Low-Level Automated Light-Duty Vehicle Technologies Provide Opportunities to Reduce Fuel Consumption

Abstract

Gasoline is the main source of energy used for surface transportation in the United States. Reducing fuel consumption in light duty vehicles can significantly reduce the transportation sector’s impact on the environment. Implementation of emerging automated technologies on vehicles could result in fuel savings. This study examines the effect of automated vehicle systems on fuel consumption using stochastic modeling. Automated vehicle systems examined in this study include warning systems such as blind spot warning, control systems such as lane keeping assistance, and information systems such as dynamic route guidance. We have estimated fuel savings associated with reduction of accident and non-accident related congestion, aerodynamic force reduction, operation load, and traffic rebound. Results of this study show automated vehicles could reduce 6% to 23% of light duty vehicle fuel consumption in the U.S. This reduction could save $60 to $266 annually for the owners of vehicles equipped with automated technologies. Also, adoption of automated vehicles could benefit all road users (i.e. conventional vehicle drivers) up to $35 per vehicle annually (up to $6.2 billion per year).

Keywords: Automated vehicle, fuel consumption, Automated vehicle technologies, energy impact

Full text

Low-Level Automated light-duty vehicle technologies provide opportunities to
reduce fuel consumption
Saeed Vasebi
PhD student
School of Systems and Enterprises
Stevens Institute of Technology
1 Castle Point Terrace, Hoboken, NJ 07030
Email: svasebi@stevens.edu
Tel: +1 (740) 591-5231
Yeganeh M. Hayeri (Corresponding Author)
Assistant professor
School of Systems and Enterprises
Stevens Institute of Technology
1 Castle Point Terrace, Hoboken, NJ 07030
Email: yhayeri@stevens.edu
Tel: +1 (484) 575-6960
Constantine Samaras
Assistant Professor
Department of Civil and Environmental Engineering
Carnegie Mellon University
Pittsburgh, PA 15213-3890
Email: csamaras@cmu.edu
Tel: +1 (412) 268-1658
Chris Hendrickson
Professor
Department of Civil and Environmental Engineering
Carnegie Mellon University
Pittsburgh, PA 15213-3890
Email: cth@cmu.edu
Tel: +1 (412) 268-1066
Word count: 5473 words
Number of figures: 4
Number of tables: 4
TRR Paper number: 18-06702
March 15, 2017

Vasebi, Hayeri, Samaras, and Hendrickson 2
ABSTRACT
Gasoline is the main source of energy used for surface transportation in the United States.
Reducing fuel consumption in light duty vehicles can significantly reduce the transportation
sector’s impact on the environment. Implementation of emerging automated technologies on
vehicles could result in fuel savings. This study examines the effect of automated vehicle systems
on fuel consumption using stochastic modeling. Automated vehicle systems examined in this
study include warning systems such as blind spot warning, control systems such as lane keeping
assistance, and information systems such as dynamic route guidance. We have estimated fuel
savings associated with reduction of accident and non-accident related congestion, aerodynamic
force reduction, operation load, and traffic rebound. Results of this study show automated vehicles
could reduce 6% to 23% of light duty vehicle fuel consumption in the U.S. This reduction could
save $60 to $266 annually for the owners of vehicles equipped with automated technologies. Also,
adoption of automated vehicles could benefit all road users (i.e. conventional vehicle drivers) up to
$35 per vehicle annually (up to $6.2 billion per year).
Keywords: Automated vehicle, fuel consumption, Automated vehicle technologies, energy impact

Vasebi, Hayeri, Samaras, and Hendrickson 3
1. INTRODUCTION
Gasoline is the main source of energy used for surface transportation in the United States.
According to the U.S. Energy Information Administration, Americans use more than a gallon of
gasoline per person each day, adding up to 392 million gallons of gasoline per day in 2016, of
which 90% is solely used for light-duty vehicles. Reducing gasoline consumption could
significantly improve air quality, health, energy security, and greenhouse gas (GHG) emissions.
Emerging automated vehicle technologies can accelerate the path to realizing some of these
benefits. Since driving patterns, commute times, and the design of metropolitan regions could all
be affected by automated vehicles, there is a great potential for saving energy as these emerging
technologies gain market share.
In this study, we examine the impacts of various low-level automated vehicle technologies
on fuel consumption. In addition, we examine the time that would take to realize such impacts
based on market penetration rate of the technologies.
The scope of this study includes three main elements:
1) Different groups of people that could benefit from automated technologies (e.g. owners
of the vehicles equipped with automated technologies)
2) Benefit and cost categories (e.g. fuel saving benefits due to accident reduction)
3) Types of automated technologies (e.g. adaptive cruise control)
The scope of this study is illustrated in FIGURE 1, which considers fuel saving
possibilities from two perspectives defined as affected groups: 1) those driving a vehicle that is
equipped with automated technologies and 2) all other road users including those driving a
conventional vehicle. The first group, automated vehicle owners, benefits directly from fuel saving
potentials of the technologies. This is also the group that incurs cost of purchasing the automated
packages. The second group, all road users, potentially benefit from the efficient flow of traffic
(e.g. less accident caused by human error, or more efficient route guidance). The first group is an
enabler for the second groups’ benefits.
In order to estimate the net benefit of automated technologies, we analyze benefits and
costs associated with each technology specified in FIGURE 1. For each automated technology, we
estimate benefits resulting from congestion reduction (accident and non-accident related), and
aerodynamic force reduction. To estimate costs, we develop a model to include 1) operational load,
which is the energy that a vehicle consumes for the operation of computers, sensors and other
automated equipment installed on the vehicle, and 2) rebound effects, which in this paper is
considered to be induced demand and fuel efficiency. Automated vehicle owners are affected by all
these benefit and cost categories. Other vehicles are affected only by the accident and non-accident
congestion reduction categories and rebound effects.
We categorize automation technologies into three groups: 1) warning systems, 2) control
systems and 3) information systems (FIGURE 1). Automated warning systems are those
generating warnings (e.g. sounds, lights) inside the vehicle to alert the driver of a potential hazard.
Warning systems considered in this study include lane departure warning (LDW), forward
collision warning (FCW), blind spot warning (BSW), speed limit detection, and traffic warning.
Control systems not only warn drivers but also, they may take control of vehicle partially to
pass a threat or improve driving experiences. Control systems included in this study are adaptive
cruise control (ACC), cooperative adaptive cruise control (CACC), lane keeping assistance
(LKA), forward collision avoidance (FCA), and active braking.

Vasebi, Hayeri, Samaras, and Hendrickson 4
Information systems provide useful driving information to reduce travel time and delay.
Information systems considered in this study include parking aid system and dynamic route
guidance. The complete list of automated technologies and their detailed definition is available
from the authors.
In the next section, we provide a thorough review of the existing impact analyses of
automated vehicle systems conducted to date. The following section explains our model for fuel
saving analysis of automated vehicle technologies. We continue by providing estimates of net
energy benefit of automated light-duty vehicle technologies as well as uncertainties involved with
our estimates and discussion of the results. In the last section, we provide steps for future
enhancement of our model. This paper’s contribution is to realize the near to mid-term energy
impact of automated vehicle technologies.
2. EXISTING IMPACT ASSESSMENTS
A few recent studies have shed lights on potential benefits and costs of automated vehicles’
technologies (2-5) (TABLE ). Harper et al determined if all light duty vehicles in the U.S. would be
equipped by LDW, FCW, and BSM systems, it would result in $4 billion to $202 billion annual net
social/safety benefit cumulatively (4). Jermakian reviewed accident records of National
Automotive Sampling System and the General Estimates System and Fatality (2). He estimated
LDW individually could prevent 3% of crashes and save up to 7,529 lives and BSW could reduce
accidents by 6.8% and save 393 lives annually. Findings of other general and safety impact studies
can be found in Table 1.
Most fuel and GHG emission studies have only concentrated on fully autonomous vehicles
(AV) (6-8) or shared autonomous vehicles (SAV) (9, 10) (TABLE ). Brown et al. examined
potential energy impact of AV(6). They concluded that AV may increase fuel demand by 173% or
decrease by 95%. MacKenzie & Wadud also analyzed energy impact of light duty and heavy duty
AV in the U.S. (7, 8). Due to uncertainty in future prediction of AV, they developed four scenarios.
The most optimistic scenario estimates 40% reduction in total fuel demand of transportation sector
and the most pessimistic scenario predicts significant growth (i.e. more than 100%) in fuel
demand. Stephens et al. studied potential impacts of connected autonomous vehicle (CAV) and
vehicle ridesharing on fuel consumption (10). They estimated -10.68% to 13.59% increase in fuel
demand for partial AV (i.e. driver assistant feature with human involvement) and -58.25 to
204.85% and -64.08% to 194.17% increase in fuel consumption for fully CAV and fully rideshared
CAV respectively.
Although safety benefits of automated vehicle technologies have been investigated in
previous studies (2-5), energy impacts of low level automated features are mostly uncertain. Most
fuel impact studies have only concentrated on autonomous vehicles (6-8), which may not be
commercialized with a strong enough market share until 2070s (11). Our study is the first energy
impact investigation focusing on lower levels of automation and individual technologies that are
either already introduced or will be in the next few years.
3. METHOD FOR ESTIMATING ENERGY IMPACTS OF AUTOMATED
TECHNOLOGIES
The paper considers five main energy impact categories resulting from the use of automated
vehicle technologies:

Vasebi, Hayeri, Samaras, and Hendrickson 5
1- Accident-related congestion reduction impact,
2- Non-accident-related congestion reduction impact,
3- Aerodynamic force reduction impact,
4- Equipment operation load impact,
5- And traffic rebound impact.
We estimate savings and costs associated with each technology as a percentage of average
gasoline consumption, which is about 508 gallons per vehicle annually in the U.S. (14).
This paper only analyzes energy impact of automated vehicles. Since vehicle
manufacturing energy use is small relative to driving energy (15), it will not be considered in this
paper. We evaluate direct vehicle fuel savings, as these are the most tangible to consumers and
policymakers. While automation may affect different types of drivers and vehicles differently, our
analysis pertains to average implications in the U.S. We focus on cumulative effect of each
technology listed on FIGURE 1. We estimate potential fuel saving benefits for automated vehicle
owners (AVO) and all road users (ARU).
3.1. Fuel Saving For Automated Vehicle Owners (AVO)
We estimate net fuel consumption change of the automated vehicles as:
 
 

 

 
 

Where  stands for total net fuel saving for an AVO,  is accident related
congestion reduction fuel saving,  is non-accident related congestion reduction fuel saving,

 is aerodynamic force reduction fuel saving, is equipment’s operation load fuel
consumption, is traffic rebound fuel consumption, and  indicates average annual fuel
consumption of each vehicle in US. In general,  stands for the categories’ fuel consumption,  is
the technologies’ fuel saving impacts,  is models’ parameters,  is the market penetration rate,
and random variables (stochastic parameters) are identified using the  symbol. Also, this paper’s
superscripts are AC (accident related congestion reduction), NAC (non-accident congestion),
NACA (non-accident congestion of all road users), NACAV (non-accident congestion of
automated vehicles drivers), AFFS (aerodynamic force reduction), EQP (equipment’s operation
load), TR (traffic rebound), IDR (induced rebound), and FER (fuel efficiency rebound). Each of
these items will be described and calculated based on the following subsections.
3.1.1. Accident-Related Congestion Reduction
It is estimated that 21% of the U.S. traffic congestion delay is a result of accidents, and an average
vehicle consumes about 3.8 gallons of fuel annually due to accident-related congestion (16-18).

Vasebi, Hayeri, Samaras, and Hendrickson 6
Most of the automated technologies have the potential to reduce the number and severity of traffic
accidents, which would reduce traffic congestion and consequentially energy consumption. The
reduction of accident-related congestion benefits all road users - those who drive conventional cars
or automated cars. Using estimates from the literature, the impacts of each of the specified
technologies on safety and energy savings were estimated for the technologies. The change in fuel
consumption from accident-related congestion reduction of technology  is estimated as:

Where  is the change in fuel consumption due to accident related congestion
reduction capability of automated technology,  is US fuel consumed as a
result of congestion (16),  is the portion of congestion delays due to accidents
(16-18),  is the portion of accident-related congestion reduced by technology , and  is the
market penetration rate of technology  . We assume that when technologies are combined, the
portion of accidents avoided (from the accidents remaining after other technologies have been
implemented) is independent of the presence of other technologies, so total fuel savings ()
from implementation of a set of technologies is determined as:

 
3.1.2. Non-Accident Related Congestion Reduction
All vehicles on road benefit from non-accident related congestion reduction of automated vehicles.
Non-accident congestion reduction reduces traffic load by better navigation, less travel time, or
more roadway capacity. Parking aid systems and dynamic route guidance systems can both reduce
non-accident congestions by improving the efficiency of the national roadway system. In addition,
CACC increases road capacity by reducing the vehicles headway on a platoon (19).
Non-accident-related congestion of ARUs is handled similarly:

Where  stands for non-accident congestion reduction of technology.
In addition to ARUs’ non-accident congestion benefits, AVOs directly benefit by energy
saving capabilities of their cars when using parking aid system or dynamic routing guidance. The
parking aid technology assists drivers to find parking lot faster than conventional automobiles.
Also, the drivers utilize dynamic routing guidance to find faster routes with less congestion.
Non-accident-related congestion fuel saving potential of the vehicles owners is estimated as:


Where  is the fuel saving capability of technology  for the vehicle driver due to
less driving. Although, non-accident congestion reduction for ARUs and AVOs are independent,
arithmetic summation of the values is used. Since the values are small, their correlation is assumed
to be negligible. So, the non-accident congestion reduction fuel saving for technology  is
calculated as:

Vasebi, Hayeri, Samaras, and Hendrickson 7




The total fuel savings of non-accident congestion reduction ( ) for a set of
technologies is estimated as:


 
 
It is important to note that in estimating total fuel savings of accident and non-accident
congestion reduction all individual technology effects are essentially aggregated.
3.1.3. Aerodynamic Force Related Fuel Saving
Aerodynamic force reduction has been estimated to cut fuel consumption significantly (20, 21).
Technologies such as ACC and CACC have the capability to capture a portion of these savings by
improvement of efficient driving. The aerodynamic related fuel saving capability of ACC and
CACC is calculated as:

 
Where  is the portion of VMT traveled on the highway (22), and  is the
fuel efficiency parameter for the vehicle driver with technology i.
3.1.4. Operation Load
Automated features aid drivers by warning, controlling, or providing of information; however,
these features consume energy to provide these functions. Operation load is defined as the energy
used by sensors, radars, steering, or braking systems. Operation of automation features requires
use of energy-consuming equipment (e.g. sensors, processors, and actuators). Based on BOSCH
automotive handbook, higher comfort and safety features have significantly increased electric
consumption of vehicles (23). Hence, theses operation loads should be considered as a negative
energy impact for automated features. The following formula is utilized to determine operation
load.



 
Where  is the change in gasoline consumption due to operation of equipment type
in technology ,  is the average power consumption of equipment type  in technology 
while active,  is the amount of time spent driving (22, 24), 
 is the energy density of gasoline (21), 
 is the portion of driving time when
equipment type  in technology  is active, and  stands for fuel efficiency of light duty
internal combustion engineers (25).
3.1.5. Traffic Rebound
Reduction of traffic congestion and travel cost can potentially induce new travel demands(26).
Rebound effect is defined as new VMT generated due to improved mobility or travel cost

Vasebi, Hayeri, Samaras, and Hendrickson 8
reduction(27). As an illustration, travel patterns show when gas prices decrease, people tend to
drive more and VMT increases(26, 28). Similarly, when mobility improves, which could result in a
more pleasant driving experience due to less congestion and consequentially saving gas, people
tend to drive more (induced demand) (27, 29, 30). Automated vehicle features have the potential to
result in traffic rebound for the following reasons: 1) congestion reduction and 2) fuel efficiency.
Safety and information systems could reduce traffic congestion. Hence, ARUs spend less time in
traffic congestion and could consequentially drive more. Moreover, when automated features (e.g.
CACC and ACC) increase fuel efficiency, it reduces driving cost per mile and AVOs are likely to
drive more.
Improvement of traffic flow induces more people to driver their own cars. We estimate
induced demand’s fuel consumption as:



 
Where  is the change of fuel consumption due to induced traffic demand,  is
elasticity of induced demand for mobility improvement,  is the average annual VMT, and  is the
average vehicle efficiency (miles per gallon). Because, so the formula simplifies to

On the other hand, the automated features increase fuel efficiency. The features decrease
cost of driving per mile and could generate new travels. Since fuel efficiency impact of low-level
automated vehicle is estimated to be relatively small, we used a linear approximation; however,
large fuel efficiency changes could have non-linear impact on the demand. We estimate the effect
of fuel efficiency rebound on fuel consumption as:


Where  is the change in fuel consumption due to the fuel efficiency rebound, 
is the elasticity fuel efficiency, and  is the price of gasoline. Since, so the
expression simplifies to

In conclusion, net traffic rebound of automated vehicles is estimated in our model as:

Details of the model including all the above-mentioned parameters as well as the underlying
assumptions and sources used in this analysis are available as part of supporting documents upon
request. Error! Reference source not found. provides a summary of the key parameter values by
category and technology. While due to space limitation documentation on how these parameters
were driven is omitted from the main text, Error! Reference source not found. provides a general

Vasebi, Hayeri, Samaras, and Hendrickson 9
idea on magnitude of these values and the importance of each technology and category.
3.2. Fuel Saving For All Road Users
Automated vehicle technologies not only save fuel for their owners but they could also save fuel
for all road users even conventional vehicle drivers. Automated vehicles could influence fuel
consumption of ARUs by accident related congestion reduction, non- accident congestion
reduction, and induced traffic demand. Total change in fuel consumption (saving) of automated
vehicles for all road users is determined as:
 
 
 


Where  is total fuel consumption change of automated vehicles for all road users.
4. RESULTS
We used stochastic simulation to trace the changes with respect to uncertain variables within our
model. Triangular probability distributions were used based on minimum, maximum and mode
values. We estimated these values from the thorough review of the existing literature. We also
considered market penetration rates in our model and analysis, as it can significantly change the
magnitude of automated features’ impacts. To simplify market penetration analyses, penetration
rate of all technologies is assumed to be equal (e.g. 30%penetration rate for all automated
technologies). We simulated our model several times using different penetration rates.
4.1. Fuel Saving For Automated Vehicle Owners
FIGURE 2 shows fuel saving impacts of automated technologies for AVOs based on 50%
penetration rate (i.e. when half of all vehicles on road are equipped with all automated
technologies considered in this study). Impacts of warning systems, lane keeping, collision
detection breaking, and active braking system are negligible (i.e. less than ±0.12%). While these
features save a small amount of fuel by reducing accident related congestion, their operation load
offsets most of the gained benefit. Parking aid system and dynamic route guidance (i.e.
information systems) could save 3.5% and 4.5% of total fuel consumption respectively. These
savings result from non-accident related congestion. The original savings in this category (4.1%
and 5.2% for parking aid system and route guidance respectively) are reduced by rebound effect by
13.5%. ACC and CACC have the highest energy saving potentials for AVOs; 6.8% and 4.6%
respectively. The original fuel savings are estimated to be 7.1% and 4.9% for ACC and CACC
respectively. However, ACC and CACC savings are reduced by traffic rebound and operation load.
We conducted trend analysis to realize the effect of market penetration rate of technologies
on energy savings. FIGURE 3 illustrates the energy saving trends for the automated technology
groups (i.e. warning systems, information systems, and control systems). Warning systems fuel
consumption is higher than their fuel saving. Their net energy saving varies from -0.51% to
-0.38% for 1% to 100% penetration rate of warning technologies. Information systems could save
6% to 9.9% for AVOs based on different penetration rates of information systems. Due to overlap
of ACC and CACC technologies in the control system category (i.e. CACC and ACC cannot
operate simultaneously), which might cause double counting, we distinguished between the two

Vasebi, Hayeri, Samaras, and Hendrickson 10
control systems;
1) Control systems with ACC (excluding CACC)
2) Control systems with ACC in a connected environment (CACC)
Fuel saving of control systems with ACC is not associated with the number of vehicles
equipped with these features. It provides 6.5% fuel saving at any penetration rate. In contrast, fuel
saving of ACC in a connected environment is strongly correlated with adoption of the
technologies. Whereas, its energy saving is -0.3% on 1% penetration, it could reduce fuel
consumption up to 15.2% when connected control systems are installed on all vehicles. Error!
Reference source not found. presents fuel saving for the two groups of ACC and CACC. All
automated features with ACC has a smooth trend (i.e. varies only from 13.3% to 13.8% for 1% to
100% penetration rates respectively) (Error! Reference source not found.– part a). In contrast,
total energy saving of automated vehicles in a connected environment that would include CACC is
strongly correlated with penetration rate (i.e. varies from 7% to 16.2% saving for 1% to 100%
penetration rates respectively) (Error! Reference source not found.– part b). Hence, vehicles
equipped with all automated features considered in this study (excluding CACC) should expect
13.3% to 13.8% of net fuel saving, regardless of the number of automated vehicles in the market.
However, vehicles equipped with all automated features considered in this study (excluding ACC)
could expect higher fuel saving potential as the number of automated vehicles increases in the
market (i.e. from 7% to 16.2% net saving).
4.2. Fuel Saving For All Road Users
The fuel saving impacts of automated vehicles on ARUs is shown in Error! Reference source not
found. (Parts c-d) for different market penetration rates. Fuel saving of ARUs is highly influenced
by market penetration rate of the automated vehicles technologies. Fuel saving of automated
features with ACC (i.e. excluding CACC) is less than 0.2% on low penetration rate (i.e. less than
25%); however, the automated vehicles could save up to 0.76% for ARUs on higher penetration
rates (Error! Reference source not found.-part c). In addition, automated features in a connected
vehicle to vehicle (V2V) environment (i.e. excluding ACC) could save up to 1.9% of total fuel for
all vehicles on road (Error! Reference source not found.– part d).
5. DISCUSSION
Combined low-level automated vehicle technologies (those considered in this study) could
reduce fuel consumption of light-duty vehicles by 5% to 23%. This saving is a result of accident
and non-accident related congestion reduction, aerodynamic force reduction, traffic rebound, and
operation load. TABLE 2 presents fuel saving ranges for these benefit-cost categories for AVOs.
Moreover, low-level automated vehicles could reduce fuel consumption of all vehicles on road.
Our results demonstrate low-level automated vehicles could save up to 9.8 gallon/vehicle/year for
all road users. TABLE 2 shows ARUs’ potential fuel saving for each benefit-cost category and
compares AVOs and ARUs saving ranges. Error! Reference source not found. summarizes these
studies while comparing their results with the result of our study. The table describes what type of
benefit – cost analysis each study conducted. It shows which study considered fully autonomous
vehicles vs. individual automated vehicle systems. We also identified what types of impacts were

Vasebi, Hayeri, Samaras, and Hendrickson 11
included in each study, followed by the results. In our study, low-level automated vehicles owners
(those equipped with all technologies considered in this study) could reduce fuel consumption 27
to 119 gallon/year/vehicle. This saving is equal to 6% to 23% of average fuel consumption in the
U.S. and it could save $60 to $266 USD2016 each year for the vehicle owners (31).
In comparison with other studies, our result demonstrates that regardless of the penetration
rate, automated vehicles equipped with a combination of warning, control and information
systems always result in fuel savings. Other studies’ result includes potential negative and positive
impacts (6-8, 10). Their analyses predict automated/autonomous vehicle might increase annual
fuel consumption due to average vehicle miles traveled (VMT) increase. Autonomous vehicles
also might induce new user groups (e.g. elder people, or teenagers). Our study demonstrates that
although low-level automated vehicles increase traffic rebound and annual VMT, this increase is
less than the fuel saving potentials associated with low-level automated vehicles. It is important to
note that automated vehicles, which have limited control abilities, are not able to attract new user
groups as autonomous vehicles might do.
In this study, we conducted a comprehensive literature review on energy and safety impacts
of automated features. This thorough assessment provided precise data for our analyses. As a
result, the uncertainty of calculations is significantly reduced in comparison with similar studies.
Our estimation (i.e. annual $60 to $266 saving) narrows down uncertainty of energy impact
prediction in comparison with similar studies (Error! Reference source not found.).
6. CONCLUSION
In this paper, we assessed the energy impact of low-level automated vehicle technologies. We
conducted a comprehensive literature review of automated features’ safety, congestion, fuel
efficiency, operation load, and traffic rebound impacts including research papers, real experiments,
and road testing data associated with individual automated technologies (2, 4, 6, 13, 17, 19, 20,
32-35). The thorough literature review made detailed modeling of the impacts possible. We
developed a stochastic method to consider both induced demand and fuel efficiency rebound
effects, which is one of the main contributions of the paper. In addition, we modeled operation load
of automated features. These types of potential negative impacts of automated vehicles have not
been thoroughly considered in previous studies. We also used market penetration rate of automated
technologies to predict the technologies current, and near to mid-term impacts. Analysis of
automated features trends based on penetration rate assist decision makers to estimate the
technologies fuel saving at any given time.
We estimate that low-level automated vehicles could reduce fuel consumption from 27 to
119 gallon/year/vehicle. These technologies could save $60 to $266 USD2016 annually for
automated vehicle owners. Automated vehicles could also reduce fuel consumption of other
vehicles. It could save up to about 10 gallons/vehicle/year which values up to 6 billion dollars
annually. Our study demonstrates positive impact of automated features on energy savings and this
impact becomes more significant as penetration increases. Low-level automated vehicles provide
environmental benefits for the society (i.e. reduce GHG emission). Hence, authorities should
promote development and deployment of low-level automated vehicle technologies.
This study has a few limitations and future work opportunities. First, we do not consider
interactions of technologies and their dependencies. Automated features might have interactions
with each other that could affect the overall saving potential (perhaps lowering it). For instance,
LDW and LKA might have some interactions in their impact levels. Interactions and dependencies
should be considered in the next versions of the model. Exact timeline of market penetration of

Vasebi, Hayeri, Samaras, and Hendrickson 12
automated technologies and their penetration trends are not clear and they were just estimated by a
few studies (34,36). We considered equal penetration rates for all technologies. However, different
levels of penetration of automated technologies could cause varying results (e.g. negative total fuel
impact). These scenarios should be investigated in future studies. Moreover, we only analyzed
energy impacts of automated vehicles. Ideally, the model should be expanded to include a holistic
analysis of automated vehicles; one that could capture not only energy saving potentials but also
social, economic, mobility and safety impacts.
7. Acknowledgement
The authors sincerely thank Professor Jeremy Michalek from Carnegie Mellon University, who
co-developed some of the early models adopted for this study.

Vasebi, Hayeri, Samaras, and Hendrickson 13
8. REFERENCES
1. Use of Gasoline, U.S. Energy Information Administration, 2017 [cited 2017 June]; Available from:
https://www.eia.gov/energyexplained/index.cfm?page=gasoline_use.
2. Jermakian Jessica S, Crash avoidance potential of four passenger vehicle technologies. Accident
Analysis & Prevention,Vol 43, No.3, 2011, pp. 732-740.
3. Bayly Megan, Brian Fildes, Michael Regan, and Kristie Young, Review of crash effectiveness of
Intelligent Transport Systems. Emergency,Vol 3, 2006, pp. 14.
4. Harper Corey D, Chris T Hendrickson, and Constantine Samaras, Cost and benefit estimates of
partially-automated vehicle collision avoidance technologies. Accident Analysis & Prevention,Vol 95,
2016, pp. 104-115.
5. Carsten Oliver MJ and FN Tate, Intelligent speed adaptation: accident savings and cost–benefit analysis.
Accident Analysis & Prevention,Vol 37, No.3, 2005, pp. 407-416.
6. Brown Austin, Jeffrey Gonder, and Brittany Repac, An analysis of possible energy impacts of
automated vehicle, in Road vehicle automation, 2014, Springer, pp. 137-153.
7. MacKenzie Don, Zia Wadud, and Paul Leiby, A first order estimate of energy impacts of automated
vehicles in the United States, in Transportation Research Board annual meeting: Washington, D.C.,
USA, 2014.
8. Wadud Zia, Don MacKenzie, and Paul Leiby, Help or hindrance? The travel, energy and carbon impacts
of highly automated vehicles. Transportation Research Part A: Policy and Practice,Vol 86, 2016, pp.
1-18.
9. Greenblatt Jeffery B and Samveg Saxena, Autonomous taxis could greatly reduce greenhouse-gas
emissions of US light-duty vehicles. Nature Climate Change,Vol 5, No.9, 2015, pp. 860-863.
10. Stephens S, Jeff Gonder, Yuche Chen, Z Lin, C Liu, and D Gohlke, Estimated bounds and important
factors for fuel use and consumer costs of connected and automated vehicles, National Renewable
Energy Laboratory 2016.
11. Shladover S. , Research and Policy Actions Needed to Support Implementation, Transportation
Research Board annual meeting: Washington, D.C., USA, 2017.
12. National Highway Traffic Safety Administration (NHTSA), Preliminary Statement of Policy
Concerning Autonomous Vehicles, in U.S. Department of Transportation, 2013. Available from:
http://www.nhtsa.gov/About+NHTSA/Press+Releases/U.S.+Department+of+Transportation+Releases
+Policy+on+Automated+Vehicle+Development
13. Anderson R, TP Hutchinson, B Linke, and G Ponte, Analysis of crash data to estimate the benefits of
emerging vehicle technology. Centre for Automotive safety Research, The University of Adelaide,Vol,
2010.
14. How much does the United States consume?, U.S. Energy Information Administration (EIA), 2016
[cited 2017 July]; Available from: http://www.eia.gov/tools/faqs/faq.cfm?id=23&t=10.
15. MacLean Heather L and Lester B Lave, Peer reviewed: A life-cycle model of an automobile.
Environmental Science & Technology,Vol 32, No.13, 1998, pp. 322A-330A.
16. Schrank David, Bill Eisele, Tim Lomax, and Jim Bak, 2015 urban mobility scorecard. 2015.
17. Blincoe L. J., Miller, T. R., Zaloshnja, E., & Lawrence, B. A. The economic and societal impact of motor
vehicle crashes, 2010. (Revised), DOT HS 812 013, Department of Transportation, National Highway
Traffic Safety Administration, 2015, May, Available from:
https://crashstats.nhtsa.dot.gov/Api/Public/ViewPublication/812013.
18. National Highway Traffic Safety Administration (NHTSA). Traffic Safety Facts 2015, Department of
Transportation, FARS GES Annual Report, 2016, Available from:
https://crashstats.nhtsa.dot.gov/#/DocumentTypeList/12.
19. Shladover Steven, Dongyan Su, and Xiao-Yun Lu, Impacts of cooperative adaptive cruise control on
freeway traffic flow. Transportation Research Record: Journal of the Transportation Research
Board,Vol, No.2324, 2012, pp. 63-70.

Vasebi, Hayeri, Samaras, and Hendrickson 14
20. Gonder Jeffrey, Matthew Earleywine, and Witt Sparks, Analyzing vehicle fuel saving opportunities
through intelligent driver feedback. SAE International Journal of Passenger Cars-Electronic and
Electrical Systems,Vol 5, No.2012-01-0494, 2012, pp. 450-461.
21. Where the Energy Goes: Gasoline Vehicles, Department of Energy, [cited 2017 October], Available
from: http://www.fueleconomy.gov/feg/atv.shtml.
22. Federal Highway Administration (FHWA). Highway Statistics 2015: Vehicle-miles of travel, by
functional system, 1980-2014, Department of Transportation, 2015, Available from:
https://www.fhwa.dot.gov/policyinformation/statistics/2015/.
23. Robert Bosch GmbH, Bosch Automotive Handbook. 9th ed. 2014: SAE.
24. Triplett T., Santos, R., Rosenbloom, S., & Tefft, B. , American Driving Survey: 2014–2015. 2016.
Available from: https://www.aaafoundation.org/sites/default/files/AmericanDrivingSurvey2015.pdf.
25. Department of Energy (DOE). Quadrennial Technology Review 2015: Internal Combustion Engines,
Department of Energy, 2015, Available from:
https://energy.gov/sites/prod/files/2015/11/f27/QTR2015-8C-Internal-Combustion-Engines.pdf.
26. National Highway Traffic Safety Administration (NHTSA). Final Regulatory Impact Analysis:
Corporate Average Fuel Economy for MY 2017-MY 2025 Passenger Cars and Light Trucks,
Department of Transportation, 2012, Available from:
http://www.nhtsa.gov/staticfiles/rulemaking/pdf/cafe/FRIA_2017-2025.pdf.
27. Hymel K M Small K A and Dender K V, Induced demand and rebound effects in road transport
Transportation Research Part B: Methodological Vol 44 2010, pp. 1220-1241.
28. Goodwin Phil, Joyce Dargay, and Mark Hanly, Elasticities of road traffic and fuel consumption with
respect to price and income: a review. Transport reviews,Vol 24, No.3, 2004, pp. 275-292.
29. Barr L C, Testing for the significance of induced highway travel demand in metropolitan areas.
Transportation Research Record: Journal of the Transportation Research Board,Vol 1706, 2000, pp.
1-8.
30. Haughton S Sarkar, Gasoline tax as a corrective tax: estimates for the United States 1970-1991. The
Energy Journal,Vol 2, 1996, pp. 103-126.
31. Gasprices. Gas prices. 2017 [cited 2017 10 Jan ]; Available from: http://gasprices.aaa.com/.
32. Fulton L. M., Noland, R. B., Meszler, D. J., & Thomas, J. V. , A statistical analysis of induced travel
effects in the US mid-Atlantic region. Journal of Transportation and Statistics,Vol 3, No.1, 2000, pp.
1-14.
33. Kunze R., Haberstroh, M., Ramakers, R., Henning, K., & Jeschke, S. , Automated truck platoons on
motorways–a contribution to the safety on roads Automation, Communication and Cybernetics in
Science and Engineering 2009/2010, 2011, pp. 415-426.
34. Kyriakidis Miltos, Carlo van de Weijer, Bart van Arem, and Riender Happee, The deployment of
advanced driver assistance systems in Europe. 2015.
35. Regan M. A., Mitsopoulos, E., Haworth, N., & Young, K., Acceptability of in-vehicle intelligent
transport systems to Victorian car drivers. Monash University Accident Research Centre,Vol, 2002.
36. Navigant Research, Autonomous Vehicles Self-Driving Vehicles, Autonomous Parking, and Other
Advanced Driver Assistance Systems, Global Market Analysis and Forecasts, 2013. Available from:
http://www.navigantresearch.com/wp-content/uploads/2013/08/ASDV-13-Executive-Summary.pdf
37.Shoup D C 2006 Cruising for parking. Transport Policy, 13(6), 479-486.
38.Wilmink I., et. al (2008). Impact assessment of Intelligent Vehicle Safety Systems. eIMPACT
Deliverable D4. Version 1.0, April 2008.
https://trimis.ec.europa.eu/sites/default/files/project/documents/20130401_140253_77423_eIMPACT_
D9_D10_v2.0.pdf
39.Kulmala, Risto, Pirkko Rämä, and Niina Sihvola, Safety impacts of cooperative systems, In Proceedings
of 21st ICTCT workshop, pp. 288-293. 2008. Available from:
http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.533.5412&rep=rep1&type=pdf
40.Luo, Li-hua, et al. Model predictive control for adaptive cruise control with multi-objectives:
comfort, fuel-economy, safety and car-following. Journal of Zhejiang University SCIENCE A 11.3

Vasebi, Hayeri, Samaras, and Hendrickson 15
(2010): 191-201. Available from: https://doi.org/10.1631/jzus.A0900374
41.Park, B., Malakorn, K., & Lee, J., Quantifying benefits of cooperative adaptive cruise
control toward sustainable transportation system. Center Transportation Studies, Univ. Virginia,
Charlottesville, VA, USA, 2011.
42.Lind, H., An efficient visual forward collision warning display for vehicles (No.
2007- 01-1105). SAE Technical Paper, 2007.
43.Abele, C. Kerlen, S. Krueger, H Baum, et al. Exploratory study on the potential socio-
economic impact of the introduction of intelligent safety systems in road vehicles. Final
report, SEISS, Teltow, Germany, January 2005.
44.Cervero, R., & Hansen, M., Induced travel demand and induced road investment: a
simultaneous equation analysis. Journal of Transport Economics and Policy (JTEP), Vol. 36, No. 3,
2002, 469-490.

Vasebi, Hayeri, Samaras, and Hendrickson 16
LIST OF TABLES
TABLE 1 Existing assessments of automated and autonomous vehicles' safety and energy impacts…….21
TABLE 2 Key parameters used for the analysis of low-level automation technologiesError! Bookmark
not defined.
TABLE 3 Comparison of all road users and automated vehicle owners' fuel savings ................................ 23
TABLE 4 Comparison of energy impact analysis of automated and autonomous vehicles systems. ......... 24
LIST OF FIGURES
FIGURE 1 Scope and boundaries of this study ........................................................................................... 17
FIGURE 2 Automated technologies fuel saving potential on 50% market penetration rate with 90%
confidence interval ............................................................................................................................... 18
FIGURE 3 percent fuel saving for AVOs based on automated technology groups (with 90% confidence
interval) ................................................................................................................................................ 19
FIGURE 4 Percent fuel savings of automated vehicles owners and all road users for different penetration
rates of the technologies. ...................................................................................................................... 20

Vasebi, Hayeri, Samaras, and Hendrickson 17
FIGURE 1 Scope and boundaries of this study

Vasebi, Hayeri, Samaras, and Hendrickson 18
FIGURE 2 Automated technologies fuel saving potential on 50% market penetration rate with 90%
confidence interval
-2%
2%
6%
10%
Lane departure warning
Collision warning
Blind spot warning
Speed limit detection
Traffic warning
Cooperative adaptive cruise control
Adaptive cruise control
Lane keeping
Collision detection braking
Active braking
Parking aid system
Dynamic route guidance
Percent Energy Savings Potential
Traffic Rebound
Operational Load
Aerodynamic force reduction
Non-accident related congestion
reduction
Accident related congestion reduction
Total

Vasebi, Hayeri, Samaras, and Hendrickson 19
FIGURE 3 percent fuel saving for AVOs based on automated technology groups (with 90%
confidence interval)

Vasebi, Hayeri, Samaras, and Hendrickson 20
FIGURE 4 Percent fuel savings of automated vehicles owners and all road users for different
penetration rates of the technologies. (a) fuel saving of all automated features with ACC for
automated vehicles owners; (b) fuel saving of all automated features in a connected
V2Venvironment for automated vehicles owners; (c) fuel saving of all automated features
with ACC for all road users; (d) fuel saving of all automated features in a connected V2V
environment for all road users.

Vasebi, Hayeri, Samaras, and Hendrickson 21
TABLE 1 Existing assessments of automated and autonomous vehicles' safety and energy impacts
Study
Scope
Model
Results
Harper et al. (4)
Safety/social benefits of
automated vehicle
features
Studied safety/social impacts of LDW, FCW, and
BSW based on crash and insurance data.
If all light duty vehicles in the U.S. would be equipped by these warning systems, it will
result in $4 billion to $202 billion annual net benefit cumulatively per year.
Bayly et al. (3)
Safety/social benefits of
automated vehicle
features
Reviewed crash avoidance impact of 138 vehicle
safety features in European Union; including passive,
active, and automated safety features.
The study claims that automation safety features could prevent most human errors and
reduce accidents, injuries, and fatalities.
Jermakian (2)
Safety/social benefits of
automated vehicle
features
Reviewed accident recodes of National Automotive
Sampling System and the General Estimates System
and Fatality.
LDW individually could prevent 3% of crashes and save up to 7,529 lives and BSW could
reduce accidents by 6.8% and save 393 lives per year.
Anderson et al.
(13)
Safety/social benefits of
automated vehicle
features
Analyzed safety benefit/cost of FCA, alcohol
interlocks, and fatigue management systems.
These technologies could avoid 31% of passenger cars fatal in Australia. This study
estimates the technologies provide 0.5 to 1.3 benefit-cost ratios.
Carsten et al. (5)
Safety/social benefits of
automated vehicle
features
Studied installation of speed limit detection
technology in advisory, voluntary, and mandatory
level
Speed limit detection technology in advisory, voluntary, and mandatory level could reduce
fatal accidents by 18-24%, 19-32%, and 37-59% respectively. Their safety cost and benefit
ratios are between 5.0 to 16.7.
Greenblatt et al.
(9)
SAV energy impact
Studied SAV influence on fuel demand and GHG
emissions in the U.S.
If SAV is substituted for 10% of vehicle mile travel (VMT) in the U.S., it will result in 65
to 75 MTCO2 (metric tons of carbon dioxide) and 25.2 million gallons gas reduction
annually. This reduction equals to $56.7 million fuel saving or $0.21 saving per vehicle per
year.
Stephens et al.
(10)
Light duty partial
autonomous (i.e. driver
assistant feature with
human involvement),
CAV, and shared CAV’s
energy impact
Studied potential impacts of connected autonomous
vehicle (CAV) and vehicle ridesharing on VMT, fuel
consumption, and costs of the users.
They estimated -10.68% to 13.59% increase in fuel demand for partial AV and also -58.25
to 204.85% and -64.08% to 194.17% increase in fuel consumption for fully CAV and fully
rideshared CAV respectively.
Brown et al. (6)
Light duty AV energy
impact
Considered several AV’s potential impacts (including
air drag reduction (i.e. platooning), efficient routing,
efficient driving, more travel, vehicle size
optimization, and fast parking).
They concluded that fully autonomous vehicles (NHTSA levels 3 and 4 (12)) may increase
fuel demand by 173% or decrease by 95%.
MacKenzie et
al. & Wadud et
al. (7, 8)
Heavy duty and light
duty AV energy impact
Analyzed AV impacts on VMT and energy demand.
They considered several parameters (e.g. eco-driving,
higher speed travel, vehicle right size, congestion
reduction, and crash avoidance). Due to uncertainty in
future prediction of AV (e.g. acceptance of AV
technology, traffic elasticity, traffic policies), they
developed four scenarios.
The most optimistic scenario estimates 40% reduction in total fuel demand of
transportation sector and the most pessimistic scenario predicts significant growth (i.e.
more than 100%) in fuel demand when AV reaches levels 3 & 4 of automation. However, if
AV is stopped in NHTSA’s level 2 automation (12), it will reduce total fuel consumption
on road transportation by 9%.
Our study
Light duty automated
vehicles’ energy impact
Analyzed impacts of each automated technology in
detail for the vehicles owners and all road users.
Detailed results and comparison with existing literature in Table 4.

Vasebi, Hayeri, Samaras, and Hendrickson 22
TABLE 1 Key parameters used for the analysis of low-level automation technologies
Shaded areas assumed negligible impact by the technology on fuel consumption.
: induced traffic elasticity of automated vehicles assumed values of  (27, 29, 30, 32, 44).
 Automated features increase fuel efficiency therefore reduce cost of driving. As a result, more trips can be generated. In regulatory impact assessment
of new corporate average fuel economy (CAFÉ) regulations, NHTSA (18) reviewed recent rebound research showing a declining rebound effect over time and used a baseline
rebound value of 10 percent, with a sensitivity analysis using 5 percent and 25 percent bounds.
% of accident-related congestion reduced by
technology i
AC
% of non-accident-related congestion
reduced by technology i
NAC
% of fuel saving capability of
technology 
NACAV
% of fuel saving capability of
technology 
AFFS
Automated vehicles’ equipment
average energy consumption rate
(watt), and Portion of driving time
when equipment is active
(EQP,
ACT)
Low
Mode
High
Rationale and source
Low
Mode
High
Rationale and
source
Low
Mode
High
Rationale
and source
Low
Mode
High
Rationale
and source
Camera
Radar
Steering
Brake
Lane
departure
warning
0%
1.5%
3%
1.21% (4); 7% of fatal
crashes (13) equal to
0.035% of all crashes
(18); 3% (2)
15, 1
Collision
warning
2%
3%
4%
7% of rear-end (equal to
1.9% of all accidents)
(35); 3.97%(4)
20, 1
Blind spot
warning
0.5%
3.65
%
6.8%
0.53%( 4); 6.8% (2)
20, 1
Speed limit
detection
2.8%
3.2%
3.6%
10-13% of injury crashes
and 18-24% of fatal
crashes (5); equal to 2.8-
3.6% of all crashes (18)
15, 1
Traffic
warning
0%
0.5%
1%
0.5% (42)
Adaptive
cruise
control
1.7%
4.5%
7.25
%
4-7% (13); 5.9% of rear-
end equal to 1.71% of all
(3); 25% of rear-end equal
to 7.25% of all (43)
7%
13.5%
20%
7.59-
19.69%
(40)
13.38 %
(41)
20,
0.55
60, 0.55
Cooperative
adaptive
cruise
control
1.7%
4.75
%
8.75
%
A conservative range
between ACC and
platooning accident
reduction capabilities
(3,13,33,43)
0%
50%
100%
It is varying from
0% to 100% for
different market
penetration rates
(19)
0%
18.5%
37%
Up to 37%
fuel saving
(41)
20,
0.55×
Mi
60,
0.55×Mi
Lane
keeping
2.5
3.2
4.2
8.9 - 15.2% fatality and
injury crashes; equal to
2.5-4.2% of all crashes
(18, 38) 2.9% all crash
(39)
15, 1
60,
negligible
Collision
detection
braking
3
3.5
4
7% fatality and injury of
rear-end and fixed
obstacle crashes equal to
3.5% of all crashes (38)
20, 1
60,
negligible
Active
braking
5%
15%
25%
Range aggregated 5-25%
(3)
20, 1
60,
negligible
Parking aid
system
1%
6%
11%
15% of all VMT
on CBD (22,37).
2%
4%
6%
4% fuel
saving (6).
Dynamic
route
guidance
10%
15%
20%
(3)
3%
5%
7%
5% fuel
saving (6,
20).

Vasebi, Hayeri, Samaras, and Hendrickson 23
TABLE 2 Comparison of all road users and automated vehicle owners' fuel savings
Benefit/Cost Category
All Road Users
(percentage saving)
Automated
Vehicle Owners
(percentage saving)
Reduced accident related congestion
0 – 0.3
0 – 0.3
Reduced non-accident related congestion
0 – 2.1
8.8 – 10.7
Aerodynamic Force Reduction
0
7.2 – 16.3
Operational Load
0
-1.3 – -1.7
Traffic Rebound
-0.5 – 0
-1 – -3.2

Vasebi, Hayeri, Samaras, and Hendrickson 24
TABLE 3 Comparison of energy impact analysis of automated and autonomous vehicles systems.
Study
Type of Impact
Analysis
Scope
Analysis factors
Cost Savings in
$2016/VMT
This Study
Energy Saving
Individual
Low-level
Automated
Light Duty
Vehicle
Accident congestion reduction
Non- accident congestion reduction
Air drag reduction
Operation load
Travel rebound
$60 to $266
saving per year
per vehicle
Brown et al.
(2014) (6)
Energy Saving
Autonomous
Light Duty
Vehicles
Efficient driving
Air drag reduction
Travel rebound
Efficient routing
Travel by underserved population
Vehicle size optimization
Fast parking
Higher occupancy
Vehicle electrification
$ -2063 to 1132
saving per year
per vehicle
MacKenzie et
al. (2014) &
Wadud et al.
(2016) (7, 8)
Energy Saving
Autonomous
Heavy Duty and
Light Duty
Vehicles
Improved crash avoidance
Congestion mitigation
Eco-driving
Air drag reduction
De-emphasized performance,
Vehicle right-sizing
Higher highway speed
Reduction in generalized costs
New user groups
Car-sharing
Increased features
Potential for low carbon transition
$-975 to 542
saving per year
per vehicle
Stephens, et al.
(2016) (10)
Energy Saving
Partial and fully
Connected
shared/non-shared
autonomous light
duty vehicles
(SAV)
Fast travel
air drag reduction
V2I communication
collision avoidance
smooth traffic flow
vehicle resize
empty vehicles
ridesharing
search for parking spot
easier travel
travel mode change
new users
$-764 to 2442
saving per year
per vehicle
Greenblatt et
al. (2015) (9)
Energy Saving
SAV
VMT covered by shared AV
$0.21 per road
user
Harper (2016)
(4)
Safety
Individual
Low-level
Automated
light Duty Vehicle
Crash reduction benefits
Equipment installation costs
$20 to 877
saving per year
per vehicle