additional wood harvests on forest carbon stocks and the

consequence on total GHG emissions in the carbon cycle [10].

Campbell et al. [11] conclude that bioelectricity production

for electric mobility needs significantly less land than produc-

tion of biofuels. However, they assess technical aspects such as

crop production potentials and conversion efficiencies and do

not take into account the costs of the various options. Steenhof

et al. [12] presentadetailedmodel but donot explicitly calculate

land use effects or costs of various options. Other studies

[5],[13], compare costs andGHGemissions of various renewable

transport options but do not explicitly address the issue of

changes in land use and forest carbon stocks.

We link a land use optimization model with a spatially

explicit energy system model to evaluate the use of first and

second generation biofuels and bioelectricity in the trans-

portation sector with respect to land use change, GHG emis-

sions, and fossil fuel substitution in Austria. We track the

effect of biofuel policies on the amount of food and feed crops

replaced by energy crops and also track changes in forest

wood harvests. The techno-economic characteristics of future

biofuel production as well as of electric cars are not well

known yet. Also, high uncertainty is attached to future energy

price developments. We therefore apply a Monte-Carlo

Fig. 1 e Model co

simulation of model parameters to explicitly address uncer-

tainty in model outcomes.

The article is structured such that section 2 presents the

applied methodology, section 3 reports results, and section 4

discusses and concludes.

2. Methodology

2.1. Model components

The data and the models used in the analysis are presented in

Fig. 1. CropRota [14], EPIC [15] and PASMA [16] are used jointly

to deliver spatially explicit supply curves for agricultural

biomass. Soil, climate and management data as well as

production costs and prices of agricultural commodities are

necessary input for the model framework. A forest growth

model that spatially explicitly estimates annual stock

increases in forests is used to determine regional energy wood

potentials from forestry. For each modeled region, a biomass

supply curve is derived by combining energy wood potentials

with historic prices and price elasticities estimated from

historic data [17]. For different types of owners, i.e. small

private owners, large private owners and state owned forests,

different supply curves are derived because different price

elasticities apply to the three distinct groups. A detailed

description of the approach can be found in [18]. The heat

demand is spatially explicitly estimated using data on build-

ings (type and age) [19], which are combined with average

consumption values for such buildings. Performance data of

bioenergy production technologies are taken from a literature

mponents.




review while GHG emission factors are mainly taken from the

Austrian National Inventory Report [20].

BeWhere is a techno-economic energy system model that

integrates the output of the other models. It is a spatially

explicit,mixed integerprogramand is developedandapplied to

assess the costs, land use and GHG emissions of different bio-

energy conversion routes. The model minimizes the costs of

supplyingAustriawith transportation fuels,heatandelectricity

from either bioenergy or fossil fuels. It is static and simulates

one year of operation. The current model version considers

domesticbiomasssupplyandenergydemandonlyanddoesnot

allow imports and exports of biomass or bioenergy commodi-

ties. Themodel determineswhich bioenergy plants of a specific

size and specific location shall be built and which demand

regionsare suppliedwithbioenergy and/orwith fossil fuels (see

Fig. 2). Plantsmayproducevariousenergy commodities, e.g. the

heat produced in a combined heat and power plant (CHP) may

bedelivered todistrict heatingnetworks. By-productsof biofuel

plants are assumed to be sold as animal feed. Biomass supply

curvesendogenouslydeterminethepriceofdomestic feedstock

from forestry and agriculture, while prices of fossil fuels and

energy demand are defined exogenously.

Taxes currently applied to both fossil and bioenergy fuels

are not included in the model. The subsequent sections

describe in detail the biomass supply modeling (section 2.2),

the modeling of technologies (section 2.3), of greenhouse gas

emissions (section 2.4) and of demand scenarios (section 2.5).

A detailed description of themodel, including the formulation

of the mixed integer program BeWhere, can be found in [18].

2.2. Biomass supply

We consider biomass from forestry and agriculture as possible

feedstock for bioenergy production. Biomass supply curves

Fig. 2 e Diagram of the mixed in

are used tomodel the amount of biomass that can be provided

domestically at different prices. However, data is separately

available for forestry and agriculture. A forest growth model

that spatially explicitly estimates annual stock increases in

forests is used to determine regional energy wood production

potentials from forestry. For each modeled region, a biomass

supply curve is derived by combining energy wood potentials

with price elasticities estimated from historic data [17]. For

different types of owners, i.e. small private owners, large

private owners and state owned forests, different supply

curves are derived because different price elasticities apply to

the three distinct groups (See Fig. 3). A detailed description of

the approach can be found in [18].

Biomass growth on agricultural land for different crops

under different management options and crop rotations is

simulated with the biophysical process model EPIC (Environ-

mental Policy Integrated Climate) [15],[21]. Outputs of EPIC are

part of gross margin calculations, which serve as input to the

spatially explicit land use optimization model PASMA (Posi-

tive Agricultural Sector Model Austria) [16]. The model opti-

mizes revenue from land use by maximizing total gross

margin from agricultural activities subject to resource

endowments atmunicipal level. Prices of crops are taken from

the year 2006 [22] and linearly extrapolated to 2020, based on

the agricultural outlook of OECD & FAO [23]. We assume that

deviations of Austrian commodity prices from world market

prices observed in the past will prevail in the future. About 40

agricultural crops with three intensification levels are repre-

sented in PASMA. They have been grouped into seven cate-

gories including grains, oil seeds, forage crops, ethanol crops,

oil crops, short rotation cellulose, and others. Ethanol crops

include starchy and sugar crops that can be used for first

generation ethanol production. Oil crops (e.g. sunflower and

rapeseed) are used for first generation biodiesel production.

teger programming model.




Heat, power and second generation biofuels may be

produced from short rotation coppice (src, e.g. short rotation

poplar). Spatially explicit biomass supply curves are generated

with PASMA by steadily increasing prices for energy crops by

0%e300% while all other crop prices remain constant. Thus,

points on the supply curve are generated accounting for

intensification and land use changes, i.e. bioenergy crops

substitute food and feed. For instance, an aggregated supply

curve for short rotation lignocellulose and oil crops can be

found in Fig. 3. To indicate the level of production attained in

the model scenarios, the mean production levels attained in

themodeled scenarios is reported. The figure shows that price

increases of 100% are necessary to trigger substantial amounts

of short rotation coppice production in Austria. The increases

in short rotation coppice production cause significant

decreases in food and feed production. The figure also shows

that the potentials for oil crop production are limited, even if

prices are increased by 300%. Agronomic constraints impede

an expansion of oil crop production beyond a limit of 0.8 TWh

annually. Crop rotations limit rapeseed and sunflower

production significantly: the crop rotation model CropRota

assumes that they may be grown only every four years on the

same plot. The suitability of soils and climates in growing oil

crops further decreases the potential for oil crop production.

The substitution of food and feed crops by energy crops is

determined by calculating the difference in food and feed crop

production between biofuel policy scenarios and a baseline

scenario without policy intervention. The effect of biofuel side

products is considered as well. Distiller’s dried grain and

solubles (DDGS) is a side product of first generation ethanol

production while oil cake is produced in first generation bio-

diesel plants. Both are usually used as animal feed. DDGS

substitutes corn while oil cake is a conventional source of

protein. The side products are an economic asset as they can

be sold on the feed market. Biofuel side products can also

reduce the land use requirements of feed production. We

assume that the DDSG produced from one tonne of ethanol

feedstock substitutes 0.30 tonnes of corn while 0.40 tonnes of

Fig. 3 e Forest wood supply curve (left), supply curve for short r

production in scenarios. Note: The middle (right) picture shows

decrease in the production of the most important food and feed

curves are an agreggate for Austria.

oil cake are produced from one tonne of oil crops [24]. Both

products are sold at respective market prices. We further

assume that one tonne of the side product reduces the land

requirements of biofuel production by 0.07 ha based on the

average Austrian productivity of corn.

2.3. Technologies

We assess several bioenergy technologies which are able to

replace fossil fuels in the transportation sector along with

technologies that convert biomass to heat and power. First

generation biofuels are classified into ethanol produced from

fermentation of starchy and sugar crops (e.g. wheat and corn)

and biodiesel which is produced from vegetable oil derived

from oil crops (e.g. sunflower and rapeseed). Both technolo-

gies are commercially available and are currently used for the

production of biofuels in Austria. Ethanol is blended with

gasoline. A blend of 5% ethanol and 95% gasoline is considered

to be safe for use in all cars, while all cars sold currently on the

market are also able to handle a blend of 10% of ethanol.

Similar limitations apply to biodiesel [25].

Second generation biofuels are able to use cellulosic feed-

stock and even waste for the production of biofuels. There are

two major technological options [4]. The biomass can be

gasified and subsequently upgraded to liquid transportation

fuels such as methanol or synthetic natural gas (SNG) which

can also be used as transportation fuel. The second option is

the hydrolysis of cellulose to sugars that are fermented to

ethanol afterwards. We assess gasification only as it is esti-

mated to be economically more viable than hydrolysis with

fermentation [4],[26]. Second generation production technol-

ogies are currently under research and first pre-commercial

installations are being built. US legislation requires 572 TWh

of annual cellulosic biofuel production until 2022 [27]. There-

fore, a rapid increase can be expected in the construction of

second generation facilities. Currently, cars cannot run solely

on methanol and the amount of methanol that may be

blended to gasoline is, similar to ethanol, limited. SNG

otation coppice (src, middle) and for oil crops and mean

the increase in src (oilcrops) production and the respective

crops if prices for src (oil crops) are increased. All supply




requires significant modifications to the car, including the

installation of a gas tank.

Electric cars are currently globally under research,

however, costs and ranges of batteries are major economic

and technical obstacles to full implementation of the tech-

nology. Ranges of above 150 km are currently only achieved at

very high costs [6]. Also, electric cars need the large scale

deployment of charging stations. Metering and billing of

charging power still has to be developed. Themodel considers

investment costs for electric cars. Costs associated with

additional infrastructure necessary for electric cars are not

included. With respect to power production, the model allows

two technologies: steam engines and biomass integrated

gasification combined cycle (BIGCC) plants. While steam

engines are well established in Austria and the installed

capacity exceeded 300 MW in 2007 [1], BIGCC is a technology

that is still under research. It allows higher electrical conver-

sion efficiencies than steam engines but capital costs are also

significantly higher. We assume that power is either used to

fuel electric cars or is sold on the electricity market at a fixed

price. Heating technologies include fuel wood furnaces, pellet

furnaces and heating plants for district heating networks.

2.3.1. Total cost of ownership e carsThe costs for the various transportation options differ with

respect to the fuel costsbut investment costsandoperationand

maintenance (O&M) costsmay have an important influence on

total costs of operating a car, particularly if the difference in

investment costs between ICE and BEV are considered. We

assess Gasoline, Diesel and Gas ICE and BEV. For cargo trans-

portation, only Diesel ICE is considered. To be able to compare

thedifferent options in theoptimizationmodel, the investment

and O&M costs of all cars have to be determined.

We use the concept of total cost of ownership (tco) to assign

different costs to different cars in the model. Costs for fuels

are endogenously determined by the model and are therefore

not included in the calculations of tco. The tco per km is

described by equations (1)e(3):

tco ¼

�Cþ B

� ið1þ iÞtð1þ iÞt�1km

þ om (1)

t ¼ min

�maxKm

km; 10

�(2)

B ¼Xtb˛y

bc

ð1þ iÞtb(3)

The tco is determined by the annuity of capital costs C of the

car, assuming an interest rate i and a lifetime t. For electric

cars, the battery cost B is additionally considered as explained

below. Total necessary annual investment costs are divided by

the kilometers km driven annually. Technical characteristics,

investment and battery costs for cars are taken from literature

[5],[6],[12]. Learning curve effects are assumed for all cars until

2020, however, declining costs are expected to be highest for

electric cars. O&M costs per km of om are added. These costs

are assumed to be lower for electric cars because the electric

motor needs less maintenance than an internal combustion

engine (ICE) [6]. The lifetime of the car is limited to ten years,

however, if the car is driven a lot (i.e. more than maxKm), the

lifetime is further reduced as indicated by equation (2). The

lifetime of a battery is significantly less than that of the

carriage. A change of the battery within the lifetime is there-

fore probable and is modeled by equation (3): the annuity of

battery costs is derived by adding up the discounted battery

costs over the whole lifetime, assuming that one battery costs

bc. The battery is changed in year y when the driven kilome-

ters since the last change exceed the lifetime of the battery.

The tco depends significantly on the annual transportation

distance per car [25]. A higher distance implies lower specific

capital costs per km. We therefore estimate ten classes of

annual private car utilization (P1-P10) based on data provided

by OAMTC, the biggest Austrian Automobile Association.

OAMTC checks approximately 10% of all Austrian cars for

their technical liability each year. The total driven kilometers

and the year of the first registration of the car are collected in

the examination of the cars. An approximate estimate of the

annually driven kilometers can be derived from this data. We

classified the cars by the driving distance into ten classes (see

Fig. 4). For each class, the mean of the annual transportation

distance and the sum of the transportation distance of all cars

in the class are determined. The sum of driven kilometers is

linearly extrapolated from the OAMTC data with statistics of

total Austrian car ownership from Statistik Austria. This

allows a national estimate of driven kilometers as OAMTC

data only covers around 10% of all registered cars. The

procedure yields a total transportation distance for private

cars of 59.73 billion kilometers driven annually. A study on the

total distance driven in Austria comes to a similar result [28].

We do not compare different technological options for

cargo transportation but assume that ICE trucks fuelled by

diesel are used for transportation. Therefore, only a single

total distance for cargo transportation is assumed (class C1).

Table 1 reports the mean driving distances and the sum of

driving distances for all classes.

2.4. Greenhouse gas emissions

Weaccount for GHGemissions,which are producedwithin the

modeled regionofAustria. The followingdirectGHGemissions

are considered: N2O emissions from fertilizer application in

agriculture, CO2 emissions from biomass and commodity

transportation,CO2emissions fromenergyconsumption in the

biomass conversion process and CO2 emissions of fossil fuel

combustion. GHG emissions from the combustion of biomass

and biofuels are not considered.

The production plants that use lingo-cellulosic feedstock,

i.e. power plants and second generation biofuel plants,

produce the electricity and heat that is required in the

production process internally from the biomass. No CO2

emissions from fossil fuel consumption accrue in the

production process therefore. First generation ethanol and

biodiesel plants, however, have to rely on external energy

sources for electricity and heat. We assume that heat as well

as electricity are produced in a natural gas plant and include

the resulting CO2 emissions in the calculations.

Increased harvests of forest wood have an impact on the

carbon stock in forests. This effect should be considered in



Annual driving distance (km)

Num

ber o

f car

s

0e+00 2e+04 4e+04 6e+04 8e+04 1e+05

050

010

0020

0030

00

P 1 P 2 P 3 P 4 P 5 P 6 P 7 P 8 P 9 P 10

Fig. 4 e Histogram of annual driving distances of cars inspected by OAMTC.


bioenergy studies [29]. However, many factors such as the

timing of harvesting [30], the observed time period [10], the

type and magnitude of management practices [31], the likeli-

hood of non-human disturbances such as forest fires [32], and

the still discussed problem if carbon equilibrium is reached in

old grown forests [33] determine the carbon effect of increased

forest harvesting. Detailed dynamic forest carbon cycle

models are necessary to assess these effects. Such a model is

beyond the scope of the article. Instead, we propose to use

sensitivity analysis to assess the impact of this effect on the

overall GHG emission balance of biofuels. A sensitivity anal-

ysis is also used to assess the impact of indirect GHG emis-

sions from dislocation of agricultural food and feed

production (see section 3.3).

2.5. Demand scenarios

Theheat, thepower, and the transportationsectoraremodeled

in BeWhere. The demand in each category is modeled in

a scenario for 2020. For heat, the spatially explicit heat demand

Table 1 e Classes of transportation distances.

Class Average transportationdistance (km)

Total transportationDistance (106 km)

P1 6555 9130

P2 14,418 32,900

P3 23,477 13,500

P4 33,598 2910

P5 43,840 756

P6 54,080 266

P7 64,610 94

P8 74,790 77

P9 85,599 42

P10 95,497 50

C1 e 24,000a

a Distance of C1 are in 106 tonne-km.

estimation is updated by assuming rates for building retrofit-

ting and for the demolition and new construction of dwellings.

Spatially explicit estimations of population growth by OROK

[34] are used to distribute growth of building areas spatially in

Austria. Due to increased insulation, the heat demanddeclines

significantly. While the model allocates biomass resources to

various conversion routes depending on energy prices and

production costs, the minimum amount of biomass heating is

assumed tonot fall below17TWh.This is apossible declineof 5

TWh from current consumption levels. Setting a lower bound

for biomass consumption for heating is reasonable because

adjustments of individual heating devices to new economic

conditions usually have a time lag.

We assume a linear trend for the increase of power

demand from all sectors besides transportation from 58 TWh

to 68 TWh until 2020. This is a simple linear extrapolation

from the 2000-2010 trend. In 2009, around 41 TWh of power

were produced by big hydro power plants, around 20 TWh by

thermal power plants using fossil fuels and another 3 TWh by

power generating facilities using other renewables i.e. wind,

photovoltaic, small hydro power, geothermal and biomass

[35]. Imported and exported power balance the mismatch of

Austrian demand and supply. It has been estimated that

between 12 and 28 TWh of annual non-biomass renewable

power production (wind, PV, small hydro power) may be

available up to 2020 [36],[37].

This implies that the increase in demand may be entirely

supplied by renewables (under the assumption of a strict

carbon policy) and that thermal power production (currently

19.5 TWh) will not expand significantly, although the inter-

mittency of renewables may imply that a certain share of

thermal power production has to remain in place. We assume

that thermal power production may not fall below 20 TWh

therefore. This amount of power may be supplied by fossil

fuels and biomass.

The model assumes that the additional demand for power

caused by an increase in electric mobility will be exclusively

supplied by electricity production from biomass. This



Table 3 e Main model parameters and uncertaintyranges.

LowerBound

UpperBound

Prices

Price of oil (V MWh�1) 40 60

Price of gas (V MWh�1) 30 50

Price of gasoline (V MWh�1) 42 62

Price of electricity (V MWh�1) 54 74

Price of carbon (V MWh�1 tCO2�1) 21 55

Agricultural prices

(multiplier from baseline)

1 1.4

Plant Characteristics

Fuel Conversion

1st generation Ethanol 0.43 0.46

1st generation Biodiesel 0.62 0.62

2nd generation Methanol 0.54 0.60

2nd generation SNG 0.63 0.70

Power Conversion

CHP-Steam 0.29 0.31

BIGCC 0.41 0.44

Heat Conversion

CHP-Steam 0.47 0.52


assumption is made to make domestic land use effects of

electric mobility and biofuels comparable. It can, however, be

assumed that the assumption is conservative due to two

reasons: (i) additional demand from electric mobility may

partly be in off-peak hours when, at least partly, abundant

non-biomass renewable power supply may be available and

an increase in marginal power production is therefore not

necessary; (ii) the introduction of electric mobility may even

decrease the peak demand in the grid if batteries of electric

cars are used as distributed storage device for power [38].

As outlined in section 2.3.1, current transportation demand

is estimated from OAMTC data. We assume that the demand

for transportation remains constant until 2020. We assume

a total of 60 billion annual kilometers for personal trans-

portation and a total of 24 billion tonne kilometers for cargo

transportation by truck. Although transportation fuel

consumption has historically increased in the last years, the

cause stems mainly from fuel tourism due to lower fuel taxes

in Austria compared to neighboring countries. We exclude

demand from fuel tourism from our analysis and also assume

that public transportation will take a higher share of the

overall transportation supply, thus allowing road trans-

portation to remain constant. Table 2 shows a comparison

between 2008 demand values and the 2020 projections.
BIGCC 0.43 0.47
2nd generation Methanol 0.05 0.11

2nd generation SNG 0.05 0.11

Investment costs (Million V 100 MWbiomass�1 )

1st generation ethanol 35 45

1st generation biodiesel 9 18

2nd generation methanol 85 110

2nd generation SNG 55 70

CHP-Steam 50 60

BIGCC 65 95

Performance data cars

Costs

Battery costs/car (1000 V) 4.00 6.50

Investment costs electric cars

(w/o battery) (1000 V)

14.00 16.50

Investment costs gasoline cars (1000 V) 16.00 16.00

Investment costs diesel cars (1000 V) 17.00 17.00

Investment costs gas cars (1000 V) 17.50 17.50

O&M costs electric car (V km�1) 0.02 0.025

Technical characteristics

Replacement distance battery (1000 km) 70 90

Conversion Gasoline (1000 km MWhfuel�1 ) 2.00 2.20

Conversion Diesel (1000 km MWhfuel�1 ) 2.25 2.45

2.6. Policy scenarios

We model one baseline scenario that assumes no policy

intervention at all, and seven policy scenarios. Three of the

scenarios assume that 5% (S5), 10% (S10) and 15% (S15) of the

transportation sector are supplied by bioenergy, allowing all

technologies to be selected by the model. These scenarios are

chosen, because a blending share of 5.75% is in place in

Austria while the increase of the share to 10% is planned until

2020. The other four scenarios examine the impact of a 10%

target of renewable transportation fuels, if only single tech-

nologies (i.e. first generation ethanol (eth), second generation

methanol (met), second generation sng (sng), electric mobility

(emo)) are allowed. First generation biodiesel is not modeled

in a single technology scenario because domestic feedstock

production is too low to supply 10% of the transportation

sector with biofuels (see Fig. 3). There are no additional policy

interventions such as technology subsidies included except

of a fixed share of renewable transportation and a CO2 price

(see Table 3 for levels). All scenarios are analysed for the year

2020.

Table 2 e Historic and projected demands for electricity,heat and transportation in 2008 and 2020.

2008 2020

Electricity demand (TWh) 58.00 68.00

Water power 41.00 41.00

Thermal power (biomass & fuel) 19.50 20.00

Renewable without biomass 3.00 17.00

Heat demand (TWh) 71.50 61.00

Personal transportation demand (km) 60E109 60E109

Cargo transportation demand (km) 24E109 24E109

Conversion electricity (1000 km MWhelec�1 ) 5.60 7.00

2.7. Uncertainty

Uncertainties on the performance and costs of various tech-

nologies as well as uncertainty about future energy prices

remain high. We explicitly address this issue by performing

Monte-Carlo simulations of the MIP model and conducting an

extensive sensitivity analysis. We first define plausible ranges

for the uncertain parameters reviewing literature and assume

that the parameters are distributed uniformly within that




range. For energy and CO2 prices, correlation between the

prices of oil, gas, gasoline and CO2 are determined from

historical spot prices. The input data for the Monte-Carlo

simulation is generated by performing a Latin Hypercube

Sampling procedure and combining it with the Iman-Conover

method to guarantee correlation of correlated parameters in

the procedure [39]. Latin Hypercube Sampling is used to

guarantee that the whole parameter range is covered in the

Monte-Carlo simulations. Results are given in form of proba-

bility distributions and a stepwise regression analysis of

model inputs on model outputs is performed to examine the

sensitivity of results to input parameters. The assumption on

the distribution of the stochastically modeled parameters is

reported in Table 3.

S5 S10 S15 eth met sng emo

02

46

8

1st Gen Ethanol

Etha

nol (

billio

n km

)


02

46

810

12

2nd Gen Methanol

Met

hano

l (bi

llion

km)


−30

−20

−10

010

E−Mobility

E−M

obilit

y(bi

llion

km)


−8−6

−4−2

02

Private Heat

Hea

t (TW

h)

Fig. 5 e Boxplots of differences in utilization of biomass conver

scenarios in the year 2020. Note: The gray bars show the 99% c

3. Results

3.1. Technologies

The cost-minimal set of technologies for attaining renewable

energy targets in the transportation sector are determined in

scenarios S5, S10 and S15 for three levels of renewable energy

shares in transportation (5%, 10%and15%, respectively) for the

year 2020. In all scenarios, a mix of first generation biodiesel,

second generation methanol, and a small amount of electric

mobility is chosen. Fig. 5 shows the difference in the produc-

tionof fuels, heat and electricity between the baseline scenario

and the policy scenarios. Biodiesel is cost-competitive with


01

23

4

1st Gen Biodiesel

Biod

iese

l(billi

on k

m)


02

46

8

2nd Gen SNG

SNG

(billi

on k

m)


−4−3

−2−1

01

2

Power

Pow

er (T

Wh)


−8−6

−4−2

0

Biomass Power and Heat Production

Pow

er a

nd H

eat(T

Wh)

sion technologies between the baseline and the policy

onfidence interval of the mean.




second generation biofuels. However, the amount of domestic

production is limited by the availability of oil crops (see section

2.2). Second generation methanol is the cheapest option for

substituting gasoline and outperformsfirst generation ethanol

in terms of costs. Second generation SNG is not deployed

although fuel production costs are lower than for methanol.

However, additional investment costs for cars accrue if SNG is

used in ICE. A small amount of electric mobility is chosen,

further expansion of electric mobility is prohibited by car

investment costs. Fig. 6 shows which transportation distance

classes are supplied by the various transportation fuel options.

SNG is chosen for classes with high transportation distances.

This is a consequence of tco: the higher investment costs for

cars that are fuelled by SNG in relation to ICE matter less in

classes with higher annual transportation distances because

Fig. 6 e Percentage of transportation distance class supplied

the share that investment costs take on total costs, including

fuels, declines. Interestingly, electricmobility is deployed in all

driving classes. Investment costs for the car without battery

are comparable to ICE cars. However, investment costs for

batteriesdependmainlyon thedrivingdistanceas theyhave to

be changed after a certain amount of charging cycles. Higher

total driving distances therefore also imply higher costs for

batteries. A relative advantage in relation to ICE does therefore

not occur for higher driving classes.

When renewable transportation targets are introduced, the

production levels of electricity and heat from biomass are

reduced because of increased competition for biomass. Higher

shares of renewable energy in the transportation sector cause

a declineof biomasspowerandheat production.A comparison

of S5, S10, and S15 in Fig. 5 allows drawing this conclusion.

by renewable fuels (Mean of Monte-Carlo simulations).




A comparison of the single technology scenarios shows

that electric mobility and ethanol production have the lowest

impact on the production of biomass power and heat. Electric

mobility uses little biomass resources in comparison to the

other conversion chains due to the high conversion efficiency

of electric cars. The conversion efficiency of ethanol along the

whole production chain is lower than for any other trans-

portation option. However, first generation ethanol produc-

tion relies on agricultural feedstock. Therefore, feedstock

competition with bioenergy conversion chains that use lingo-

cellulosic biomass is lower than for the second generation

biofuels that mainly use forestry products. For the ethanol

case, the competition is limited to agricultural land that may

be used to produce ligno-cellulosic biomass. In eth, competi-

tion for land increases if CO2 prices increase because addi-

tional agricultural land is used for the production of lingo-

cellulosic biomass for power and heat production.

The spread in non-transportation bioenergy production

between the baseline scenario and the ethanol scenario

therefore increases with rising CO2 and energy prices: while

agricultural land is used for the production of lingo-cellulosic

biomass for power and heat production in the baseline

scenario, the land is already used for ethanol production in eth.

This effect is less pronounced for the second generation fuel

scenarios because feedstock competition is high even in the

lowCO2 price scenarios. The higher variance of non-bioenergy

production in eth compared tomethand sng canbeexplainedby

this effect. The error bars included in thefigures allowdeciding

if significant differences can be identified between themeanof

the distributions of results. With the exemption of meth and

sng, the mean results of all scenarios differs significantly from

the other scenarios. However, due to the use of similar feed-

stock and similar conversion efficiencies, the results for meth

and sng do not differ significantly with respect to the deploy-

ment of power and heat technologies.

3.2. GHG emissions, land use change and energysystem costs

Fig. 7 shows the difference in GHG emissions, fossil fuel

substitution, and costs between the baseline scenario and the

policy scenarios. With the exemption of emo, mean domestic

GHG emissions increase, i.e. the introduction of renewable

transportation policies increases GHG emissions, even

Fig. 7 e Boxplot of differences in GHG emissions, fossil fuel sub

scenarios. Note: The gray bars show the 99% confidence interva

without the consideration of indirect GHG emissions. The

increased GHG emissions accrue partly in the biofuel

production process and partly as consequence of reducing

biomass utilization in electricity and heat production. With

the exemption of emo, no significant differences in the mean

of GHG emissions can be identified between the scenarios as

indicated by the error bars. Uncertainty is considerable,

particularly in eth. The high variation in bioenergy deploy-

ment results also in a high variation in GHG emissions. The

same applies to fossil fuel substitution. Although GHG emis-

sions and fossil fuel substitution of electric mobility are

promising, Fig. 7 explains why a large scale deployment of

electric mobility may not be expected by 2020: the energy

system costs are significantly higher than the energy system

costs of any alternative scenario.

A detailed look at the GHG emissions (Fig. 8) in the various

conversionchainsshows thatfirst generationethanolproduces

high GHG emissions in the conversion process due to external

energy needs. The other technologies do not need external

inputsbecause process heat andpowerare internally produced

from biomass. Fig. 8 shows that GHG emissions from the

transportation of biomass and biofuel are very small in

comparison to other GHG emissions in the supply chain. Mean

transportation distance of biomass is 45 km. Differences in the

GHG emissions produced in agriculture are of minor impor-

tance. Increasing crop production for first generation ethanol

leads to intensification of agricultural land uses. Therefore

slightly more nitrogen fertilizer is applied than in the case

without biofuel production and N2O emissions increase.

However, the total amount of additional GHG emissions is low.

Productionof short rotationcoppiceonagricultural landcauses

fewer GHG emissions than the production of the substituted

crops because less application of nitrogen fertilizer is neces-

sary.The landuseeffectsarehigher iffirst generationethanol is

produced while the land use effects of second generation fuels

are lowerandofelectricmobilityare lowest (seeFig. 9) although

no significant difference in themean of land use change can be

detected for scenarios S5, S10, S15, meth and sng. There are two

simple reasons for the high land use of ethanol:

(I) Fig. 10 shows the total amount of biomass that is

necessary for biofuel production. It is lowest for electric

mobility and highest for ethanol. Low conversion efficiencies

of the first generation ethanol production process in

comparison to second generation biofuel production and low

stitution and costs between baseline scenario and policy

l of the mean.



Fig. 8 e Difference in GHG emissions between the baseline and the technology scenarios for different parts of the supply

chain. Note: Total is the sum of all four subcategories (transport, conversion, agriculture, fossil substitution).


conversion efficiencies of the ICE engine in comparison to an

electric engine are the reason for the high amount of biomass

requirements.

(II) First generation ethanol production is in need of crops

produced on agricultural land and cannot use forestry

resources. Therefore, agricultural land for energy crop

production increases if ethanol production is expanded. There

is also a positive land use effect of ethanol production: the by-

product can reduce the need for feed production. By-products

are however not able to offset the total increase in land use

effect caused by ethanol production as shown in Fig. 10.

3.3. Sensitivity of GHG emissions to land use change andcarbon sequestration in Forests

The results on land use and on forest wood utilization can be

used to test the influence of assumptions on indirect land use

GHG emissions and on forest carbon stocks on total GHG

emission performance in the biofuel scenarios.


-1e+

050e

+00

1e+0

52e

+05

Aggregated Energy LandΔ

Δ Ar

ea (h

a)

Fig. 9 e Boxplots of differences in land uses between baseline an

for the production of bioenergy crops, right: changes in the amo

Note: The gray bars show the 99% confidence interval of the m

(I) Additional GHG emissions may arise because the

substituted food or feed crop has to be grown in other

areas. GHG emissions may stem from the conversion of

natural land to agricultural crop areas, and they may also

be created in the agricultural production process (e.g.

fertilizer application). GHG emissions of around 0.3 tCO2e

tproduct�1 are the standard value of corn and wheat

production when applying the standard values for agri-

cultural GHG emissions that are reported in the EU

directive 2009/28/EC. Those values take into account GHG

emissions caused by agricultural inputs (i.e. fertilizer,

seeds, agro-chemicals) but do not take into account any

additional land use change due to increased agricultural

production.

(II) The long term carbon stock in forests may be reduced if

additional wood is harvested. This is the carbon oppor-

tunity cost of increasing bioenergy production from

forests (see section 2.4 for details). We test the sensitivity

of the results to varying assumptions for forest carbon

Δ

Δ


-250

000

-150

000

-500

000

5000

0

Aggregated Food and Feed Land

Are

a (h

a)

d policy scenarios, left: changes in the amount of land used

unt of land used for the production of food and feed crops.

ean.



Fig. 10 e Boxplots of differences between baseline and policy scenarios in the use of biofuel feedstock (left), forest wood

utilization (middle) and the effect of by-products on land substitution (right). Note: The gray bars show the 99% confidence

interval of the mean.


stock changes. The carbon content of wood is estimated

to be around 0.09 tCarbon MWh�1 (5.3 MWh

tonneHHVrywood�1 , around 47% of carbon in wood on weight

basis [40]) which corresponds to around 0.3 tCO2). This is

an indication of how much carbon is removed from

forests when harvesting wood. The effect of harvests on

subsequent carbon sequestration of forests is not

included in this figure.

We calculated the mean net GHG emission effect (see

Fig. 7), the mean amount of substituted agricultural products,

and the mean amount of additional wood harvests from

forests (see Fig. 10). Fig. 11 shows the results of (i) increasing

GHG emissions per tonne of substituted agricultural

product (from 0 to 0.5 tCO2 tProduct�1), taking into consid-

eration the by-products of first generation ethanol and bio-

diesel production, (ii) decreasing carbon sequestration in the

forests if forest wood harvests are increased (from 0 to

0.5 tCO2 MWhwood�1), and (iii) changing the two

simultaneously.

0.0 0.1 0.2 0.3 0.4 0.5

−0.4

−0.2

0.0

0.2

0.4

0.6

0.8

GHG Emissions (tCO2) tAgriculturalProduct−1

ethS10methsngemo

0.0 0.1 0.2

−0.4

−0.2

0.0

0.2

0.4

0.6

0.8

Loss in carbon sequ

GH

G E

mis

sion

s (M

tCO

2e)

GH

G E

mis

sion

s (M

tCO

2e)

Fig. 11 e Effects of iLuc GHG emissions (left), forest carbon stoc

emissions among policy scenarios.

While no significant changes in GHG emissions can be

detected for electric mobility, GHG emissions from the first

generation ethanol process increase significantly if additional

GHG emissions from agriculture are taken into account.

Second generation fuels are mainly influenced by changes in

the assumptions on forest carbon sequestration. The mixture

of first and second generation fuels and electric mobility in

scenario S10 causes a relatively low increase in total GHG

emissions when assumptions are changed. In any case, net

GHG emissions of all biofuel conversion chains are higher

than GHG emissions in the reference scenario without bio-

fuels. Only electric mobility may reduce total GHG emissions.

3.4. Sensitivity analysis

The influence of model parameter variation on total output is

assessed in the sensitivity analysis. For that purpose, the

output of the Monte-Carlo simulations of S10 is used to build

a linear meta model by means of regression analysis, i.e. the

variation in model outputs is explained by a linear

0.3 0.4 0.5estration tCO2MWh

wood

−10.0 0.1 0.2 0.3 0.4 0.5

−0.4

−0.2

0.0

0.2

0.4

0.6

0.8

Agricultural GHG Emissions increase / Carbon Sequestration decrease

GH

G E

mis

sion

s (M

tCO

2e)

k changes (middle), and both (right) on total net GHG



Table 4 e Results of the sensitivity analysis for GHGemissions and the land used for energy crop production.

Parameter Coefsa p-valueb Coefsa p-valueb

GHG Emissions Land use

Carbon price �0.40 0.000 0.41 0.000

Gas price e e 0.22 0.001

Oil price �0.36 0.000 0.26 0.000

Power price �0.27 0.000 e e

n 200 200

R2 0.78 0.64

a Standardized coefficients of the regression.

b p-value of regression coefficient: probability that coefficient is

not different from 0.


combination of model parameters. S10 was chosen because

the scenario allows free choice of transportation technologies.

In that way, the influence of parameters on the choice

Table 5 e Results of the sensitivity analysis for the deployment of second generation methanol and electric mobility.

Parameter Coefsa p-valueb Coefsa p-valueb

Deployment of methanol Deployment of electric mobility

Carbon price e e 0.12 0.019

Electric car, investment costs 0.43 0.000 �0.40 0.000

Battery electric car, replacement costs 0.56 0.000 �0.53 0.000

Gasoline price 0.17 0.000 e e

Electric car, operation & management costs 0.13 0.008 e e

Battery electric car, distance until replacement �0.10 0.04 0.15 0.006

Conversion efficiency BIGCC plant - heat 0.09 0.046 e e

Conversion efficiency SNG plant e SNG �0.09 0.048 e e

Investment costs methanol plant �0.09 0.046 e e

n 200 200

R2 0.58 0.47

a Standardized parameters of the regressions.

b p-value of regression coefficient: probability that coefficient is not different from 0.

between biofuels and electric mobility can be assessed

directly. The scenario is also in line with the official 10%

renewable transportation target. Table 4 shows the results of

the analysis for GHG emissions and land use by energy crop

production. The carbon and energy prices have the most

important influence on total output. They are positively

correlated with land use and negatively with GHG emissions.

We also assessed the competition between second gener-

ation fuels and electric mobility with help of the sensitivity

analysis. The results are reported in Table 5. The amount of

second generationmethanol versus electricmobility ismainly

determined by technical parameters of electric cars e costs of

the electric car are positively correlated with the amount of

methanol deployed and negatively with electric mobility. The

amount of methanol production is also reduced by increasing

investment prices formethanol plantswhile little competition

with second generation SNG production exists, implied by the

negative sign of the coefficient of SNG conversion efficiency.

The CO2 price has no influence on the deployment of meth-

anol production, but electric mobility is positively correlated.

The results of the sensitivity analysis show clearly that the

performance and costs of the battery impede the large scale

deployment of electric cars.

The carbonprice ismore important for electric cars than for

second generation methanol, because electric cars reduce

significantlymoreGHGemission thansecondgeneration fuels.

4. Discussion and conclusions

4.1. Discussion

Themodel results of our analysis are in line with other studies

that estimate lower land use impacts for electric cars fuelled

by bioelectricity than for ICE fuelled by biofuels [11]. They are

also in line with studies that come to the conclusion that

battery replacement costs are currently the biggest economic

barrier to the large scale introduction of electric mobility in

the transportation sector [5],[6].

We assume that electricity for electric cars is produced in

biomass plants. This assumption was made to be able to

compare land use changes implied by different transportation

options. In reality, end consumers rely on grid electricity that

comes fromdifferent power production technologies. This has

consequences for theprice the consumerpays: thepowerprice

may be higher or lower than power production from biomass.

Additionally, the possibility to use the battery of the electric

car as energy storage for the grid operator may actually allow

negative costs, depending on the charging profile. Greenhouse

gas emissions of electricmobility are also highly influenced by

the assumptions on marginal electricity and on the balancing

effect of distributed storage in car batteries. With respect to

costs, the choice of biomass power production technology

causes that costs for electricity are likely overestimated,

therefore decreasing the competitiveness of electric cars

slightly. Even if tco of electric cars are lower than for ICE,

specific technical constraints of electric cars may be a barrier

for the large scale introduction: in fact, charging times and

range may be factors that have a much larger influence on

consumer’s decision than simple tco considerations.




The discussion on the net GHG emission effect of biofuels

and bioenergy is manifold. Quantitative assessments of the

effects need global models to allow estimating the global

consequences of local agricultural and energy policies on land

use. When relying on increasing forest wood harvests,

detailed models of forest carbon stock changes due to har-

vesting are necessary. Large uncertainties are associated with

indirect land use changemodels [41] and carbon sequestration

in forests, which is usually determined by various factors as

outlined in section 2.4. We opted for a different assessment of

the net GHG emissions that can be locally applied: we deter-

mine the amount of additional forest wood harvests and

which food and feed crops are likely to be substituted by

energy crop production. In a sensitivity analysis, we assess the

influence of associating GHG emission factors with extended

forest wood harvests and substituted agricultural crops,

assuming that they have to be, at least partly, produced

elsewhere. The analysis shows that even conservative

assumptions that do not account for land use change related

GHG emissions increase net GHG emissions of biofuel

production significantly.

The applied model is static. The diffusion of new technol-

ogies is therefore not assessed dynamically and assumptions

have to be made on how currently employed energy technol-

ogies are substituted bynewbioenergy technologies.We chose

to allow complete substitution in all sectorswith exemption of

the private heating sector. Assuming a lower adoption rate of

new technologies would further increase the pressure on

extending the production of energy crops and on increasing

forest harvests because biomass resources could not be devi-

ated from existing bioenergy uses to new ones in that case.

Assessing future technological development is inherently

uncertain. The applied uncertainty analysis allows showing

the range of possible outcomes, based on current knowledge

about the technologies as well as applying a global sensitivity

analysis. This approach allows identifying significant differ-

ences between technology scenarios even if uncertainty in

input parameters is high but also indicates that the choice

between some technologies may not be possible yet because

the available information is not sufficient.

4.2. Conclusions

We have presented a spatially explicit energy model for the

assessment of renewable transportation options from bio-

energy production. An agricultural production model is

coupled to an energy system model to allow assessing the

effects of bioenergy policies on land use change and the

production of food and feed crops. Prices and future devel-

opment of technologies are inherently uncertain. These

uncertainties are explicitly addressed by conducting Monte-

Carlo simulations.

The model results show substantial differences when

comparing different technology groups, i.e. first generation

biofuels with second generation biofuels and with electric

mobility: second generation biofuels have less impact on

agricultural land use than first generation ethanol due to two

reasons: yields of biofuels per hectare are higher for agri-

cultural land and the feedstock may additionally come from

forests. First generation biodiesel has relatively high yields

per hectare, but the total domestic potential is limited at

a low level. The lowest land use impacts are implied by the

utilization of electric cars, which, at currently expected

technological developments, are still very costly in compar-

ison to cars fuelled by liquid fuels in the year 2020. Results

also indicate that the net GHG emissions increase if biofuels

are deployed due to substitution effects within the energy

sector, i.e. power and heat production is substituted by bio-

fuel production. Only electric mobility can reduce GHG

emissions in comparison to fossil fuelled transportation. Due

to the high uncertainty ranges of the results, no significant

conclusions can be drawn about the comparison of second

generation biofuels, i.e. SNG and methanol, with respect to

land use change, greenhouse gas emissions and fossil fuel

substitution. However, costs of SNG seem to be considerable

higher due to additional investment costs in cars. The net

GHG emissions of first generation ethanol versus second

generation fuel production depend on changes in the carbon

sequestration of forests due to additional forest harvests and

on GHG emissions from additional agricultural production

caused by the substitution of food and feed crops by energy

crops. Electric mobility is almost not affected by the choice of

these parameters.

With respect to policies for promoting second generation

biofuel production, it has to be considered that investments in

second generation biofuel production will have a long-term

effect on the utilization of biomass resources. The results of

the study indicate, however, that the gains in efficiency in

relation to first generation fuels are relatively low while signif-

icant efficiency increases can only be expected when devel-

oping a transportation systembasedonelectricity.A large scale

introductionof secondgeneration biofuels has tobe considered

very carefully as well as in the light of a possible total restruc-

turing of the transportation sectorwithin the next 20e30 years.

Acknowledgements

This article has been supported by the “Energiesysteme der

Zukunft” project “Energieversorgung aus Land- und For-

stwirtschaft in Osterreich unter Berucksichtigung des Klima-

und Globalen Wandels in 2020 und 2040 (Energ.Clim)” and by

the FP7 project “Climate Change e Terrestrial Adaptation &

Mitigation in Europe (ccTame)” funded by the European

Commission. We want to thank the team of the Vienna

Scientific Cluster (VSC) computer grid for the support when

running the model scenarios on the VSC. Comments received

from an anonymous referee are gratefully acknowledged.

r e f e r e n c e s

[1] Kranzl L, Haas R. Strategien zur optimalen Erschließung derBiomassepotenziale in Osterreich bis zum Jahr 2050 mit demZiel einer maximalen Reduktion an Treibhausgasemissionen[Strategies for the optimal development of biomasspotentials in Austria until the year 2050 aiming at a maximalreduction of green house gas emissions]. Vienna: EnergyEconomics Group, University of Technology; 2008.




[2] Winter R. Biokraftstoffe im Verkehrssektor in Osterreich2008 (Biofuels in the transportation sector in Austria 2008).Umweltbundesamt; 2010.

[3] Havlık P, Schneider UA, Schmid E, Bottcher H, Fritz S,Skalsky R, et al. Global land-use implications of first andsecond generation biofuel targets. Energ Policy; 2010. doi:10.1016/j.enpol.2010.03.030.

[4] Lange J- P. Lignocellulose conversion: an introduction tochemistry, process and economics. Biofuels, Bioprod andBiorefin 2007;1:39e48.

[5] Sandy Thomas CE. Transportation options in a carbon-constrained world: Hybrids, plug-in hybrids, biofuels, fuelcell electric vehicles, and battery electric vehicles. IntJ Hydrogen Energy 2009;34:9279e96.

[6] Thiel C, Perujo A, Mercier A. Cost and CO2 aspects of futurevehicle options in Europe under new energy policy scenarios.Energ Policy;In Press, Corrected Proof.

[7] Searchinger T, Heimlich R, Houghton RA, Dong F, Elobeid A,Faboisa J. Use of U.S. croplands for biofuels increasesgreenhouse gases through emissions from land use change.Science 2008;319:1238e40.

[8] Bryan BA, King D, Wang E. Biofuels agriculture: landscape-scale trade-offs between fuel, economics, carbon, energy,food, and fiber. GCB Bioenergy 2010;2:330e45.

[9] Dauvergne P, Neville KJ. Forests, food, and fuel in the tropics:the uneven social and ecological consequences of theemerging political economy of biofuels. J Peasant Stud 2010;37:631.

[10] Reijnders L, Huijbregts MAJ. Choices in calculating life cycleemissions of carbon containing gases associated with forestderived biofuels. J Clean Prod 2003;11:527e32.

[11] Campbell JE, Lobell DB, Field CB. Greater transportationenergy and GHG offsets from bioelectricity than ethanol.Science 2009;324:1055e7.

[12] Steenhof PA, McInnis BC. A comparison of alternativetechnologies to de-carbonize Canada’s passengertransportation sector. Technol Forecast and Soc Change2008;75:1260e78.

[13] van Vliet OPR, Kruithof T, Turkenburg WC, Faaij APC.Techno-economic comparison of series hybrid, plug-inhybrid, fuel cell and regular cars. J Power Sources 2010;195:6570e85.

[14] Schonhart M, Schmid E, Schneider UA. CropRota - A croprotation model to support integrated land use assessments.Eur J Agron 2011;34:263e77.

[15] Izaurralde RC, Williams JR, McGill WB, Rosenberg NJ,Jakas MCQ. Simulating soil C dynamics with EPIC: modeldescription and testing against long-term data. Ecol Model2006;192:362e84.

[16] Schmid E, Sinabell F. On the choice of farm managementpractices after the reform of the Common Agricultural Policyin 2003. J Environ Manage 2007;82:332e40.

[17] Schwarzbauer P. Austria, in Demand and supply analyses ofroundwood and forest productsmarkets in Europe, B. Solberg,Ed.Helsinki: Proceedingsof the1stWorkshopof theConcertedAction Project AIR3-CT942288, 1997.

[18] Schmidt J, Leduc S, Dotzauer E, Schmid E. Cost-effectivepolicy instruments for greenhouse gas emission reductionand fossil fuel substitution through bioenergy production inAustria. Energ Policy 2011;39:3261e80.

[19] Statistik Austria. Buildings- and dwellings census 2001: Mainresults; 2004. Gebaude- und Wohnungszahlung 2001:Hauptergebnisse). Statistik Austria.

[20] Umweltbundesamt Austria. Austria’s National InventoryReport 2010. Austria: Umweltbundesamt; 2010.

[21] Williams JR. The EPIC model. In: Singh VP, editor. Computermodels of watershed hydrology. Highlands Ranch, Colorado:Water Resources Publications; 1995. p. 909e1000.

[22] Statistik Austria. Price of products in agriculture and forestry2003-2009 (Land und forstwirtschaftliche Erzeugerpreise2003-2009). Statistik Austria; 2010.

[23] OECD. Oecd-Fao Agricultural Outlook 2009. OECD Publishing;2009.

[24] Ozdemir ED, Hardtlein M, Eltrop L. Land substitution effectsof biofuel side products and implications on the land arearequirement for EU 2020 biofuel targets. Energy Policy 2009;37:2986e96.

[25] van Vliet O, van den Broek M, Turkenburg W, Faaij APC.Combining hybrid cars and synthetic fuels with electricitygeneration and carbon capture and storage. Energy Policy;2011:31.

[26] Bram S, De Ruyck J, Lavric D. Using biomass: a systemperturbation analysis. Appl Energy 2009;86:194e201.

[27] Environmental Protection Agency. Regulation of fuels andfuel Additives: changes to renewable fuel standard program.FR 2010.

[28] Herry M, Sedlacek N, Steinacher I. Traffic in Numbers(Verkehr in Zahlen). Bundesministerium fur Verkehr.Innovation und Technologie; 2007.

[29] McKechnie J, Colombo S, Chen J, Mabee W, MacLean HL.Forest bioenergy or forest carbon? Assessing trade-offs ingreenhouse gas mitigation with wood-based fuels. EnvironSci Technol 2011;45:789e95.

[30] Kirschbaum M.U.F. To sink or burn? A discussion of thepotential contributions of forests to greenhouse gas balancesthrough storing carbon or providing biofuels. Biomass andBioenergy;24:297e310.

[31] Newsletter 4 Jandl R, Rasmussen K, Tome M, Johnson DW.Issue 4. ForestManagement andCarbon Sequestration. IUFROe The Global Network for Forest Science Cooperation; 2006.

[32] Gough CM, Vogel CS, Schmid HP, Curtis PS. Controls onannual forest carbon storage: Lessons from the past andPredictions for the future. BioSc 2008;58:609e22.

[33] Luyssaert S, Schule ED, Borner A, Knohl A, Hessenmoller D,Law BE, et al. Old-growth forests as global carbon sinks.Nature 2008;455:213e5.

[34] OROKSchule ED, Borner A, Knohl A, Hessenmoller D, Law BE.(Szenarien der Raumentwicklung Osterreichs 2030[Scenarios for the spatial and regional development ofAustria in the European context]. RegionaleHerausforderungen und Handlungsstrategien, Wien 2009, III.Osterreichische Raumordnungskonferenz (OROK); 2009.

[35] E-Control. Time series of power production (Jahresreihen:Gesamte Elektrizitatsversorgung Aufbringung). E-Control;2009.

[36] Hinterberger F, Stocker A, Bohunovsky L, Kowalski K.Erneuerbare Energien in Osterreich: Modellierung moglicherEntwicklungsszenarien bis 2020 [Renewable Energies inAustria: Modeling of possible development scenarios until2020]; 2008.

[37] Prinz T, Biberacher M, Gadocha S, Mittlbock M, Schardinger I,ZocherD.EnergieundRaumentwicklung- raumlichePotenzialeerneuerbarer Energietrager [Energy and spatial development -spatial potentials of renewable energies]. OROK; 2009.

[38] Hartmann N, Ozdemir ED. Impact of different utilizationscenarios of electric vehicles on the German grid in 2030.J Power Sources 2011;196:2311e8.

[39] Helton JC, Davis FJ. Sampling-based methods. In: Saltelli A,Chan K, Scott ME, editors. Sensitivity analysis. John Wiley &Sons Ltd; 2001.

[40] Lamlom SH, Savidge RA. A reassessment of carbon contentin wood: variation within and between 41 North Americanspecies. Biomass Bioenerg 2003;25:381e8.

[41] Edwards R, Mulligan D, Marelli L. Indirect Land Use Changefrom increased biofuels demand. Joint Research Center,Institute of Energy.



Land use changes, greenhouse gas emissions and fossil fuel substitution of biofuels compared to...

Documents

Transcript of Land use changes, greenhouse gas emissions and fossil fuel substitution of biofuels compared to...