Environ. Sci. Technol. 2008, 42, 6973–6979
Parametric Assessment of Climate Change Impacts of Automotive Material Substitution ROLAND GEYER* Donald Bren School of Environmental Science and Management, University of California, Santa Barbara, California 93106-5131
Received February 5, 2008. Revised manuscript received May 5, 2008. Accepted May 13, 2008.
Quantifying the net climate change impact of automotive material substitution is not a trivial task. It requires the assessment of the mass reduction potential of automotive materials, the greenhouse gas (GHG) emissions from their production and recycling, and their impact on GHG emissions from vehicle use. The model presented in this paper is based on life cycle assessment (LCA) and completely parameterized, i.e., its computational structure is separated from the required input data, which is not traditionally done in LCAs. The parameterization increases scientific rigor and transparency of the assessment methodology, facilitates sensitivity and uncertainty analysis of the results, and also makes it possible to compare different studies and explain their disparities. The state of the art of the modeling methodology is reviewed and advanced. Assessment of the GHG emission impacts of material recycling through consequential system expansion shows that our understanding of this issue is still incomplete. This is a critical knowledge gap since a case study shows that for materials such as aluminum, the GHG emission impacts of material production and recycling are both of the same size as the use phase savings from vehicle mass reduction.
1. Introduction This article is concerned with assessing the impact that automotive material choice has on climate change, i.e., the relationship between automotive material selection and life cycle GHG emissions of vehicles. Traditionally, material selection in product design was dominated by the tradeoff between technical properties and economic cost. Today, however, the increasing importance of pollution prevention requires that the environmental impacts of material choice are already considered at the design stage. This is particularly true for the climate change impacts of material selection as the world communities increasingly look for ways to reduce GHG emissions. The transportation sector, in general, and on-road vehicles, in particular, are a major source of GHG emissions, especially in highly industrialized nations. The United Nations Framework Convention on Climate Change (UNFCCC) has recently emphasized that “in particular, transport remains a sector where emission reductions are urgently required but seem to be especially difficult to achieve” (1). Legislators such as California and the EU thus attempt to reduce GHG emissions from vehicle use through regulation or voluntary agreements (2–5). * E-mail:
[email protected]. 10.1021/es800314w CCC: $40.75
Published on Web 08/14/2008
2008 American Chemical Society
One possibility to decrease GHG emissions from vehicle use is to reduce vehicle mass (6). Relevant design strategies are smaller vehicle platforms, more efficient packaging, and material substitution, i.e., the use of so-called lightweight materials. Vehicle size reduction is a simple yet powerful design strategy, since it is likely to reduce the GHG emissions at all stages of the vehicle life cycle. However, in countries like the United States the average size of passenger vehicles has been increasing over the last couple of decades (7). Material substitution, on the other side, potentially achieves use phase reductions at the possible expense of increased emissions at other life cycle stages, such as material production. Even though vehicle use (including fuel production) makes up 70-90% of the life cycle GHG emissions of a vehicle, assessing material substitution requires a life cycle perspective, i.e., the consideration of production, use, and disposal of the vehicle (8, 9). A review of existing life cycle energy or GHG assessments of vehicle mass reduction based on material substitution shows that there is substantial disparity between study results, caused by significant differences in input data, modeling choices, and model assumptions (9–19). To facilitate consensus building regarding modeling choices and model assumptions, while avoiding the ongoing controversy regarding input data, the presented assessment model is completely parameterized. The advantages of parameterized assessment of this issue have been pointed out previously (20). The assessment methodology is informed by current best practices and advances them by making several contributions. The presented model can be used for original assessments, as shown in a case study, and as a meta-model for benchmarking and comparison of existing studies.
2. Methods and Data 2.1. Scope and System Boundaries. The presented research is based on life cycle assessment (LCA) according to ISO 14040/44 (2006), but contains two digressions from standard LCA practice. The first is that only one impact category, climate change, is used in impact assessment. This is done for clarity’s sake rather than necessity, since it is straightforward to include more impact categories, given that the necessary data are available and the system boundaries are revisited. The second digression is the focus on parametric modeling of the inventory stage, i.e., the use of parameters and variables rather than data to develop the life cycle inventory. This has two important advantages: (1) The computational structure of the model is separated from its input data. (2) Sensitivity analysis of the model results with respect to input data and modeling choices is facilitated. The functional unit, FU, of the study is defined as follows: transportation services of passenger cars of equivalent size, utility, equipment, and power train configuration over their total vehicle life. To assess the impact of mass reduction based on material substitution, the FU is translated into the reference flows of a baseline vehicle RFb and one or more contenders RFc in which part of the baseline materials are replaced with contender materials. The resulting reference flows are passenger cars of equivalent size, utility, equipment, and power train configuration but with different material composition and thus fuel energy demand. For the modeling purposes of the study, the reference flows of the vehicles are fully characterized by a vector containing their material composition and fuel energy demand: VOL. 42, NO. 18, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
6973
RFx ) (m1x, . . ., mnx, EDx) mix in kg, EDx in MJ/100 km, and x ) b, c (1)
For simplicity, let us consider only one contender vehicle for now. The total mass of each vehicle RFx is simply the sum over all its mass fractions, i.e., VMx ) Σin) 1mix. Vehicle life cycles are very complex product systems (21). In the presented methodology, boundaries are chosen based on the objective to assess the differences in life cycle GHG emissions between vehicles, rather than to comprehensively capture their life cycle GHG emissions. Figure 1 shows all included processes in black boxes with solid lines. They are primary and secondary production of the automotive materials, their further processing into finished material products, use of the vehicle, and fuel production. Material finishing includes processes such as rolling, galvanizing, extruding, and casting. It excludes subsequent forming and joining operations, such as forging, stamping, welding, and bonding. The main excluded process groups are vehicle manufacturing (forming and joining), vehicle end-of-life management (vehicle shredding and material separation), and transportation steps between processes. The GHG impact of manufacturing and end-of-life scrap, however, is accounted for through consequential system expansion. Based on contribution analyses available in literature, this boundary choice captures between 93% and 96% of the life cycle GHG emissions of passenger cars (8, 22–24). Processes can be excluded from the analysis if they are not significantly affected by automotive material choice. GHG savings from distributing a lighter vehicle, for example, are estimated to be at best a few per mille of total life cycle GHG emissions of the vehicle (22). The difference in GHG emissions from end-of-life management (without the impact of scrap generation) due to different material composition of the vehicles is estimated to be no more than one or a few per mille of total life cycle GHG emissions (23, 24). Vehicle manufacturing (without the impact of scrap generation) is estimated to account for 4-7% of total life cycle GHG emissions of passenger cars (8, 22–24). GHG emission differences due to the different manufacturing processes required for the different materials might thus conceivably be in the order of 1% of total life cycle GHG emissions (17, 25). This is estimated to be the largest error source in the methodology and should thus be the first system boundary issue addressed in any future refinements. 2.2. Modeling the Mass Reduction Potential of Automotive Materials. The first step of the inventory model is to derive the reference flow RFc of the contender vehicle, i.e., its material composition and fuel energy demand, from the
baseline reference flow RFb and the material replacement characteristics. The change in material composition can be characterized by the following set of parameters: ∆M ) totalmass of replaced material (in kg); kij ) material replacement coefficient of material i with regard to material j (in kg/kg); s ratio of secondary mass savings to primary mass savings (in kg/kg); πj ) material composition (in %) of replaced material (∑j πj ) 1); µij ) fraction of replaced material πj∆M being replaced with material i (∑ij µij ) 1); νi ) material composition (in %) of secondary mass savings (∑i νi ) 1). The material composition vector of the contender vehicle is then calculated as mic ) mib - πi∆M +
∑ k µ π ∆M - ν s(1 - ∑ k µ π )∆M ij ij j
i
ij ij j
j
ij
(2)
Occasionally, the material-specific replacement coefficients kij are derived from the mechanical properties of the materials (18, 26). An example would be to set kij equal to the mass ratio at which beams of the two materials i and j have equal tensile strength, i.e. σiAi ) σjAj w kij : )
mi FiAil σjFi ) ) mj FjAjl σiFj
(3)
Here, σi is the tensile strength of material i, i.e., the tensile force per unit area at which the material breaks, and Fi is its density, while Ai is the cross sectional area, and l is the length of the beams. There are two drawbacks to this approach (26). First, automotive materials typically have to deal with more than just one type of stress, such as tension, compression, bending, and shearing. It is thus not obvious which material property to use for calculating kij. Second, primary mass savings are not just a function of replaced and replacing materials but also of the design in which they are employed. There is, for example, a significant difference between using open or closed profiles in mechanical design (26, 27). This makes it impossible to determine the exact values of the relevant replacement coefficients kij in eq 2 without knowledge of the employed design principles. It is therefore more common to model less complex material replacement scenarios, e.g., situations in which all replaced materials are assumed to have the same replacement characteristics, i.e., the replacement coefficients only depend on the replacing material, kij ) kik ) ki. Equation 2 then simplifies to
(
mic ) mib - πi∆M + kiµi∆M - νis 1 -
∑ k µ )∆M i i
(4)
i
FIGURE 1. Life cycle of automotive material i including consequential system expansion for scrap recycling (parameters quantifying scrap flows are defined in the section on materials recycling). 6974
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 18, 2008
where µi: ) Σj µij πj is the fraction of ∆M that is replaced with material i. If all replacing materials have the same replacement characteristics, i.e., ki ) kj ) k, eq 4 further simplifies to mic ) mib - πi∆M + kµi∆M - νis(1 - k)∆M
(5)
It is useful to define an overall replacement coefficient k for the more complex material replacement scenarios of eqs 2 and 4, which are k: )
∑ k µ π for eq 2 and k : ) ∑ k µ for eq 4 ij ij j
i i
ij
(6)
i
Using the definitions from equation 6, the total mass of the contender vehicle can be expressed through the generic equation VMc )
∑ m ) VM - ∆M + k∆M - s(1 - k)∆M c i
b
(7)
i
The parameter k characterizes the overall replacement coefficient of the design of contender vehicle RFc relative to the baseline vehicle RFb, i.e., what fraction of the replaced mass is required by contender design and materials in order to achieve a functionally equivalent vehicle. The parameter s describes the secondary mass savings, which are typically modeled as a linear function of k. The reasoning behind s is that primary mass reductions enable the downsizing of other vehicle parts and components. The net mass savings of the mass-reduced vehicle RFc relative to RFb are thus ∆VM ) VMb - VMc ) (1 + s)(1 - k)∆M
(8)
2.3. GHG Emissions from the Vehicle Use Phase. To complete the reference flow RFc of the contender vehicle, its fuel energy demand EDc has to be derived, EDc ) EDb - ∆ED
(9)
The relationship between vehicle mass reduction ∆VM and reduced fuel energy demand ∆ED of contender vehicle RFc is modeled as a linear function, i.e. ∆ED ) ES · ∆VM ) ES(1 + s)(1 - k)∆M
(10)
where ES denotes the fuel energy savings per mass savings in MJ/(100 km · 100 kg). That the energy demand at the wheels, EDwheels, is a linear function of vehicle mass VM follows directly from the fact that the power required to move the vehicle, PDwheels, is a linear function of VM, i.e. PDwheels ) PDacceleration + PDpotential + PDrolling resistance + PDaerodynamic drag ) VM · a · v + VM · g · v · sin φ + Fair fr · VM · g · v · cos φ + cw A · v3 2 EDwheels )
∫
time
PDdt ) VM
∫
time
(av + gvsin φ +
frgvcos φ)dt +
∫
time
(
)
Fair 3 cw Av dt (11) 2
Here, v, a, A, fr, and cw are velocity, acceleration, frontal area, and rolling and aerodynamic resistance coefficient of the vehicle, g is the gravitational acceleration, φ is the gradient of the road, and Fair is the density of the air. If the energy conversion efficiency of the power train is not affected by the mass reduction or adjusted to equal the baseline efficiency, the change in fuel energy demand is proportional to the change in vehicle mass, as assumed in eq 10. For each reference flow RFx, x ) b,c, GHG emissions of x , are now calculated as the vehicle use phase, Guse x ) GWTW · EDx · TM, x ) b, c Guse
The three parameters of eq 12 are defined as follows: EDx ) fuel energy demand for given fuel type and driving cycle (in MJ/100 km); GWTW ) well-to-wheel GHG emissions of the fuel (in kg CO2eq/MJ); and TM ) total life of the vehicle, assumed to be the same for both vehicles (in km). Using fuel energy demand EDx (in MJ/100 km) instead of the traditionally used volumetric fuel economy FEx (in L/100 km) enables the use of one generic equation for all fuel types, including biofuels, hydrogen, and electricity. EDx can be readily converted into FEx if the volumetric energy density eV of the fuel is known, since FEx ) eV-1EDx. GWTW accounts for all GHGs emitted during production, delivery and use of 1 MJ of fuel energy. Its value depends on both fuel type and production pathway (28–30). For combustion fuels, fuel energy is typically given as net calorific value, since the enthalpy of condensation is not utilized by the power train. The parameter EDx gives the energy demand that results from a given velocity profile v(t) (called driving cycle), the relevant vehicle characteristics (frontal area, rolling and aerodynamic resistance coefficients), and the energy conversion efficiency of the power train. In the literature, mass-reduction-related fuel economy improvements are often calculated using the concept of a mass elasticity of the fuel economy, i.e., the ratio of relative fuel savings per relative mass savings ∆FE · VM/FE · ∆VM. This ratio of percentage changes is typically assumed to be constant, e.g., 0.5, which means that mass savings of y% generate fuel savings of 0.5 · y%. The use of the economic concept of elasticity is problematic for at least two reasons: First, a relative fuel economy improvement of y% yields different absolute fuel economies, depending on whether the unit of measurement used is mpg or L/100 km. Second, functions with constant elasticity are of the form f(x) ) a · xb, i.e., generally nonlinear. However, if ∆FE · VM/FE · ∆VM ) constant, then the fuel economy is a linear function of the vehicle mass, which directly contradicts the assumption of a constant elasticity. These problems are avoided in this model since the fuel energy savings due to mass savings, ES, are given in absolute instead of relative terms as described above. ES is a critical model parameter, which has to summarize the behavior of a complex system (31, 32). It depends on vehicle design, mass and power train configuration, driving cycle, and type of power train adjustment after mass reduction, and thus has to be modeled as a function of these parameters (33). Sometimes, use phase GHG emissions are allocated to individual vehicle parts or subassemblies based on mass. For the replaced vehicle parts of this study, this would yield x :) Guse,∆M
(13)
However, such an approach is not suited to assess the use phase GHG emission impacts of automotive material substitution since generally b c - Guse * Guse
∆M b k∆M c G Guse b use VM VMc
(14)
where ∆M and k∆M are the masses of the considered parts of the vehicle before and after material substitution. Left and right side of equation (14) are only identical in the unlikely case that s ) 0 and ES ) EDb/VMb. 2.4. GHG Emissions from Materials Production. The next step of the parametric assessment is to calculate the GHG emissions from the production of the automotive materials, x , for each reference flow RFx, x ) b,c. Based on the material Gmat x is calculated as composition of the vehicles, Gmat x Gmat )
∑ i
(12)
∆M x G VM use
mixGimat )
∑ i
mix [(1 - ricon)Gip + riconGis + Gif], x ) γi b, c (15)
VOL. 42, NO. 18, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
6975
TABLE 1. Contribution Analysis of Life Cycle GHG Emission Impacts of Automotive Material Substitution
material production use phase material recycling
replaced material
replacing material
mat +Gout
-kGmat in
s(1 -
rec +Gout
-kGrec in
s(1 - k)∑i νi Girec
The definitions of all new parameters are as follows: Gimat ) GHG emissions from material production per kg of material i in the vehicle; γi ) manufacturing yield of material i; ricon ) % of shipped material i coming from secondary production; Gpi ) cradle-to-ingot GHG emissions from primary production of 1 kg of material i; Gis ) cradle-to-ingot GHG emissions from secondary production of 1 kg of material i; and Gif ) ingot-to-finished-material GHG emissions from finishing 1 kg of material i. The GHG emissions for automotive material i are typically calculated by multiplying the cradle-to-gate inventory of one unit of material, e.g., 1 kg, with the total required amount, mxi · γ-1 i , called shipped material. This is more than the material content mxi since the manufacturing processes of automotive components have limited yields, i.e., γi < 1, and thus generate manufacturing scraps (1 - γi)mix, which are either recycled or disposed of. Many important automotive materials have primary or secondary production routes that produce functionally equivalent material with very different GHG emissions. Examples are steel, aluminum, magnesium, and some thermoplastic resins. Equation 15 thus accounts for the fractions of unfinished material i that come from the respective production routes. The GHG emissions of the finishing processes, however, do not depend on the recycled content of the material. Some processes, such as sheet rolling can have technical issues with high recycled content, one example being limited tolerance for impurities. This is reflected in the respective values of ricon, which will be low or zero for finishing processes that currently use little or no material from the secondary production route. Equation 15 uses the standard assumption in LCA that process inventories bI can be modeled as linear functions of their output levels, i.e., bI(m · u) ) m ·bI(u), where m and u are total and unit quantity of the main economic output of process bI. The only remaining issue is then the appropriate choice of bI(u), which is usually the average or marginal technology for a certain geographical and temporal coverage, depending on whether the inventory modeling is attributional or consequential (34). However, the linearity assumption may not hold if the aim is to assess the GHG impacts of a largescale replacement of automotive materials, i.e., if the studied change in output level is likely to have a significant impact on production capacity, technology mix, economies of scale, and technological development and learning. In such a case, the functional unit of the assessment should be changed from one single vehicle to the total number of vehicles that would be affected by the studied material replacement scenario. Additionally, the inventory modeling should explicitly account for the GHG impacts that large scale changes in the output levels of the material production processes are likely to have, since it is likely that in such cases bI(mbefore · u) - bI(mafter · u)*(mbefore - mafter) ·bI(u). 2.5. GHG Emission Impacts from Materials Recycling. The final, and most challenging, step of the inventory model is to assess the GHG implications of the use and generation of secondary material feedstock, called scrap in the remainder of this text. Scrap is here defined as any material output in a product life cycle that is not generated as an intentionally produced economic good but as potential production, manufacturing, or end-of-life waste. Scrap is considered recycled if it is converted into (secondary) economic material 6976
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 18, 2008
secondary mass savings
use phase
k)∑i νi Gimat
GWTWES(1 + s)(1 - k)TM
used in subsequent product life cycles. From an LCA perspective scrap flows are significant since their recycling turns them from waste flows into product flows that typically cross the boundaries of the product life cycle and should thus be treated as coproducts. Coproduction (also called multifunctionality) creates a well-known allocation problem in LCA. Coproduct allocation has been extensively discussed in LCA literature, an overview of which can be found in Guine´e (35). It causes different challenges in attributional and consequential assessments, which is not reflected in ISO 14044 (2006), since it does not distinguish between the two methodologies. ISO 14044 recommends avoiding coproduct allocation through subdivision of the coproducing processes or system expansion. For methodological rigor, a distinction should be made between attributional and consequential system expansion (35). Subdivision is not an option for scrapgenerating processes, which is why the standard approaches in literature are either to use consequential system expansion (also called avoided burden method) or to treat scrap as waste flows (sometimes called recycled content method) (35, 36). The presented inventory model uses consequential system expansion to avoid allocation, i.e., it treats the scrap flows across the vehicle system boundaries as changes in scrap inputs and outputs and models the consequences they have outside of the vehicle life cycle. The definitions of all new parameters (all for material i) are as follows: Grec i ) GHG emission impacts of scrap recycling per kg of material content in the vehicle; sis ) scrap input into secondary material production (in kg/kg); sip ) scrap input into primary material production (in kg/kg); sin i ) total amount of scrap input into vehicle life cycle (in % of shipped material); siout ) total amount of scrap output from vehicle life cycle (in % of shipped material); Ri ) impact of recycling on scrap collection outside of the vehicle life cycle (∈[0,1]); ceim ) collection efficiency of automotive manufacturing (prompt) scrap; seim ) separation efficiency of automotive manufacturing (prompt) scrap; ceieol ) collection efficiency of automotive end-of-life (eol) scrap; and seeol i ) separation efficiency of automotive end-of-life (eol) scrap. For each reference flow RFx, x ) b,c, the GHG emission x , are calculated as impacts of materials recycling, Grec x Grec )
∑
mixGirec )
i
∑ i
mix β (Gs - Gip), x ) b, c γi i i
(16)
where βi denotes the change in secondary production of material i outside of the vehicle life cycle due to the net change in scrap flows across the vehicle system boundary (dotted line in Figure 1). The value of βi is derived by balancing the scrap flows of material i shown in Figure 1: siout - siin - Ri(siout - siin) - sisβi + sipβi ) 0 w βi ) (1 - Ri)
siout - siin sis - sip
(
) (1 - Ri)
siout - sip sis - sip
)
- ricon (17)
Parameters Ri quantify the extent to which changes in scrap input and output to and from the vehicle life cycle impact scrap collection elsewhere. Each additional unit of scrap leaving the vehicle life cycle reduces scrap collection elsewhere by Ri units, each additional unit of scrap entering the vehicle life cycle increases scrap collection elsewhere by
Ri units. In the limit case of Ri ) 0, the changes in scrap flows from and to the vehicle life cycle have no impact on scrap collection outside of the vehicle life cycle. This could, e.g., be the case if scrap collection rates outside of the vehicle life cycle are already close to 100%, and scrap demand consistently exceeds supply. In the limit case of Ri ) 1, the changes in scrap flows from/to the vehicle life cycle cause decreases/ increases in scrap collection outside of the vehicle life cycle of equal size. This could, e.g., be the case if scrap collection rates outside of the vehicle life cycle are low and there are significant disincentives for increased scrap collection, such as low profitability and market demand. Several simplifying assumptions have been used to derive eqs 16 and 17. The first and most important one is the assumption that the changes in secondary production of material i that take place outside of the vehicle life cycle due to the changes in scrap flows impact only primary production of the same material type. This is a common assumption in LCA research and frequently taken as fact, even in ISO 14044 (2006) (36, 37). However, since the causal relationship of displaced production is socioeconomic rather than physical, it is likely that increased secondary production of a material impacts the production of all competing materials and not just its primary production route. A second assumption is that displaced primary production is of the same size as the increase in secondary production, i.e., that the total change in production of material i outside of the vehicle life cycle is zero. However, increased secondary production may lead to an increase in total production, e.g., through lowering cost of production and increasing market competition. Unfortunately, even in consequential system expansion socioeconomic relationships are typically dealt with through accounting rules rather than rigorous study of its dynamics which limits the scientific value of LCA. Some research exists but more work is needed to close this important gap in LCA research (38, 39). Through the use of parameters Ri eqs 16 and 17 advance current LCA practice regarding materials recycling. The equations should be further refined in the future through the use of more sophisticated socioeconomic models of displaced production.
3. Results and Discussion Together, eqs 12, 15, and 16 sum up the significant climate change impacts that automotive materials have over their life cycle: x x x x ) Gmat + Grec + Guse ) Glifecycle
mix
∑γ i
Gif + (1 - Ri)
siout - siin sis - sip
i
[
(1 - ricon)Gip + riconGis +
]
(Gis - Gip)
+ GWTW · EDx · TM (18)
In the limit case that all Ri ) 1, eq 18 simplifies to x ) Glifecycle
∑ i
mix p [G + ricon(Gis - Gip) + Gif] + GWTW · EDx · TM γi i
vehicle, which is the assumption of the standard avoided burden approach for metals (36). Consider now the case of single material substitution, i.e., one material type (i ) out) being replaced with one different material type (i ) in). Using eqs 5, 10, 15, 16, and 18, the change in GHG emissions per kg of replaced material is c b - Glifecycle Glifecycle rec mat + Gout ) GWTWES(1 + s)(1 - k)TM + Gout ∆M mat rec + Gin k(Gin ) + s(1 - k)
∑ ν (G i
mat + Girec i
)
The material substitution reduces life cycle GHG emissions when eq 21 is positive. To examine how the different stages and elements of the material substitution contribute to the life cycle GHG emission changes, it is useful to arrange the terms in eq 21 in the format shown in Table 1. At the material production stage, the emissions savings from the replaced material and the secondary mass savings need to be measured against the emissions of producing the replacing material. The use phase emission reduction is a linear function of the GHG-intensity of the fuel, the fuel energy savings per mass savings, primary and secondary mass savings, and total driven distance. Low-carbon fuels, such as certain biofuels, and increased power train efficiency, e.g., through hybrid designs and regenerative braking, will thus decrease use phase GHG savings from vehicle mass reduction. The material substitution is furthermore likely to change GHG emissions from material recycling. This is the least understood part of the equation and thus deserves special attention. The complete parameterization of the GHG emission calculations facilitates sensitivity analysis. This is illustrated by calculating the derivatives of eq 21 with respect to replacement coefficient k, secondary mass savings s, scrap recycling parameter Rin, and vehicle life TM:
(
)
b c ∂ Glifecycle - Glifecycle ) ∂k ∆M mat rec -GWTWES(1 + s)TM - (Gin + Gin )-s
(
∑ ν (G i
mat + Girec i
)
i
)
c b ∂ Glifecycle - Glifecycle ) ∂s ∆M
GWTWES(1 - k)TM - (1 - k)
(
)
∑ ν (G i
mat + Girec i
)
i
b c ∂ Glifecycle - Glifecycle ) ∂Rin ∆M
[k - s(1 - k)νin] γ1
(
in
out in sin - sin s p sin - sin
(Gins - Ginp)
)
b c ∂ Glifecycle - Glifecycle ) ∂TM ∆M
GWTWES(1 + s)(1 - k) (22)
(19) i.e., life cycle GHG emission impacts are independent of prompt and eol scrap recycling. This result is identical with the earlier-mentioned recycled content method (36). In the limit case that all Ri ) 0, eq 18 simplifies to x ) Glifecycle
mix
∑γ i
i
[
Gip +
siout - sip sis - sip
(Gis - Gip) + Gif
]
+
GWTW · EDx · TM (20)
This means that in this case life cycle GHG emission impacts are independent of the secondary material content of the
(21)
i
Equations 21–22 and Table 1 are now employed to conduct the following case study: Mild steel sheet, used, e.g., in bodyin-whites or closures, is replaced with advanced high strength steel (AHSS) sheet or rolled aluminum. The complete set of required input data, including source documentation, is listed in the Supporting Information. Table 2 shows that, based on the assumptions made to derive eq 21 and the input data from the Supporting Information, both material replacements reduce automotive life cycle GHG emissions by 6.3-6.4 kg CO2eq per kg of replaced mild steel sheet. However, the ways in which these emission reductions are achieved are very different. For rolled VOL. 42, NO. 18, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
6977
TABLE 2. Life Cycle GHG Emission Impacts of Substituting 1 kg of Mild Steel Sheet with 0.6 kg of Rolled Aluminum and 0.75 kg of AHSS Steel Sheet
material production use phase material recycling total material production use phase material recycling total
replaced material
replacing material
secondary mass savings
1 kg mild steel +3.93
0.6 kg aluminum -13.94
0.12 kg +0.80 +8.80
-1.98 +1.95 1 kg mild steel +3.93
+9.21 -4.73 0.75 kg AHSS -2.95
-1.98 +1.95
+1.49 -1.46
b c - Glifecycle k ) 0.6 ( 0.05 w (Glifecycle )⁄
b c Glifecycle - Glifecycle
∆M ) 6.44 ( 1.64 kgCO2 eq
the two material substitutions. Based on this case study it is thus impossible to determine which material substitution generates higher life cycle GHG emission savings. The case study also demonstrates that in the case of vehicle mass reduction through materials substitution, automotive GHG reduction policies require the adoption of a life cycle perspective. This has been recognized by policy makers but has yet to be implemented (2–5, 40). Policies based on a life cycle perspective require rigorous assessment methods. This paper thus presents the first fully parameterized life cycle GHG assessment of automotive material substitution. It is based on a detailed model of primary and secondary mass savings. To support multifuel assessments, use phase emissions are derived from fuel energy demand rather than volumetric fuel economies. Most importantly, the impact of scrap use and generation is modeled through a comprehensive and novel application of consequential system expansion, which should be developed further in future research.
Supporting Information Available This material is available free of charge via the Internet at http://pubs.acs.org.
b c - Glifecycle TM ) 2 · (105 ( 104) km w (Glifecycle )⁄
∆M ) 6.44 ( 0.88 kgCO2 eq
Literature Cited
For AHSS: b c - Glifecycle k ) 0.75 ( 0.05 w (Glifecycle )⁄
∆M ) 6.26 ( 1.25 kgCO2 eq b c s ) 0.3 ( 0.2 w (Glifecycle - Glifecycle )⁄
b c Rin ) 0.1 ( 0.1 w (Glifecycle - Glifecycle )⁄
+5.50
+1.48 +5.50 -0.72 +6.26
(23)
∆M ) 6.44 ( 1.00 kgCO2 eq
∆M ) 6.26 ( 1.02 kgCO2 eq
-0.23 +0.27
-9.21 +8.80 +6.85 +6.44
I thank Professor Atsushi Inaba, Professor Greg Keoleian, Dr. John Sullivan, and Dr. Gerald Rebitzer for their critical review of the methodology and the International Iron and Steel Institute for partial financial support of this research.
)⁄
b c Rin ) 0.1 ( 0.1 w (Glifecycle - Glifecycle )⁄
(24)
∆M ) 6.26 ( 0.16 kgCO2 eq b c - Glifecycle TM ) 2 · (105 ( 104) km w (Glifecycle )⁄
∆M ) 6.26 ( 0.55 kgCO2 eq Based on the assumed error margins for k, s, Rin, and TM, the results appear robust even though their error may be as high as 25%. Based on this case study, substituting automotive mild steel sheet with aluminum and AHSS reduces life cycle GHG emissions at an estimated rate of 5-8 kg CO2eq per kg of replaced material. On the other hand, it can be seen that the error margins of the results are up to 1 order of magnitude larger than the difference between the GHG reductions of 9
+8.80
total
Acknowledgments
∆M ) 6.44 ( 1.55 kgCO2 eq
6978
-0.38 +0.42 0.075 kg +0.50
+5.50
aluminum the increase in GHG emissions from material production is larger than the GHG savings from vehicle weight reduction, since all material comes from GHG-intensive primary production. To achieve net emission reductions, rolled aluminum relies on the emission savings due to prompt and eol scrap recycling. Table 2 shows that the avoided GHG emissions from aluminum recycling are actually larger than the use phase savings. AHSS, on the other hand, reduces the GHG emissions from both vehicle use and material production, since the emission difference between AHSS and mild steel sheet production is insignificant, yet less AHSS sheet is required to achieve the same function. The material substitution slightly decreases the avoided GHG emissions from prompt and eol scrap recycling, since less scrap is now available to displace primary production. The derivatives of eq 22 are now used to gain some sense of the uncertainty of the GHG emission impacts of the material substitutions described above. For aluminum:
s ) 0.3 ( 0.2 w (
use phase
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 18, 2008
(1) UNFCCC. Greenhouse Gas Data 2006; UNFCCC: Bonn, Germany, October 30, 2006. (2) CARB. Report to the Legislature and the Governor on Regulations to Control Greenhouse Gas Emissions from Motor Vehicles; California Air Resources Board (CARB): Sacramento, CA, December 2004. (3) European Commission. Commission Recommendation on the reduction of CO2 emissions from passenger cars. Off. J. Eur. Commun. 1999, 40, 49–50. (4) European Commission. Commission Recommendation on the reduction of CO2 emissions from passenger cars (KAMA). Off. J. Eur. Commun. 2000, 100, 55-56. (5) European Commission. Commission Recommendation on the reduction of CO2 emissions from passenger cars (JAMA). Off. J. Eur. Commun. 2000, 100, 57-58. (6) IFEU. Energy Savings by Light-Weighting; Institut fu ¨ r Energieund Umweltforschung (IFEU): Heidelberg, Germany, 2003. (7) EPA. Light-duty automotive technology and fuel economy trends: 1975 though 2003; EPA420-S-03-004; U.S. Environmental Protection Agency: Washington, DC, 2003. (8) Sullivan, J. L.; Cobas-Flores, E. Full Vehicle LCAs: A Review; SAE Paper 2001-01-3725; SAE International: Warrendale, PA, 2001. (9) Schmidt, W.-P.; Dahlqvist, E.; Finkbeiner, M.; Krinke, S.; Lazzari, S.; Oschmann, D.; Pichon, S.; Thiel, C. Life Cycle Assessment of Lightweight and End-of-Life Scenarios for Generic Compact Class Passenger Vehicles. Int. J. LCA 2004, 9 (6), 405–416.
(10) Birat, J.-P.; Gue´rin, V.; Rocchia, L.; Tuchman, M. Ecodesign of Automobiles Based on the Environmental Properties of Body Materials; SAE Paper 2004-01-0250; SAE International: Warrendale, PA, 2004. (11) Dhingra, R.; Overly, J. G.; Davis, G. A. Environmental Evaluation of Materials in New Generation Vehicles, Center for Clean Products and Clean Technologies; University of Tennessee: Oak Ridge, TN, 2001. (12) Hayashi, T.; Kumata, M.; Aoki, K. Reduction of Life Cycle CO2 Emissions - The Example of Honda Insight; SAE Paper 200101-3722; SAE International: Warrendale, PA, 2001. (13) Das, S. The Life-Cycle Impacts of Aluminum Body-in-White Automotive Material. JOM 2000, 41–44. (14) Saur, K.; Schuckert, M.; Beddies, H.; Eyerer, P. Foundations for Life Cycle Analysis of Automotive Structures - The Potential of Steel, Aluminium and Composites; SAE Paper 951844; SAE International: Warrendale, PA, 1995. (15) Stodolsky, F.; Vyas, A.; Cuenca, R.; Gaines, L. Life-Cycle Energy Savings Potential from Aluminum-Intensive Vehicles; SAE Paper 951837; SAE International: Warrendale, PA, 1995. (16) Aluminium applications and society - Life cycle inventory of the worldwide aluminium industry with regard to energy consumption and emissions of greenhouse gases; Paper 1 - Automotive; International Aluminium Institute (IAI): London, 2000. (17) Ecobilan. LCA of the Automotive Steel Body and Comparison with an Aluminium Body; SM.100; Ecobilan: Paris, February 1997. (18) Hakamada, M.; Furuta, T.; Chino, Y.; Chen, Y.; Kusuda, H.; Mabuchi, M. Life cycle inventory study on magnesium alloy substitution in vehicles. Energy 2007, 32, 1352–1360. (19) Tharumarajah, A.; Koltun, P. Is there an environmental advantage of using magnesium components for light-weighting cars? J. Cleaner Prod. 2007, 15 (2007), 1007–1013. (20) Sullivan, J.; Hu, J. Life Cycle Energy Analysis for Automobiles; SAE Paper 951829; SAE International: Warrendale, PA, 1995. (21) Keoleian, G. A.; Lewis, R. B.; Coulon, R. B.; Camobreco, V. J.; Teulon, H. P. LCI Modeling Challenges and Solutions for a Complex Product System: A Mid-Sized Automobile; SAE Paper 982169; SAE International: Warrendale, PA, 1998. (22) Estudillo, K.; Koehn, J.; Levy, C.; Olsen, T.; Taylor, C. Reducing GHG emissions with hybrid-electric vehicles: An environmental and economic analysis; Group Project Report; Bren School of Environmental Science and Management: UC Santa Barbara, 2005. (23) DBJ. How life cycle assessment (LCA) can enhance the fight against global warming; Research Report No.45; Economic and Industrial Research Department: Development Bank of Japan (DBJ), 2004. (24) Sullivan, J. L.; Williams, R. L.; Yester, S.; Cobas-Flores, E.; Chubbs, S. T.; Hentges, S. G.; Pomper, S. D. Life Cycle Inventory of a Generic U.S. Family Sedan: Overview of Results USCAR AMP Project; SAE Paper 982160; SAE International: Warrendale, PA, 1998.
(25) Geyer, R. Life Cycle Greenhouse Gas Emission Assessments of Automotive Materials. The Example of Mild Steel, Advanced High Strength Steel and Aluminium in Body in White Applications; Bren School of Environmental Science and Management: UC Santa Barbara, 2007. (26) FKA. Communication Module Mass Reduction; Report 56690; Forschungsgesellschaft Kraftfahrwesen (FKA): Aachen, Germany, 2006. (27) Shaw, J. Auto/Steel Partnership light weight front structure: Hydroform solution and cost analysis; Presented at International Auto Body Congress, Novi, MI, September 20, 2006. (28) IEA. Biofuels for Transport - An International Perspective; International Energy Agency (IEA), OECD: Paris, 2004. (29) Delucchi, M. A. Lifecycle Analyses of Biofuels; UCD-ITS-RR-0608; Institute of Transportation Studies (ITS): UC Davis, 2006. (30) GREET. The Greenhouse Gases, Regulated Emissions, and Energy Use in Transportation (GREET) Model 1.8a; Argonne National Laboratory: Argonne, IL, 2007. (31) Wallentowitz, H. Untersuchungen des Zusammenhangs zwischen Pkw-Gewicht und Kraftstoffverbrauch - Messungen an 11 Fahrzeugen auf dem dynamischen Rollenpru ¨ fstand; Research Report P374; Studiengesellschaft Stahlanwendung; Du ¨ sseldorf, Germany, 2000. (32) An, F.; Santini, D. J. Mass Impacts on Fuel Economies of Conventional vs. Hybrid Electric Vehicles; SAE paper 2004010572; SAE International: Warrendale, PA, 2004. (33) FKA. Determination of Weight Elasticity of Fuel Economy for Conventional ICE Vehicles, Hybrid Vehicles and Fuel Cell Vehicles; Report 55510; Forschungsgesellschaft Kraftfahrwesen (FKA): Aachen, Germany, 2006. (34) Ekvall, T.; Weidema, B. P. System Boundaries and Input Data in Consequential Life Cycle Inventory Analysis. Int. J. LCA 2004, 9 (3), 161–171. (35) Guine´e, J. B., Ed. Handbook on Life Cycle Assessment, Operational Guide to the ISO Standards; Kluwer Academic Publishers: Dordrecht, The Netherlands, 2002. (36) Atherton, J. Declaration by the Metals Industry on Recycling Principles. Int. J. LCA 2007, 12 (1), 59–60. (37) Ekvall, T.; Finnveden, G. Allocation in ISO 14041 - a critical review. J. Cleaner Prod. 2001, 9, 197–208. (38) Ekvall, T. System Expansion and Allocation in Life Cycle Assessment; PhD Thesis; Chalmers University of Technology, Go¨teborg, Sweden, 1999. (39) Weidema, B. P. Market Information in Life Cycle Assessment; Environmental Project No. 863 2003; Danish EPA: Copenhagen, Denmark, 2003. (40) EC. Integrated Product Policy, Building on Life-Cycle Thinking; COM (2003) 302 Final; European Commission (EC): Brussels, Belgium, 2003.
ES800314W
VOL. 42, NO. 18, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
6979
