Energy 2013 - Energy Supply Infrastructure LCA Model For Electric and Hydrogen Transportation Systems

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 11

Energy xxx (2013) 1e11

Contents lists available at SciVerse ScienceDirect

Energy
journal homepage: www.elsevier.com/locate/energy

Energy supply infrastructure LCA model for electric and hydrogen


transportation systems
Alexandre Lucas a, *, Rui Costa Neto b, Carla Alexandra Silva b
a
MIT Portugal Program, IST-Technical University of Lisbon, Avenida Prof. Cavaco Silva, Campus IST, TagusPark, Room 16.6, 2780-990 Porto Salvo, Portugal
b
Institute of Mechanical Engineering, IST-Technical University of Lisbon, Av. Rovisco Pais 1, Pav. Mec. I, 2 andar, 1049-001 Lisboa, Portugal

a r t i c l e i n f o a b s t r a c t

Article history: Many transportation environmental life cycle analyses neglect the contribution of the energy supply
Received 11 October 2012 infrastructures. In alternative light duty vehicle technologies, it has been shown through case studies
Received in revised form that this can be a relevant factor. However, no model that can generalise the evaluation of energy and
22 March 2013
emissions from construction, maintenance and decommissioning of such infrastructure to analyse
Accepted 24 April 2013
Available online xxx
different scenarios currently exists. A model is proposed, focussing on electricity and on hydrogen supply
through centralised steam methane reforming (H2(a)) and on-site electrolysis (H2(b)). The model outputs
are in gCO2eq/MJ and MJeq/MJ of the final energy. Model main inputs are the region’s electricity mix, the
Keywords:
Hydrogen
annual distance driven, supply chain losses and the number of vehicles per station or chargers. The
Electric vehicles evaluation of the number of vehicles served per each charger/station as a function of annual distance
Life cycle analysis driven is presented. The uncertainty is estimated by using the pedigree matrix, impact uncertainty and
Infrastructure literature estimates. The model shows consistency in the results and uncertainty range. Charging policies
Uncertainty that minimise the electricity infrastructure burden should incentivise approximately 37% of normal
charging. H2(a) pipeline lifetime should be extended. Efforts in the electrolyser should be undertaken to
approximate the ratio of vehicles per station with a conventional one.
Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction and pathways are also well covered in the literature. Gross et al.
(2007) [4] estimates future costs of hydrogen fuel and the scale of
In the road transportation sector, the Life Cycle Analysis (LCA) distribution facilities. They demonstrate that a hydrogen infra-
cradle-to-grave (CTG) concept is usually separated into a Well-to- structure that could support large-scale deployment of fuel cell
Wheel (WTW) fuel analysis and a materials cycle analysis. How- electric vehicles could be commercially viable. Huang et al. (2006)
ever, the infrastructure required to manufacture, transport, and [5] simulated ten hydrogen pathways using petroleum-based
distribute the actual fuel has been left unattended. Tank-to-Wheel naphtha, natural gas (NG), electricity and coal as feedstock. They
(TTW) reference studies by Weiss et al. (2000, 2006) [1,2] present concluded that all pathways have significant reductions in WTW
the expected energy use and emissions for 2020 from passenger car petroleum use except the H2 pathways from naphtha.
transportation. The studies cover pure combustion (gasoline, The final report of the Hysociety Project [6] shows an extensive
diesel) engines (ICE), combustion hybrids, fuel cell hybrid electric amount of research work related to LCA studies of hydrogen applied
vehicles (FCHEV), and battery-powered electric cars (EV). Although to the transportation sector. No study includes the materials for the
analysed over different time horizons, both electricity and energy supply infrastructure, including final refuelling stations or
hydrogen are regarded as important alternatives to mitigate envi- electricity chargers. However, as addressed by Frischknecht et al.
ronmental concerns. Edwards et al. (2011) [3] presents a compre- (2007) [7], depending on the sector, capital goods or infrastructure
hensive report of WTW cycles for several hydrogen, electricity, and may have a relevant impact on the total environmental burden. In
conventional fuels pathways, as well as local or centralised pro- previous studies, Lucas et al. (2012) [8,9] analysed the energy
duction philosophies. Studies for hydrogen fuelling infrastructure supply infrastructure component separately and included it in the
vehicle LCA. Research findings report that this stage may have a
total LCA contribution as high as 12%.
* Corresponding author. Tel.: þ351 961741327.
The environmental impacts of power plants have been studied
E-mail addresses: alexandre.lucas@ist.utl.pt, Alexandre.Lucas@efacec.com in several contexts [10e13], and, although for conventional tech-
(A. Lucas), costaneto@ist.utl.pt (R.C. Neto), carla.silva@ist.utl.pt (C.A. Silva). nologies a consensus has been reached, the uncertainty is still very

0360-5442/$ e see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.energy.2013.04.056

Please cite this article in press as: Lucas A, et al., Energy supply infrastructure LCA model for electric and hydrogen transportation systems,
Energy (2013), http://dx.doi.org/10.1016/j.energy.2013.04.056
2 A. Lucas et al. / Energy xxx (2013) 1e11

high in other technologies (e.g., renewable sources and nuclear). As part of the feedstock logistics infrastructure, the NG pipeline
There are several main reasons for this uncertainty: the variety in was included in the Natural Gas Combined Cycle (NGCC) plants. All
the technologies used, location, scale, plant’s lifetime, capacity other types of feedstock (coal, oil, fuel, etc.) were considered to be
factors considered; or simply different allocation methods used in supplied by mobile transportation (ships and trucks), so their
the LCA [14]. For H2 production, Spath et al. (2001) [15] developed contributions were already accounted for in the part of the fuel
inventories for an H2 plant including the pipeline infrastructure, cycle in the Well-to-Tank (WTT) stages or the vehicle’s materials.
and it can be seen that the main variables influencing the result per Maintenance activities were only considered for power plants, H2
unit of final energy are the same. plant and H2 stations.
Regarding the electricity transmission grid, the range of the Following the boundaries in Fig. 1, the model estimates the
environmental burden associated with its construction, mainte- emissions and energy use related to the construction, maintenance
nance and decommissioning has been accepted by the scientific and decommissioning of electricity and H2 pathways. Emissions and
community [9,12,16e18]. The variations in the estimates are related energy use are based on Global Warming Potential for 100 years
to different lifetimes considered, the high use of SF6 insulated (GWP100y) and Cumulative Energy Demand (CED) methods [20],
substations [17], high densities of the transmission grid per square with output functional units of gCO2eq/MJ and MJeq/MJ, respectively.
kilometre and the high usage of underground cables [16]. The model, as expected, is more abstract than the system it
Regarding electricity chargers and H2 refuelling stations, esti- represents. From one point of view, abstraction, and the assump-
mates [8,9,19] have shown that, within their own pathways, they tions made to achieve it, eliminates unnecessary detail and allows
are the main energy and emission contributors and also that the concentration on elements within the system that are most rele-
number of refuelling/charging stations per vehicle is the main vant to define. On the other hand, such an abstraction process in-
factor of uncertainty. However, by establishing scenarios it is troduces uncertainty in the results.
possible to study the impact of each choice and advise future The nature of uncertainty can be separated into two categories:
decisions. aleatory uncertainty, generally referred to as variation, that is
Because there is no tool to assist in the evaluation of infra- inherent and naturally unpredictable variation, and epistemic un-
structure deployment in terms of environmental impact caused by certainty, generally referred to as uncertainty, that is related to
construction, maintenance and decommissioning activities, and inaccurate measurements, lack of data, and missing information. In
due to its potential relevance, it is important to derive a generalised this model, only the latter is addressed. Typically, the uncertainty
model that could allow such analysis. addressed in LCA can be divided into three groups: parameter
uncertainty (input data), model uncertainty (mathematical rela-
2. Methodology tionship) and scenario uncertainty (choices). Because we are pre-
senting a model and the scenario inputs are left to the user, only the
System boundaries are shown in Fig. 1 following the authors’ parameter uncertainty was considered. There can be two types of
previous studies [8,9]. They refer directly to the supply chains parameter uncertainty: measurement uncertainty, which is related
divided into generation, transportation and end distribution facil- to imperfections, assumptions or the inability to take an exact
ities. The electricity pathway for EV’s includes the infrastructure for measurement when the actual inventory is being developed, and
the feedstock logistics (only an NG pipeline was considered), Power uncertainty related to the data quality of the inventories used. The
Plants (Coal, Oil, NG, Nuclear, Hydro, Geothermal, Wind and Solar), Ecoinvent [21] Pedigree matrix was used to account for such un-
the electricity transmission grid and normal, home and fast char- certainty, defining distributions using individual raw data points
gers. Energy supply infrastructures used by FCHEV include an NG and attributing uncertainty levels according to the nature of the
pipeline associated with the steam methane reforming centralised uncertainty.
H2 plant, H2 distribution pipelines and H2 refuelling stations Another source of uncertainty considered in this study accounts
(subsequently referred to as the H2(a) pathway). Another possible for the actual emission or energy impact to be considered for a
H2 pathway used by FCHEV includes the same set of power plants certain material. This is not due to the uncertainty of the identified
and incoming feedstock infrastructure used in the electricity impact value of the specific material (considered negligible in this
pathway, the electricity transmission grid, and H2 refuelling sta- study), but due to an under-specification of the material. When a
tions equipped with an electrolyser system for on-site hydrogen certain type of material is inventoried, it may be difficult to identify
production (subsequently referred to as H2(b)). Fuel cell plug-in its exact composition or type, for example, plastics or alloys. For
hybrid electric vehicles (FCPHEV) use both pathways. that reason, there is the need to either use a similar product to
compensate for its absence in the used database (application un-
certainty) or to incorporate all possible values for that impact. For
the inventories developed and compiled in this study (for the
chargers and H2 refuelling stations), the best fit distribution was
used to incorporate all different values (quantity of emissions or
energy) found in the Simapro 7.1 database [12] for variations of the
same material. The same approach was used for the NG and H2
pipelines but using a uniform distribution. This analysis can be
found in Tables 1e4 of the Support Information (SupInf) section.
Fig. 1 of the SupInf exemplifies how both parameters (quantity and
impact) are related and addressed in this study.
Regarding power plants, to incorporate the whole range of
possibilities within a certain generation technology, uniform dis-
tributions were considered in order, to include all possible inves-
tigated impacts. All uncertainty values can be found in Table 5 of
the SupInf.
Fig. 1. LCA scope definition of the three pathways: electricity, hydrogen centralised In terms of how uncertainty is considered to influence the
production H2(a), hydrogen local production H2(b). model, it is important to note that the values of a, b, g used to

Please cite this article in press as: Lucas A, et al., Energy supply infrastructure LCA model for electric and hydrogen transportation systems,
Energy (2013), http://dx.doi.org/10.1016/j.energy.2013.04.056
A. Lucas et al. / Energy xxx (2013) 1e11 3

replace the last term in the electricity pathway and F, J, l used to values found in the literature review (Table 5 of the SupInf). The
replace the last term in the H2(a) pathway (see SupInf) are esti- uncertainty values and capacity factors considered are shown in
mated assuming average inventory values. In the event of a varia- Table 7 of the SupInf.
tion in the inventory, the same percentage of variation should be Parameters Nc, Hc and Fc, are calculated byEqs. (3)e(5) and
applied to each parameter of the quadratic functions. reflect the relation of (GWP from materials)/(Total energy flowing):

3. Model formulation 187:04Eþ03  %NchSR  SRTotal


Nc ¼ (3)
AvrcarMJ=year  %Nch  LTimeNch
The model is presented in a set of parameterised equations with
two equations (emissions and energy) per pathway. The structure
42:04Eþ03  %HchSR  SRTotal
of each equation is similar, with each parameter referring directly Hc ¼ (4)
to a supply chain stage: generation, transmission or end distribu- AvrcarMJ=year  %Hch  LTimeHch
tion facilities.
In each stage, the value of gCO2eq/MJ or MJeq/MJ is corrected by 3146:91Eþ03  %FchSR  SRTotal
the loss factor of the downstream stages so that it can reflect a unit Fc ¼ (5)
AvrcarMJ=year  %Fch  LTimeFch
based on the final energy; i.e., a unit of final energy is equal to the
energy supplied to the vehicle (1 MJ), accounting for the energy Nc in Eq. (3) is given by the carbon intensity of the normal
losses in all stages in between. Losses are given in percentages by charger (187.04 gCO2/MJ inventoried based on Refs. [8,9] and on the
ðMJex=MJf Þ=ðMJ Stage InputÞ, which is the energy spent in an in- uncertainty developed in Tables 1 and 2 of SupInf), divided by two
dividual step of the chain per unit of final energy divided by the (the number of sockets in each satellite), multiplied by the per-
energy that goes into that stage, which is also per unit of final centage of normal chargers (%NchSR) of the total number of
energy. An example is shown in Table 6 of the (SupInf) section. For chargers-per-vehicle and multiplied by the chargers-per-vehicle
the case studies analysed, data from Edwards et al. (2011) [3] was ratio (subsequently referred to as the total service ratio or SRTotal).
used for all losses except for the transmission grid, which were The numerator is then divided by the total energy that flows
based on Ref. [22]. L%TG refers to the losses in the electric trans- through the chargers, which is calculated by multiplying the
mission grid, L%pline to losses in the H2 distribution pipeline, L%CO to average energy spent by a car per year (AvrcarMJ/year ¼ TTW  km/
the losses in the H2 refuelling station compressor, L%H2Plant to the year) with the percentage of charge actually being performed in
losses in the H2 SMR plant and L%El&CO to the losses in the elec- this type of charger (%Nch) during its lifetime period (LTimeNch). Hc
trolyser and compressor system at the H2 refuelling station. Losses (4) and Fc (5) follow exactly the same idea, only with different in-
considered are only assumed to be related to electricity. ventory values and ratios.
The infrastructure LCA of the correspondent pathway is ob-
tained by adding up all the terms in the general equations. To 3.1.1. Evaluation of SRTotal and charging percentages
convert the output of the model to a MJ/km unit, the total should be If SRTotal and the percentages of chargers are unknown, the
simply be multiplied by the TTW value of the vehicle; no other methodology below can be applied. Using a vehicle’s TTW and
adjustments are needed. annual distance driven, one can estimate the amount of energy
required from each of these chargers. Eq. (6) can be used to
3.1. Electricity pathway calculate the total service ratio (SRTotal) by inputting the average
number of kilometres travelled per year of a vehicle (x) in the
The electric infrastructure emissions LCA is represented in scenario under study.
gCO2eq/MJ in Eq. (1) and is a function of parameters that refer
directly to the supply chain in Fig. 1. SRTotal ¼ 8:8931E06 x þ 9:7934E01 (6)

Mc To develop the projection of what the total Service Ratio might


LCACO2eq Electricity ¼ þ 0:07 þ Nc þ Hc þ Fc (1) be in each scenario when the annual mileage changes, we consid-
ð1  L%TG Þ
ered the reference study [23] values for France, shown in Table 1.
The parameter given by variable Mc refers to the carbon in- The ratio of each type of charger (home, normal and fast) per
tensity from the construction, maintenance and decommissioning vehicle (further referred as partial service ratios) and its corre-
of all power plants, which can be calculated by (2) according to the spondent percentage can also be estimated, given by Eq. (7)
region’s electric mix. The 0.07 gCO2eq/MJ constant refers to the best
estimate of the electricity transmission grid carbon intensity per %NchSR ¼ 7:1826E06 x þ 1:2008E02 (7)
unit of final energy. This value has reached some consensus in the
literature [9,12,16e18], even though some variation is reported, To make the projection, one can assume an annual mileage of
hence its inclusion in the uncertainty analysis. Nc, Hc and Fc define 14,500 km per year [23] and that 2.5% of level 1 chargers (typical
the emissions life cycle analysis of normal, home and fast chargers, home charger) will be installed on the street (with physical char-
respectively, which are further developed using Eqs. (3)e(5). The acteristics similar to those of level 2 chargers or typical normal
energy losses in the chargers are considered to be negligible. chargers). Additionally, the mean amount of charging per type of
charger considered is 10% with normal charging, 84% at home and
Mc ¼ 1:5M1 þ 1:2M2 þ 1:1M3 þ 1:67M4 þ 3:1M5 þ 4:2M6
þ 6:2M7 þ 15:1M8 (2) Table 1
Percentage of chargers of total service ratio and partial service ratios.
Sub-parameters Mi in Formula (2) indicate the contribution to
Normal Home Fast
the electric mix of each of the electricity generation sources (Oil,
Coal, Natural Gas, Nuclear, Hydro, Geothermal, Wind and Solar, Partial service ratios 0.105 0.975 0.003
P8 % of Total service ratioa 11.73% 88.00% 0.3%
respectively), where i¼1 Mi ¼ 100%. The multiplying factors in
a
gCO2eq/MJ used for each Mi variable are the mean and median Considering 2.5% of home chargers level on the street.

Please cite this article in press as: Lucas A, et al., Energy supply infrastructure LCA model for electric and hydrogen transportation systems,
Energy (2013), http://dx.doi.org/10.1016/j.energy.2013.04.056
4 A. Lucas et al. / Energy xxx (2013) 1e11

5.5% using fast chargers [23]. Applying Eq. (3), we reach reference to the high weight of the third parameter gc (Fig. 3c). Eqs. (4)e(6) of
values of 5.2 gCO2eq/MJ and, with a similar approach, 0.11 MJeq/MJ the SupInf are extracted from the curves in Fig. 3 and can be
of normal chargers intensity. inputted in Eq. (8), replacing Nc, Hc and Fc shown in Eq. (1).
Because the functional unit is a unit based on final energy, the To identify the charging policy (percentages of charging among
LCA values will vary if the amounts of charging performed in each chargers) that minimises the burden caused by the infrastructure
type of charger also changes. Although fast chargers are not likely on an MJ-based unit, the derivative of the third parameter should
nor desirable to have a higher use than indicated for environ- be equal to zero. As seen in Fig. 3c, although unlikely, if the share of
mental, electric constraints and load diagram management prob- the total charge performed by normal chargers reaches 37%, the
lems [23,24], home and normal charging may change considerably marginal effect of an energy flow increase is zero. In fact, if it be-
depending on the region’s number of garages, public or private comes higher than 37%, the intensity values starts to increase. This
parking spaces or population density. For that reason, both the is because, while the normal charging share increased, the share of
mileage and the amount of charging (critical to define the amount home charging decreased to a point where, with the same fixed
of energy that flows through the charger) have to be included. partial home service ratio (e.g., 0.975), the quantity of energy that
Therefore, by changing the number of kilometres, maintaining the flows through the home charger is so low that it is no longer
ratio of the number of fast chargers and fixing the reference in- beneficial to the system to reduce it (or increasing the normal
tensities values, a new total service ratio is obtained. charging share).
In addition to maintaining the fast chargers ratio, the ratio of Following the exact same methodology, the infrastructure en-
home chargers is also considered to be unchangeable (as it is un- ergy intensity equations were estimated. Eq. (9) contains three
likely that a user will install additional chargers at home depending types of parameters.
on the distance driven); hence, the only variable changing will be
the normal chargers ratio, which is assumed to change linearly. Eq. Me
LCAMJeq Electricity ¼ þ 1:52E03 þ Ne þ He þ Fe (9)
(7) provides the percentage of normal chargers service ratio as a ð1  LTG Þ
function of annual kilometres per vehicle. The percentage of home
chargers service ratio is then given by taking the total service ratio First, the parameter given by Me refers to the energy intensity
and subtracting the normal and fast chargers percentage. from the construction, maintenance and decommissioning of po-
To estimate the carbon and energy intensity of the charging wer plants which can be calculated by (10), according to the re-
infrastructure, an example is given considering the values pre- gion’s electric mix. Second, the constant 1.52E03 MJeq/MJ refers to
sented in Table 1. The carbon and energy intensity will vary ac- the best estimate of the transmission grid energy intensity per unit
cording to a quadratic function. Fig. 2 shows that an increase of the of final energy. This value has reached some consensus in the
average number of kilometres per year will result in a decrease of literature [9,12,16e18], even though some variation is reported;
the carbon intensity of the charging infrastructure per unit of final hence, it is included in the uncertainty analysis. Third, Ne, He and Fe
energy (assuming only the normal chargers service ratio changes). define the energy intensity of the chargers, which can be further
The same behaviour is verified in the energy use LCA estimations. estimated by using Eqs. (11)e(13). Energy losses in the chargers are
Such behaviour is verified for Table 1 inputs; however, the pa- considered to be negligible.
rameters are not fixed and the annual average distance driven per
vehicle is not the only variable. Additionally, the amount of normal Me ¼ 1:9E03 M1 þ 1:9E03 M2 þ 3:5E03 M3 þ 1:2E03 M4
and home charge can vary. Hence, by changing the amount of þ 5:7E02 M5 þ 6:4E02 M6 þ 4:4E02 M7 þ 2:6E01 M8
charge to be provided per each type of charger, the curve shown in
Fig. 2 will increase or decrease its positioning in relation to the (10)
abscissa axes according to a general quadratic curve. This curve can Parameters Ne, He, and Fe, are calculated with Eqs. (11)e(13) and
completely replace the Nc, Hc, and Fc parameters of Eq. (1) by reflect the relation of (Energy Use from materials)/(Total energy
inputting the average annual distance driven (x), expressed now in flowing):
Eq. (8).
3975:894  %NchSR  SRTotal
Mc Ne ¼ (11)
LCACO2eq Electricity ¼ þ 0:07 þ ac x2  bc x þ gc (8) AvrcarMJ=year  %Nch  LTimeNch
ð1  L%TG Þ
The three parameters of the quadratic function must be found 856:03  %HchSR  SRTotal
by inputting the percentage of normal charging given by %Nch. Fig. 3 He ¼ (12)
AvrcarMJ=year  %Hch  LTimeHch
shows how each parameter evolves when the percentage of normal
charging increases (maintaining fast charging and adjusting home
charging accordingly). Overall, the carbon intensity decreases due 60016  %FchSR  SRTotal
Fe ¼ (13)
AvrcarMJ=year  %Fch  LTimeFch

In case SRTotal and the percentages of chargers are unknown, Ne,


He and Fe can be replaced by the general quadratic curve, in MJeq/MJ,
presented in Eq. (14), where x is the average distance driven for a
vehicle in km/year.

Me
LCAMJeq Electricity ¼ þ 1:52E03 þ ae x2  be x þ ge (14)
ð1  LTG Þ
Following the same methodology as in the emission analysis,
from the resulting curves, the corresponding Eqs. (7)e(9) of the
Fig. 2. Life cycle emissions variation of the charging system with average annual SupInf for energy use can be extracted and inputted into Eq. (14),
driven distance (x). replacing Ne, He and Fe of Eq. (9).

Please cite this article in press as: Lucas A, et al., Energy supply infrastructure LCA model for electric and hydrogen transportation systems,
Energy (2013), http://dx.doi.org/10.1016/j.energy.2013.04.056
A. Lucas et al. / Energy xxx (2013) 1e11 5

Fig. 3. Carbon intensity quadratic curves of each parameter ac (a), bc (b), gc (c) of the chargers general equation.

3.2. Hydrogen (a) pathway substantially from region to region; hence, this last term has to be
analysed further. The curve shown in Fig. 4 will increase or decrease
The H2(a) infrastructure emissions LCA in gCO2eq/MJ is repre- its positioning in relation to the abscissa axes, according to a gen-
sented by Eq. (15) and is given by parameters that refer directly to eral quadratic curve. By introducing the average annual distance
the supply chain. The first and third parameters account for the NG driven per vehicle (x), this curve can replace completely the last
pipeline infrastructure that supplies the SMR plant and that dis- parameters of Eq. (15), resulting in Eq. (16).
tributes H2 to the refuelling stations, respectively, and are adjusted
with their corresponding losses. The 0.354 gCO2eq/MJ constant is 0:354
the best estimate for these parameters based on the literature [12], LCACO2eq H2ðaÞ ¼      
but due to the variety of estimations in the database, a uniform 1  L%H2 Plant  1  L%pline  1  L%Co
distribution was used to characterise it. The second parameter is 0:17=CF 0:354 (16)
related to the SMR hydrogen plant. Because the capacity factor þ   þ
1  L%pline  1  L%Co ð1  L%Co Þ
greatly influences the result, the constant 0.17 gCO2eq of output
energy, based on Ref. [15], was considered using a 100% capacity þ fc x2  Jc x þ lc
factor, which must be adjusted with the real capacity factor by
introducing the variable CF. The fourth parameter refers to the H2 Simulating several ratios of vehicles per station and extracting
refuelling station. The numerator is the emissions LCA in gCO2/year all three terms of each quadratic function that was found, the
based on Ref. [19] inventory. An uncertainty analysis was conducted corresponding curves in Fig. 5 are obtained, from which the cor-
on this variable, which is presented in Table 3 of the SupInf. This responding Eqs. (10)e(12) of the SupInf can be extracted.
factor is then divided by the total energy that flows through the Following the exact same methodology, the infrastructure
refuelling station per year. energy intensity equations were estimated:

LCACO2eq H2ðaÞ ¼   
0:354
   2:59E02
LCAMJeq H2ðaÞ ¼      
1  L%H2 Plant  1  L%pline  1  L%Co 1  L%H2 Plant  1  L%pline  1  L%Co
0:17=CF 0:354 3:03E03 = CF 2:59E02
þ   þ
þ   þ
1  L%pline  1  L%Co ð1  L%Co Þ ð1  L%Co Þ
1  L%pline  1  L%Co
1:84Eþ07 3:20Eþ05
þ (15) þ  (17)
ðMJsold=yearÞ
MJsold=year
Regarding the last parameter, the energy that flows through the
refuelling station may change according to the number of kilo-
metres driven per year and the number of cars per station. Using a
reference TTW value of 1.25 MJ/km [8] for a FCHEV and considering
a given ratio of 2000 vehicles per station, the curve in Fig. 4 is
obtained. As expected, the figure shows that an increase of the
average number of kilometres per year will result in a decrease of
the carbon intensity of the charging infrastructure per unit of final
energy. The same behaviour is verified in the energy use LCA
estimations.
However, the annual distance driven is not the only variable that
may vary. The amount of energy supplied by the refuelling station Fig. 4. Life cycle emissions variation of a H2(a) refuelling station with different average
also depends on the ratio of vehicles per station that can vary annual driven distance.

Please cite this article in press as: Lucas A, et al., Energy supply infrastructure LCA model for electric and hydrogen transportation systems,
Energy (2013), http://dx.doi.org/10.1016/j.energy.2013.04.056
6 A. Lucas et al. / Energy xxx (2013) 1e11

Fig. 5. Carbon intensity quadratic curves of each parameter Fc(a), Jc(b), lc(c) of the H2(a) refuelling station by vehicles per station.

where the last term can be replaced by a general quadratic equa- to the electrolyser capacity, there is a maximum value that can be
tion, with x representing the average annual distance driven. supplied. Considering the installed capacity and the associated
Eq. (17) can hence be replaced by Eq. (18) capacity factor, which may vary, divided by the energy required by a
vehicle, Eq. (20) may be rearranged to provide the ratio of vehicles/
2:59E02 station (Eq. (21)).
LCAMJeq H2ðaÞ ¼      
1  L%H2 Plant  1  L%pline  1  L%Co
Vehicles EInst:cap:  CF
¼ (21)
3:03E03 = CF 2:59E02 (18) Station TTW  x
þ   þ
1  L%pline  1  L%Co ð1  L%Co Þ Fig. 6 shows the ratio of vehicles per stations when the average
annual distance driven in the vehicle changes based on the
þ fe x  Je x þ le
2
considered capacity and inventory [19]. This is caused by a linear
The terms of the quadratic function refer to energy intensity relation between the number of kilometres and the number of
values, whose values can be estimated by Eqs. (13)e(15) of the stations required because the energy output is limited by the pro-
SupInf in a manner similar to the procedure in the emissions duction ratio of the electrolyser.
analysis. Considering a 92.88% capacity factor, the ratio of vehicle per
H2(b) refuelling station can be estimated using Eq. (22), which in-
troduces the annual distance driven x.
3.3. Hydrogen (b) pathway
Vehicles
¼ 2:1969E06 x2  9:0998E02 x þ 1:1463Eþ03 (22)
The H2(b) infrastructure emissions LCA in gCO2eq/MJ is repre- Station
sented by Eq. (19). The Mc variable corresponds to the electric mix Following the same methodology, the Energy LCA can be esti-
and should be calculated according to Eq. (2). The second term mated by (23) in MJeq/MJ, where Me can be calculated using (10).
numerator is the same constant of the electricity pathway and ac- The losses considered are only related to electricity.
counts for the transmission grid. The third term refers to the H2
refuelling station. The numerator is the station’s carbon intensity in Me 1:52E03
gCO2/year based on Ref. [19] inventory. This value is then divided by LCAMJeq H2ðbÞ ¼ þ
ð1  L%TG Þ  ð1  L%El&Co Þ ð1  L%El&Co Þ
the total energy that flows through the refuelling station per year.
3:55Eþ05
Mc 0:07 þ  (23)
LCACO2eq H2ðbÞ ¼ þ MJsold=year
ð1  L%TG Þ  ð1  L%El&Co Þ ð1  L%El&Co Þ
2:033Eþ07
þ  (19)
MJsold=year

The energy sold per year is given by the number of vehicles per
station multiplied by the average TTW value of the vehicle and the
average annual distance driven (x) (Eq. (20)).

Vehicles
MJSupplied ¼  TTW  x (20)
Station
Naturally, the question to know is how many vehicles per
station to consider. Because the production capacity is limited due Fig. 6. Vehicles per refuelling station ratio H2(b) according to annual driven distance.

Please cite this article in press as: Lucas A, et al., Energy supply infrastructure LCA model for electric and hydrogen transportation systems,
Energy (2013), http://dx.doi.org/10.1016/j.energy.2013.04.056
A. Lucas et al. / Energy xxx (2013) 1e11 7

4. Model verification and validation Table 3


Input values considered for electricity pathway (chargers).

There are two steps to evaluate how coherent a model is with N Intc Nch ¼ 187.0a TTW ¼ 0.54
respect to the system. One must ascertain whether the model im- Inte Nch ¼ 3975.89b LTimeNch ¼ 6 years [8]
plements the assumptions and is built correctly (model verifica- Scenario/inputs %NchSR SRTotal AvrcarMJ/year %Nch
France 11.62% 1.108 7830 15%
tion) and whether the assumptions that have been made are Portugal 10.47% 1.094 6966 15%
reasonable with respect to the real system (model validation) [25]. USA 14.85% 1.148 10,260 15%
Regarding verification, the model structure is very simple,
H Intc Hch ¼ 42.0c
which is typical of a LCA and is just a sum of the impacts that have Inte Hch ¼ 856.03d LTimeHch ¼ 15 years [8]
been verified for each stage along the chain. The difficulties in the Scenario/inputs %HchSR SRTotal AvrcarMJ/year %Hch
developed model are the treatment of the support data, the correct France 88.11% 1.108 7,830 79.5%
presentation of the functional unit, assuring that there is no double Portugal 89.26% 1.094 6966 79.5%
USA 84.88% 1.148 10,260 79.5%
counting and checking that the most relevant system parameters
are addressed. In addition, it is necessary to make the model easy to F Intc Fch ¼ 3146.9e
Inte Fch ¼ 60016f LTimeFch ¼ 12 years [8]
use but also to allow the users to input specific scenario data. The
Scenario/inputs %FchSR SRTotal AvrcarMJ/year %Fch
Structured walk-through/One-step Analysis [25] technique was France 0.27% 1.108 7830 5.5%
used with the support of other researchers. Because a one-step Portugal 0.27% 1.094 6966 5.5%
analysis can take a long time, it is often applied to such simplified USA 0.27% 1.148 10,260 5.5%
models. Other researchers’ contributions facilitated the building of a, c, e, b, d, f
Refer to carbon and energy intensity of normal, home and fast chargers
the required simple and flexible model. In addition to this method, respectively. TTW is the energy in MJ/km for an average EV [8].
the Continuity Test [25] technique was applied. The model was
tested several times for slightly different values of input parameters
(charging shares, annual distances driven, electricity mix, and 4.1. Case scenarios
number of vehicles per stations). A slight change in the input
generally produced only a slight change in the output best estimate Three countries were chosen to apply the model. Portugal is
values; no discontinuities or sudden changes were observed in the used as a confirmation scenario because the present study is based
output values. on previous studies [8,9] and due to its developed stage of infra-
Model validation essentially consists of demonstrating that the structure deployment. France has a particular electric mix, highly
model is a reasonable representation of the actual system. It may be dependent on nuclear sources, and has been used as a reference for
difficult in practice to achieve full validation of the assumptions, studies addressing end distribution infrastructure. In the USA,
input parameters and distributions or output values of the model, several vehicle technologies are being analysed, and it presents one
especially if the system being modelled does not yet exist. For this of the most developed hydrogen infrastructures in the world,
reason, the main assumptions of the service ratios and the number especially in California [26]. The differences between the average
of H2 (a) stations were assumptions that were based on expert distance driven in a vehicle per year for each scenario will enhance
judgments with no current possibility of being validated by real any impact the final energy demand has over each pathway.
measurements. Table 2 presents the inputs to the electricity pathway model. The
However, there are several approaches [25] used to validate a electric mix by country [27e29], required to calculate M, refers to
model and any combination of these approaches may be consid- the year 2011. The annual distances driven considered are based on
ered appropriate for a particular model; hence, three techniques average real country data [8,23,30].
were used. First, the Parameter Variability technique [25] was used, Table 3 presents all the inputs to calculate N, H and F using Eqs.
which consists of applying the model and changing its input values (3)e(5), (11)e(13) in the electricity pathway. Eq. (6) was used to
and internal parameters to determine the effect upon the model’s estimate the SRTotal in the Portugal, France and USA case studies.
behaviour or output. Three case scenarios with very different real Table 4 presents the input values used when applying H2(a) Eqs.
life inputs were used: Portugal (high renewable energy sources (15) and (17). The inventory values are presented in units based on
penetration in electric mix), France (high vehicle/station ratio) and energy output, not final energy. Because the ratio of vehicles per
USA (high annual distance driven). This technique may be used for station variable is crucial for the estimations, it was assumed that
both verification and validation purposes. they were in the same proportion as conventional fuel stations in
Second, the Face Validity or result analysis [25] technique was each country. This ratio was obtained by dividing the number of
used, where other researchers, individuals who were knowledge-
able about the system, were asked whether the model and/or its
behaviour was reasonable and adjustments were made accordingly.
Finally, a Comparison with other data was performed. No other Table 4
Input values considered for H2(a) pathway.
models were identified, so the results were compared with the
existing literature [8] that used Portugal as a case study. Intapline 3.54E01 IntcH2Plant 0.17 InteH2station 1.84Eþ07 TTW ¼ 1.25
Intbpline 2.59E02 IntdH2Plant 3.03E03 IntfH2station 3.20Eþ05 CFg ¼ 0.9
Inputs L%H2plant L%pline L%Co km travelled/ MJ sold/ Vehicles/
year station/ Station
year
Table 2
France 21% 2% 16% 14,500 5.65Eþ07 3118
Input values considered for electricity pathway (electric mix by country).
Portugal 21% 2% 16% 12,900 3.60Eþ07 2234
M1 M2 M3 M4 M5 M6 M7 M8 USA 21% 2% 16% 19,000 4.69Eþ07 1974
a, b
Electric mix % Refer to carbon and energy intensity of NG and H2 pipeline.
c, d
Coal Oil NGCC Nuclear Hydro Geo Wind Solar km/year Refer to carbon and energy intensity of SMR plant.
e,f
2.5% 1.5% 6.5% 77.7% 9.3% 0.0% 2.2% 0.3% France 14500 Refer to carbon and energy intensity per year of H2 refuelling stations respec-
15.2% 1.0% 32.0% 0.0% 28.5% 1.0% 21.5% 0.8% Portugal 12900 tively. Units are in gCO2eq/MJ and MJeq/MJ respectively.
g
42.0% 1.0% 25.0% 19.0% 8.0% 1.0% 3.0% 1.0% USA 19100 Refers to the capacity factor of the SMR plant. TTW is the energy in MJ/km for a
FCHEV [8].

Please cite this article in press as: Lucas A, et al., Energy supply infrastructure LCA model for electric and hydrogen transportation systems,
Energy (2013), http://dx.doi.org/10.1016/j.energy.2013.04.056
8 A. Lucas et al. / Energy xxx (2013) 1e11

Table 5 value. Total values are presented in gCO2eq and MJeq by unit of final
Input values considered for H2(b) pathway. energy with the minimum, maximum and most likely values.
TTW 1.25 CF H2 statione 92.88%
Int H2 Stationa gCO2eq/year 2.033Eþ07 Int TGc gCO2eq/MJ 0.07
Int H2 Stationb MJeq/year 3.56Eþ05 Int TGd MJeq/MJ 1.52E03 4.2. Results and analysis
Inputs L%TG L%ElþCO MJ sold Vehicles/
station The model was applied demonstrating appropriate structure,
France 6% 43% 5.27E þ06
290.56 logic as well as correct causal relationships. Extreme real values
Portugal 8% 43% 5.27Eþ06 326.59 were considered for the electricity mix, the ratio of vehicles to
USA 6% 43% 5.27Eþ06 221.74 station and the average annual distance driven per vehicle.
a, b
Refer to carbon and energy intensity per year of H2 refuelling stations with Regarding the scenario analysis, the French scenario reports
electrolyser module. lower LCA emissions and energy use for both hydrogen pathways.
c, d
Are the carbon and energy intensities of the transmission and distribution
In terms of the potential use of the electric vehicle, the USA scenario
electric grid. Units are in gCO2eq/MJ and MJeq/MJ respectively.
e
Refers to the capacity factor of the electrolyser. TTW is the energy in MJ/km for a reports lower LCA values and the Portuguese scenario reports the
FCHEV [8]. highest values in all pathways. The comparison of the three sce-
narios studies is presented in Fig. 7.
A characteristic that stands out is the high uncertainty in elec-
light duty vehicles in each country by the number of existing tricity and H2(b) pathways of the Portuguese scenario. This is
conventional fuel stations [31e33]. The disparity in the values can explained by a high contribution of hydroelectricity to the electric
be explained by the different sizes of the station itself (number of mix when compared with the other two scenarios. The revised
dispensers), among other reasons. However, this question was literature presents an extreme variability of results for this elec-
considered to be negligible in the study, as the goal is to demon- tricity source, as both the scale and location of the plant are critical
strate the model application. factors. The reduction of uncertainty in the result by improving the
Regarding the H2(b) pathway, input values and losses are pre- quality of these data or distribution characterisation is recom-
sented in Table 5. Inventory values are presented in a unit based on mended. In addition to variability, this scenario also presents
the energy output, not the final energy. The hydrogen refuelling higher values for both emissions and energy use. In the electricity
station inventory is based on Ref. [19]; the electrolyser has a and H2(b) pathways, in addition to the generation mix, a high value
maximum capacity rate of 47,250 kgH2/year. The Low Heating of contribution from fast chargers is also verified in comparison
Value was used to convert kgH2 to MJ units. with the other two scenarios. This can be explained by the inferior
Table 6 shows the results for the energy supply infrastructure distance driven per year of the average vehicle, which, despite
LCA for the pathways under study after applying general Eqs. (1), resulting in a lower total service ratio, is penalised by the non-
(9), (15), (17), (19) and (23). The results are presented for each of linear relation with the required energy from the charger, which
the terms in the equations (left to right). Overall, the Portuguese is inferior.
scenario reports higher values both of emissions and energy use The second highest scenario in terms of uncertainty is France.
from the infrastructure. This can be explained by the high pene- This is due to the 77% nuclear power contribution to the electric
tration of renewable energy sources, in particular hydro and wind. mix. Nuclear has also a high variability of results, as there are many
The USA scenario presents very low impact from the infrastructure technologies that can be employed and addressed in the studies
with low variability. This is due to its coal- and NG-based electric revised. An improvement of the quality of this input is also advised
mix because both technologies are mature and well documented. to reduce the resultant total uncertainty amplitude. The USA sce-
It also benefits from the high annual distance driven per vehicle nario benefits from low intensity and low variability of the power

Table 6
Results of energy and emissions LCA from energy supply infrastructure in France, Portugal and USA.

gCO2eq/MJ-minemax (most likely) MJeq/MJ-minemax (most likely)

France Portugal USA France Portugal USA

Electricity
Mix 2.0 3.2 2.0 9.03E03 3.24E02 1.17E02
Grid 0.1 0.1 0.1 1.52E03 1.52E03 1.52E03
N 3.4 3.4 3.5 7.26E02 7.26E02 7.34E02
H 0.4 0.5 0.3 8.95E03 1.01E02 6.82E03
F 1.8 2.1 1.5 3.53E02 3.91E02 2.79E02

Total 5.6e26.4 (7.8) 6.2e36.6 (9.3) 6.0e17.7 (7.3) 0.08e0.20 (0.13) 0.10e0.32 (0. 16) 0.07e0.20 (0.12)

H2(a)
NG pipeline 0.5 0.5 0.5 3.99E02 3.99E02 3.99E02
H2 SMR plant 0.2 0.2 0.2 4.11E03 4.11E03 4.11E03
H2 pipeline 0.4 0.4 0.4 3.08E02 3.08E02 3.08E02
H2 station 0.3 0.5 0.4 5.66E03 8.88E03 6.82E03

Total 0.6e2.2 (1.5) 0.7e2.4 (1.7) 0.6e2.3 (1.6) 0.02e0.12 (0.08) 0.02e0.13 (0.08) 0.02e0.12 (0.08)

H2(b)
Mix 2.0 3.2 2.0 1.54E02 5.64E02 2.04E02
Grid 0.1 0.1 0.1 2.64E03 2.64E03 2.64E03
H2 station 3.9 3.9 3.9 6.74E02 6.74E02 6.74E02

Total 4.4e24.3 (6.0) 5.3e34.1 (7.2) 5.0e16.3 (6.0) 0.06e0.18 (0.08) 0.07e0.36 (0. 13) 0.06e0.18 (0.09)

Please cite this article in press as: Lucas A, et al., Energy supply infrastructure LCA model for electric and hydrogen transportation systems,
Energy (2013), http://dx.doi.org/10.1016/j.energy.2013.04.056
A. Lucas et al. / Energy xxx (2013) 1e11 9

Fig. 7. Emissions (a) and energy (b) LCA of energy supply infrastructures for France, Portugal an USA scenarios.

Fig. 8. Model results for Portugal case scenario comparison of with literature [8].

plants infrastructure LCA, as it bases its generation mix in mature higher in the model than in the case study under comparison.
conventional technologies. However, differences show consistency as the case study values
Losses are not a significantly distinct factor between the sce- are within the model’s range. Regarding the best estimate value,
narios, as the transport and end distribution infrastructures along differences are reported in different directions; however, this is not
the pathways are fairly similar. Regarding the H2(a) pathway, the due to lack of accuracy.
variable contributing the most to scenario differentiation is the For the electricity pathway, this fact is explained by the inclu-
number of vehicles per station considered. The average ratio of sion of impact uncertainty in the chargers inventories, which also
vehicles per conventional fuel station in each country was considers lower impact values than the ones used in the case study;
considered to represent the ratio of vehicles per H2(a) refuelling hence, the average value of both emissions and the energy use
station. An high disparity is verified within the three scenarios infrastructure LCA is decreased.
ranging from 3118 vehicles in France to 1974 vehicles in the USA per Regarding the H2(a) pathway, emissions and energy use impact
station. However, not only is this ratio important, but the energy vary in the same direction. An increase of the best estimate is
supplied by each station should also be analysed. On average, in the verified in comparison with the case study values. This change is
French scenario more energy would be supplied per station in a mostly given by inventory data treatment using the pedigree ma-
year despite having an inferior annual distance driven. This causes trix [21], which not only changed the uncertainty range but also the
both emissions and energy use to be inferior to the other two best estimate value. Concerning the H2(b) differences, the emis-
scenarios. The Portuguese scenario is penalised by having a lower sions in the model show a decrease in comparison with the emis-
ratio of vehicles per station than France and, consequently, by sions in the case study. On the contrary, the energy use results
supplying less energy on each one. report an increase. This is given by changes in the pondering factors
The output behaviour is reasonable in terms of the direction and of the power plants compiled by all the literature estimates. This
amplitude of results. A model is considered valid for a set of difference in direction also occurs in the electricity pathway, but it
experimental conditions if the model’s accuracy is within its is concealed by a higher influence caused by the impact uncertainty
acceptable range, which is the amount of accuracy required for the inclusion. The accuracy is acceptable and depends on the type of
model’s intended purpose. Although high uncertainty is verified in variables inputted. Higher accuracy could be obtained through data
all pathways, it is a trade-off with the simplicity of the model’s quality improvement. The model, however, correctly fulfils the
structure. purpose of the analysis.

4.3. Comparison to literature results 5. Conclusions

Fig. 8 shows a comparison of the model results with the ones A model is proposed that focuses on the electricity and
found in the literature [8]. As expected, uncertainty values are hydrogen supply through centralised steam methane reforming

Please cite this article in press as: Lucas A, et al., Energy supply infrastructure LCA model for electric and hydrogen transportation systems,
Energy (2013), http://dx.doi.org/10.1016/j.energy.2013.04.056
10 A. Lucas et al. / Energy xxx (2013) 1e11

and on-site electrolysis. Verification and validation were performed GWP Global Warming Potential
mainly by Face Validity, a Scenario Application using real data from Hc home charger emissions impact
France, Portugal and USA and by literature Comparison techniques. He home charger energy impact
The model shows consistency in the results and in the uncertainty ICEV Internal Combustion Engine Vehicle
range. The evaluation of the service ratios of electric chargers and Intpline pipeline carbon/energy intensity
stations as a function of the annual distance driven is also IntH2Plant H2 plant carbon/energy intensity
presented. IntH2station H2 station carbon/energy intensity
The model is consistent with the values from a previous analysis IntTG transmission Grid carbon/energy intensity
performed by the authors [8], which used Portugal as a case study. Intc Fch fast charger carbon intensity
Uncertainty values, as expected in a generalisation study, have Inte Fch fast charger energy intensity
increased, mainly due to the inclusion of the estimations found in Intc Hch home charger carbon intensity
the literature of the different power plants, the inclusion of the Inte Hch home charger energy intensity
impact uncertainty and the application of the pedigree matrix in Intc Nch normal charger carbon intensity
the inventories. Improving data quality or the characterisation of Inte Nch normal charger energy intensity
the statistical distributions and future real measurement validation LCA Life Cycle Analysis
is advised. LTimeNch normal charger lifetime
The French scenario reported lower LCA emissions and LTimeHch home charger lifetime
energy use for both hydrogen pathways, with 0.6e2.2 gCO2eq/MJ LTimeFch fast charger lifetime
and 0.02e0.12 MJeq/MJ for H2(a) and 4.4e24.3 gCO2eq/MJ and L%TG losses in transmission grid
0.06e0.18 MJeq/MJ for H2(b). Regarding the potential use of the L%pline losses in pipeline
electric vehicle, the USA scenario reports lower LCA values with L%H2Plant losses in H2 plant
6.0e17.7 gCO2eq/MJ and 0.07e0.20 MJeq/MJ. The Portuguese scenario L%El&CO losses in electrolyser and compressor
reported the highest values in all pathways, with 6.2e36.6 gCO2eq/MJ L%Co losses in the H2 compressor
and 0.10e0.32 MJeq/MJ for the electricity pathway, 0.7e2.4 gCO2eq/ MJex energy expended
MJ and 0.02e0.13 MJeq/MJ for the H2(a) pathway and 5.3e MJf final energy
34.1 gCO2eq/MJ and 0.07e0.36 MJeq/MJ for H2(b). This reveals that a Mc infrastructure carbon intensity of electric mix
country with high penetration of renewable sources has a tendency Me infrastructure energy intensity of electric mix
to have higher contribution to LCA from the infrastructure. Nuclear Nc normal charger emissions impact
energy on the other hand, although having higher impacts from the NG Natural Gas
infrastructure than conventional technologies, improves the overall Ne normal charger energy impact
burden as it has negligible generation emissions. NGCC Natural Gas Combined Cycle
The model indicates that, with the considered assumptions, the SMR Steam Methane Reforming
environmental infrastructure burden of the electricity pathway is TTW tank-to-wheel
minimised with a policy which incentivises approximately 37% of USA United States of America
the charge from normal chargers. Regarding the H2(a) pathway, the WTT well-to-tank
pipeline infrastructure is the most emission and energy intensive. WTW well-to-wheel
Efforts to increase its lifetime should be exerted to reduce the overall x annual average distance driven per vehicle
impact. Regarding H2(b), as the ratio of vehicles per station is limited %NchSR normal chargers share of service ratio
by the electrolyser production, increasing the installed capacity %HchSR home chargers share of service ratio
would decrease the impact per MJ supplied. Efforts should be made %FchSR fast chargers share of service ratio
to approximate this ratio to one of a conventional fuel station. SRTotal total service ratio (chargers/vehicle)
AvrcarMJ/year vehicle average energy required per year
Acknowledgements %Nch average share of normal charging
%Hch average share of home charging
The authors would like to thank the MIT-PP and the support of %Fch average share of fast charging
all companies that made this study possible, in particular EFACEC
for the financial support. Special thanks to Elsevier Language
Editing Services. References

Appendix A. Supplementary material [1] Weiss M, Heywood J, Drake E, Schafer A, AuYeung F. On the road in 2020 e a
life-cycle analysis of new automobile technologies. Cambridge (MA): MIT
Energy Laboratory; 2000 Oct. p. 160. Report No.: MIT EL 00e003.
Supplementary material related to this article can be found at [2] Weiss M, Heywood J, Schafer A. Future fuel cell and internal combustion
http://dx.doi.org/10.1016/j.energy.2013.04.056. engine automobile technologies: a 25-year life cycle and fleet impact
assessment. Energy 2006;31(12):2064e87.
[3] Edwards R, Larive JF, Beziat JC. Well-to-wheels analysis of future automotive
Nomenclature fuels and powertrains in the European context. Version 3c. Concawe: JRC,
EUCAR; 2011 Jul. p. 74.
[4] Gross B, Sutherland I, Mooiweer H. Hydrogen fuelling infrastructure assess-
CED Cumulative Energy Demand ment. Mound Road (MI): General Motors Research Development Center; 2007
CF Capacity Factor Dec. p. 43. Report No.:R&D11,065.
CTG Cradle-to-Grave [5] Huang Z, Zhang X. Well-to-wheels analysis of hydrogen based fuel-cell vehicle
pathways in shanghai. Energy 2006;31(4):471e89.
EV Full Electric Vehicle
[6] Geerken T, Timmermans V, Lassaux S. Hydrogen and it’s applications: review
EInst.cap. energy installed capacity of life cycle assessment studies and well-to-wheel studies. In: 10th European
FCHEV Fuel Cell Hybrid Electric Vehicle Roundtable on sustainable consumption and production, Antwerp (Belgium),
FCPHEV fuel cell plug-in hybrid electric vehicles Hysociety project, 5-6-7; 2005 Oct.
[7] Frischknecht R, Althaus H, Bauer C, Doka G, Heck T, Jungbluth N, et al. The
Fe fast charger energy impact environmental relevance of capital goods in life cycle assessments of products
Fc fast charger emissions impact and services. Int J LCA 2007:1e11.

Please cite this article in press as: Lucas A, et al., Energy supply infrastructure LCA model for electric and hydrogen transportation systems,
Energy (2013), http://dx.doi.org/10.1016/j.energy.2013.04.056
A. Lucas et al. / Energy xxx (2013) 1e11 11

[8] Lucas A, Alexandra Silva C, Costa Neto R. Impact of energy supply infra- [20] Forster P, Ramaswamy V, Artaxo P, Berntsen T, Betts R, Fahey D, et al. Changes
structure in life cycle analysis of hydrogen and electric systems applied to in atmospheric constituents and in radiative forcing. Climate Change 2007:
the Portuguese transportation sector. Int J Hydrogen Energy 2012;37: the Physical Science Basis. IPCC Fourth Assessment Report. UK, New York,
10973e85. USA: Joint Publication of Cambridge University Press; 2007 Feb. p. 106.
[9] Lucas A, Alexandra Silva C, Costa Neto R. Life cycle analysis of energy supply Working Group I: [Chapter 2].
infrastructure for conventional and electric vehicles. Energy Policy 2012;41: [21] Frischknecht R, Jungbluth N, Althaus HJ, Doka G, Dones R, Heck T, et al.
537e47. Overview and methodology. Ecoinvent report No. 1, 2007 Dec., vol. 2.0.
[10] Sullivan J, Clark C, Han J, Wang M. Life-cycle analysis results of geothermal Dubendorf, Switzerland: Swiss Centre for Life Cycle Inventories; 2007. p. 60.
systems in comparison to other power systems. Chicago (IL): DOE; 2010 Oct. [22] Worldbank.org [Internet]. Electric power transmission and distribution losses
p. 72. Argonne Laboratories, Energy Systems Division Report No.: ANL/ESD/10-5. (% of output) 2009. Available from: http://data.worldbank.org [updated 2012;
Contract No.: DE-AC02-06CH11357. Supported by U.S. Department of Energy. cited 11.05.12].
[11] Hondo H. Life cycle GHG emission analysis of power generation systems: [23] Groupe de travail sur le: Livre vert pour le développement des infrastructures
Japanese case. Energy 2005;30:2042e56. de charge électrique. Paris: MEEDDM; 2010 Apr. p. 198 [in French].
[12] PRé Consultants [Internet]. Simapro tool. Available from: http://www.pre.nl/ [24] Schroeder A, Traber T. The economics of fast charging infrastructure for
simapro [updated 2012 Apr.; cited 2012 Apr.]. electric vehicles. Energy Policy 2012;43:136e44.
[13] Weisser D. A guide to life-cycle greenhouse gas (GHG) emissions from electric [25] Sargent RG. Verification and validation of simulation models. In: Johansson B,
supply technologies. Energy 2007;32(9):1543e59. Jain S, Montoya-Torres J, Hugan J, Yucesan E, editors. Proceedings of the 2010
[14] Soimakallio S, Kiviluoma J, Kiviluoma SL. The complexity and challenges of winter simulation conference. Piscata-way, New Jersey: IEEE; 2010. p. 166e82.
determining GHG (greenhouse gas) emissions from grid electricity con- [26] Fuelcells.org [Internet]. Fuel cell databases. Available from: http://www.
sumption and conservation in LCA (life cycle assessment) e a methodological fuelcells.org; 2012. n.d [updated 2012, cited 11.05.12].
review. Energy 2011;36(12):6705e13. [27] Bilan électrique 2011. Paris (France): RTE Réseau de transport d’électricité;
[15] Spath PL, Mann MK. Life cycle assessment of renewable hydrogen production 2011. p. 40 [in French].
via natural gas steam reforming. Golden (CO): NREL; 2001. p. 33. Report No.: [28] Dados Técnicos 2011. Lisboa (Portugal): REN Redes Energéticas Nacionais;
NREL/TP-570-27637. Contract No.: DEA3699GO10337. Supported by the 2011. p. 36 [in Portuguese].
Department of Energy. [29] Eia.gov [Internet]. Electricity in the US. Available from: http://www.eia.gov;
[16] Itten R, Frischknecht R, Stucki M. Life cycle inventories of electricity mixes and 2010 [updated 2012 May; cited 11.05.12].
grid. Uster (Switzerland): Paul Scherrer Institute; 2012 Jul. p. 229. [30] Fhwa.dot.gov [Internet]. Annual vehicle distance travelled in kilometers and
[17] Harrison GP, Maclean ENJ, Karamanlis S, Ochoa LF. Life cycle assessment of the related data e 2009 1/by highway category and vehicle type. Available from:
transmission network in Great Britain. Energy Policy 2010;38:3622e31. http://www.fhwa.dot.gov [updated 2012 June; cited 2012 Aug].
[18] Working Group B2.15. Life Cycle Assessment (LCA) for overhead lines. Paris: [31] DBK Informa. Estudio sectores Portugal, Estaciones de servicio. 6a ed. 2011
Cigre; 2005 Jan. p. 168. Report No.:265. Feb. p. 163 [in Spanish].
[19] Maack M. Generation of the energy carrier hydrogen in context with elec- [32] Census.gov [Internet]. United States Census Bureau 2011. Industry Statistics
tricity buffering generation through fuel cells. NEEDS: New Energy External- Sampler, NAICS 447 Gasoline Stations 2007. Available from: http://www.
ities Developments for Sustainability; 2008. p. 47. Report No.: 8-5 RS1a. census.gov [updated 2011 May; cited 05.06.12].
Project No.: 502687 for Icelandic New Energy. Co-sponsored by the European [33] Datamonitor.com [Internet]. Service station retailing in France 2011. Available
Commission within the Sixth Framework. from: http://www.datamonitor.com [updated 2012 May; cited 11.05.12].

Please cite this article in press as: Lucas A, et al., Energy supply infrastructure LCA model for electric and hydrogen transportation systems,
Energy (2013), http://dx.doi.org/10.1016/j.energy.2013.04.056

You might also like