Life Cycle Assessment of Vehicle Lightweighting: A ... - ACS Publications

Nov 15, 2013 - Life Cycle Assessment of Vehicle Lightweighting: A Physics-Based. Model of Mass-Induced Fuel Consumption. Hyung Chul Kim* and Timothy ...
0 downloads 0 Views 2MB Size
Article pubs.acs.org/est

Life Cycle Assessment of Vehicle Lightweighting: A Physics-Based Model of Mass-Induced Fuel Consumption Hyung Chul Kim* and Timothy J. Wallington Systems Analytics and Environmental Sciences Department, Ford Motor Company, Mail Drop RIC-2122, Dearborn, Michigan 48121-2053, United States S Supporting Information *

ABSTRACT: Lightweighting is a key strategy used to improve vehicle fuel economy. Replacing conventional materials (e.g., steel) with lighter alternatives (e.g., aluminum, magnesium, and composites) decreases energy consumption and greenhouse gas (GHG) emissions during vehicle use, but often increases energy consumption and GHG emissions during materials and vehicle production. Assessing the life-cycle benefits of mass reduction requires a quantitative description of the mass-induced fuel consumption during vehicle use. A new physics-based method for estimating mass-induced fuel consumption (MIF) is proposed. We illustrate the utility of this method by using publicly available data to calculate MIF values in the range of 0.2−0.5 L/(100 km 100 kg) based on 106 records of fuel economy tests by the U.S. Environmental Protection Agency for 2013 model year vehicles. Lightweighting is shown to have the most benefit when applied to vehicles with high fuel consumption and high power. Use of the physics-based model presented here would place future life cycle assessment studies of vehicle lightweighting on a firmer scientific foundation.

1. INTRODUCTION Life cycle assessment (LCA) is used to measure the environmental impacts associated with the complete life cycle of a vehicle. The use-phase, that is, vehicle operation, is the most energy consuming and accounts for approximately 60− 90% of the total life cycle energy use of conventional vehicles.1−3 Lightweighting is an effective measure to reduce the use-phase environmental impacts. However, the production and processing of lightweight materials tends to require more energy and thus generates more greenhouse gases (GHGs) than for conventional steel and iron based alloys.1,2,4 An accurate estimation of the impact of mass reduction on the usephase fuel consumption is required to assess the life cycle benefit of vehicle lightweighting. The fuel consumption reduction resulting from lightweighting is typically estimated based on a fuel-mass correlation (e.g., fuel consumption reduction value (FRV)).5−7 Accurately describing the mass-induced fuel consumption is particularly challenging in vehicle component LCAs. Current methods to estimate mass-induced fuel consumption give a wide range of results8 depending on assumptions regarding driving cycle, vehicle design, powertrain type, and whether the powertrain is rematched for performance equivalence. LCA studies in the literature use FRV values in the range 0.20−0.48 L/(100 km 100 kg) for internal combustion engines vehicles (ICEVs). This range of FRV values translates into a 2−3 fold range in the usephase fuel consumption savings. The choice of FRV significantly affects the estimated life cycle benefits of lightweighting for ICEVs.2 Only a few studies have evaluated © 2013 American Chemical Society

the lightweighting effect on the fuel consumption for advanced powertrain vehicles such as hybrid electric vehicle (HEV), battery electric vehicle (BEV), and fuel cell vehicle (FCV). Harmonized general conclusions are not available regarding the correlation of mass and fuel economy for such advanced powertrain vehicles.8−10 Fuel-mass correlation parameters are usually not available to LCA practitioners for specific vehicle models. Often, simple generic values are used which are based on literature heuristics (e.g., 6% fuel consumption reduction per 10% mass reduction). The currently available data and associated models generally do not provide guidance on estimating mass-induced fuel consumption for specific vehicle models currently in use. In this study, we present a novel physics-based model of mass-induced fuel consumption which can be used in vehicle LCAs. The model is transparent and can be used with publicly available information from the U.S. Environmental Protection Agency’s (EPA) fuel economy certification data11 to estimate the mass-induced fuel consumption for specific ICEV models available in the U.S. We provide guidance for LCA practitioners on estimating mass-induced fuel consumption based on the EPA’s fuel economy test data. The model is used to explore the relationship between mass-induced fuel consumption and Received: Revised: Accepted: Published: 14358

July 6, 2013 November 14, 2013 November 15, 2013 November 15, 2013 dx.doi.org/10.1021/es402954w | Environ. Sci. Technol. 2013, 47, 14358−14366

Environmental Science & Technology

Article

vehicle mass, fuel economy, and power. In addition, we will discuss potential extension of this model to advanced powertrain vehicles.

Table 1. Physical Constants and Vehicle Parameters Used to Estimate Fuel Consumption for the Combined U.S. FTP City/Highway Cycle from Eqs 1−5 parameter/ constant

2. VEHICLE FUEL CONSUMPTION We first summarize the elements of fuel consumption in a vehicle, which consist of the portion that provides useful work and that lost without providing useful work.12,13 Fuel economy is typically measured using standardized drive cycles such as the U.S. Federal Test Procedure (FTP) and New European Driving Cycle (NEDC). Integrating the instantaneous power demand over the driving cycle and factoring in the lower heating value of fuel, indicated efficiency (= thermodynamic efficiency, i.e., fraction of fuel energy converted to work), and transmission efficiency gives the total fuel consumption F (L) to operate a vehicle over a driving cycle: F = Fw + Fx + Ff + Fl

(1)

The fuel consumption, F (L), has contributions from the mass-induced load (Fw), aerodynamic resistance and accessory power demand (Fx), mechanical losses in the engine friction and pumping (Ff), and losses outside of the engine (Fl) such as in the differential, axle, and bearings. The contributions to fuel consumption (L) in a driving schedule are described as follows:12,13 Fw =

1 av(1 + ε)M dt Hf ηiηt 1 + CRg (1 − φ)vM dt Hf ηiηt

1 2Hf ηiηt

Ff =

1 2Hf ηi

Fl =

1 Hf ηi

∫ ρCDAFv3dt + ⎛

∫ β dt

ηt ηi

transmission efficiency indicated (thermodynamic) engine efficiency friction mean effective pressure engine displacement gear ratio; N = average engine speed (rps) V = average vehicle speed (m/s) vehicle mass fuel consumption

D N/V

(2)

∫ α dt

value

reference

0.1 9.8 m/s2 0.009 1.2 kg/m3 0.3 0.75 kW 2.4 m2 0.145 2220 m2/s2 17 200 m 6 345 000 m3/s2 312 300 m2/s 1,375 s 32.3 MJ/L (gasoline) 0.88 (automatic) 0.41 (gasoline)

13

13 13 13 14 14 14 14 14 14

150 kPa

12,13

3.0 L 1.33 r/m

13

13

13 13

1800 kg 1.35 L (based on fuel economy of 7.8 L/100 km)

3. MASS-INDUCED FUEL CONSUMPTION- EXISTING MODELS In fuel consumption models currently used in vehicle LCA studies, the fuel consumption during lifetime vehicle travel distance, TD, is typically presented as a simple linear function of vehicle mass, M as follows:6,7

(3)



∫ fmep D⎝ NV ⎠vdt ⎜

1 Hf ηi

rotational mass factor gravitational acceleration rolling resistance coefficient air density aerodynamic drag coefficient power demand for accessories frontal area fraction of idling time = ∫ avdt = ∫ vdt = ∫ v3dt = ∫ v2dt = ∫ dt lower heating value of fuel

M F



Fx =

ε g CR ρ CD α AF φ I1 I2 I3 I4 I5 Hf

f mep



definition



(4)

F = (R wM + γ ) × TD (5)

(6)

where F is the use-phase fuel consumption in L, Rw is the fuel consumption reduction value (FRV) corresponding to the mass induced fuel consumption factor in L/(km·kg), M is vehicle mass in kg, TD is travel distance in km, and constant γ is the fuel consumption factor in L/km related to aerodynamic resistance and accessories including parasitic losses.3,6,8 The FRV, Rw lies in the range 0.15−0.7 L/(100 km 100 kg) for conventional ICEVs.8 Larger values of FRV are usually associated with scenarios where powertrain parameters such as engine size are downscaled along with lightweighting for equal vehicle performance, which can be measured by various indicators such as time to reach 0- to-60 miles per hour and gradeability. In LCAs of vehicle components, researchers need to assign fuel consumption in proportion to component mass assuming that the component has no effect on the vehicle aerodynamics. In other words, only mass-induced fuel consumption should be accounted for. Thus, the use-phase fuel consumption (f) for a component corresponds to

where v = vehicle speed (m/s), a = vehicle acceleration (m/s2), t = time in a driving cycle (s), α = power demand for accessories, and β = power losses outside of the engine. The definition and representative values for the constants and parameters along with integration values, I1−I5 needed to calculate each of the components of fuel consumption in eqs 2−5 are listed in Table 1 for a generic vehicle. Note that some values are hypothetical (e.g., CR). The detailed formulas to derive vehicle load, that is, power demand and loss factors are available in the Supporting Information. The generic vehicle considered here has a 3.0 L engine, mass of 1800 kg, and city/ highway combined fuel economy of 7.8 L/100 km (30 miles per gallon). Since the value of β is not available, the losses outside of the engine, Fl, are estimated as a balance from the total fuel consumption, F (=1.35 L). In this example, Fw = 0.58, Fx = 0.31, Ff = 0.39 and Fl = 0.07 (L), indicating that 66% (= (0.58 + 0.31)/1.35) of the mechanical energy generated by the engine is used to provide the power demand needed to operate the vehicle while the balance is lost.

f = R wm × TD 14359

(7)

dx.doi.org/10.1021/es402954w | Environ. Sci. Technol. 2013, 47, 14358−14366

Environmental Science & Technology

Article

Table 2. FRVs Used in LCA Literature vehicle mass (kg)

fuel consumption (L/ 100 km)

Sullivan & Hu 19951

1395

9.7

0.39 (0.57)

0.27 (0.40)

Stodolsky et al 19954

1441

9.4

0.66

0.43

Saur et al 199715 Kiefer et al 199816 Keoleian et al 199817 Shen et al 199918 Thiel & Jenssen 200019 Keoleian and Kar 200320 Schmidt et al 200421 Birat et al 200422 Tharumarajah & Koltun 200723 Du et al 201024 Dubreuil et al 201025 Keoleian and Sullivan 20123

1800 1018 2766 2426 990 1471 1000 1050 1300

14.0 6.7 14.3 17.4 5.8 7.5 8.1 5.5 8.5

0.5 0.35; 0.45; 0.55 0.44 0.32 0.6 0.4

0.39 0.23; 0.30; 0.36 0.23 0.23 0.35 0.20 0.38 0.26 0.39

1445 1595 1500

10.6 11.1

0.65 0.66

study

fuel-mass coefficient (c)

0.5 0.6

where f is the use-phase fuel consumption of a component in L and m is the component mass in kg. An alternative method of characterizing the correlation between vehicle weight and fuel consumption is to measure the change in fuel use for a given change in vehicle mass, using a fuel-mass coefficient5 c: c=

⎛ ΔF ⎞⎛ M ⎞ ⎛ ΔF ⎞⎛ M ⎞ Cc ⎛M⎞ ⎟ = ⎜ ⎟ = R ⎜ ⎟ ⎟⎜ ⎟⎜ =⎜ w⎝ ⎝ ⎠ ⎝ ⎠ ⎝ ⎠ ⎝ ⎠ Cw F ΔM ΔM F F⎠

0.48 0.46 0.372

source; note from formula (with powertrain adjustment) generic estimate with powertrain adjustment literature estimated for G passenger Van industry estimate literature estimated for Ford Contour literature literature literature literature; with powertrain adjustment simulation

induced fuel consumption of the baseline scenario varies with the choice of FRV (Rw) resulting in 0.3 versus 0.15 L for the case m = 100 kg and TD = 100 km. Therefore, the choice between the two versions of FRV in the baseline creates substantial uncertainties in lightweighting benefits. In addition, this is a contradiction to the premise that a baseline scenario involves no weight reduction; therefore, only one version of mass-induced fuel consumption should exist. Second, for LCA practitioners, it is very difficult to obtain the FRV or fuel-mass coefficient for a specific vehicle model under study. Only a small number of publications exist on FRVs by vehicle manufacturers for a few models.17,20 Current studies do not account for the specific model parameters shown in eqs 2−5, such as rolling resistance (CR), rotational mass factor (ε), and aerodynamic coefficient (CD). Instead, most LCA studies use a simple heuristic fuel-mass coefficient in the range 0.3−0.7, which means that most studies assume that a 10% mass reduction results in a 3−7% decrease in vehicle fuel consumption per distance. Alternatively, some studies rely on very sophisticated simulation tools.3,10,26 For example, a recent study by Keoleian and Sullivan (2012) used a FRV of 0.372 L/ (100 kg 100 km) based on a simulation model to estimate lightweighting impact on LCA. For other simulation studies, it is difficult to directly compare their fuel-mass relation with the FRV range in the literature because their simulation typically includes secondary mass compounding.10,26 Third, and perhaps most importantly, the absence of a uniform approach has led to a wide range of values of FRV (Rw) used in the literature studies (see Table 2). To address these three concerns we propose here a new physics-based model for estimating the mass-induced vehicle fuel consumption. Publicly available data for vehicle performance from the US EPA was used in this model to provide physicsbased estimates for the mass-induced vehicle fuel consumption.

(8)

where CC and CW correspond to the change of fuel consumption (%) and weight (%), respectively. As shown in eq 8, FRV, Rw is a part of the fuel-mass coefficient, c. Then, the use-phase fuel consumption f (L) of a component with mass m (kg) is determined by ⎛m⎞ f = R wm × TD = F ⎜ ⎟c × TD ⎝M⎠

FRV [L/(100 km 100 kg)] (Rw)

(9)

Fuel-mass coefficients, c, cited in the literature lie in the range 0.3−0.7. LCA studies have been conducted to estimate the benefit of replacing conventional materials such as steel and iron with lighter alternatives such as aluminum, magnesium, and fiber reinforced plastics. Table 2 lists FRVs used in such studies. When the fuel-mass correlation was given in the form of a fuelmass coefficient, c, we calculated the corresponding FRV (Rw) using eq 8. The FRVs in the literature span the range 0.2−0.5 L/(100 km 100 kg) reflecting the use of different boundary conditions in different studies. The lower values, that is, 0.2− 0.3 L/(100 kg 100 km) are for scenarios with lightweighting without powertrain adjustment, the higher estimates are for scenarios with powertrain adjustment. Typically, the FRVs used in LCAs are not evaluated for a specific vehicle but are instead generic values deemed relevant to the class of vehicle studied. There are three fundamental issues with the existing methods of assessing mass-induced fuel consumption. First, the mass-induced fuel consumption of the baseline scenario is not clearly defined7 as it depends on the choice of FRV (Rw). As discussed above, previous studies assume that FRVs are much higher when engine parameters are adjusted with lightweighting than when they are not adjusted, e.g., 0.3 versus 0.15 L /(100 kg 100 km). Applying eq 9, the mass-

4. MASS-INDUCED FUEL CONSUMPTION: PROPOSED MODEL Figure 1 gives a schematic of our proposed mass-induced fuel consumption model. The total fuel consumption consists of the fuel consumption to overcome vehicle load, Fload (=Fw + Fx) and that related to lost energy in the engine and drivetrain, Floss 14360

dx.doi.org/10.1021/es402954w | Environ. Sci. Technol. 2013, 47, 14358−14366

Environmental Science & Technology

Article

Fw(M ) =

M [(1 + ε)I1 + (1 − φ)C R gI2] = R wM Hf ηiηt (10)

In the present model, as shown in Figure 1, a fraction of the gross mechanical losses, (Ff + Fl)(Fw/(Fw + Fx)), is also massinduced while the rest, (Ff + Fl)(Fx/(Fw + Fx)), is induced by the aerodynamic resistance and power demand for accessories and are independent of vehicle mass. Thus, the “gross” massinduced fuel consumption, F*w is ⎛ Fw ⎞ ⎛ F + Fx + Ff + Fl ⎞ Fw* = Fw + (Ff + Fl)⎜ ⎟ = Fw ⎜ w ⎟ Fw + Fx ⎝ Fw + Fx ⎠ ⎝ ⎠ ⎛ F ⎞ F R M = Fw ⎜ ⎟ = w = w = R w*M ηm ηm ⎝ Fw + Fx ⎠

(11)

Here, ηm corresponds to the gross vehicle mechanical efficiency, which is the fraction of the total work that is delivered by the engine to the transmission (= (Fw + Fx)/F), while we call Rw* (= Rw/ηim) “Mass-Induced Fuel Consumption (MIF)” or “gross FRV” in this study in comparison to FRV (= Rw). The key difference between our model and the models discussed above is the inclusion of ηm that was ignored in the latter. The usephase fuel consumption (L) for a vehicle component with a mass, m, is represented as

Figure 1. Schematic diagram of fuel use in a driving cycle. Gradient resistance is deemed negligible. Fw= fuel consumption due to massinduced loads; Fx = fuel consumption due to aerodynamic resistance and accessory power loads; Ff = fuel consumption due to mechanical losses in the engine; Fl = fuel consumption due to mechanical losses outside of the engine; Fload = fuel consumption used to overcome vehicle loads; Floss = fuel consumption due to mechanical losses.

R wm × TD = R w*m × TD ηm

f w* =

(12)

Note that ηm is a function of vehicle mass (M). From eq 11, if the mass reduction is a small fraction of the total vehicle mass (ΔM ≪ M), the gross mechanical efficiency ηm would remain nearly unchanged with lightweighting. For example, using the generic parameters in Table 1, a 200 kg mass reduction results in only 1% of decrease in ηm. Thus, taking ηm as constant across weight reduction is a reasonable simplifying assumption. 4.2. Mass Reduction with Powertrain Adjustment. In this approach, powertrain parameters such as gear ratio and engine displacement are adjusted to match the reduced vehicle weight for performance equivalence with the baseline vehicle. There are multiple performance indicators such as acceleration capability, maximum torque, or maximum engine power. In our model, the combined performance parameter (D/M) (N/V) is held constant. The parameter (D/M) represents the normalized engine displacement, while (N/V) indicates the gear ratio (N = average engine speed in rps; V = average vehicle speed in m/s). As the vehicle weight, M, decreases, the engine displacement volume D or gear ratio (N/V) is adjusted. The frictional engine energy loss is proportional to vehicle mass, M:

(=Ff + Fl). The mass-induced fuel consumption from acceleration and rolling resistance, Fw, is defined in eq 2 and used in literature LCA modeling studies as described in eq 6. In our model, we also consider the mass-induced fuel consumption associated with the energy lost without doing useful work. We separate the wasted fuel consumption (Floss) into the mass-induced part, Floss (Fw/(Fw + Fx)), and the balance, Floss (Fx/(Fw + Fx)), based on the energy breakdown for the vehicle load (see Figure 1). Then, we allocate the massinduced wasted fuel consumption to the mass. The rationale behind this is that the wasted fuel consumption stems from inefficiencies in delivering the engine power to overcome massrelated vehicle loads. As a result, the mechanical energy losses, Floss (Fw/(Fw + Fx)), decrease with lightweighting along with the fuel consumption from mass induced-loads, Fw. This is a clear distinction from the principle of existing models described in eq 6, where mechanical energy losses are either unaccounted for or deemed unrelated to vehicle mass. 4.1. Mass Reduction without Powertrain Adjustment. In this approach, vehicle mass is reduced without adjusting powertrain parameters. In the absence of powertrain adjustment the performance (e.g., acceleration rate) of the vehicle will increase after lightweighting. To preserve approximate functional equivalence, that is, performance, in the vehicle (or component) lightweighting scenario, the mass reduction without powertrain adjustment approach is appropriate only for a certain degree of vehicle mass reduction. From eq 2, the fuel consumption from mass-induced loads for a driving cycle Fw corresponds to

⎛ D ⎞⎛ N ⎞ M f ⎜ ⎟⎜ ⎟ 2Hf ηi mep ⎝ M ⎠⎝ V ⎠

Ff+ =

∫ v dt = R f M

(13)

Then, the fuel consumption from mass-induced loads includes this frictional loss term: Fw+ = Fw + Ff+ = +

M [(1 + ε)I1 + (1 − φ)C R gI2] Hf ηiηt

⎛ D ⎞⎛ N ⎞ M fmep ⎜ ⎟⎜ ⎟I2 = (R w + R f )M = R w+M ⎝ M ⎠⎝ V ⎠ 2Hf ηi (14)

R+w

Here, (=Rw + Rf) is the FRV with powertrain adjustment, which corresponds to 0.31 L/(100 km 100 kg) based on the 14361

dx.doi.org/10.1021/es402954w | Environ. Sci. Technol. 2013, 47, 14358−14366

Environmental Science & Technology

Article

Figure 2. Fuel consumption breakdown for a 200 kg weight reduction based on the generic vehicle parameters in Table 1 for the U.S. FTP combined cycle. Fw= fuel consumption due to mass-induced loads; Fx = fuel consumption due to other loads; Floss= total fuel consumption due to mechanical losses; Floss(Fw/Fw + Fx) = mass-induced mechanical losses; Floss(Fx/Fw + Fx) = other mechanical losses; ηm and η+m, = gross mechanical efficiencies; Rw and R+w = fuel reduction values (FRVs); Rw* and R+w* = mass-induced fuel consumptions (MIFs). “*” corresponds to our proposed model and “+” represents powertrain adjustment.

⎛ ⎞ ⎛ Fw + Fx + Ff + Fl ⎞ Fw + Fx =⎜ ⎟×⎜ ⎟ + ηm Fw + Fx ⎝ Fw + Fx + Ff + Fl ⎠ ⎝ ⎠ F + Fx + Ff + Fl F = w = + ≥1 Fw + Fx + Ff+ + Fl F (17)

ηm+

parameter values in Table 1, compared to Rw of 0.19 L/(100 km 100 kg). However, further modification of the formula is needed to conform to the energy flow breakdown in Figure 1. That is, F+f includes fuel consumption that is not induced by vehicle mass, F+f (Fx/(Fw + Fx)), which should be subtracted from the mass-induced fuel consumption. Part of the mechanical loss outside of the engine, Fl (Fw/(Fw + Fx)), is considered as mass-induced and added to the mass-induced fuel consumption. The gross mass-induced fuel consumption, F+w* corresponds to:

Here, the equal sign corresponds to the baseline vehicle. The difference between η+m and ηm would be small for a practical range of mass reduction. For a small mass reduction δ, the difference between F and F+ is δFf. Then, ⎛F ⎞ F + + δFf F = 1 + δ ⎜ +f ⎟ + = ⎝F ⎠ F F+

⎛ Fx ⎞ ⎛ Fw ⎞ Fw+ * = Fw + Ff+ − Ff+⎜ ⎟ + Fl ⎜ ⎟ ⎝ Fw + Fx ⎠ ⎝ Fw + Fx ⎠ ⎛ F + Fx + Ff+ + Fl ⎞ ⎛ F+ ⎞ F = Fw ⎜ w ⎟ = Fw ⎜ ⎟ = w+ ηm Fw + Fx ⎝ ⎠ ⎝ Fw + Fx ⎠ R M = w+ = R w+ *M ηm (15)

Thus, for a small range of δ, ηm+ ≅ ηm , R w+ * ≅ R w*

R wm × TD = R w+ *m × TD ηm+

(19)

Figure 2 gives the breakdown of fuel consumption, FRVs and MIFs estimated from our model in comparison with the existing model for an example of 200 kg weight reduction from the generic vehicle in Table 1. For a 200 kg reduction scenario, the FRVs without and with adjustment of powertrain, that is, Rw and R+w are 0.19 and 0.31 L/(100 km 100 kg), respectively, while the MIFs, that is, R*w and R*w + derived from our model are 0.28 L /(100 km 100 kg). The gross mechanical efficiency, ηm and η+m remains 0.66−0.67. For a specific vehicle model, these figures need to be evaluated based on measured vehicle parameters. We note that other definitions of performance equivalency can be used to describe the lightweighting scenarios with powertrain adjustment. An adequate formulation of the gross mechanical efficiency in accordance with the defined performance equivalency can be adapted to our model. For example, a regression model in the U.S. EPA study correlates the 0−60 mph acceleration time with vehicle mass and power as follows:27

Here R+w* and ηm+ correspond to the mass-induced fuel consumption (MIF) and gross vehicle mechanical efficiency, respectively, while F+ is the total fuel consumption with powertrain adjustment. The use-phase fuel consumption for a component with mass m (kg) is f w+ * =

(18)

(16)

From eqs 13 to 15, η+m is a function of vehicle mass M. There are two practical notes useful for lightweighting LCA. First, similar to the case without powertrain adjustment, the sensitivity of η+m over a practical range of mass reduction is small. In the current example, 200 kg of mass reduction (from 1800 to 1600 kg) results in a 1% reduction in η+m (Table 1 and Figure 2). Second, η+m is greater than or equal to ηm as shown below: 14362

dx.doi.org/10.1021/es402954w | Environ. Sci. Technol. 2013, 47, 14358−14366

Environmental Science & Technology

Article

Figure 3. Fuel reduction values (FRVs, Rw) and mass-induced fuel consumption (MIFs, Rw*) estimated from the 2013 model year EPA fuel economy test data for 106 vehicles.

⎛ M ⎞j ⎟ t0 − 60mph = k ⎜ ⎝ HP ⎠

load parameters of light duty vehicles are available in the U.S. EPA certification data measured by the Federal Test Procedure (FTP). The gross vehicle mechanical efficiency, ηm, is estimated based on the fuel economy and load parameters. Data from a homogeneous cohort of cars with automatic transmissions and gasoline engines were evaluated (vehicles with hybrid powertrains or turbo chargers were excluded). After deleting duplicate tests and tests with minimal modifications in options, the massinduced fuel consumption was evaluated for a total of 106 test records. The detailed calculation steps are explained in the following equations. The vehicle load for specific vehicles for the FTP city and highway test cycle, Pload, is available in the EPA certification database for light-duty vehicles (www.epa.gov/otaq/tcldata. htm).11 The load is estimated from dynamometer test data using the following equation:

(20)

where M = vehicle mass (kg), HP = engine horsepower (hp), k = 0.529, and j = 0.805 for vehicles with automatic transmissions and gasoline engines. In this model, the performance is maintained by holding the 0−60 mph acceleration time constant while changing the vehicle mass M. The engine power (HP) is correlated with engine displacement (D). Assuming a simple proportional relationship between HP and D, the powertrain adjustment scenario corresponds to that of our model described in eqs 13 and 14. One simple way of modeling the powertrain adjustment scenarios may be to utilize the fuel economy of the lightweighted vehicle with powertrain adjustment, F+, if available. From eq 17, η+m can be calculated from F+ since ηm is given from eq 11. More sophisticated indicators of vehicle functional equivalency are available elsewhere.28

Pload = Av + Bv 2 + Cv 3 + avM + α

(21)

Here, the “coast down” coefficients, A (N), B (N/(m/s)), and C (N/(m/s)2) are the rolling, rotating and aerodynamic resistive coefficients, respectively. These coast down coefficients (also called “target coefficients”) for each vehicle are available in the EPA database.11,13 The fuel consumption from massinduced loads, Fw (L) can be determined from the following equation:

5. APPLICATION TO SPECIFIC VEHICLE MODELS We used the model to estimate the MIFs (= gross FRVs) for 2013 Model Year ICEVs using the U.S. EPA’s fuel economy certification data.11 In our model, estimation of mass-induced fuel consumption is parsed into two steps. First, FRV (Rw) is estimated based on vehicle load parameters. Second, the gross vehicle mechanical efficiency (ηm) is estimated from fuel consumption and energy loss profiles. As shown in Figure 2, the MIF estimated in this method can be used for both baseline and lightweight scenarios with or without powertrain adjustment for a practical range of lightweighting in which mechanical efficiency (ηm) is not substantially changed. The

Fw(L) = = 14363

1 (Av + Bv 2 + avM )dt Hf ηiηt AI2 + BI4 + MI1



106Hf ηiηt

(22)

dx.doi.org/10.1021/es402954w | Environ. Sci. Technol. 2013, 47, 14358−14366

Environmental Science & Technology

Article

All the necessary parameters to calculate Fw are listed in Table 1. The constant 106 in denominator is for unit conversion. Note that the all the parameters including coast down coefficients are in SI units in eq 22. Then, the FRV, Rw is estimated from eqs 10 and 22. Fuel consumption related to aerodynamic and accessory load, Fx (L) can be calculated from Fx(L) =

1 Hf ηi



⎛ C v3 ⎞ CI3 + 103ηtαI5 ⎜⎜ + α⎟⎟dt = 106Hf ηη ⎝ ηt ⎠ i t

Still, a moderate correlation between the normalized fuel economy and the FRVs is shown in Figure 3(b). It is probably related to the fact that the coast-down coefficients measured in the EPA test contain information on vehicle efficiency. In other words, more-efficient cars, for example, with superior aerodynamic design have lower coast down coefficients and experience lower load during the test. We also observe that engine power and thus displacement is closely related to the MIFs as fuel economy is usually a function of maximum power (Figure 3(c)). However, it is found that vehicle mass does not have a strong correlation with the MIFs (Figure 3(d)). This is because the test data set includes luxury vehicles which have higher engine power and higher fuel consumption than, but similar mass to, nonluxury vehicles.

(23)

The total fuel consumption, F in eq 1 can be derived from the unadjusted fuel economy data in the EPA database. The city/ highway combined average fuel economy in miles per gallon, Ecomb is estimated from the FTP city- and highway-fuel economy, Ecity and Ehwy respectively, listed in the database. Ecomb(mile/gal) =

6. DISCUSSION In summary, we propose use of MIF values, R*w to calculate the mass-induced fuel consumption for the use-phase inventory analysis in vehicle component LCAs. There are two steps to find the MIFs: calculating the FRV (Rw) and the gross mechanical efficiency (ηm) then applying eq 11 to calculate MIF = Rw* = Rw/ηm. Typically in LCA analyses FRVs are inferred from the vehicle load equation for a test cycle or estimated using heuristic approaches. We present a method to calculate the FRV for specific vehicle models from the EPA fuel economy test data. The gross mechanical efficiency is modeled as a function of vehicle mass and can also be calculated from the fuel economy data. The resulting MIFs correlate with vehicle fuel economy (Figure 3a and b) and engine power (Figure 3c). Our model provides important theoretical and practical advantages to the conventional method of estimating massinduced fuel consumption that rely on FRVs. First, the massinduced fuel consumption of a component for the baseline scenario is clearly defined. The FRV is typically measured in two versions: with and without powertrain adjustment. The former FRV is larger than the latter because it entails powertrain resizing for performance equivalency. Therefore, the baseline mass-induced fuel consumption is greater when the lightweighting scenario entails powertrain adjustment than when the scenario does not assume the adjustment. This poses a significant methodological uncertainty in lightweighting LCA as discussed in section 3. On the other hand, our model shows that Rw = R+w for the baseline case since ηm = η+m at the baseline as shown by eq 17. Thus the baseline mass-induced fuel consumption remains the same regardless of powertrain adjustment Another important benefit of our model is that the gross mechanical efficiencies with and without powertrain adjustment are almost equal within a practical range of lightweighting, that is, η+m ≅ ηm allowing use of the MIF without powertrain adjustment as a surrogate for that with powertrain adjustment, that is, R+w* ≅ R*w as shown in eq 19. Estimating the latter value is particularly difficult as it involves modeling engine friction losses which vary widely with engine technology. It is of interest to compare conventional FRV with our MIF values. As shown in Figure 2, the conventional FRVs without powertrain adjustment Rw significantly (20−50%) underestimate the mass-induced fuel consumption compared with MIFs, R*w because they ignore the mechanical energy losses induced by mass. For a given mass, the fuel consumption based on Rw and Rw* corresponds to Fw and [Fw + Floss(Fw/(Fw + Fx))], respectively, in Figure 1. On the other hand, we found

1 0.55 Ecity

0.45 E hwy

+

(24)

Note that this is the laboratory fuel economy used for the corporate average fuel economy standard and is higher than the adjusted value used in the fuel economy label. Then, the total fuel consumption over the combined FTP cycle is Fcomb(L) =

2.35 Ecomb



v dt =

2.35 × 10−3I2 Ecomb

(25)

Here, Fcomb is the total fuel consumption during the FTP combined cycle in L and the constant 2.35 is included for unit conversion between miles per gallon and liters per km. Finally, the MIF, Rw* (L/100 kg 100 km) is calculated based on the EPA database from eqs 11 and 15: R w*[L /(100kg100km)] = =

F R Rw = comb w ηm Fw + Fx

107FcombFw 2.35 × 104Fw = (Fw + Fx )I2M Ecomb(Fw + Fx )M

(26)

Here, Fcomb, Fw, and Fx are in L while Rw and R*w are in L/(100 kg 100 km). Values for Ff and Fl are not available in the database but are not necessary to estimate Rw* as shown in eq 26. We assume that ηt =0.88, ηi = 0.41, Hf = 32.3 MJ/kg, and α = 0.75 kW (Table 1). The mass, M (kg), listed in the EPA database corresponds to the equivalent test weight (ETW) used in the dynamometer testing of a vehicle that is based on loaded vehicle weight. As noted before, one can assume Rw* = R+w* for a practical range of mass reduction. Figure 3 gives the FRVs (Rw) and MIFs (R*w ) for the 106 test records plotted against vehicle parameters. The FRVs and MIFs lie in the ranges 0.15−0.26 L/(100 km 100 kg) and 0.21−0.48 L/(100 km 100 kg), respectively, and are consistent with results reported in literature. Our model shows that the MIFs (R*w ) have a strong linear correlation with the fuel consumption of a vehicle while the FRVs (Rw) are unrelated or insignificantly related to it. Figure 3 (a) and (b) shows that a unit mass reduction applied to a less efficient vehicle saves more fuel than the same mass reduction applied to a more efficient vehicle. This has not been observed using the FRVs. As discussed before, the FRVs (Rw) used in LCAs do not carry information on vehicle parameters; therefore, they have been universally applied to all types of vehicles. Thus, LCAs have not captured the real benefit of lightweighting for specific vehicle models. 14364

dx.doi.org/10.1021/es402954w | Environ. Sci. Technol. 2013, 47, 14358−14366

Environmental Science & Technology that the conventional FRVs based on powertrain adjustment, R+w in eq 14, can be either greater or smaller than MIF, R*w depending on the characteristics of a vehicle including engine size and energy loss profile. For a given mass, the fuel consumption based on R+w and R*w corresponds to [Fw + Ff] and [Fw + Floss(Fw/(Fw + Fx))], respectively, in Figure 1. The difference seems to be insignificant compared with the former comparison as shown in Figure 2 although further analysis is beyond the scope of this analysis. The key policy implication of our model is that the benefit of lightweighting depends on vehicle fuel economy and power. Applying our model to the EPA fuel economy test results shows that a unit mass reduction saves more fuel from less fuel efficient vehicle models than from more fuel efficient models and from higher power rather than lower power vehicles. The existing models based on FRVs do not provide such insight because FRVs are not related significantly to vehicle fuel economy or maximum engine power as shown in Figure 3b and c. Vehicles with an advanced powertrain such as hybrid electric vehicle (HEV), battery electric vehicle (BEV), and fuel cell vehicle (FCV) would have a different range of efficiency parameters such as thermodynamic and mechanical efficiency. Moreover, the distinct powertrain configuration of these vehicles would pose a challenge of modeling additional factors such as efficiencies of motor and regenerative brake. However, a top-down approach corresponding to eq 26 can be applied to these vehicles. That is, the ratio of mass-induced to total fuel consumption (= Fw/(Fw + Fx)) together with fuel economy determines MIF (R*w ). Incorporating additional parameters such as motor and battery efficiencies, the present model can readily be adapted for the advanced powertrain vehicles. This would be particularly pertinent for HEVs and PHEVs whose test records are available in the EPA database. Detailed analysis on estimating MIFs for these vehicles is left for future research. Another area of model refinement would be the formulation of secondary weight savings within our model. The powertrain adjustment in parallel with the initial lightweighting may result in a reduced cylinder size and other mass savings in the powertrain. Therefore, further reduction factors in vehicle loads or engine losses need to be developed for a more accurate analysis. Our model is designed to use publically available EPA vehicle test database. LCA practitioners can estimate the mass-induced use-phase fuel consumption for a component or vehicle following the steps in this paper. In addition, this approach provides great flexibility of modeling for LCA practitioners in terms of vehicle choice and lightweighting boundary conditions.



ACKNOWLEDGMENTS



REFERENCES

We thank John Ginder, Sandy Winkler, Thomas Megli, and Patrick Phlips at Ford, and Claire Boland, Robb De Kleine, and Greg Keoleian at University of Michigan for valuable discussions and comments. While this article is believed to contain correct information, Ford Motor Company (Ford) does not expressly or impliedly warrant, nor assume any responsibility, for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, nor represent that its use would not infringe the rights of third parties. Reference to any commercial product or process does not constitute its endorsement. This article does not provide financial, safety, medical, consumer product, or public policy advice or recommendation. Readers should independently replicate all experiments, calculations, and results. The views and opinions expressed are of the authors and do not necessarily reflect those of Ford. This disclaimer may not be removed, altered, superseded or modified without prior Ford permission

(1) Sullivan, J. L.; Hu, J. Life cycle energy analysis for automobiles. In Total Life Cycle Conference and Exposition; Society of Automotive Engineers: Vienna, Austria, October 16−19, 1995. (2) Kim, H. C.; Wallington, T. J. Life-cycle energy and greenhouse gas emission benefits of lightweighting in automobiles: Review and harmonization. Environ. Sci. Technol. 2013, 47, 6089−6097. (3) Keoleian, G. A.; Sullivan, J. L. Materials challenges and opportunities for enhancing the sustainability of automobiles. MRS Bull 2012, 37 (April), 365−372. (4) Stodolsky, F.; Vyas, A.; Cuenca, R.; Gaines, L. Life-cycle energy savings potential from aluminum-intensive vehicles. In Total Life Cycle Conference and Exposition; Society of Automotive Engineers: Vienna, Austria, October 16−19, 1995. (5) Ridge, L. EUCARAutomotive LCA guidelinesPhase 2. Total Life Cycle Conference and Exposition, Graz, Austria, December 1−3; Society of Automotive Engineers, 1998. (6) Geyer, R. Parametric assessment of climate change impacts of automotive material substitution. Environ. Sci. Technol. 2008, 42, 6973−6979. (7) Koffler, C.; Rohde-Brandenburger, K. On the calculation of fuel savings through lightweight design in automotive life cycle assessments. Int. J. Life Cycle Assess. 2010, 15, 128−135. (8) Wohlecker, R.; Johannaber, M.; Espig, M. Determination of Weight Elasticity of Fuel Economy for ICE, Hybrid and Fuel Cell Vehicles. SAE World Congress, Detroit, MI, April 16−19; Society of Automotive Engineers, 2007. (9) An, F.; Santini, D. J. Mass impacts on fuel economies of conventional vs. hybrid electric vehicles. In SAE World Congress; Society of Automotive Engineers: Detroit, MI, March 8−11, 2004. (10) Moon, P.; Burnham, A.; Wang, M. Vehicle-cycle energy and emission effects of conventional and advanced vehicles. In SAE World Congress; Society of Automotive Engineers: Detroit, MI, April 3−6, 2006. (11) U.S. EPA Emissions and Fuel Economy Test Data. http://www. epa.gov/otaq/tcldata.htm (accessed March 2013). (12) Ross, M. Fuel efficiency and the physics of automobiles. Contemp. Phys. 1997, 38 (6), 381−394. (13) Nam, E. K.; Giannelli, R. Fuel Consumption Modeling of Conventional and Advanced Technology Vehicles in the Physical Emission Rate Estimator (PERE), EPA420-P-05-001; U.S. Environmental Protection Agency, 2005. (14) U.S. EPA Dynamometer Drive Schedules. http://www.epa.gov/ nvfel/testing/dynamometer.htm (accessed March 2013). (15) Saur, K.; Hesselbach, J.; Stichling, J.; Wiedemann, M. Environmental friendly and cost efficient body designs. In International

ASSOCIATED CONTENT

S Supporting Information *

Supporting Information includes equations for vehicle loads and fuel consumptions, and acronyms and symbols used in this manuscript. This material is available free of charge via the Internet at http://pubs.acs.org.





Article

AUTHOR INFORMATION

Corresponding Author

*(H.C.K.) Phone: 313-323-9745; e-mail: [email protected]. Notes

The authors declare no competing financial interest. 14365

dx.doi.org/10.1021/es402954w | Environ. Sci. Technol. 2013, 47, 14358−14366

Environmental Science & Technology

Article

Body Engineering Conferences and Exposition; Automotive Technology Group, Inc.: Stuttgart, Germany, September 30 to October 2, 1997. (16) Kiefer, B.; Deinzer, G.; Haagensen, J. Ö .; Saur, K. Life cycle engineering study of automotive structural parts made of steel and magnesium. In Total Life Cycle Conference and Exposition; Society of Automotive Engineers: Graz, Austria, December 1−3,1998. (17) Keoleian, G. A.; Spatari, S.; Beal, R. T.; Stephens, R. D.; Williams, R. L. Application of Life Cycle Inventory Analysis to Fuel Tank System Design. Int. J. Life Cycle Assess. 1998, 3 (1), 18−28. (18) Shen, D.; Phipps, A.; Keoleian, G.; Messick, R. Life-cycle assessment of a powertrain structural component: Diecast aluminum vs. hypothetical Thixomolded® magnesium. In SAE International Congress and Exposition; Society of Automotive Engineers, Inc.: Detroit, MI, March 1−4, 1999. (19) Thiel, C.; Jenssen, G. B. Comparative life cycle assessment of aluminium and steel bumper carriers. In Total Life Cycle Conference and Exposition; Society of Automotive Engineers: Detroit, MI, April 26− 28, 2000. (20) Keoleian, G. A.; Kar, K. Elucidating complex design and management tradeoffs through life cycle design: Air intake manifold demonstration project. J. Cleaner Prod. 2003, 11, 61−77. (21) 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. Life Cycle Assess. 2004, 9 (6), 405−416. (22) Birat, J.; Guérin, V.; Rocchia, L.; Tuchman, M. Ecodesign of Automobiles based on the environmental properties of body materials. In SAE World Congress and Exhibition; Society of Automotive Engineers: Detroit, MI, March 8−11, 2004. (23) Tharumarajah, A.; Koltun, P. Is there an environmental advantage of using magnesium components for light-weighting cars? J. Clean. Prod. 2007, 15, 1007−1013. (24) Du, J.; Han, W.; Peng, Y. Life cycle greenhouse gases, energy and cost assessment of automobiles using magnesium from Chinese Pidgeon process. J. Clean. Prod. 2010, 18, 112−119. (25) Dubreuil, A.; Bushi, L.; Das, S.; Tharumarajah, A.; Gong, X. A comparative life cycle assessment of magnesium front end autoparts. In SAE World Congress; Society of Automotive Engineers: Detroit, MI, April 13−15, 2010. (26) Kim, H.-J.; McMillan, C.; Keoleian, G. A.; Skerlos, S. J. Greenhouse gas emissions payback for lightweighted vehicles using aluminum and high-strength steel. J. Ind. Ecol. 2010, 14 (6), 929−946. (27) U.S. EPA Light-Duty Automotive Technology, Carbon Dioxide Emissions, and Fuel Economy Trends: 1975 Through 2012; EPA-420-R13-001; U.S. Environmental Protection Agency, 2013. (28) An, F.; DeCicco, J. Trends in technical efficiency trade-offs for the U.S. light vehicle fleet. In SAE World Congress; Society of Automotive Engineers: Detroit, MI, April 16−19, 2007.

14366

dx.doi.org/10.1021/es402954w | Environ. Sci. Technol. 2013, 47, 14358−14366