Content uploaded by Conor C O Reynolds

Author content

All content in this area was uploaded by Conor C O Reynolds on Mar 12, 2014

Content may be subject to copyright.

IOP PUBLISHING ENVIRONMENTAL RESEARCH LETTERS

Environ. Res. Lett. 2(2007) 014003 (8pp) doi:10.1088/1748-9326/2/1/014003

How hybrid-electric vehicles are different

from conventional vehicles: the effect of

weight and power on fuel consumption

CReynolds

1and M Kandlikar2,3,4

1Institute for Resources, Environment and Sustainability, University of British Columbia,

2202 Main Mall, Vancouver, BC, V6T 1Z4, Canada

2Institute of Asian Research, University of British Columbia, 1855 West Mall, Vancouver,

BC, V6T 1Z2, Canada

3Liu Institute for Global Issues, University of British Columbia, 6476 NW Marine Drive,

Vancouver, BC, V6T 1Z2, Canada

E-mail: c.reynolds@ires.ubc.ca and mkandlikar@ires.ubc.ca

Received 26 January 2007

Accepted for publication 7 March 2007

Published 28 March 2007

Online at stacks.iop.org/ERL/2/014003

Abstract

An increasingly diverse set of hybrid-electric vehicles (HEVs) is now available in North

America. The recent generation of HEVs have higher fuel consumption, are heavier, and are

signiﬁcantly more powerful than the ﬁrst generation of HEVs. We compare HEVs for sale in

the United States in 2007 to equivalent conventional vehicles and determine how vehicle weight

and system power affects fuel consumption within each vehicle set. We ﬁnd that heavier and

more powerful hybrid-electric vehicles are eroding the fuel consumption beneﬁt of this

technology. Nonetheless, the weight penalty for fuel consumption in HEVs is signiﬁcantly

lower than in equivalent conventional internal combustion engine vehicles (ICEVs). A 100 kg

change in vehicle weight increases fuel consumption by 0.7l/100 km in ICEVs compared with

0.4l/100 km in HEVs. When the HEVs are compared with their ICEV counterparts in an

equivalence model that differentiates between cars and sports-utility vehicles, the average fuel

consumption beneﬁt was 2.7l/100 km. This analysis further reveals that a HEV which is

100 kg heavier than an identical ICEV would have a fuel consumption penalty of

0.15 l/100 km. Likewise, an increase in the HEV’s power by 10 kW results in a fuel

consumption penalty of 0.27 l/100 km.

Keywords: hybrid-electric vehicles, fuel consumption, performance, vehicle weight, system

power

1. Introduction

Hybrid-electric vehicles (HEVs) use batteries, electric motors,

regenerative braking and reduction of engine idling time to

enhance a conventional internal combustion engine. This

approach is a proven means of reducing the fuel consumption

and/or improving the performance of light-duty passenger

vehicles [1,2]. The fuel consumption of HEVs is lower for

4Author to whom any correspondence should be addressed.

a number of reasons: the electric motor provides a portion of

the power for propulsion, especially at high-load conditions

when the vehicle is accelerating; some of the vehicle’s inertial

energy can be recaptured through regenerative braking systems

and stored in vehicle batteries; an engine-stop feature reduces

idling fuel consumption whenever the car is coasting, braking,

or stopped5. It has been argued that policies are needed

5This feature is also employed by so-called ‘mild hybrids’, which are

essentially conventional vehicles with oversized starter motors. Mild hybrids

such as the Saturn VUE and the GM Silverado are excluded from this study.

1748-9326/07/014003+08$30.00 1 ©2007 IOP Publishing Ltd Printed in the UK

Environ. Res. Lett. 2(2007) 014003 C Reynolds and M Kandlikar

to reduce transportation fuel consumption, as part of a suite

of approaches to reduce carbon emissions worldwide [3–6].

Hybrid-electric technology is expected to play a signiﬁcant

role in achieving this. Yet these studies estimate potential fuel

savings by simply scaling the ﬂeet average fuel consumption

by a given percentage. In order to more accurately calculate

the magnitude of fuel savings, a more robust quantitative

analysis of the relationship between weight, performance and

fuel consumption of HEVs vis-`a-vis conventional internal

combustion engine vehicles (ICEVs) is needed.

Previous studies have successfully modeled the fuel

consumption, weight, and performance tradeoffs of HEVs

based on an assessment of one or two HEV models [7,8].

This study aims to add to that body of work by analyzing

speciﬁcations and fuel consumption data for 2007 model

year hybrid-electric vehicles and their ICEV counterparts in

the North American market. We begin by examining the

expanding hybrid ﬂeet in North America, and present data on

the changing nature of the US HEV ﬂeet as hybrid-electric

technology is adopted across a more diverse array of light-

duty passenger vehicle platforms. Using linear regression

models, we analyze how vehicle weight and power affect the

fuel consumption of HEVs and ICEVs independently, and then

we compare the set of currently available HEVs (2007 model

year) against a functionally equivalent set of conventional

ICEVs. The ‘equivalent set’ are ICEVs of the same make

and model, and with similar power. These analyses provide a

more accurate quantitative picture of the relationship between

fuel consumption, vehicle weight and system power, using

empirical data from commercially available vehicles.

2. Diversifying hybrid-electric vehicle ﬂeet

Automotive manufacturers are offering an increasing variety

of HEV model types. In 2007, there were nine light-duty HEV

models available in the US, and this number is projected to

double in the next two to three years [9]. Figure 1shows

the annual and cumulative sales of HEVs in the US between

1999 and 2006, as well as the number of car and sports-

utility vehicle (SUV) models available in each year. SUVs

made up approximately 30% total hybrid sales in 2006. In

addition to the growing diversity of model offerings, HEV

annual sales continue to grow both in numbers and as a

proportion of the total light-duty vehicle sales. 2006 HEV

sales in the US represent approximately 1.6% of the year’s

new vehicle sales [10,11], with predictions of continued

increase [12]. Vehicle manufacturers publish performance

speciﬁcations for individual HEV models, including gasoline

engine size (displacement), vehicle curb weight, net hybrid-

electric system power6, and acceleration. These HEV

data are presented for all of the HEVs investigated in this

study (table 1). An investigation of air pollutant emissions

performance of HEVs is beyond the scope of this paper.

However, it should be noted that since HEVs were introduced

6Torque is another characteristic that can be used to compare performance,

since it inﬂuences vehicle acceleration. However, we have chosen system

power as it is more commonly presented as a measure of powertrain

performance, and it is directly related to torque through engine speed.

0

100

200

300

400

500

600

700

800

1999 2000 2001 2002 2003 2004 2005 2006

(1/0) ( 2/0) ( 2/0) (2 /0) (2 /0)

(6/3)

(4/2)

(3/1)

HEV Sales, 1000s

2004:

Toyota Prius 2nd

generation introduced;

Also, first hybrid-electric

SUV available for sale

(Ford Escape/Mercury

Mariner)

2006:

Nine HEV models

available;

SUV sales reach

29% of total 2006

sales

1999:

First HEVs sold

(17 Honda

Insights)

Cars: annual sales

SUVs: annual sales

Cumulative HEV sales

Figure 1. Car and sports-utility vehicle (SUV) shares of HEV annual

sales in the United States; annual SUV sales have been increasing

rapidly since their introduction in 2004. Cumulative HEV sales are

also shown, which are assumed to closely represent the number of

HEVs currently in use. Numbers in parentheses show numbers of

car/SUV models available for sale in a given year [11].

in North America, they have been lower-emitting than their

ICEV counterparts.

The combined fuel consumption rate is from US

Environmental Protection Agency (EPA) 2-cycle test results,

and refers to 45% highway and 55% city driving. The

EPA has recently proposed a new 5-cycle methodology that

will more accurately reﬂect actual driving behaviour and

corresponding fuel consumption [13]. The EPA predicts that

this methodology will increase combined fuel consumption

by 8% for conventional light-duty passenger vehicles, and by

16% for hybrid-electric vehicles. It is likely that the new

methodology will be implemented for the 2008 model year

in the US, so this study uses fuel consumption ﬁgures that

have been corrected by these factors (table 1). This approach

has been used before [14], and is supported by the reported

difference between real-world fuel consumption for the Toyota

Prius and its EPA 2-cycle fuel consumption rating [15].

From 2000 to 2006, the sales-weighted average hybrid-

electric vehicle in the US ﬂeet has changed signiﬁcantly, driven

largely by the introduction of new sports-utility and high-

performance HEV models. The average curb weight has

increased by 30%. Propelling this larger weight is a hybrid-

electric system that delivers 60% more power. The gasoline

engine component of this system is 43% larger in terms of

engine displacement. Some of the observed net power increase

is explained by the need to provide a larger vehicle with

acceptable performance. Over the same period, however, the

manufacturer-reported acceleration times also increased: the

average HEV in 2004 reaches 96.6kmh

−1(60 miles per hour)

from a standing start in 20% less time than the average in 2003.

Because vehicle weight and power both strongly inﬂuence fuel

consumption, it is not surprising that average fuel consumption

has gone up by 15% with the shift towards higher-performance

HEVs (ﬁgure 2).

2

Environ. Res. Lett. 2(2007) 014003 C Reynolds and M Kandlikar

Tab l e 1. Speciﬁcations of the HEV models considered in the analysis (ref: manufacturer speciﬁcations). The ﬁrst-generation Toyota Prius

and the Honda Insight were discontinued in 2003 and 2006, respectively, and the reported fuel consumption for the Honda Accord was

improved for the 2007 model year.

Combined fuel

consumptionc

Hybrid-electric

vehicles Model

year Drive

typea

Trans-

missionb

Engine

displ.

(l)

Curb

weight

(kg)

Net

power

(kW)

Accel’n.

(0–96.6

km h−1)

(s)

EPA

2-cycle

l/100 km

(mpg)

Corrected

l/100 km

(mpg)

Cars

Toyota Prius (1st Gen.) 2003 FWD ECVT 1.5 1237 75 12.7 4.9 (48) 5.7 (41)

Toyota Prius (2nd Gen.) 2006/7 FWD ECVT 1.5 1310 82 10.2 4.3 (55) 5.0 (47)

Toyota Camry 2006/7 FWD ECVT 2.4 1669 140 8.9 6.0 (39) 7.0 (34)

Honda Insight 2006 FWD CVT 1.0 896 53 10.7 4.2 (56) 4.9 (48)

Honda Civic 2006/7 FWD CVT 1.3 1304 82 11.0 4.7 (50) 5.5 (43)

Honda Accord 2006 FWD Auto 3.0 1627 189 6.5 8.4 (28) 9.7 (24)

Honda Accord 2007 FWD Auto 3.0 1635 189 6.5 7.6 (31) 8.8 (27)

Lexus GS 450h 2006/7 RWD ECVT 3.5 1875 254 5.2 9.0 (26) 10.5 (22)

Nissan Altima 2007 FWD ECVT 2.5 1588 148 — 6.0 (39) 7.0 (34)

Sports-utility vehicles

Toyota Highlander 2006/7 AWD Auto 3.3 1925 200 7.2 8.1 (29) 9.4 (25)

Ford Escape/Mercury Mariner 2006/7 AWD ECVT 2.3 1718 116 9.0 7.6 (31) 8.8 (27)

Lexus RX 400h 2006/7 AWD ECVT 3.3 1980 200 7.3 8.1 (29) 9.4 (25)

aFWD =front-wheel drive; RWD =rear-wheel drive; AWD =all-wheel drive.

bECVT =electronically controlled continuously variable transmission; CVT =continuously variable transmission;

Auto =automatic transmission.

cCombined fuel consumption represents 55% city, 45% highway driving.

2000 2002 2004 2006

80%

60%

40%

20%

0%

-20%

Relative Change, %

Curb Weight

Net hybrid system peak power

Combined fuel consumption

Figure 2. Vehicle performance trends over time for the average

HEV: curb weight, net hybrid system peak power, and fuel

consumption. Variability across models is indicated by the error bars

representing one (sales-weighted) standard deviation.

The changing nature of HEV technology can be gauged

by two further parameters, namely, the ratio of net system

power to vehicle curb weight and the ratio of power/engine dis-

placement. These parameters have been chosen because they

are primarily a function of the propulsion system, and they

offer insight into how hybrid-electric system technology has

evolved over time (ﬁgure 3). In comparison, fuel consump-

tion (another metric of performance) is inﬂuenced by vehicle

aerodynamics and curb weight, as well as the choice of test-

ing drive-cycle. Between 2000 and 2003, both parameters

stayed approximately constant, since only two vehicles, the

2004 20062000 2002

40

50

60

70

80

90

(Units indicated in legend)

Net Power / Vehicle Weight (W/kg)

Net Power / Engine Displacement (kW/litre)

Figure 3. Trends in net system power per vehicle weight (a measure

of vehicle performance) and net system power per engine volumetric

displacement (a measure of the technological development of the

propulsion system). Data dispersion is indicated by the weighted

standard deviation.

Toyota Prius and the Honda Insight, were offered for sale. In

2004, three new HEVs were released (the second-generation

Toyota Prius, the Honda Civic and the Ford Escape) and the

ﬁrst-generation Prius was discontinued. The increased power

per unit vehicle weight between 2003 and the present indi-

cates a trend towards new vehicles having higher performance.

This is similar to what has been observed across the

whole light-duty vehicle ﬂeet. For hybrid-electric propulsion

systems, power per engine displacement reveals technological

development; in other words, more system power is made

3

Environ. Res. Lett. 2(2007) 014003 C Reynolds and M Kandlikar

Tab l e 2. Vehicle weight, net system power, and corrected fuel consumption differences between 2007 model year HEVs and their ICEV

counterparts.

Hybrid-electric vehicles ‘Equivalent’ ICEV Vehicles

Delta weight

(kg)

Delta power

(kW)

Delta fuel

consumption

(l/100 km)

Carsa

Toyota Prius

2nd Gen. (I4) Toyota Matrix (I4) 61 −12.0−3.25

Toyota Camry (I4) Toyota Camry (I4) 169 21.5 −2.40

Honda Civic (I4) Honda Civic (I4) 84 −22.4−2.21

Honda Accord (V6) Honda Accord (V6) 110 6.7 −1.27

Lexus GS 450h (V6) Lexus GS 430 (V8) 175 37.3 −1.60

Nissan Altima (I4) Nissan Altima (I4) 150 17.2 −1.75

Sports-utility vehiclesa

Toyota Highlander (V6) Toyota Highlander (V6) 166 39.5 −2.69

Ford Escape/Mercury Mariner (I4) Ford Escape/Mercury Mariner (I4) 165 1.5 −2.75

Lexus RX 400h (V6) Lexus RX 350 (V6) 125 −1.5−2.69

aInternal combustion engine type is indicated as follows: I4 =inline 4-cylinder engine; V6 and V8 refer to 6- and 8-cylinder

engines in ‘V’ conﬁguration, respectively.

available for a given internal combustion engine size. This

parameter drops between 2000 and 2004 because the ﬁrst-

generation Toyota Prius, which has a lower power per engine

displacement, captured an increasing proportion of sales.

Between 2003 and 2004 the power to engine displacement

ratio increased by about 10%, indicating that the newer HEV

technology was signiﬁcantly improved.

3. Weight, power and fuel consumption

A number of factors can independently affect the fuel

consumption of conventional vehicles, including differences

in powertrain type and peak power, vehicle weight,

aerodynamics, rolling resistance, and accessory power

demand. In general, heavier and more powerful vehicles

have higher fuel consumption rates. Due primarily to

the signiﬁcant difference in powertrain type, engineering

models have shown that conventional ICEVs have a different

relationship between vehicle inertial weight, net system power

and fuel consumption than HEVs [7,16]. Until recently there

have been insufﬁcient HEV models to perform a statistical

comparison with conventional ICEVs that differ only in terms

of powertrain, so other analysts have used hypothetical hybrids

based on the best understanding of the technology [16]. To

evaluate the impact of current commercially available hybrid-

electric technology on light-duty vehicle fuel consumption,

we compare a set of nine model year 2007 HEVs to a set of

nine equivalent 2007 vehicles powered by internal combustion

engines. For the purposes of this study, ‘equivalence’ for any

given ICEV is deﬁned as being the same make and model as

the HEV. Where different engine options are available for the

ICEV, the engine with the closest power is used7(table 2).

7The analyses shown in this paper were also performed with three other

assumptions for equivalent ICEs where different engine options were available.

These included equivalents based on similar engine size, smallest IC engines,

largest IC engines. Since there are not three powertrain options for each

equivalent ICEV, these different choices made little or no difference to the

conclusions of this study.

Only the Toyota Prius has no direct equivalent, so the Toyota

Matrix (a mid-size hatchback of similardimensions) is selected

as its counterpart. Automatic transmissions were chosen for

the equivalent ICEVs in the set, and each has the same drive

type (two-wheel or all-wheel drive) as their HEV counterpart.

Vehicle weight and engine power data were obtained from

the published manufacturer speciﬁcations, and combined fuel

consumption estimates were used, corrected using the EPA

factors as described earlier.

First, we examine how vehicle weight and net system

power affects the fuel consumption within each vehicle set

(HEVs and ICEVs). A simple regression model is used,

with fuel consumption (Fc in table 3) as the dependent

variable. Weight (100 kg) and power (10 kW) are examined

independently in turn as predictors of fuel consumption. Each

analysis yields signiﬁcant results at the 95% conﬁdence level

for both HEVs and ICEVs. The results of this regression

analysis are presented in table 3. A 100 kg increase in weight

(not considering power) results in 0.72 l/100 km increase in

fuel consumption for HEVs, and 0.77 l/100 km increase in

fuel consumption for ICEVs. Similarly, a 10 kW power

increase (not considering weight) results in 0.29 l/100 km

increase in fuel consumption for HEVs, and an almost identical

increase for ICEVs. When weight and power are evaluated

simultaneously as predictors of fuel consumption, weight is

found to be a signiﬁcant predictor of fuel consumption for

ICEVs, but not power. However, we ﬁnd that the HEV

sample does not produce signiﬁcant results for weight and

power coefﬁcients at the 95% conﬁdence interval. This is not

surprising since weight and power are correlated (correlation

coefﬁcient =0.84). Tolerance tests using inﬂated variation

factors [17] did not detect signiﬁcant multi-collinearity, and the

Cook’s-D test [18] did not reveal any outliers.

One way of coping with high correlation between weight

and power is to directly estimate the relationship between

differences in fuel consumption and differences in weight and

power within each vehicle class (HEVs and ICEVs). This

signiﬁcantly increases the sample size (N=36), and helps

4

Environ. Res. Lett. 2(2007) 014003 C Reynolds and M Kandlikar

Tab l e 3. Regression analyses results for HEVs and equivalent ICEVs. N=9 for the simple regressions, and N=36 for the pairwise

difference regression. Fuel consumption is the predicted variable.

Regression coefﬁcients

Predicted variableaPredictor variablesbCW P 95% conﬁdence intervals R2

HEVs

Fc C,W−3.94 0.72c—(C:−8.6, 0.7) (W: 0.44, 0.99) 0.84

Fc C,P3.34c0.29c(C: 1.2, 5.5) ( P: 0.16, 0.42) 0.8

Fc C,W,P−1.5 0.43 0.14 (C:−7,4) (W:−0.03,0.9)

(P:−0.6, 0.33)

0.89

Fc (pairwise

difference)

C,W,P0.16 0.4c0.14c(C:−0.2, 0.51) (W: 0.23, 0.58)

(P: 0.07, 0.2)

0.85

Fcw C,P0.34 0.008c(C: 0.25, 0.44) (P: 0.002, 0.014) 0.6

ICEVs

Fc C,W−1.60.77

c(C:−6, 2.9) (W: 0.49, 1.07) 0.85

Fc C,P5.68c0.31c(C: 2.2, 9.16) (P: 0.09, 0.54) 0.61

Fc C,W,P−1.12 0.70c0.04 (C:−6.8, 4.5) (W: 0.17, 1.24)

(P:−0.2, 0.3)

0.86

Fc (pairwise

difference)

C,W,P0.13 0.72c0.04 (C:−0.51, 0.25) (W: 0.53, 0.92)

(P:−0.04, 0.13)

0.83

Fcw C,P0.74 −0.002 (C: 0.6, 0.88) (P:−0.01, 0.007) 0.025

aFc =fuel consumption (l/100 km);Fcw=fuel consumption per 100 kg vehicle weight (l/100 km/100 kg).

bC=regression constant; W=weight coefﬁcient (100 kg); P=power coefﬁcient (10 kW).

cSigniﬁcant at the 95% conﬁdence level.

mitigate the effects of high correlation between power and

weight. Here we ﬁnd that HEVs and ICEVs perform quite

differently. Both weight and power difference are signiﬁcant

for the HEVs, with a 100 kg weight increase and 10 kW

power increase resulting in 0.4l/100 km and 0.14 l/100 km

increase in fuel consumption, respectively. However, only a

change in weight is a signiﬁcant predictor of fuel consumption

for the ICEVs, with a 100 kg increase in weight resulting in

a0

.72 l/100 km increase in fuel consumption. Additionally,

the coefﬁcient for power is not signiﬁcant at the 95% level.

Increasing the sample size of ICEVs to include all engine

options available does not change this result. In order to

conﬁrm our ﬁndings we also tested an alternative model using

weight-corrected fuel consumption (Fcw in table 3) to control

for the effect of correlation between weight and power. We

again found that a change in power is a signiﬁcant predictor of

weight-corrected fuel consumption only for HEVs.

HEVs tend to have lower fuel consumption in urban

driving (with more stop–start driving) than highway driving

due to regenerative braking technology. This is contrary to fuel

consumption ﬁndings in conventional vehicles, which have

signiﬁcantly lower fuel consumption under highway driving

conditions. Therefore we separately analyze fuel consumption

in city and highway driving as a function of weight and

power (table 4). Under city driving conditions, differences

in weight are not a signiﬁcant predictor of differences in

fuel consumption for HEVs, but power is important. This

is because braking and acceleration are common during city

driving, and the full fuel-saving beneﬁts of hybrid-electric

technology are realized. In contrast, during city driving

weight is a signiﬁcant predictor of fuel consumption for ICEVs

because conventional technology cannot recover the energy

lost during deceleration. For highway conditions, on the

other hand, differences in weight are a strong determinant of

differences in fuel consumption for both HEVs and ICEVs.

In highway driving mode, both types of vehicle operate

primarily using their internal combustion engine, and hence

differences in weight (rather than peak system power) dictate

fuel consumption.

In summary, the analysis has demonstrated that differ-

ences in fuel consumption between different HEVs are ex-

plained by both weight and power, while for the equivalent

ICEVs they are explained primarily by weight. This is an im-

portant ﬁnding: despite the fact that hybrid-electric propulsion

systems use less fuel than conventional internal combustion

engines, heavier and more powerful vehicles are eroding the

fuel consumption beneﬁt. Also, due to the way that hybrid-

electric systems recapture energy during vehicle braking, in-

creased weight is less of a factor in HEVs than in ICEVs.

4. Comparing HEV and ICEV fuel consumption

The share of HEVs in the North American light-duty passenger

vehicle ﬂeet is expected to continue increasing, driven by their

potential for fuel savings and steadily reducing cost. A 2004

study forecasted that HEVs could capture 4–7% of the light-

duty vehicle market by 2008 and 10–15% by 2012 [12]. This

raises the question of how best to estimate the change in fuel

consumption that a given vehicle would realize, were it to use

a hybrid-electric system rather than a conventional engine. We

approach this problem by comparing the set of HEVs against

the sample of equivalent ICEVs, rather than within each set.

We regress fuel consumption differences using three ‘models’

of equivalence between HEVs and ICEVs (table 5).

5

Environ. Res. Lett. 2(2007) 014003 C Reynolds and M Kandlikar

Tab l e 4. Regression analyses results for HEVs and equivalent ICEVs with city and highway fuel consumption (pairwise comparison,

N=36).

Regression coefﬁcients

Predicted variableaPredictor variablesbCW P 95% conﬁdence intervals R2

HEVs

Fc (City) C,W,P0.11 0.15 0.26c(C:−0.33, 0.56) (W:−0.06, 0.36)

(P: 0.17, 0.34)

0.77

Fc (Highway) C,W,P0.32 0.75c−0.02 (C: 0.13,0.53) (W: 0.67, 0.87)

(P:−0.06, 0.014)

0.93

ICEVs

Fc (City) C,W,P0.16 0.29c0.26 (C:−0.53, 0.68) (W: 0.02,0.61)

(P:−0.13, 0.38)

0.65

Fc (Highway) C,W,P−0.32 1.21c−0.18c(C:−0.91, 0.28) (W: 0.9, 1.5)

(P:−0.3,−0.05)

0.73

aFc =fuel consumption (l/100 km).

bC=regression constant; W=weight coefﬁcient (100 kg); P=power coefﬁcient (10 kW).

cSigniﬁcant at the 95% conﬁdence level.

Tab l e 5. Across-group comparison of fuel consumption differences between HEVs and ICEVS. Crepresents the difference in fuel

consumption between HEVs and ICEVs due to the difference in propulsion system technology.

Regression coefﬁcients

Dependent variableaIndependent variablesbCWPConﬁdence intervals R2

Model 1: Each HEV versus equivalent ICEV (N=9)

Fc C,W3.06c0.5 — (C: 1.68, 4.4) (W:−0.48, 1.48) 0.17

Fc C,P2.5c0.1 (C: 1.66, 3.46) (P:−0.1, 0.3) 0.17

Fc C,W,P2.8c0.25 0.06 (C: 0.4, 5.2) (W:−1.75, 2.26)

(P:−0.35, 0.47)

0.18

Model 2: HEV versus ICEV (group comparison N=81)

Fc C,W3.4d0.74d—(C: 3.15, 3.63) (W: 0.67, 0.81) 0.85

Fc C,P2.7d0.30d(C: 2.24, 3.0) (P: 0.25, 0.34) 0.70

Fc C,W,P3.25d0.56d0.1d(C: 3.01., 3.48) (W: 0.45, 0.68)

(P: 0.04, 0.14)

0.87

Model 3: HEV versus ICEV (group comparison with car and SUV sub-groups N=162)

Fc C,W3.3d0.86d—(C: 2.97, 3.7) (W: 0.75, 1) 0.86

Fc C,P2.5d0.32d(C: 2.34, 2.67) (W: 0.3, 0.35) 0.96

Fc C,W,P2.7d0.15c0.27d(C: 2.43, 2.87) (W: 0.02, 0.3)

(P: 0.22, 0.32)

0.97

aFc =fuel consumption (l/100 km).

bC=regression constant; W=weight coefﬁcient (100 kg); P=power coefﬁcient (10 kW).

cSigniﬁcant at the 95% conﬁdence level.

dSigniﬁcant at the 99% level.

In each model, the dependent variable is the fuel

consumption difference between an HEV and an equivalent

ICEV, and corresponding differences in weight and power

are the independent variables. In each case, the regression

constant, C, is the difference in fuel consumption (in units

of l/100 km) between HEVs and ICEVs. This difference

can be attributed to the effect of using hybrid-electric

technology instead of conventional internal-combustion-

engine technology for the vehicle’s propulsion system. Model

1 is based on the assumption that consumers will switch from

a conventional vehicle to its hybrid counterpart; model 2

allows for any HEV to substitute for any ICEV; and model

3 embeds the assumption that consumers will only switch

from conventional cars to hybrid-electric cars, and likewise

for SUVs. In model 1, each HEV is compared only with its

equivalent (N=9). In model 2, we compare HEVs and

ICEVs as a class, i.e., each HEV was compared with every

ICEV (N=81). In model 3 we split each HEV and ICEV set

into two groups, SUV and cars, and compare all vehicles within

each group (N=45). This is because car and SUV HEVs are

likely to contribute to different relationships between weight,

power and fuel consumption.

6

Environ. Res. Lett. 2(2007) 014003 C Reynolds and M Kandlikar

Equivalence model 1

Only ‘technology’ (i.e. the use of a hybrid-electric propulsion

system rather than an internal combustion engine) was found to

be a signiﬁcant predictor of the difference in fuel consumption.

On an average, the use of hybrid-electric technology resulted in

a fuel consumption beneﬁt of 2.8l/100 km. Both weight and

power differences were found not to be signiﬁcant at the 95%

level. The mean differences in power and weight for an HEV

and its ICEV equivalent were small (5% and 9%, respectively),

and the impact of these changes cannot be observed in the

face of substantial differences in fuel consumption. Though

signiﬁcant, the constant has a large conﬁdence interval and the

model has a low R2(0.18). This suggests wide variation in

the implementation of hybrid technology from model to model,

resulting in large variation in fuel consumption beneﬁts.

Equivalence model 2

When HEVs and ICEVs are compared as a class, the model

shows highly signiﬁcant results for each of the three predictors

(at the 99% conﬁdence level). An HEV of the same weight and

power uses approximately 3.2l/100 km less fuel than its ICEV

counterpart. Further, an increase in weight of 100 kg results in

a fuel consumption increase of 0.56 l/100 km, and an increase

in power of 10 kW results in a fuel consumption increase

of 0.10 l/100 km. On an average 2007 HEVs are 136 kg

heavier than equivalent ICEVs, resulting in a weight-related

fuel consumption penalty of 0.75 l/100 km. Further, 2007

HEVs have 10 kW more average power than equivalent ICEVs,

resulting in a fuel consumption penalty of 0.1l/100 km. The

total penalty amounts to a reduction in the fuel consumption

beneﬁts of HEVs by 27%.

Equivalence model 3

This model of equivalence allows us to differentiate between

cars and SUVs. Here we ﬁnd signiﬁcant effects (at the

95% conﬁdence level) for all three predictors. The average

gain from using hybrid technology was 2.7l/100 km, while

the weight and power penalties were 0.15 l/100 km and

0.27 l/100 km, respectively. This is an interesting result,

because the weight coefﬁcient is almost half the magnitude

of the power coefﬁcient. Compare this result to model 2,

where the weight coefﬁcient is over ﬁve times as large as the

power coefﬁcient. The reason for the change in magnitude of

the coefﬁcients is because this model compares like vehicles

with like, and cars are different from SUVs in a number of

important ways. The three hybrid-electric SUVs use all-wheel

drive rather than two-wheel drive, which results in an extra fuel

saving since all of the four wheels provide regenerative braking

capability [7]. Also, SUVs generally have a higher coefﬁcient

of drag than cars and have a larger frontal area.

5. Conclusions

In summary, based on this statistical analysis of 2007 model

year HEVs and their equivalent ICEVs sold in the United

States, this study presents the following conclusions.

(1) Heavier and more powerful hybrid-electric vehicles are

eroding the fuel consumption beneﬁt of this technology.

(2) The fuel consumption penalty imposed by increased

vehicle weight is signiﬁcantly lower in HEVs than in

equivalent ICEVs. A 100 kg change in vehicle weight

increases fuel consumption by only 0.4l/100 km in

HEVs, compared with 0.7l/100 km in ICEVs.

(3) Three different equivalence models (based on different

comparisons of HEVs against ICEVs) yielded fuel

consumption beneﬁts ranging from 2.7 to 3.25 l/100 km,

with varying effectsof changes in weight and power.

(4) When the HEVs are compared with their ICEV

equivalents (model 3, grouped into cars and SUVs),

the average fuel consumption beneﬁt of an HEV was

2.65 l/100 km. This analysis reveals that an HEV

that is 100 kg heavier than an identical ICEV, holding

everything else constant, has a fuel consumption penalty

of 0.15 l/100 km. Likewise, an HEV that is 10 kW more

powerful than the ICEV results in a fuel consumption

penalty of 0.27 l/100 km.

Hybrid technology is new, and it will evolve as the technology

matures and diffuses within the automobile ﬂeet. We have

shown that, even in the relatively short time that HEVs

have been commercially available, there have been signiﬁcant

changes in the dimensions, performance and fuel consumption

of the ﬂeet. The number of models available is small, and this

small sample size will be an inevitable challenge for analyses

of the ﬂeet done today. As the number of HEV types increases,

we will no doubt get a more robust picture of the technology.

Acknowledgments

We thank Kimberly Jones for her assistance assembling vehicle

sales and speciﬁcations data, and Gordon McTaggart-Cowan

and Julian Marshall for valuable comments on a draft of this

article. The University of British Columbia Bridge Program,

the Natural Science and Engineering Research Council of

Canada, and the Auto21 Network of Centres of Excellence

provided support for this research.

References

[1] An F, Vyas A, Anderson J and Santini D 2001 Evaluating

commercial and prototype HEVs SAE Technical Paper

2001-01-0951

[2] An F, Stodolsky F and Santini D 1999 Hybrid options for

light-duty vehicles SAE Technical Paper 1999-01-2929

[3] Hoffert M I et al 2002 Advanced technology paths to global

climate stability: energy for a greenhouse planet Science

298 981–7

[4] Pacala S and Socolow R 2004 Stabilization wedges: solving the

climate problem for the next 50 years with current

technologies Science 305 968–72

[5] Jackson R B and Schlesinger W H 2004 Curbing the U.S.

carbon deﬁcit Proc. Natl Acad. Sci. 101 15827–9

[6] Romm J 2006 The car and fuel of the future Energy Policy

34 2609–14

[7] An F and Santini D J 2004 Mass impacts on fuel economies of

conventional vs hybrid electric vehicles SAE Trans. 113

258–76

7

Environ. Res. Lett. 2(2007) 014003 C Reynolds and M Kandlikar

[8] An F and Santini D 2003 Assessing tank-to-wheel efﬁciencies

of advanced technology vehicles SAE Technical Paper

2003-01-0412

[9] Berman B 2006 Hybrid vehicles in the US: available and

expected models available from http://www.hybridcars.com/

cars.html

[10] Heavenrich R M 2006 Light-duty automotive technology and

fuel economy trends: 1975 through 2006 report no.

EPA420-R-06-011, Ofﬁce of Transportation and Air Quality,

US Environmental Protection Agency

[11] Millikin M 2006 Green car congress: energy, technologies,

issues and policies for sustainable mobility available from

http://www.greencarcongress.com

[12] Greene D L, Duleep K G and McManus W 2004 Future

Potential of Hybrid and Diesel Powertrains in the US

Light-Duty Vehicle Market report no. ORNL/TM-2004/181

(Oak Ridge, TN: Oak Ridge National Laboratory)

[13] US Environmental Protection Agency 2006 Fuel economy

labeling of motor vehicles: revisions to improve calculation

of fuel economy estimates report no EPA420-D-06-002,

Ofﬁce of Transportation and Air Quality, US Environmental

Protection Agency

[14] Lipman T E and Delucchi M A 2006 A retail and lifecycle cost

analysis of hybrid electric vehicles Transport Res. D

11 115–32

[15] B C Climate Exchange 2005 The hybrid experience report:

Toyota Prius fuel performance available from http://www.

hybridexperience.ca/Toyota Prius Fuel Performance.htm

[16] Santini D, Vyas A D and Anderson J L 2002 Fuel economy

improvement via hybridization vs vehicle performance level

SAE Technical Paper 2002-01-1901

[17] Mansﬁeld E R and Helms B P 1982 Detecting multicollinearity

Am. Stat. 36 158–60

[18] Cook R D 2000 Detection of inﬂuential observation in linear

regression Technometrics 42 65–8

8

- A preview of this full-text is provided by IOP Publishing.
- Learn more

Preview content only

Content available from Environmental Research Letters

This content is subject to copyright. Terms and conditions apply.