ThesisPDF Available

Increasing Electric Vehicle Range with a Recommendation App providing ContextSpecific Trip Rankings

Thesis

Increasing Electric Vehicle Range with a Recommendation App providing ContextSpecific Trip Rankings

Abstract and Figures

Electric vehicles (EVs) and in particular battery-electric vehicles (BEVs) represent a scalable and yet sustainable way of future personal transportation. While the limited driving range, which is an obstacle for BEV adoption, can be directly improved by technological inventions on the long run, we are aiming for indirect and immediate ways like optimizing energy consumptions by more efficient driving styles, or increased awareness for range-limiting factors. Within more than two years, driving data of several thousand BEV trips was recorded, processed and analyzed, in order to reveal dependencies of outdoor temperature, route topography, tires, and many more. Based on this data a novel in-car dashboard was developed to provide the driver with extended live-feedback, recommendations, and a ranking to compare the energy consumption with other drivers and historical trips on the same route and under similar conditions. Within a real-world user study it was evaluated whether such gamification-based applications can unobtrusively motivate humans and change behavior towards more energy efficient driving.
Content may be subject to copyright.
Submitted by
Benjamin Pichler
Submitted at
Institute of
Pervasive Computing
Supervisor
Priv.-Doz. DI Dr.
Andreas Riener
November 2016
JOHANNES KEPLER
UNIVERSITY LINZ
Altenbergerstraße 69
4040 Linz, ¨
Osterreich
www.jku.at
DVR 0093696
Increasing Electric
Vehicle Range with a
Recommendation App
providing Context-
Specific Trip Rankings
Master Thesis
to obtain the academic degree of
Diplom-Ingenieur
in the Master’s Program
Computer Science
ABSTRACT
Electric vehicles (EVs) and in particular battery-electric vehicles (BEVs) represent a scalable and yet
sustainable way of future personal transportation. While the limited driving range, which is an obstacle
for BEV adoption, can be directly improved by technological inventions on the long run, we are aiming
for indirect and immediate ways like optimizing energy consumptions by more efficient driving styles,
or increased awareness for range-limiting factors. Within more than two years, driving data of several
thousand BEV trips was recorded, processed and analyzed, in order to reveal dependencies of outdoor
temperature, route topography, tires, and many more. Based on this data a novel in-car dashboard was
developed to provide the driver with extended live-feedback, recommendations, and a ranking to compare
the energy consumption with other drivers and historical trips on the same route and under similar condi-
tions. Within a real-world user study it was evaluated whether such gamification-based applications can
unobtrusively motivate humans and change behavior towards more energy efficient driving.
KUR ZFAS SUNG
Elektrofahrzeuge und im speziellen Batterie-Elektrofahrzeuge stellen eine skalierbare und trotzdem nach-
haltige Art des k¨
unftigen Personenverkehrs dar. Die geringen Reichweiten, welche momentan als großes
Hindernis f¨
ur deren Verbreitung gelten, werden auf lange Sicht durch technologische Fortschritte direkt
verbessert. Diese Arbeit widmet sich jedoch der indirekten aber daf ¨
ur sofortigen M¨
oglichkeit einer Re-
duktion des Energieverbrauchs durch effizienteres Fahren und erweitertes Wissen ¨
uber jene Faktoren die
Einfluss auf die Reichweite haben. ¨
Uber mehr als zwei Jahre wurden Daten zu tausenden Fahrten eines
Elektroautos aufgezeichnet, verarbeitet und analysiert, um m ¨
ogliche Zusammenhange mit Außentem-
peraturen, H¨
ohenprofilen, Reifen, uvm. herstellen zu k¨
onnen. Darauf aufbauend wurde ein neuartiges
Armaturenbrett (Dashboard) vorgestellt, welches dem Fahrer Live-Feedback anzeigt und Hinweise gibt.
Zus¨
atzlich erm¨
oglicht es einen Vergleich (Ranking) des eigenen Energieverbrauchs mit jenem von an-
deren Fahrern bzw. bereits aufgezeichneten Fahrten, welche auf derselben Strecke und unter ¨
ahnlichen
Bedingungen stattfanden. In einer Nutzerstudie wurde dann sowohl quantitativ als auch qualitativ er-
hoben, ob sich dieser spielerische und kompetitive Ansatz eignet, Menschen unterschwellig zu einem
sparsameren Fahrstil zu motivieren.
ACKNOWLEDGMENTS
I would like to thank my adviser Andreas Riener, whose mentoring and ongoing support was more than
excellent. Special thanks also to Johannes Freudenthaler and IBIOLA Mobility Solutions GmbH [50] for
access to both the car-sharing network database and real-time data as well as for their continuous support.
I am also very thankful to Norbert Rainer who made it possible to conduct a user study with electric
vehicles in Krenglbach, Austria. Further I want to thank my family and friends for their support as well
as Tina F¨
ureder for proof reading. Thank you so much.
-i-
Contents
CONTENTS
1 Introduction 1
1.1 MotivationandProblem .................................. 1
1.2 The Future of Electric Vehicles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.2.1 Types of Electric Vehicles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.2.2 Costs ........................................ 6
1.2.3 Efficiency and Environmental Impacts . . . . . . . . . . . . . . . . . . . . . . . 7
1.2.4 Motor........................................ 11
1.2.5 Batteries ...................................... 11
1.2.6 Charging and Infrastructure . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
1.2.7 UsageAnalysis................................... 20
1.2.8 Incentives...................................... 22
1.2.9 TrendsandForecasts................................ 22
1.3 Behavior Change and Gamification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
1.4 Increasing Efficiency and Range . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
1.5 OutlineofthisThesis.................................... 26
2 Related Work 28
2.1 EVRangeProblems..................................... 28
2.2 Range-Limiting Factors of EVs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.2.1 External and Environmental Factors . . . . . . . . . . . . . . . . . . . . . . . . 29
2.2.2 Vehicle-Dependent Factors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
2.2.3 PersonalFactors .................................. 34
2.3 AnalyzingEVTrackData.................................. 39
2.4 Behavior Change for Efficient Driving . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
2.5 Summary and Distinction of our Work . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
3 Track Processing and Analysis 44
3.1 CollectingDrivingData................................... 44
3.2 Preprocessing........................................ 45
3.3 RouteCorrection ...................................... 45
3.4 EnhancingTracks...................................... 49
3.4.1 WeatherData.................................... 50
3.4.2 ElevationData ................................... 51
3.4.3 TireData ...................................... 52
3.5 AnalysisofTracks ..................................... 52
3.5.1 Track Data Representation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
3.5.2 Interactive Exploration Tool . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
3.5.3 Visual Analysis with Different Data Views . . . . . . . . . . . . . . . . . . . . 55
4 Dashboard Application 57
4.1 Hardware and Vehicle Installation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
- ii -
Contents
4.2 Software and Interface Implementation . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
4.3 ComparisonofTracks.................................... 65
4.4 Recommendations for Efficient Driving . . . . . . . . . . . . . . . . . . . . . . . . . . 69
5 Results and Evaluation 72
5.1 Basic Statistics of EV Tracks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
5.2 EvaluationofPotentials................................... 79
5.3 UserStudy ......................................... 87
5.3.1 Setup ........................................ 87
5.3.2 Quantitative Evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
5.3.3 Qualitative Evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
6 Summary and Future Work 99
A Appendix 101
A.1 Survey............................................ 101
References 104
- iii -
List of Figures
LIST OF FIGURES
1.1 The historical progression of total automobile registrations in Germany highlights the
huge popularity of petrol and diesel vehicles (top). The pie charts (bottom) are visualizing
the detailed shares and proportions of different drivetrain types in 2014, which shows that
BEVs accounted for just 0.0003 % in total [76]. . . . . . . . . . . . . . . . . . . . . . . 2
1.2 The Well-to-Wheel (WtW) analysis along the energy path includes all stages from the
fuel production (or energy generation) to the conversion into kinetic energy within the
vehicle. ........................................... 8
1.3 The Well-to-Wheel (WtW) emissions of BEVs and other vehicle types consist of different
Well-to-Tank (WtW) and Tank-to-Wheel (TtW) ratios. BEV emissions are largely depen-
dent on electricity generation mix (e-mix) that is used to charge batteries (data from 2014
[27,73]). .......................................... 10
1.4 Approximate gravimetric and volumetric energy densities of several available battery
chemistries. Lithium-ion (Li-Ion) batteries are most popular in today’s BEVs, as they
need less weight and space to store energy [22, 4]. Abbreviations: lead-acid (Pb-acid),
nickel-cadmium (Ni-Cd), nickel-zinc (Ni-Zn), nickel-metal hybrid (Ni-MH), sodium-nickel
chloride(Na-NiCl)...................................... 13
1.5 Shows the average reduction of battery capacity (relative to initial capacity) over time.
The technological progress of four Li-Ion battery generations is indicated based on their
year of production (with forecasts for 2020 and 2025). The guarantee of battery manufac-
turers is typically based on the time when battery life is considered to be at the end (under
80%withinvehicles)[83].................................. 15
1.6 Schematic progression of battery capacity over the full lifetime. After the first life of
batteries within EVs, the remaining storage capacity might still be sufficient for a second
life with different requirements [83]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
1.7 Popular lithium-based battery chemistries and their properties are illustrated by means
of a polar chart, in which more distance to center means better (which corresponds to
cheaperonthecostscale). ................................ 17
1.8 Average combined electric power demand over one day (based on data from New England,
2010)[55]. ......................................... 20
1.9 Typical location/activity of vehicles throughout the week (data from 2001, USA [22]).
Although the long parking periods indicate a bad utilization, they represent an interesting
opportunity to recharge EVs, especially at night times. . . . . . . . . . . . . . . . . . . 21
1.10 Aggregated average vehicle location/activity on weekdays or weekends (data from 2009,
Germany [99]). Actually, vehicles are driven only for a very small percentage of the day,
with weekends being no exception (red slice). . . . . . . . . . . . . . . . . . . . . . . . 21
2.1 The life of Li-Ion batteries (cycle life) strongly depends on the operating temperature.
Aging is also influenced by the rate of charge/discharge, which is depicted for 1C, 2C,
and3C[74].......................................... 30
2.2 Classification of rolling resistances for the EU efficiency label on passenger vehicle tires. 33
- iv -
List of Figures
2.3 The energy required to move a typical passenger car (at constant speed without slope) is
distributed as a function of the vehicle speed. The main forces are rolling resistance (Frr ),
internal friction (Fv) and air drag (Fd) which becomes the predominant factor at higher
drivingspeeds[9]....................................... 35
2.4 The estimated driving range at different speed levels also depends on auxiliary loads.
The optimal speed to maximize range increases from disabled heating to medium heating
(2 kW) up to maximum heating (4.2 kW) settings [57]. . . . . . . . . . . . . . . . . . . 36
3.1 Processing steps from collecting driving data within vehicles to analyzing those tracks
(trips) with an interactive exploration tool. . . . . . . . . . . . . . . . . . . . . . . . . . 44
3.2 The route correction algorithm is needed to retrieve the fine-grained ”correct” driving
route (below) based on the scarce recorded track point locations (above). . . . . . . . . . 46
3.3 Small inaccuracies of recorded GPS coordinates can lead to erroneous computations of
the driving route (left side). Intelligently excluding the problematic raw coordinates (mag-
nified area) results in successful route corrections (right side). . . . . . . . . . . . . . . . 47
3.4 Illustration of the ”Divide and Conquer” algorithm to find bad (inaccurate) coordinates.
The routing API is queried multiple times with smaller sets of raw coordinates in order to
narrow down the problematic area. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
3.5 Mapping of track point data from raw route coordinates (left) to the nearest points within
the corrected route (right). Missing data between these points is interpolated (blue values). 49
3.6 For spatial interpolation of temperature and precipitation measurements, multiple weather
stations around the location of interest are queried. A weighted average is computed based
on the distance to each weather station (weights are illustrated as percentage values). . . 51
3.7 User interface of web-based tool for interactive exploration. Track data is visualized
based on three interconnected views: (A) list, (B) map and (C) visual analysis view. . . . 54
4.1 Smartphone-based dashboard mounted next to the steering wheel inside the BEV (right-
handtrafc)......................................... 60
4.2 Processing diagram with all steps to compute driving characteristics and comparisons
(rankings) that are displayed to the driver with the dashboard application. . . . . . . . . 62
4.3 User Interface of our novel dashboard application (English translation). . . . . . . . . . 63
4.4 Loading/preparation screen of the dashboard application while the vehicle is driven (En-
glish translation). Distance-dependent computations would yield inaccurate data that
should not be displayed until a few kilometers were driven. . . . . . . . . . . . . . . . . 64
4.5 The route of the current track (left) next to a different/historical track’s route (right) illus-
trated on the same map segment. Only those parts of other tracks, that are overlapping
with legs of the current track (indicated by color), are considered for comparisons. . . . . 67
4.6 The ranking visualization of the dashboard application shows comparison results regard-
ing the average consumption. On the left side, the driver is moving more energy efficiently
than other drivers. And on the right side, it gets indicated that other drivers performed
better on the current route (and under similar conditions). . . . . . . . . . . . . . . . . . 68
5.1 The quantity of BEV tracks (trips) over a full year summarized per calendar week. The
type of tires that was typically applied (summer or winter tires) is indicated by different
colors............................................. 73
-v-
List of Figures
5.2 The quantity of BEV tracks summarized per weekday (Monday to Sunday). . . . . . . . 73
5.3 The histogram shows the frequency of BEV tracks regarding their duration (in range
classes)............................................ 74
5.4 The histogram shows the frequency of BEV tracks regarding their distance (in distance
classes)............................................ 75
5.5 The histogram shows the frequency of BEV tracks regarding their average outdoor tem-
perature values (in temperature classes). . . . . . . . . . . . . . . . . . . . . . . . . . . 76
5.6 The histogram shows the logarithmic frequency of BEV tracks regarding their average
precipitation rates (in classes). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
5.7 The histogram shows the frequency of BEV tracks regarding their altitude sum (in altitude
classes)............................................ 77
5.8 The histogram shows the frequency of BEV tracks regarding their altitude gain (in altitude
classes)............................................ 77
5.9 The histogram shows the frequency of BEV tracks regarding their average energy con-
sumption(inclasses)..................................... 78
5.10 The frequency of BEV tracks regarding their average energy consumption (in classes) is
illustrated with two histograms, once for tracks with summer tires (upper) and once with
winter tires (lower). Their combination can be seen in Figure 5.9. . . . . . . . . . . . . . 78
5.11 The quantity of trips each driver made over the full observation period (about 24 months). 79
5.12 Shows a significant correlation of energy consumption and outdoor temperatures of BEV
tracks (with color-coded tire type). The averaged dependency is indicated by a linear
regression curve (with LOESS). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
5.13 Correlation of energy consumption and altitude gain of BEV tracks. The averaged depen-
dency is indicated by a linear regression curve (with LOESS). . . . . . . . . . . . . . . 82
5.14 Correlation of energy consumption and altitude sum of BEV tracks. The averaged depen-
dency is indicated by a linear regression curve (with LOESS). . . . . . . . . . . . . . . 82
5.15 Correlation of energy consumption and driving distance of BEV tracks. The averaged
dependency is indicated by a linear regression curve (with LOESS). . . . . . . . . . . . 83
5.16 Correlation of energy consumption and the duration of BEV tracks. The averaged depen-
dency is indicated by a linear regression curve (with LOESS). . . . . . . . . . . . . . . 84
5.17 Correlation of energy consumption and average driving speed of BEV tracks. The aver-
aged dependency is indicated by a linear regression curve (with LOESS). . . . . . . . . 85
5.18 Correlation of energy consumption and average precipitation (rain) during BEV tracks.
The linear regression curve (with LOESS) indicates that there is almost no dependency. . 85
5.19 Boxplots summarizing the average energy consumptions of BEV trips for each driver. On
average, the worst-efficient driver (left boundary) consumes more than double the energy
of the most-efficient driver (right boundary). . . . . . . . . . . . . . . . . . . . . . . . . 86
5.20 Consumption comparison of BEV tracks during reference period (red) and test period
(blue) at corresponding outdoor temperatures. The linear regression curves (with LOESS)
indicate small tendential improvements in the test period. . . . . . . . . . . . . . . . . . 89
- vi -
List of Figures
5.21 Answers provided by interviewed drivers (on a 7-level Likert scale). A: To what extent
were you satisfied overall with the system? B: To what extent did you feel that the system
presented real data of your current driven trip? C: To what extent did you feel that the
system was something fun you were experiencing? D: To what extent were you satisfied
with the speed of the system?. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
5.22 Answers provided by interviewed drivers (on a 7-level Likert scale). A: To what extent
was the visual representation of the system appealing to you? B: How satisfied were you
with the readability and font size? C: How much effort did you put into controlling the
system? D: To what extent did the system distract you from driving?. . . . . . . . . . . . 94
5.23 Answers provided by interviewed drivers (on a 7-level Likert scale). A: How satisfied are
you with the unit ”% per kilometer” to denote energy consumption? B: Was the wording
ofthesystemfamiliar?.................................... 94
5.24 Answers provided by interviewed drivers (on a 7-level Likert scale). A: To what extent
do you like the idea of competing with other drivers regarding consumption? B: To what
extent did you like the presentation of rankings? C: How often did the system present
rankings with similar trips?. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
5.25 Answers provided by interviewed drivers (on a 7-level Likert scale). A: To what extent
did you feel you need more time in order to make progress? B: To what extent are you
interested in further developments of such systems? C: Would you like to use the system
again?............................................. 96
5.26 Answers provided by interviewed drivers (on a 7-level Likert scale). A: To what extent
did you feel like you were reducing your average consumption during the whole experi-
ment? B: To what extent did you feel as though you were able to influence the displayed
energy consumption? C: To what extent do you think such systems are useful to reduce
consumption?......................................... 96
5.27 Answers provided by interviewed drivers (on a 7-level Likert scale): To what extent do
you think certain factors can influence driving efficiency?. . . . . . . . . . . . . . . . . 97
5.28 Answers provided by interviewed drivers (on a 7-level Likert scale). A: To what extent
will you sacrifice comfort in order to reduce energy consumption? B: To what extent are
you willing to be charged for driving according to your driving efficiency? C: Would you
drive the same way if the vehicle was owned by you instead of the car sharing community?. 98
- vii -
List of Tables
LIST OF TABLES
1.1 Environmental impacts of different power plant technologies (average accross Europe) [30]. 9
1.2 Present and forecasted Li-Ion EV battery cost estimates. Technological progress and
economy- of-scale effects will allow a substantial price decrease of today’s most expan-
siveBEVpart[22]. ..................................... 14
1.3 Battery types and characteristics of several modern EVs [22, 68, 15]. . . . . . . . . . . . 17
2.1 Shows the impact of different driving styles and settings for HVAC (heating, ventilation,
and air conditioning) on BEV energy consumption, recorded within a Nissan Leaf. [30]. 38
4.1 English translations of recorded recommendations (tips) with short description and tran-
scribedauditoryoutput.................................... 70
5.1 Descriptive statistics of the several available features for each BEV track: duration, dis-
tance, outdoor temperature, precipitation (rain), altitude sum (added up and down move-
ments), altitude gain (up minus down movements), and average energy consumption. . . 74
5.2 Shows the statistical correlation between individual BEV track features and the average
energy consumption. The features are sorted by their absolute correlation coefficients,
according to Pearson and Spearman. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
5.3 Shows feature statistics for comparable BEV tracks within the reference period (stage 1)
and the test period (stage 2). To provide more expressive comparisons, the same basic
filters were applied to retrieve data of both periods (e.g. exclude tracks with summer
tires). The last column shows the significance (two-side) of a t-test for mean-equality. . . 88
- viii -
Chapter 1 - Introduction
1 INTRODUCTION
1.1 Motivation and Problem
Automobiles have proven to generate tremendous benefits for individuals as well as our society. On the
contrary, they cause negative effects like pollution, noise emission, and traffic accidents around the globe.
Nevertheless, the worldwide growth in popularity remains strong, as per 1000 inhabitants around 123
automobiles are available on average (by 2014). In highly developed economies there are even between
400 and 600, which is about half of the population [76]. Thus, the number of automobiles surpassed 1
billion worldwide, and it is assumed that it will double until 2030 [100].
With this rapid progress we cannot afford to forget about our environment. In the EU, for instance, the
overall transportation sector is responsible for 30 % of all fossil fuel emissions [31], with the vast majority
coming from road traffic [100]. Due to the widespread availability, it is no wonder that personal vehicle
emissions are the largest single contributor to household/individual emissions [8]. Part of these emissions
is the greenhouse gas carbon dioxide (CO2), which is known to amplify greenhouse effects and accelerate
global warming. Furthermore, the direct emissions of traditional gasoline-powered vehicles were found to
be a major source of urban air pollution and its negative health effects [18]. To reduce the environmental
impact in the long term, mandatory goals for lower total emissions were defined on a national level in
many countries. It is a goal of the EU, for instance, to reduce transportation-related emissions by 30 %
between 2005 and 2030 [76]. This includes also sanctions regarding automobile manufacturers and their
vehicle’s tailpipe emissions (e.g. CO2 emissions). To evaluate and estimate fuel consumption as well as
tailpipe emissions, different laboratory measurements are obtained over predefined test cycles, like the
NEDC (”New European Driving Cycle”) [63]. Unfortunately, it was found out that this type-approval
data and real-world CO2 values diverge more and more for newer vehicles. While the average real-world
values of vehicles built in 2001 were about 7 % higher, the discrepancy widened to 30 % in 2013 [63].
Facing this alarming trend and the negative side effects of today’s unrestricted mobility, the question of
a scalable and yet sustainable way of transportation arises. Traditional vehicles with internal combustion
engines (ICE) are on the way to be replaced by a combination of alternative concepts. One interesting
alternative is electric mobility and in particular the battery-electric vehicle (BEV). They promise to have
no (zero) tailpipe emissions, less energy consumption with more efficient drivetrains, lower noise and
vibration levels, low costs of operation, less vehicle components, and better traffic safety. On an energy-
centric point of view they might help to utilize local energy sources and reduce the dependence on specific
single energy sources like crude oil or natural gas, which are not only limited geographically but are also
non-renewable [100, 90, 96, 55, 31].
From a consumer perspective drivers have recognized even more advantages of BEVs, like the smooth
operation and the ability to conveniently charge the vehicle at home (e.g. over-night). In addition, the
electric drivetrain allows for a better acceleration with nearly instantaneous torque, independent from the
motor’s rotational speed [22].
Electric vehicles are thus seen as the main answer to the transport sector’s current problems and they have
tremendous potential to directly benefit society as a whole [55]. But of course EVs cannot improve the
climate footprint of transportation overnight. This is rather a long process and after all it seems we are just
-1-
Chapter 1 - Introduction
at the beginning of technological developments and serious interest of the broader public [100]. Looking
at the stock of German automobiles (total registrations), for instance, underlines this argument (see Figure
1.1). ICE vehicles powered by petrol or diesel accounted for over 99 % of all passenger cars in 2014 [76].
Hence, in Europe the overall share of EVs was under 1 % in 2015, and despite rapid growth rates, BEVs
accounted for less than 0.1 % [5]. It is estimated thus, that a sufficient market penetration of EVs will take
until 2030, when most of the existing ICE vehicles are replaced by environmentally friendly alternatives
[100]. To accelerate the adoption of EVs and especially BEVs, it is important to increase the awareness
of consumers and drivers. They should learn more about BEV characteristics in order to reduce certain
fears, refrain from prejudices, and understand specific dependencies.
1960 1966 1972 1978 1984 1990 1996 2002 2008 2014
0
5
10
15
20
25
30
35
40
45
50
dieselpetrol alternative
registered automobiles (millions)
Pkw-Bestand 2014 nach Antriebsarten, in Mio.
petrol
68.3 %
diesel
30.1 %
alternative
1.6 % LPG 73.7 %
hybrid 12.6 %
CNG 11.6 %
battery electric (BEV) 1.8 %
others 0.3 %
Figure 1.1: The historical progression of total automobile registrations in Germany highlights the huge
popularity of petrol and diesel vehicles (top). The pie charts (bottom) are visualizing the detailed shares
and proportions of different drivetrain types in 2014, which shows that BEVs accounted for just
0.0003 % in total [76].
-2-
Chapter 1 - Introduction
To improve certain weaknesses of BEVs, it is also important to find out more about the concerns and
fears of customers. One main issue was found to be the initial vehicle cost, which is substantially higher
than that of comparable ICE vehicles [55, 96]. The reason for this price differential is particularly due
to the large battery packs, which are not only heavy but also rather expensive today. Unfortunately it
is not only the cost of batteries, but also their performance that is problematic. Their energy capacity
directly limits the available driving range, which is typically much lower than with gasoline-powered
competitors [100, 96, 55]. Furthermore, when driving with ICE vehicles, we are used to having plenty of
gas stations around, which enable refueling within just a few minutes. BEVs on the other hand, will take
several hours and the availability of (public) charging stations is comparatively low [55, 100]. So it seems
technical boundaries and infrastructural circumstances are by far not optimal today and all obstacles for
BEV adoption are somehow related to battery constraints (cost, range, charging).
But is a driving range of about 100 kilometers indeed too little for most people? Actually, it was found
out that 80 % of vehicles in Germany travel less than 50 kilometers per day, and 95 % less than 100
kilometers [34]. In the United States just about 1 % of all trips are longer than 160 kilometers [22].
In fact, it was shown that more than 50 % of all journeys lie within a distance of five kilometers in
Austria [3]. So as the vast majority are short trips, it can be concluded that BEVs could already fulfill
the transportation needs of most people’s daily routine [100, 99]. But interestingly for many drivers
the lower range buffers of BEVs are causing range anxiety, which is the fear of running out of charge
[92, 31]. Various studies have shown that the preferred range of people is significantly higher than the
actual needed/used range. European drivers, for instance, responded on average to ”require” (prefer) a
range of 308 kilometers [34]. Further it was shown, that range anxiety is provoked and intensified by EV
range displays which are inaccurate and not always reliable [35, 92]. Most often the remaining driving
range is displayed as a single figure that is calculated based on variables and logics that are not apparent
to the drivers [57]. Our surveys have confirmed that one should not have blind faith in range displays, and
that some kind of learning and adaption may be necessary to better estimate the ”true” remaining distance
(e.g. decrease the numbers in colder winter months, or increase them in summer). Although range anxiety
might be reduced with driving experience over time, better dashboard displays can speed-up that process
[92]. One main aspect of better range estimations is to include dependencies and characteristics that are
not directly visible to the driver. This might include driving style and speed as well as climate control,
outdoor weather conditions, and many more. Helping drivers to better understand certain range-limiting
factors is particularly important from a psychological point of view, as it allows them to better utilize the
limited range and gain more confidence in BEVs [57, 35].
So in general there are two approaches to improve limited BEV range, either directly via technological
developments, or indirectly when the driver is able to better utilize the remaining battery capacity. Al-
though we expect technological optimizations from time to time, we should not underestimate the role of
human beings. Driving style, for instance, has a tremendous influence on the remaining range, because
high driving speeds and frequent accelerations eat up more energy. But despite the general knowledge of
more energy-efficient driving of ICE vehicles, there are several differences in regard to EVs [47]. One
example is that smoother accelerations can save much more energy on BEVs, because electric motors en-
able peak efficiencies regardless of the current rotational speed [47]. But also the opportunities of slower
regenerative braking are important to consider. Overall it is estimated that economic driving can reduce
-3-
Chapter 1 - Introduction
energy consumption by 10 % on the long term (sustained) and up to 25 % on a very focused (temporary)
basis [8]. Hence, the actions individuals can take to reduce environmental impacts are significant on
aggregate. But also for the driver itself they are relevant, because when driving BEVs inefficiently, the
increased energy consumption weighs particularly heavy due to the lower total range.
As such, the motivation of this work was to help eliminating obstacles regarding the adoption of BEVs,
which are seen as a very promising approach of a more energy-efficient and sustainable way of personal
transportation. BEVs might reduce the environmental impact by decreasing overall emissions, reduce
dependence on (imported) fossil fuels, and improve urban air quality.
To boost EV sales it is necessary to help consumers deal with limited range and gain more confidence.
Range anxiety is partly caused by unreliable range displays today, which do not consider most of the
range-limiting factors of EVs. Hence, it was an additional objective of this work to unfold such relation-
ships and provide a tool to better reason about impacts on energy consumption (and thus remaining range).
Such insights might be helpful to mitigate information deficits of drivers and gain better knowledge of
range-limiting factors.
Motivating people to drive more energy-efficient is finally considered a necessary complement to signifi-
cantly save energy, reduce emissions, and avoid catastrophic climate change [8]. But not only on a global
scale an optimized driving style is beneficial, because on a driver-centric point of view the range of BEVs
increases, recharging gets less urgent, operational costs are lower, and also the durability of batteries gets
improved.
1.2 The Future of Electric Vehicles
While active transport like walking or cycling has declined progressively since the 1950s, motorized
transportation has become extremely popular. At a global scale a reversal of this trend is not yet in sight
and thus more environmentally friendly ways of road traffic are important. To consider electric vehicles as
a replacement of traditional vehicles in the long term, we have to look at several characteristics of current
technologies as well as novel trends and forecasts. Therefore, this chapter covers the most important
aspects in a holistic consideration over the full lifetime of an automobile. Although several electric
vehicles are commercially available today, the most important obstacles for wide-spread adoption were
identified as little range, long charging periods and high initial costs. Analyzing and eliminating these
drawbacks is thus a major goal for a multitude of research topics.
1.2.1 Types of Electric Vehicles
Although the focus of this work is directed towards battery-electric vehicles, a comparison of different
powertrain concepts and energy sources is necessary to classify existing kinds of vehicles.
In traditional vehicles with an internal combustion engine (ICE), chemical energy is converted into kinetic
energy to turn the wheels. Most often such specialized gasoline or diesel engines are based on fuels like
petrol, bioethanol, compressed natural gas (CNG), liquefied petroleum gas (LPG) or diesel [55, 76]. But
-4-
Chapter 1 - Introduction
lately a trend of electrified powertrains is observable – either as main or as additional drive. An electric
vehicle (EV) can be classified based on its design and dependent energy source [76, 31]:
HEV (Hybrid Electric Vehicle):
A popular mixed form where both, an electric motor and internal combustion engine (ICE), are
supplementing each other to power the vehicle. Therefore both engines are connected to the wheels
via a torque coupler.
PHEV (Plug-In Hybrid Electric Vehicle):
An evolution to HEV, but this vehicle could be plugged in to the grid in order to charge the internal
batteries that supply the electric motor. Again, both engines are used to actuate the vehicle. Ac-
cording to stricter definitions, like those of the German Government, only such vehicles that could
be plugged in, are denoted as electric vehicles (EV).
REEV (Range Extended Electric Vehicle):
A vehicle that is only (directly) powered by an electric motor which obtains electricity from batter-
ies. In case the batteries get depleted to a certain threshold, a very small and lightweight ICE is
activated to generate power in order to recharge the batteries while driving. Compared to HEV or
PHEV, some parts like a (variable) transmission are not needed, as the additional engine combusts
permanently in the same optimal operating point. Due to the relatively light ICE, less automotive
parts and smaller batteries, the overall weight and initial costs of the vehicle are reduced. When
operating in pure battery-electric mode, the reduced weight of a REEV also results in a more
energy-efficient movement [81].
FCEV (Fuel Cell Electric Vehicle):
In principal similar to REEV, but instead of a little combustion engine a fuel cell is used to generate
electricity with compressed hydrogen stored in the tank. This electricity is then used to charge
the batteries and power the electric motor. Although hydrogen allows higher energy densities in
regard to weight, its production and storage is more complex which makes fuel cell electric vehicles
generally less energy-efficient than pure battery electric vehicles.
BEV (Battery Electric Vehicle):
The drivetrain of a full battery electric vehicle consists solely of an electric motor powered by large
batteries. In comparison to other presented EV designs there is no supplementary type of engine
and most often not even a (variable) transmission is necessary. BEVs are therefore also referred to
as all-electric vehicles. Batteries can be recharged primarily by plugging in to the grid, but charging
periods are typically much longer than traditional petrol fueling.
Regarding current developments and trends, one can further classify BEVs according to their total range
available on a single battery charge. One possible classification (without hard boundaries) might be:
Long-range BEV (e.g. 2014 ”Tesla Model S” with a range of 426 kilometers):
Offers several hundreds of kilometers with the goal to be on par with traditional ICE vehicles.
Limited-range BEV (e.g. 2014 ”Nissan Leaf” with a range of 135 kilometers):
Smaller batteries result in lower investment costs but also limited range. Nevertheless the range is
acceptable for many commuters and mundane short trips.
-5-
Chapter 1 - Introduction
In recent years more and more automobile manufacturers are jumping on board and introduce new EVs
onto the market. The US-based company Tesla for instance produces solely battery electric vehicles and
has somehow ”glamorized” them by eliminating common prejudices and drawbacks of EVs. Very often
they were seen as generally unaesthetic, impractical or slow. But associated with the marketing slogan
”Zero Emissions. Zero Compromises.” Tesla seemingly managed to change that image and make EVs
more attractive and appealing for a broader audience [70, 100]. On the other side of the world, the Chinese
company BYD (”build your dreams”) is focused on electrically powered vehicles in several conventional
fields like transit buses, coach buses, taxis, logistic vehicles, construction vehicles and consumer vehicles
as well. As world’s largest manufacturer of rechargeable batteries BYD has a tremendous advantage
when building automobiles with large (and usually costly) battery packs. The fact that they are the only
manufacturer that makes their own battery, helps them to become the largest EV manufacturer in the
world (as from 2016) [70].
From here on this work will focus primarily on all-electric vehicles with batteries (BEVs), but of course
their promising future is not guaranteed. Although the electrification of the drivetrain is advantageous,
there seems to be a lot of uncertainty about the best energy source of the future. Energy density is seen
as a crucial factor and thus (Lithium-Ion) batteries or even hydrogen might be problematic. According
to [90], methanol is seen as a very promising and yet environmentally friendly alternative. The proposed
methanol engine concepts of the MIT (Massachusetts Institute of Technology), for instance, reached peak
efficiencies of up to 60 % and might represent a valuable addition to electric drivetrains in range extended
electric vehicles (REEVs).
1.2.2 Costs
In contrast to traditional ICE vehicles, which have constantly evolved over several decades, the initial
investment costs of vehicles with alternative drivetrains are comparatively expensive. For BEVs the main
contributor of such high costs is the large battery mounted in the vehicle. Due to this dependence on the
battery price, more information is given in subchapter 1.2.5. According to forecasts [76] the acquisition
cost of BEVs will remain more expensive than comparable ICE vehicles in the year 2020. This price
differential will remain a major obstacle for many consumers, because to switch to an EV, they have to
believe that they get a widespread upgrade from their conventional petroleum-powered vehicle [12].
But of course the total cost of a vehicle is not defined through investment costs alone. There is also the
factor of operational costs, and this is actually an advantage BEVs have. It starts with lower maintenance
costs due to less mechanical parts inside the vehicle. BEVs need neither a fuel tank, exhaust system,
clutch and catalyst, nor a starter and alternator (dynamo). Usually, even the five/six speed gearbox is
replaced by a single speed [31]. Instead a BEV just needs batteries (with highest possible energy density),
an integrated charging system, and a battery management system with several electronic components to
regulate the battery’s temperature, the drivetrain, and recuperation. Even brake pads, which need to be
changed regularly on traditional ICE vehicles, won’t have to be changed for most EVs at all [70].
In addition to much lower maintenance costs, battery electric vehicles can also compete on the market
with lower fuel costs (driving costs) [13]. When comparing electricity to petroleum as direct energy
source, the most important factor is the price of crude oil, which seemed to stabilize at 100 $/bbl and then
-6-
Chapter 1 - Introduction
declined to 60 $/bbl in recent years [12]. Although many forecasters predict oil to remain at this price
level for several years, such price volatility makes it hard to compare driving costs for the longer term. A
rather conservative projection for the year 2020 estimates the total driving cost of gas-powered vehicles
to 9 e/100km and BEVs to 6.5e/100km in Central Europe [76]. At a more rapid progressive rate of
technology and falling electricity prices, BEV costs may even drop to 4 e/100km. But nevertheless, at
a cost-focused perspective over the full life cycle of an automobile, electric vehicles will probably not
catch up with ICE vehicles in the near future, mainly due to the higher investment costs. Even at very
optimistic predictions, BEVs will only be more profitable at a total driving distance starting from 100000
to 200000 kilometers [76].
However, to boost EV sales and make them more cost attractive for potential buyers, there is also the
possibility of lowering the barriers through governmental regulations and financial incentives, which is
also covered in subchapter 1.2.8.
1.2.3 Efficiency and Environmental Impacts
To analyze the environmental impact of EVs and compare them to ICE vehicles, a multitude of factors
has to be considered. Besides energy efficiency, which plays an important role, it is also absolutely
essential to consider the upstream chain and production of fuels/electricity. Because although a BEV
causes no pollution while driving, there might be a much larger environmental impact due to the electricity
generation to charge its batteries.
Thus a more comprehensive analysis must be a ”Well-to-Wheel” (WtW) inspection, which includes all
stages from fuel production and delivery until the actual use within a vehicle [80]. This consideration rang-
ing from well to wheel, depicted in Figure 1.2, consists of two consecutive stages: ”Well-to-Tank” (WtT),
which includes the provisioning and fuel production out of raw materials and resources, and ”Tank-to-
Wheel” (TtW), which includes everything after fueling/charging a vehicle. Usually, when manufacturers
advertise ”zero emission” vehicles, they are referring to the TtW stage because they have no influence on
emissions within the WtT stage, which strongly depends on regional conditions and political decisions
regarding electricity generation.
To analyze environmental impacts of EVs, it is also necessary to consider how efficient an energy source
is used to generated electricity and how efficient it is used to actually move the vehicle. Thus we want to
know how much of the energy is wasted along this path and which percentage of the original energy is fi-
nally converted into kinetic energy. Here a comparison between EVs and ICE vehicles reveals tremendous
differences too, which are best analyzed on a Well-to-Wheel (WtW) point of view:
Well-to-Tank (WtT) Efficiency:
For conventional gasoline vehicles the WtT efficiency is about 88 % [55], which includes feedstock
recovery, processing, fuel production, transportation, storage and distribution of gasoline.
In case of electricity generation for BEVs, the efficiency varies strongly depending on the energy
source. The average plant efficiencies for petroleum, coal and natural gas are ranging from 32 to
42 % [55]. But although the efficiency of renewable energy sources might be in the same range, it
is generally assumed to be 100 % for such comparisons, as the raw material is considered unlimited
-7-
Chapter 1 - Introduction
(e.g. sunlight is not ”wasted”). The overall efficiency of electricity generation depends thus on the
energy mix of multiple sources, and in 2012 this average efficiency was about 36 % in the United
States [55].
Unfortunately the electricity path into the EV batteries includes additional losses due to transmis-
sion (98 % efficiency), distribution (92 % efficiency) and finally the charging process (96 % effi-
ciency) [55, 30]. Independent from the energy generation, this sums up to an efficiency of about
87 %.
Thus, considering the average U.S. electricity mix of 2012, the overall WtT efficiency for BEVs is
about 30 to 32 %, which is much lower than the 88 % for gasoline vehicles [55].
Tank-to-Wheels (TtW) Efficiency:
For typical gasoline vehicles the TtW efficiency is about 14 to 18 %, which is mainly due to the low
operational efficiencies of gasoline-powered internal combustion engines (ICEs) with 28 to 30 %
on average [55, 31].
The electric drivetrain of BEVs on the other hand, achieves efficiencies of about 90 %, which
includes average gear efficiencies of 97 %, inverter efficiencies of 98 %, and motor efficiencies
of 95 % [46, 96, 31]. Because of the thermal management necessary for most battery packs, the
overall battery efficiency fluctuates between 85 to 96 % [96]. Including auxiliary loads (e.g. climate
control), this results in TtW efficiencies of BEVs ranging from 64 to 68 % [55].
Well-to-Wheels (WtW) Efficiency:
Observing the entire energy path on a WtW scale, the efficiencies of gasoline vehicles are about 12
to 16 %, while the efficiency of BEVs are believed to be around 19 to 22 % (with a 2012 U.S. energy
mix) [55]. Thus, BEVs are generally more energy-efficient than ICE vehicles, but the difference is
often smaller than one would expect when comparing just motor efficiencies. This is mainly due to
the fact that electricity for driving EVs is still generated via the detour of thermal (caloric) energy
in many power plants [76].
Well-to-Wheel (WtW)
production
processing
transportation
storage
distribution
storage
operation
charging
Well-to-Tank (WtT)
WELL WHEELTANK
Tank-to-Wheel (TtW)
Figure 1.2: The Well-to-Wheel (WtW) analysis along the energy path includes all stages from the fuel
production (or energy generation) to the conversion into kinetic energy within the vehicle.
-8-
Chapter 1 - Introduction
Of course, energy efficiency does not tell the whole story when comparing environmental impacts of
vehicles. Especially when observing the WtT stages, the utilization of renewable energy sources has addi-
tional advantages. As soon as the ”waste” of non-renewable energy sources is reduced when generating
electricity, the comparison of electric and ICE vehicle efficiencies will fade into the background.
Depending on the primary energy source and the conversion technology the environmental impacts of
power plants are varying tremendously (see table 1.1). While the average brown coal power plant emits
about 1000 g CO2 to produce 1 kWh of energy, renewable energy sources like solar photovoltaic are
emitting just 50 g CO2 (considering the full life cycle) [30]. Thus, in countries where coal represents the
main source of the grid energy mix, the overall environmental impact of ICE vehicles may be comparable
to that of EVs [29]. In the north and northeast areas of China, for instance, EVs could even contribute
to a worse carbon impact, as coal represents more than 90 % of their energy mix (as from 2008) [48].
But in Europe there are some interesting outliers too: Poland, for instance, uses also about 90 % coal
while France uses about 80 % nuclear power (as from 2011) [30]. In Austria the share of renewable
energy sources is about 33 %, and even better performs Portugal, which obtains half of its energy from
renewable sources like wind, solar and hydro (as from 2014) [49, 3]. In 2013, only 52 % of the average
European electricity mix were based on fossil energy sources [96], which clearly favors the use of EVs in
regard of environmental impacts.
Table 1.1: Environmental impacts of different power plant technologies (average accross Europe) [30].
Technology Abiotic Depletion Acidification Eutrophication Global Warming
(g Sb/kWh) (g SO2/kWh) (g PO4/kWh) (g CO2/kWh)
Coal 7.8 2.8 2.3 1020
Wind 0.08 0.05 0.027 11.3
Hydroelectric 0.03 0.16 0.05 6.5
Solar PV 0.36 0.246 0.157 50.9
Nuclear 0.04 0.047 0.015 6.05
Natural Gas 3.7 0.413 0.07 434
Diesel, Oil 5.9 19 0.57 911
Regarding tailpipe emissions (TtW emissions), the EU legislation constantly forces car manufacturers to
lower their carbon footprint. Although a threshold of 95g CO2/km is in discussion for 2020 [13], research
and development to achieve such refinements of ICE vehicles is costly and will take some time. Battery
electric vehicles (BEVs) on the other hand, do not have any direct tailpipe emissions at all, which makes
them extremely environmentally friendly in this regard.
Figure 1.3 finally compares the greenhouse gas emissions of several vehicle types on a Well-to-Wheel
(WtW) perspective. While the average ICE vehicle emits more than 100 g CO2 per driven kilometer,
BEVs charged with the average European electricity mix will emit just 50g CO2, which is less than
-9-
Chapter 1 - Introduction
storage
operation
g CO2/km0 30 60 90 120 150
Vehicle type
g CO2/km
WtT TtW WtW
BEV (German e-mix)
BEV (Austrian e-mix)
BEV (Sweden e-mix)
BEV (France e-mix)
BEV (EU e-mix)
BEV (USA e-mix)
FCEV
CNG vehicle
Petrol-ICE vehicle
Diesel-ICE vehicle
70
22
3
10
50
60
74
22
23
20
0
0
0
0
0
0
0
67
121
93
70
22
3
10
50
60
74
89
144
113
WtT emissions
TtW emissions
Figure 1.3: The Well-to-Wheel (WtW) emissions of BEVs and other vehicle types consist of different
Well-to-Tank (WtW) and Tank-to-Wheel (TtW) ratios. BEV emissions are largely dependent on electricity
generation mix (e-mix) that is used to charge batteries (data from 2014 [27, 73]).
half of it [31, 73]. When driving a BEV in Sweden, for instance, not only TtW emissions are reduced
to zero, but also total WtW emissions are barely existent. This emphasizes that the carbon footprints
of BEVs are directly depending on the electricity generation mix [31]. And as more and more power
plants are backing on renewable energy sources, emissions associated with electric vehicles will globally
decrease even further in the future [29, 31]. However, even in the worst-case scenario, in which BEVs
are charged with electricity of modern coal-based power plants, their total emissions will be lower than
those of traditional gasoline powered vehicles [18]. Also, it is important to underline the location-specific
occurrence of emissions, because in cities people are usually not exposed to plant-related emissions, but
indeed to tailpipe emissions of ICE vehicles [18].
Production and Recycling
In addition to the proposed WtW analysis there are also other stages in the life cycle of vehicles, which can
raise environmental impacts. This includes especially production and recycling, which could significantly
contribute to the environmental burden associated to BEVs [96]. Assuming a vehicle lifetime driving
distance of 150000 km, the production-related CO2 emission of BEVs contribute to about 62 g/km, while
that of ICE vehicles is considerably lower at 41 g/km (as from 2012). And according to forecasts, this
proportional gap will remain even for the next decades [96].
It has been reported also, that EV production may be constrained by resources including certain rare earth
elements [90]. Required critical resources for BEVs are copper (batteries and wiring), lithium (batteries),
cer (batteries), lanthan (batteries), neodym (batteries and permanent magnets), praseodym (batteries and
permanent magnets), samarium (permanent magnets), terbium, germanium, gallium, kobalt, and what
- 10 -
Chapter 1 - Introduction
is assumed as particularly problematic is dysprosium which is needed for permanent magnets of electro
motors [99, 100, 96]. Unfortunately the vast majority of such important metals is recovered from outside
of Europe, frequently also in conflict areas like Congo [99]. Problematic seems also the fact that many
high-tech products like photovoltaic and wind power plants compete with the electric mobility to the
same raw materials [99]. Nevertheless, it is also predicted that many of these constrained resources can
be substituted in the future [96] and of course there is potential for optimization of recycling and novel
concepts for second-life of batteries [99].
1.2.4 Motor
Despite tremendous technological progress, internal combustion engines (ICEs), which represent the com-
mon vehicle drivetrain today, still have relatively low efficiencies. A large portion of the energy stored
in petrol and diesel gets ”lost” as heat and cannot be converted into kinetic energy to move the vehicle.
Thus, the typical optimal efficiency is just 37 % for petrol-ICEs and 43 % for diesel-ICEs [100, 76]. In
electric motors, on the other hand, the electric energy is converted directly into kinetic energy, which
enables much higher efficiencies of about 95 % (excluding losses at batteries, inverter and transmission)
[76, 46, 96, 31]. Unfortunately the penalty of this large number is, that there is almost no heat loss that
could be used for passenger cabin heating, which is one of the positive side effects of ICEs. An impor-
tant challenge for all-electric vehicles is therefore to minimize the weight and power consumption of an
additional heating system in order to maximize the driving range [18].
It should be noted also that ICE efficiencies are strongly dependent on current load and rotational speed
(number of revolutions), which limits their actual real-world efficiencies even further [76]. This coher-
ence, however, is hardly given for electric motors, although their efficiency is slightly decreased at lower
speeds [96]. More information about average drivetrain efficiencies is presented in subchapter 1.2.3.
Unlike the progression of electronics and computer science, the evolution of motors has been rather long
and slow. Electric motors are anything but novel and they proved to be reliable in all possible areas for
more than a century [18]. In contrast to popular DC (direct current) motors, today’s AC (alternating
current) motors have become more efficient, more reliable, less expensive and almost maintenance free
([18] provides a good overview and classification of propulsion motors). Thus, the dominant types in
modern EVs are either permanent-magnet AC motors or AC induction motors [96, 31].
Although some improvements of electric motors are foreseeable, we cannot expect huge technological
leaps within the next decades [96]. But nonetheless, even an increased efficiency of just one percent can
lead to several additional miles of driving range, which is one of the current drawbacks of BEVs.
1.2.5 Batteries
The three major obstacles of EV adoption – range, charging time, and investment costs – substantially
depend on the batteries. Hence, it is fair to argue that battery technology represents the centerpiece of
progress in this area. Beneath the rather small battery for starting, lighting and ignition (SLI), which is
already a part of all automobiles, much larger and more powerful battery packs are needed for BEVs.
Today’s batteries incorporated in commercially-available BEVs are either nickel-metal hybrid (Ni-MH),
- 11 -
Chapter 1 - Introduction
lead acid (like SLI), sodium-nickel chloride (Na-NiCl or ”Zebra”), or particularly lithium-ion (Li-Ion)
types, which are the predominant choice [96, 13]. As there is not just one dimension an EV battery
has to perform well, the goal is to find a good compromise between high energy storage capability, low
weight, small volume, high efficiency, long lifetime, low price, and safety [96, 13]. But also ecological
aspects like production and recycling are important, as they are estimated to cause 15 % of the entire
environmental impact of electric mobility [31].
Within this chapter the most important characteristics of EV batteries are discussed. For many of these
considerations (like weight and price) it is important to differentiate between a battery assessment on the
”cell level” itself, and one on the ”pack level”, which includes packaging and electronics as well.
Energy Density and Weight
The specific energy density (or gravimetric energy density) of a battery chemistry defines the amount of
energy (Wh) stored in a certain mass (kg). Thus, it determines the necessary battery weight to achieve
a given BEV range [55]. Increasing a vehicle’s weight, however, influences energy consumption and
decreases the driving range, which narrows the profits of larger batteries [13].
Besides the gravimetric density, there is also the volumetric energy density, which defines a battery’s
energy (Wh) per required unit of space (l). With high volumetric energy densities the required battery
storage space is reduced (smaller tanks in case of fuel-based ICE vehicles), or in other words, more battery
capacity can be achieved in the same space [76].
As depicted in Figure 1.4, different battery chemistries can have different properties regarding gravimetric
and volumetric energy density. Lithium and especially lithium-ion (Li-Ion) batteries seem to be the
best choice in this regard, whereas ”lithium-ion” is also just an umbrella term for a variety of possible
material combinations [22, 13]. On the cell-level some state-of-the-art Li-Ion batteries reach volumetric
energy densities of 690 Wh/l and gravimetric energy densities of about 265 Wh/kg [4]. But on the pack-
level such densities are often reduced by 35 to 40 % [96] (corresponds to about 160 Wh/kg of the same
chemistry).
Although the theoretically achievable density maximum of currently applied Li-Ion batteries might be
twice as high [96], there is still a huge gap to traditional fuels. With an energy density of about 11 000
Wh/kg, liquid fuels like petrol and diesel can store up to 70-times the energy at the same weight. Adding
the fact that they can be stored and transported easily, it is evident that they have become extremely
popular in motorized transportation [76, 96]. Thus, to achieve BEV driving ranges coming close to those
of ICE vehicles, the applied batteries have to be much larger than a full gasoline tank. The 2014 Tesla
Model S, for instance, includes a 85 kWh Li-Ion battery (the largest passenger vehicle battery so far)
with a weight of about 680kg. To deliver the same energy to the wheels of an ICE vehicle would require
only about 24 kg (34l) of gasoline, considering the lower energy efficiency of combustion engines as well
[22, 13].
In order to achieve BEV ranges comparable to those of ICE vehicles with a similar weight and form
factor, batteries will have to provide energy densities of about 500 Wh/kg at the cell-level [55]. Due to
the combination of current research efforts, it is considered possible to achieve this within the next decade.
And in the longer term, other battery chemistries might significantly improve the situation [22].
- 12 -
Chapter 1 - Introduction
0 50 100 150 200 250 300
0
100
200
300
400
500
600
700
gravimetric energy density (Wh/kg)
lighter
smaller
volumetric energy density (Wh/l)
Pb-acid
Ni-Cd
Ni-MH
Na-NiCl
Ni-Zn
Li-Ion
Li-Polymer
Figure 1.4: Approximate gravimetric and volumetric energy densities of several available battery
chemistries. Lithium-ion (Li-Ion) batteries are most popular in today’s BEVs, as they need less weight
and space to store energy [22, 4]. Abbreviations: lead-acid (Pb-acid), nickel-cadmium (Ni-Cd),
nickel-zinc (Ni-Zn), nickel-metal hybrid (Ni-MH), sodium-nickel chloride (Na-NiCl).
Cost Factor
Due to advancements in technology and manufacturing, a steady trend of decreasing battery costs could
be observed. An analysis of more than 80 cost estimates of Li-Ion batteries revealed an annual decline of
approximately 14 % (between 2007 and 2014) [67].
For market-leading BEV manufacturers it is assumed that the price of battery packs has reached about
300 $/kWh (nearly twice the cell-level costs) [67, 12]. With continuing progress the threshold of
150 $/kWh will be achieved soon and BEVs will probably get mainstream-ready then [67]. For EV
makers this offers two options: lower the purchase price of the vehicle, or extend its driving range.
But beside technological improvements, the key for lower battery costs might be economy-of-scale ef-
fects [13]. While an increased production volume from 10000 to 100 000 batteries per year results in
cost reductions of about 40 %, an increase from 100 000 to 500 000 would allow an additional cost re-
duction of up to 30 % per battery [55]. This is also consistent with estimations of Tesla Motors Inc.,
a BEV manufacturing company that has recently cooperated with Panasonic in order to build an im-
pressive large-scale battery factory (50GWh of battery packs per year) to decrease overall vehicle costs
[https://www.teslamotors.com/blog/gigafactory]. Nevertheless, even if the forecasted battery price drops
by 50 % within the next 10 years (see table 1.2), the cost will remain a limiting factor of EV adoption
[22, 81].
- 13 -
Chapter 1 - Introduction
Table 1.2: Present and forecasted Li-Ion EV battery cost estimates. Technological progress and economy-
of-scale effects will allow a substantial price decrease of today’s most expansive BEV part [22].
cell price ($/kWh) pack price ($/kWh)
2013 275-325 400-500
2016 185-230 275-350
2020 140-190 225-275
Lifetime and Aging
Typically it is said that an EV battery’s primary lifetime comes to an end when its capacity has degraded
to 80 % of the original capacity. As this implies that the maximum driving range of BEVs is reduced at
the same proportion, it is particularly desirable to extend battery life and diminish aging effects. Hence,
some main factors that influence durability of batteries need to be considered (with focus on today’s
Li-Ion types):
Calendrical aging, which is the capacity fade proportional to the square root of time [30]. Most
vehicle batteries die because of calendar aging, and obviously also resting and parking periods are
included in that degradation over time [96, 13, 25].
Cycle lifetime, which describes how many charging and discharging cycles a battery can perform
until capacity fades to a certain threshold [96, 13]. This strongly depends on the Depth of Discharge
(DoD), the average State of Charge (SoC), overcharging/overdischarging, and the current rate [13,
30].
Temperature range, which needs to be considered for both, calendrical aging and cycle life of
vehicle batteries [96]. Often a thermal management system is used to operate the battery in an
optimal temperature range, which is important as vehicles are exposed to fluctuations of climate
conditions and outdoor temperatures. While deep temperatures might temporarily reduce the power
capability, high temperatures can substantially reduce the lifetime of state-of-the-art Li-Ion batteries
[13, 30].
Through special battery management systems (BMS) or charge management electronics, EV manufac-
turers often regulate charging power according to thermal conditions, restrict how fully a battery can be
charged, and how far its charge can be depleted during operation [30, 22]. In combination, these measures
help to slow the electrochemical degradation process, which leads to a reduction of battery capacity due
to an increase of internal resistance and self-discharge rate of the battery [30].
Figure 1.5 shows the typical aging behavior of EV batteries. Manufacturers often guarantee a certain
lifetime (or total driving range) until the battery capacity fades to 80 % of the initial capacity [83]. While
this threshold was reached after about five years for BEVs built in 2010, the lifetime could be increased
steadily to about 8 years for vehicles built in 2015. Despite the neglected operating temperatures in
this chart, the impact was shown to be tremendous: While a lifetime of ten years is possible at outdoor
- 14 -
Chapter 1 - Introduction
01234567891011 12 13 14
50
60
70
80
90
100
time (years)
relative battery capacity (%)
2010
guarantee
2015
2020
2025
Figure 1.5: Shows the average reduction of battery capacity (relative to initial capacity) over time. The
technological progress of four Li-Ion battery generations is indicated based on their year of production
(with forecasts for 2020 and 2025). The guarantee of battery manufacturers is typically based on the
time when battery life is considered to be at the end (under 80 % within vehicles) [83].
temperatures around 10 C (e.g. in Minneapolis, USA), it is reduced to 5 years in regions with 30 C (e.g.
Phoenix, USA) [22].
But also the typical driving behavior and Depth of Discharge (DoD) has a substantial impact, as the
battery life (until 80 % capacity) can be extended by a factor of three, if instead of 80 % DoD, a DoD of
20 % gets not exceeded [83]. Hence, reducing the energy consumption while driving with BEVs, which
is also the aim of this work, might help to extend battery life. After all, it is also the individual driving
profile that determines the number of battery cycles needed to travel a certain distance. On the long term,
an intensive use requires more cycles and on the short term, DoD might be higher each time.
Therefore, it can be concluded that the main contributor to battery degradation is cycle aging in case of
intensive use, and calendrical aging in case of light use [30].
Second Life
After 5 to 15 years, Li-Ion batteries have typically reached the end of their useful lifespan within EVs.
But due to their remaining available storage capacity, they can still be used for other applications after
that [60]. Within this “second life” the use of batteries can be maximized in order to reduce the overall
environmental impact. But also from an economic standpoint the re-sale value might reduce the overall
costs and make EV purchases more attractive [60, 83].
As indicated in Figure 1.6, after the first life (primary use within BEV) an additional second life from
about 80 % down to 40 % of the initial available capacity might be feasible [30, 83]. After that, the battery
is no longer suitable and a proper recycling should be conducted.
- 15 -
Chapter 1 - Introduction
time/cycles
end of
first life end of
second life
eol1
eol2
relative battery capacity
first life second life failure
Figure 1.6: Schematic progression of battery capacity over the full lifetime. After the first life of
batteries within EVs, the remaining storage capacity might still be sufficient for a second life with
different requirements [83].
Beside other application areas, one promising second life scenario is the use in combination with building-
installed photovoltaic (PV). Battery-based storage capacity might reduce the need for importing electricity
from the grid when PV output is temporarily insufficient (e.g. at night). This balancing of supply and
demand at a local level helps not only to reduce costs and import/export of grid electricity, but also allows
more independence and better prerequisites for fluctuating renewable energy sources [83, 60].
Finding the Best Compromise
Considering all of the mentioned properties allows a profound comparison of different types of batteries.
However, a decision on the ”best” battery chemistry for EVs is always a compromise that goes along
with trade-offs in certain disciplines. At the moment, most BEVs are relying on Li-Ion batteries as
they are most promising regarding electrified transportation [24]. Figure 1.7 illustrates and compares
the performance of some popular lithium-based battery chemistries on the basis of six properties. In
addition, table 1.3 shows the battery types used in some modern BEVs. Comparing the performance of
lithium-nickel-aluminum (NCA), which is used by Tesla, with lithium-iron-phosphate (LFP), which is
used in BYD vehicles, shows tremendous differences in many aspects. This means, for instance, that
BYD considers that advantages of LFP outweigh the relatively low gravimetric energy density (specific
energy) [12, 24, 70].
In addition to lithium-ion enhancements, lots of research is performed to find future battery technolo-
gies. Lithium-air batteries, for instance, may easily provide five times the energy density of lithium-ion
(theoretically even up to 5200 Wh/kg) [22, 29]. And lithium-manganese-iron-phosphate, which is in de-
velopment by BYD, may increase also the rate of charge, while at the same time production costs are
decreased [70]. Unfortunately, these battery chemistries still have to face some shortcomings and are not
market-ready yet. It is predicted, that adequate maturity will take several additional years of research, but
in the meantime, conventional battery chemistries can be used more efficiently due to optimizations of
the battery management software [29].
- 16 -
Chapter 1 - Introduction
Performance
Safety
Cost
Life span
Specific
power
Specific energy
Lithium-nickel-
manganese-cobalt (NMC)
Performance
Safety
Cost
Life span
Specific
power
Specific energy
Lithium-nickel-
cobalt-aluminum (NCA)
Performance
Safety
Cost
Life span
Specific
power
Specific energy
Lithium-manganese
spinel (LMO)
Performance
Safety
Cost
Life span
Specific
power
Specific energy
Lithium titanate
(NCA)
Performance
Safety
Cost
Life span
Specific
power
Specific energy
Lithium-iron
phosphate (LFP)
Figure 1.7: Popular lithium-based battery chemistries and their properties are illustrated by means of a
polar chart, in which more distance to center means better (which corresponds to ”cheaper” on the cost
scale).
Table 1.3: Battery types and characteristics of several modern EVs [22, 68, 15].
EV model cathode supplier energy (kWh) power (kW)
Tesla Model S NCA Panasonic 85 270
Chevrolet Volt LMO LG Chem 16.5 111
Nissan Leaf LMO Nissan/NEC 24 90
Honda Fit NMC Toshiba 20 92
BYD e6 LFP BYD 80 90
Renault Zoe LMO LG Chem 22 65
- 17 -
Chapter 1 - Introduction
It can be concluded, that by this time the electric vehicle industry has become one of the driving forces
of battery research (especially Li-Ion types) [24]. But even though Li-Ion batteries have been estab-
lished in consumer electronics all over the world, the requirements are considerably higher for electrified
transportation [81]. Only by increasing performance and concurrently decreasing costs of batteries, a
breakthrough in EV adoption will be enabled [81].
1.2.6 Charging and Infrastructure
Today, charging BEVs is perceived inconvenient and slow, when compared to traditional fueling of ICE ve-
hicles. Hence, for a better acceptance of EVs, it needs to be considered a fast charging time, a widespread
network of charging stations, valid charging standards with good compatibilities, impacts on the grid, and
also the effect on battery life.
Connectors, Plugs and Standards
In recent years, various charging concepts have been developed and deployed, with their main difference
being charging power and thus charging time. Plugging in at standard power outlets, for instance, is
referred to as ”standard charging”. It allows recharging EV batteries over the night or during the day
without installing a costly charging infrastructure [69, 13]. But with about 3.7 kW (230 V single phase
with max. 16 A in Germany), it is rather slow and takes a long time for a full recharge, especially with
long-range BEVs. Taking the 90kWh battery of a Tesla Model S with a range of about 500 km, as an
example, indicates this issue [88]: Due to losses (see chapter 1.2.3) about 100 kWh of energy are needed,
and with 3.7 kW charging power, a full charge would take about 27 hours. Furthermore, when relying on
traditional 120V-supplies (e.g. in USA), this might even take twice as long. Although overnight charging
is somehow comfortable and sufficient for many BEV owners to date [13], such low charging powers are
probably insufficient for increasingly large batteries of future vehicles.
On the other hand, what is known as ”fast-charging” requires much higher powers and therefore also a
special infrastructure. Most often a dedicated DC off-board charger is used for very powerful connec-
tions directly to the battery [13]. Typical fast charging powers range from 50 to 120 kW which ensures
tremendously shorter waiting times [69, 13, 3, 12]. In case of the 90 kWh battery of the previous example,
charging to 80 % capacity may take just 40 minutes, while a full charge will take about 75 minutes with
Tesla’s 120 kW ”Supercharger” stations [12, 3, 87]. Beside this Tesla standard, the most prominent high
power charger connectors are CHAdeMO, with up to 62 kW, and SAE-Combo, with up to 80 kW [12].
Unfortunately, the various incompatible plugs and communication protocols represent a barrier to the
adoption of BEVs [22], and drivers have to rely on a public (or semipublic) charging infrastructure as it
would be cost inefficient to supply each vehicle with its own fast-charging station [13].
In contrast to standard charging with energy efficiencies of about 95 to 97 %, fast-charging achieves
typically just 91 to 94 % [31] and needs more sophisticated charging systems. Such intelligent (off-
board) chargers have to exchange data with the vehicle in order to continually adjust the charging rate to
match a battery’s ability to accept charge [18, 3]. As higher charging rates can also reduce battery life,
such systems have to find the right balance of performance and impact on battery durability [13, 18, 74].
- 18 -
Chapter 1 - Introduction
A constant monitoring and regulation of the thermal battery conditions is also very important because the
loads during charging can be much higher than during driving operations [13, 3].
Infrastructure Impacts
To assess the opportunities and limitations of BEV charging it should not be forgotten to evaluate impacts
on the infrastructure and electricity grid as well. One drawback of fast-charging is, for instance, that it
causes high peak power demand and puts greater stress on the distribution system [18, 55]. Similarly also
the timing of standard charging can be problematic because the hourly peak demand of electric power is
typically the same time when many people come home from work and may plug in their EV (see Figure
1.8) [55].
On the other hand, this fluctuation of supply and demand might offer a great opportunity for BEVs. If
timing of battery charging can be controlled intelligently, the vehicle can be loaded in periods with low
overall energy demand and therefore also lower energy costs (e.g. night-time hours) [55]. Additionally
such ”smart charging” techniques could help to flatten the spikes and valleys of electricity generation
itself [96]. Especially in combination with the fluctuating output of renewable energy sources like wind
and solar, EV batteries may represent the perfect energy storage in peak times [14, 55]. Utilizing a large
amount of batteries correctly may thus help to lower plant-related emissions and reduce overall energy
costs [55]. Furthermore, the energy consumption of BEVs is relatively low on average (e.g. in Germany
one million EVs would account for just 0.5 % of the total energy consumption) [100].
The ”vehicle to grid” (V2G) concept goes even one step beyond and considers BEVs as active elements
in the electricity grid. Via coordinated bi-directional power flow, the vehicle battery may provide supply-
related functions for more reliability and efficient regulations (e.g. as instantaneous power input source)
[55]. However, due to technical, economical, and operational challenges, V2G is not expected to become
established soon [96].
Future Technologies
Driven by the increasing interest in electrified transportation, also research around battery charging has
gained traction in recent years. One idea was, for instance, to establish battery swap stations which are
able to exchange discharged EV batteries fully automatically in just a few minutes. But until now the
large investment costs of stations and the lack of standardized battery packs shifted interest more towards
better charging techniques. With a 250 kW DC charger a company demonstrated in 2007 to fast-charge a
35 kWh battery within ten minutes [12]. And a few years later, in 2011, several EV buses were put into
operation for a fixed-route public transit service in California. With a 500 kW DC charger it took just ten
minutes to fully charge the 83 kWh batteries while also considering their durability [12].
However, it will remain difficult to foresee whether BEV charging times of 10 minutes are sufficient to
satisfy regular ICE vehicle drivers who are used to fueling times of three minutes to travel 300 miles
[12]. In comparison to traditional gas stations, also the network of EV fast-charging stations has to be
broadened and compatible charging standards/connectors are desirable. Finally, a decision to purchase an
all-electric vehicle might strongly depend on the possibility to conveniently charge the vehicle at home
without depending on public infrastructures [100, 96, 22].
- 19 -
Chapter 1 - Introduction
0 2 4 6 8 10 12 14 16 18 20 22 24
0
10
20
30
daytime (hour)
electric power demand (gigawatts)
hourly peak demand
Figure 1.8: Average combined electric power demand over one day (based on data from New England,
2010) [55].
1.2.7 Usage Analysis
To estimate the real-world potentials and benefits of BEVs, it is important to analyze the typical driving
behavior and usage characteristics of motorized vehicles. It was shown that cars are used for very short
trips in most of the times. A study of Austria, for instance, showed that more than 50 % of all journeys lie
within a distance of five kilometers and 30 % are even below two kilometers [3]. When summarizing all
distances per day, it was estimated in Germany that 80 % of all vehicles are driven less than 50 kilometers
on an average day [99]. Further it was revealed that on weekdays about 31 % of all vehicles are used to
get to work, whereby less than half of them commute more than 40 kilometers [99].
In contrast to this average or routine use, ”special journeys” like weekend trips, longer business trips, or
holiday journeys need more available vehicle range. In the USA, for instance, just about 1 % of all trips
are longer than 160 kilometers, but unfortunately, for many consumers these long-distance trips weigh
heavily in their vehicle purchase decision [22]. Concerning the limited range of most of today’s BEVs,
that would mean that they could not be used for six longer journeys per year (on average in Germany)
[99].
Especially for BEVs it is also crucial to observe how long the vehicle is standing still, because these idle
times could be used to recharge the batteries. Figure 1.9 clearly depicts that most of the time the car is
actually not moved but rather is at rest. The aggregated shares of driving and resting (parking) during
weekdays and weekends is also shown in Figure 1.10. On an average day more than 97 % of the time
(between 23 and 24 hours) the vehicle is resting either at work, at home (residence), or at other places
(e.g. shopping centers, recreation/holiday areas, churches, etc.) [96, 22, 99]. Thus, the average utilization
of cars is actually extremely poor and inefficient, which is presumably another reason why car sharing or
ride sharing concepts are becoming more popular.
- 20 -
Chapter 1 - Introduction
0 %
10 %
20 %
30 %
40 %
50 %
60 %
70 %
80 %
90 %
100 %
parking at other placesdriving parking at work parking at home
Sunday 04:00
Sunday 08:00
Sunday 12:00
Sunday 16:00
Sunday 20:00
Monday 00:00
Monday 04:00
Monday 08:00
Monday 12:00
Monday 16:00
Monday 20:00
Tuesday 00:00
Tuesday 04:00
Tuesday 08:00
Tuesday 12:00
Tuesday 16:00
Tuesday 20:00
Wednesday 00:00
Wednesday 04:00
Wednesday 08:00
Wednesday 12:00
Wednesday 16:00
Wednesday 20:00
Thursday 00:00
Thursday 04:00
Thursday 08:00
Thursday 12:00
Thursday 16:00
Thursday 20:00
Friday 00:00
Friday 04:00
Friday 08:00
Friday 12:00
Friday 16:00
Friday 20:00
Saturday 00:00
Saturday 04:00
Saturday 08:00
Saturday 12:00
Saturday 16:00
Saturday 20:00
Sunday 00:00
Sunday 04:00
Figure 1.9: Typical location/activity of vehicles throughout the week (data from 2001, USA [22]).
Although the long parking periods indicate a bad utilization, they represent an interesting opportunity to
recharge EVs, especially at night times.
parking at other placesdriving parking at work parking at home
weekdays saturdays sundays
Figure 1.10: Aggregated average vehicle location/activity on weekdays or weekends (data from 2009,
Germany [99]). Actually, vehicles are driven only for a very small percentage of the day, with weekends
being no exception (red slice).
- 21 -
Chapter 1 - Introduction
It can be concluded that despite limited range and long charging periods, BEVs could already replace
ICE vehicles for most people’s everyday life [99]. Nevertheless, for many people the car represents some
kind of ”freedom” and ”flexibility to do what they want”, which contradicts the more sophisticated long-
distance trip planning with BEVs [100]. Hence, a German survey shows that every second respondent
would only purchase an all-electric EV as additional/secondary car, when there is an ICE vehicle already
available within the household [100].
1.2.8 Incentives
To accelerate EV adoption, there is more than just technological progress and better vehicle specifications.
Additional incentives could be provided by governments and administrative entities to make BEVs more
attractive for consumers.
Financial incentives
It could be demonstrated worldwide that substantial financial incentives are effective at motivating people
to purchase BEVs [22]. There are different incentive types that could be enacted at different stages [22]:
Purchase incentives, where the buyer is granted one-time financial benefits when purchasing (e.g.
tax credits, tax exemptions, rebates)
Ownership incentives, that are granted periodically on a recurring basis regardless of use (e.g. ex-
emption from environmental taxes, exemption from vehicle inspection)
Use incentives, which are provided on an on-going basis when driving (e.g. exemptions from motor
fuel taxes, reduced roadway tolls, charging discounts, parking discounts)
Infrastructure incentives, that are granted one-time when a BEV charging station is deployed (e.g.
rebate for installing a charging station)
Non-financial incentives
Contrary to financial benefits, non-financial incentives are effective as well. Such on-going special privi-
leges could be granted only to BEV drivers. Other drivers may not have the chance to enjoy these benefits,
which makes this option also worth considering for wealthier people who care less about saving costs.
Non-financial incentives might include reserved parking zones for BEVs, access to restricted lanes (e.g.
bus-only, high-occupancy-vehicle, or high-occupancy-toll lanes), extended limits for short-stay parking,
or even special inner-city zones that can be entered with zero-emission vehicles like BEVs only [100, 22].
1.2.9 Trends and Forecasts
In many countries, like Germany, it could be observed that the total direct and indirect CO2 emissions of
transportation have more than doubled since 1960, and the vast majority of it comes from road transport
[100]. Of course, modern vehicles tend to be more energy-efficient nowadays than 10 or 15 years ago,
but these optimizations are compensated largely due to the increasing number of vehicles and the rising
- 22 -
Chapter 1 - Introduction
needs of transportation [100]. And although significant improvements of internal combustion engines
(ICEs) can be registered again and again, a physical and technological limit is approaching [76].
Clearly, we have to take actions to improve the situation, but the question of where to take the energy for
transportation while keeping an eye on environmental impacts remains. Many scientists and researchers
agree that solar energy is the answer because of its long-term viability, scalability and sustainability [90,
3]. In contrast to photovoltaics (semiconductor solar cells), which relies on several rare earth materials
today, it has been argued that other concepts like solar thermal plants are the better solution to generate
electricity [90].
Switching to renewable energy sources like solar and wind might also help to free the dependence on fossil
fuels, that are not available everywhere (geographically limited). Hence, relying on locally produced
(generated) electricity is very beneficial from a national economic point of view [96]. In contrast to
fossil fuels, electrified transportation also allows to mix different sources of energy, and thus reduces the
dependence on specific energy sources, whether renewable or non-renewable [96]. When directly using
constrained non-renewable energy sources like crude oil, natural gas or coal, it is often forgotten that we
also rely on them for embodying many crucial physical products [3].
Electric energy is thus a good choice that perfectly matches the functionality of BEVs, which are very
efficient at conversions into kinetic energy. But when it comes to BEV energy consumption there is still
potential for optimizations regarding individual driving styles. Hence, it is also a core objective of this
work to better utilize the energy we have and achieve better results with existing technology.
On a longer run we will presumably see many optimizations and evolutions of EVs (and especially BEVs)
until they outperform traditional ICE vehicles in every category. Some of them might concern a reduction
of rare earth elements, higher energy densities of batteries, better charging solutions, more comfort (with
better accessories and equipment), better scalability, and of course lower initial costs. But already today,
BEVs are scoring high in disciplines like low noise and vibration levels, low (zero) tailpipe emissions,
less vehicle components, less energy consumption, usage of local energy sources, reduced dependence
on specific energy sources, low operating costs, and good traffic safety [100, 90, 96, 31].
It should be noted also that many novel aspects and inventions of BEV design and development can be
used to optimize traditional vehicles as well. These include lightweight construction, efficient climate
control systems, thermal management, thermal insulation, better aerodynamic, less rolling resistance
(low-resistance tires), and many more [14, 81]. Thus it is likely that achievements of BEV technology
are helping indirectly to make ICE vehicles more efficient, less expensive, and more environmentally
friendly.
It will also be interesting to consider the rapid growth and technological progress of autonomous driving
systems. They might be able to interconnect with the infrastructure or other road users/vehicles and
operate far more energy-efficient than human beings. Possibly, we might also see that people will begin
to place less emphasis on possessing an own car, but rather get on into arbitrary self-driving vehicles and
enjoy the same flexibilities. This would allow for a better utilization of vehicles and reduce the overall
amount of idle times, which leads to less vehicles and less environmental impacts. Additionally, fleets of
autonomous BEVs might reduce the problems of poor charging infrastructures and long charging times,
as it is not necessary to drive the same vehicle all the time. If the battery gets empty, one might just get
picked up by the next vehicle without taking care of inconvenient charging procedures.
- 23 -
Chapter 1 - Introduction
Recently it gets indicated that predictions about falling battery costs were even too pessimistic [67]. So
finally, we might see BEVs becoming competitive and popular far sooner than expected.
1.3 Behavior Change and Gamification
One fundamental concept to optimize individual driving styles is behavior change. Motivating human
beings to change behavior and supporting their engagement belongs to trending research topics around
gamification, serious games, pervasive games, playful design and more. Therefore it represents an area
that belongs to both, the field of psychology and the field of technology.
Using pure information as an instrument to facilitate behavior change has shown only marginal effects,
especially in the pro-environmental context [20]. It is thus necessary to target more sophisticated methods
to attract attention and increase the willingness for transformation. In behavioral psychology there was
one core concept found to be very effective, namely feedback [37]. If feedback is used properly by
individually directing information to the individual person at the time of decision making (live), it creates
more connectedness and serves as powerful behavioral variable.
”Gamification” or ”gameful design” is the rather new research discipline about such principles for moti-
vational affordance. It is defined as the usage of game elements and game design techniques in non-game
contexts [23]. As games represent one of the most powerful forces for humans in all ages, it is important
to examine its mechanics and apply some aspects in more serious environments to solve problems and
engage audiences [98]. Understanding the concepts and origins of motivation is therefore essential, as it
ultimately drives the outcome. Motivation is not only a unitary phenomenon reflected in a certain level
or degree (how much motivation), but also in its orientation (type of motivation). The most basic cate-
gorization is the distinction between intrinsic and extrinsic motivation [77]. Intrinsic motivation refers
to doing something because you want it by yourself as it is interesting or enjoyable, whereas extrinsic
motivation is triggered from the outside and refers to doing something because you are moved to. In
the contexts of gameful design and education, intrinsic motivation plays therefore an important role, as it
results in high-quality learning with longer-lasting engagement. Punishment, on the other hand, contrasts
and undermines intrinsic motivations, which makes it a particularly bad choice for voluntary applications.
The goal should rather be to keep people (players) happy and let them make their own choices, feel the
challenge and have freedom and responsibility at the same time [98].
Gamification is usually based on the following five principles: goal orientation, feedback and reinforce-
ment, achievement, fun, and competition [65]. Due to the social component of competition, it was shown
to be a very powerful factor to engage individuals or groups [37]. It seems like human beings naturally
strive to be better than others, which we also considered strongly while designing our gamification-based
dashboard application. But not only the competition with others motivates humans, also the comparison
of one’s current performance to past performances (self-comparison) is an effective strategy for behav-
ioral change [37]. One important part of using competition as a motivational affordance is to take care of
fairness and comparability. If one is getting the feeling of having no chance to win (lead) or being com-
pared in an unfair manner, motivation will drop and he/she will lose the fun to participate. Implementing
gameful design elements like ”leaderboards”, ”badges” or ”points” is hence not automatically positive
and should be applied well-conceived.
- 24 -
Chapter 1 - Introduction
However, there exist numerous studies that prove the positive impacts of implementing such motivational
affordances in non-gaming contexts. They represent effective instruments to achieve targeted psycholog-
ical and behavioral outcomes [44]. In game theory we can categorize these different learning outcomes
as the following [94]: affective (attitude, motivation, preference), communicative (communicate, cooper-
ate, negotiate), cognitive (knowledge, problem solving, decision making, situation awareness) and motor
skills (acquisition, compilation). Behavior change in the automobile environment of electric vehicles will
probably relate to cognitive and skill-based learning outcomes for the most part.
In this scope of games and gamified applications we also have to highlight some related academic terms.
”Serious games”, for instance, are based on the very similar principle of using games to educate, moti-
vate and change behavior. The broad definition for such games includes therefor an arbitrary form of
interactive computer-based software for any platform [75]. Compared to gamification or gameful design,
serious games are distinct in the characteristic of being more like a full-fledged game [23]. They do not
only incorporate some parts/elements of games, but represent a game in its entirety. Serious games are
often also referred to as ”simulation games”, which find prominent use in flight simulators, emergency
management or medical simulations.
”Pervasive games” on the other side are computer games that blend a virtual/fictive world with the real
world [52]. Either they use physical spaces in the game world and/or they use layers (with virtual objects)
to enhance the physical world (e.g. augmented reality games). Typically such pervasive games are cre-
ated with the pure purpose of fun, and not with a (serious) learning outcome in mind.
Contrary to ”gameful design” also the term ”playful design” needs to be distinguished based on the two
concepts of game and play. Whereas playing refers to a more free-form and improvisational activity, and
gaming denotes a more structured (rule-based) and competitive form with pre-defined goals.
As the borders of these terms and disciplines are overlapping, it is not always trivial to distinctively
classify different approaches in this context. Nevertheless, the potentials of technologies designed for the
purpose of behavior change are huge. They can be unobtrusively integrated in our daily life to engage
ourselves in a joyful but subtle way and influence our behavior.
1.4 Increasing Efficiency and Range
Referring to the motivational introduction (see chapter 1.1), BEVs might represent a very promising
alternative towards a more sustainable way of transportation. And in order to further optimize energy
efficiency and reduce overall emissions, the most overlooked action might simply be the alteration of
current driving styles [8]. Moderate accelerations, avoidance of sudden starts and stops, constant driving
speeds (with cruise control where appropriate), and anticipatory driving may significantly reduce energy
consumption [8]. But it is not just the individual driving style that is essential, also specific knowledge
like maintaining optimum tire pressure is desirable.
In contrast to technical BEV improvements on the long term, behavior change could improve climate
impact immediately across the entire automobile fleet [8]. Operating a BEV more energy-efficient has
also positive side effects like safety benefits (fewer accidents and traffic fatalities), and a longer lifespan
of expensive batteries [8, 83]. Most importantly the lower energy consumption leads to less battery
discharge and, in consequence, more remaining range for the driver. This in turn decreases range anxiety
- 25 -
Chapter 1 - Introduction
and might diminish barriers of BEV adoption [35]. To make this happen, it is also essential to improve
the reliability of current range displays and help drivers better understand range-limiting factors of EVs
[85].
It is thus assumed that a better estimation of remaining range and a better feedback on driving economy
could be even more important than increasing range with larger batteries, for instance [35]. Hence, our
intention was to analyze and assess this aspect with an additional dashboard that was mounted in an all-
electric vehicle. Firstly, its purpose was to improve a driver’s mental model of the inner workings of
EVs [85]. The system provides auditory tips to utilize BEVs more efficiently and helps to comprehend
important relationships of range-limiting factors. Secondly, it should provide live feedback on the current
energy consumption and motivate drivers to save energy. Feedback is seen as an important instrument to
alter driving habits and maintain behaviors over time [8]. The driver can directly see the results of his/her
actions and might recognize important strategies and opportunities to extend the remaining range (either
in extreme situations or, even better, on a regular basis). In order to provide additional motivation, we
incorporated a gamification-based approach of ranking against other drivers and other trips. As it was em-
pirically proven that humans respond positively to comparison with others [79], this competitive aspect
was used as a subtle instrument to motivate people to maximize energy savings (and remaining range).
The key of our novel dashboard application was to provide ”fair” and more engaging comparisons by dif-
ferentiating between varying environmental conditions (e.g. outdoor temperature) and by just comparing
trips with the same topography and the same route (or at least parts of it).
To record certain vehicle parameters and analyze the results and improvements, a real-world user study
was conducted within a car-sharing community. The dashboard in form of a smartphone application was
mounted in a Renault Zoe (2013) which was regularly driven by several participants. In a reference period
from 2013 to 2016, data was recorded and transferred to databases and servers via Internet. Later the data
was compared to measurements within a four-month test period using our in-car dashboard application.
Finally, also qualitative evaluations in the form of individual interviews should provide insights on our
approach’s opportunity to positively influence a driver’s behavior.
1.5 Outline of this Thesis
This work is structured as follows: The first chapter covers an introductory part which starts highlighting
current EV problems and motivations to solve those (1.1). After a more detailed summary of state-of-
the-art BEV technologies and future perspectives (1.2), as well as an overview of gamification-based
approaches for behavior change (1.3), the general purpose and approach of this work is introduced (1.4).
In chapter 2 related work is presented around the topics of current EV range problems (2.1), factors that
may limit remaining EV range (2.2), measurements and analysis of EV driving data (2.3), and behavior
change techniques for more energy-efficient driving (2.4). The necessary steps and implementations to
record and process real-world BEV driving data within our work are described in chapter 3. This includes
the collection of measurements (3.1), preprocessing of those tracks (3.2), an algorithm to convert a chain
of rough coordinates into fine-grained routes (3.3), and the addition of precise weather, elevation and tire
data (3.4). Subchapter 3.5 provides details of an interactive tool that was developed to analyze and com-
pare the processed EV driving data. In chapter 4 our approach for a novel in-car dashboard application is
- 26 -
Chapter 1 - Introduction
covered with detailed information on hardware (4.1) and software implementation (4.2), as well as con-
siderations about meaningful rankings/comparisons of tracks (4.3) and tips for efficient driving (4.4). The
results and evaluations are then presented in chapter 5, starting with some basic statistics of the recorded
EV data (5.1). After reasoning about coherences and opportunities of improvement regarding energy con-
sumption (5.2), the conducted user study with quantitative and qualitative evaluations is presented (5.3).
Chapter 6 finally covers a conclusion and summary of the work as well as some thoughts on possible
future work.
- 27 -
Chapter 2 - Related Work
2 RELATED WORK
To lay the groundwork, this chapter includes findings and discussions of other research that is related
to the topic and may be relevant to understand successive chapters of this work. It starts with the con-
sideration of the core problems surrounding EVs and their consequences for drivers or potential buyers
(chapter 2.1). For proceeding investigations it is also important to understand the basic way EVs are
functioning and to explore factors that can have an impact on energy consumption (and remaining range)
of EVs (chapter 2.2). Further, chapter 2.3 will cover related work about the recording and analysis of EV
trips in order to reason about coherences of different range-limiting factors. As one factor to reduce EV
energy consumption is the driver itself, research about behavior change is then summarized in chapter 2.4.
Finally, the conclusions of related work, as well as the differentiation and novel characteristic of our work,
is covered in chapter 2.5.
2.1 EV Range Problems
Based on a survey with several focus groups in Germany, it was concluded that the main problems with
EVs are long charging periods, poor availability of charging stations, and above all the short driving
range [100]. Because of several technical limitations (see chapter 1.2), like the energy density of battery
chemistries, the range of a modern battery electric vehicle (BEV) is typically much lower than that of tra-
ditional gasoline-powered vehicles with internal combustion engines (ICE). Another drawback is battery
aging (degradation of battery capacity), which further decreases the maximum driving range over time.
According to literature in this area, the most influential factors for this degradation are environmental
temperature, discharging current, charging rate, depth of discharge, and the time intervals between full
charge cycles [74].
Thus, range anxiety (the fear of running out of charge) seems to be a fundamental issue concerning
the limited range of BEVs. But it is not only the remaining state of charge (SoC) by itself that may
cause increased range anxiety of drivers, also the charging process has drawbacks. Participants of a 12-
month field study with 50 BEVs (Mini E) in Berlin concluded that charging at public charging stations is
inconvenient and susceptible to errors [14]. Because of this limited charging infrastructure it seems that
range anxiety is perceived almost exclusively by BEV drivers [10]. Researchers illustrated that driving
a BEV could be similar to driving an ICE vehicle with an almost empty fuel tank, with the additional
problem that recharging may take hours [57].
Although the fully charged range of BEVs might be just about 100 kilometers, it could be observed that
this should easily meet most people’s daily travel needs (see chapter 1.2.7). Thus, it was concluded that
the experience of range as a barrier is perhaps mainly a psychological issue [35]. By keeping certain
range safety buffers, it is assumed that each driver develops an individual range comfort zone [36]. And
regarding inaccurate range displays due to varying energy consumptions, this overcautiousness of drivers
might be justified very well.
So, besides the limited battery capacity and bad charging infrastructure, range anxiety is also provoked by
unreliable range estimations [35, 92]. Thus, to achieve a better user experience a trustworthy distance-to-
empty indicator is essential, because it assists the driver to decide which trips can be accomplished without
- 28 -
Chapter 2 - Related Work
experiencing range anxiety [36]. According to researchers, a better presentation of the remaining driving
range might be even more important than extending the range itself [35]. In 2011, at Warwick University,
a real-world study with 11 novice EV drivers highlighted that range predictions were on average 50 %
wrong (too high) [10]. It is therefore of little surprise, that interviews with several EV drivers in Sweden
showed that this circumstance is frustrating and that many of them had figured out own ways of relating
to the driving range (e.g. rules of thumbs) [57]. Comparing several range estimation techniques made
clear that many commercial BEVs are just using the average power (energy consumption) and current
SoC to predict the remaining driving range [19]. Current range displays take this incomplete information
and present it as trustworthy to the driver [85], but there are other factors that have tremendous influence
on energy consumption and some of them are particularly important for EVs. Thus, applying advanced
modeling and estimation techniques is considered a viable solution to make range predictions more practi-
cable [33]. In [92] several interesting dashboard concepts and range displays of EVs were compared and
in order to decrease range anxiety and educate drivers, some of them also consider additional influencing
factors. Another concept was presented in [19], where a model for a range estimator is proposed. But
unless the desired journey (particular route) is not specified in advance, it is not possible to forecast the
remaining driving range as a variety of range-limiting factors are ignored. Similarly, another approach
was designed to request the desired route of the driver [58]. But instead of providing remaining range, the
display just shows the driver the maximum average consumption that is necessary to achieve the desired
distance. As certain range-limiting factors are not considered by design, the output should be taken with
caution (e.g. what if the last route segments require more energy due to uphill driving?).
As a user study with 40 BEVs in Germany concluded [35], range resources seem to be dependent on
factors that could be predicted neither by the vehicle, nor by the driver. To overcome such BEV issues
regarding inaccurate estimations and to educate drivers about strategies to decrease energy consumption,
it is important to analyze range-limiting factors in more detail (see next chapter).
2.2 Range-Limiting Factors of EVs
Manufacturers usually provide range estimations for their fully charged EVs based on drive cycles that are
used for ICE vehicles too. Such modal cycles or transient cycles like the NEDC (New European Driving
Cycle) try to quantify typical driving scenarios in a standardized way [31]. However, it seems that BEV
dynamics are different and possibly more complex than those of ICE vehicles. Certain dependencies that
play an important role in the total consumption of EVs and BEVs are not considered. In this chapter
we provide an overview of the main factors that may influence driving range of BEVs, either direct or
indirect. They can be categorized into types that are described within the following subchapters.
2.2.1 External and Environmental Factors
External and environmental factors include those that are not (directly) dependent on vehicle specifics
and cannot be actively controlled by the driver. As vehicles are typically driven outdoor (open air) with
other road users, they are exposed to a multitude of external factors that could potentially limit EV range.
- 29 -
Chapter 2 - Related Work
Outdoor Temperature
Outdoor temperature can affect the cell temperature within the vehicle’s battery packs, which strongly
influences the discharge voltage and hence the available discharge capacity [97]. Via cycle tests with
three different BEV models, a research study of 2012 [62] found out that range is limited by 20 % when
driving at outdoor temperatures of 7C instead of 20 C (without including heating loads). With cold
ambient temperatures of 20 C, the battery capacity alone is said to be reduced by 9 % compared to
20 C. Thus most BEVs include systems for battery cooling or heating to avoid extreme temperatures.
Unfortunately, with such thermo-management systems the overall battery efficiency varies between 85 %
and 96 %, and as they consume energy, range may decrease further [62, 96].
In addition, there is evidence that battery life is reduced when charged/discharged at lower as well as
higher temperatures. Figure 2.1 highlights the extent of this dependence for Li-Ion batteries, which are
widely used in modern BEVs [74]. It is important to consider battery life as a gradually process here,
because it affects the maximal possible energy capacity (when fully charged) and thus limits the driving
range durably.
-40 -30 -20 -10 010 20 30 40 50 60 70 80 90
0
250
500
750
1000
1250
temperature (°C)
charge rate
temperature range
with maximum
cycle life
cycle life (number of cycles)
1 C
2 C
3 C
Figure 2.1: The life of Li-Ion batteries (cycle life) strongly depends on the operating temperature. Aging
is also influenced by the rate of charge/discharge, which is depicted for 1C, 2C, and 3C [74].
According to a study conducted by the AAA Automotive Research Center [1], it can be concluded that
BEV driving range can be nearly 60 % lower in extreme cold and 33 % lower in extreme hot ambient
temperatures. We should not forget, however, that there are many factors that depend indirectly on the
outdoor temperature like the energy needed for cabin heating, different rolling resistance on hot or cold
pavements [9], higher tire pressure due to warmer air temperatures, and probably even more hidden or
undiscovered dependencies.
- 30 -
Chapter 2 - Related Work
Atmospheric Conditions (Precipitation)
Conditions like precipitation (rain), snow and ice can have an impact on the energy consumption because
of several coherences. The increased humidity of the air, for instance, might slightly decrease the density
of air which affects drag. Also the electrical load might increase by using lights, wipers or defrosters. But
beside this rather small impacts on energy efficiency, it is the difference in rolling resistance that counts
most. According to simulations [72], even with little rain the energy consumption might increase by up
to 15 % when driving about 100 km/h on wet roads compared to dry roads. This is due to the continuous
drag when tires divide/part the water on the road and due to the energy needed to elevate water into the
air for nearly 2 meters (spray behind).
Wind
Depending on strength (speed) and direction of the wind, a vehicle needs more or less energy to move.
While headwinds might require more force to accelerate a vehicle, tailwinds might actually help to im-
prove energy efficiency.
In principle, this does not depend on the type of vehicle (ICE or electric vehicle) but more on its aero-
dynamic shape, which is described later when talking about vehicle-dependent factors. Anyhow, we can
assume that the sum of vehicle speed and wind speed (in opposite direction) define the energy demand,
and since aerodynamic drag increases with the square of velocity [39] (see driving speed impacts), wind
should not be underestimated as a range-limiting factor.
Road Type and Road Condition
As already indicated, the surface of roads is linked to a vehicle’s rolling resistance, which of course
depends also on certain tire properties. Such surface characteristics and pavement textures are depending
on the type of material as well as the varying conditions over time.
To evaluate the influence of various surface textures on energy consumption, the Swedish Road and
Traffic Research Institute performed tests with passenger cars at constant speeds between 50 to 70km/h
[9]. They found out that in contrast to even roads, the efficiency declined by up to 12 % on uneven roads
(roughness with wavelengths longer than 500 mm). And roads with rough macro-textures (wavelengths
between 0.5 to 50 mm) may decrease the range by 7 % relative to a very smooth macro-texture. On
average it can be stated that the impact of different road surfaces on a passenger car’s energy consumption
is up to 10 % [9].
Topography (Elevation Profile)
Driving a vehicle along a certain route is, of course, not just a movement on a two-dimensional plane,
rather there is also the third dimension of altitude (the elevation above mean sea level). Because kinetic
energy needs to be converted into additional potential energy, moving the vehicle mass uphill (against
gravity) consumes more energy than moving it on a flat surface. Since the electric motor of EVs can work
as a generator when going downhill (coasting), the potential energy can be converted back in order to
charge the battery. Nevertheless, due to a certain amount of losses, driving a hilly road still requires more
energy than driving a flat road with the same distance, although EVs perform typically much better than
- 31 -
Chapter 2 - Related Work
ICE vehicles. Unfortunately, these impacts are not accounted in driving cycle assessments like the NEDC
or FTP-75 [31].
Beside the elevation profile of trips, other road parameters like the curviness and amount of sharp turns can
influence energy efficiency as well, as they may require additional energy-consuming accelerations [10].
Traffic Conditions
Traffic conditions have a huge potential to negatively influence energy consumption, as they can dictate
driving pace to a large extent. This may include traffic congestions, traffic light timings, temporary
roadblocks, and mistakes/behavior of other road users. Avoiding too much stop-and-go driving, e.g. by
traffic light assistance systems, is said to have a saving potential of about 20 % (at least for ICE vehicles)
[47]. However, in most situations, mitigating the negative impacts of traffic conditions requires some kind
of forward-looking acting, which is an aspect of individual driving styles (personal factors).
2.2.2 Vehicle-Dependent Factors
Vehicle-dependent factors that are influencing driving range are those that are provoked primarily by prop-
erties of the vehicle itself. Despite most of them are prescribed during production (technical properties
of certain components), the extent of their influence is also heavily depending on the driver. Hence, the
positive or negative energy consumption impacts of certain components may be weakened or amplified
by human behavior. The energy demand of passenger cabin heating is one example: Although the ef-
ficiencies and capabilities are dependent on the vehicle and inner components, it is also in the driver’s
hands to adjust settings in favor of lower consumption and extended range.
Aerodynamic Shape
When a vehicle is moving. an aerodynamic drag force is created. Besides momentary parameters like
speed, this force is mainly dependent on the vehicle’s shape (frontal area) and drag coefficient [80]. Thus,
manufacturers try to optimize vehicles (of all kinds) in this regard.
Empty Weight
The weight of the vehicle (without inmates and payloads) is a property that is of high relevance when
talking about BEVs. As discussed in chapter 1.2.5, it is the battery weight that accounts for a large
portion of the total vehicle weight, because the gravimetric energy density is significantly worse compared
to fossil fuels in ICE vehicles.
It is a general principle that higher mass leads to higher inertia resistance, which requires more energy
to accelerate and move an object. In terms of vehicles it is the same situation, but noteworthy a higher
vehicle weight is also responsible for a higher rolling resistance which decreases energy efficiency even
further [9].
- 32 -
Chapter 2 - Related Work
Tires
The energy consumption of vehicles can be influenced significantly by the tire and its tread pattern. The
higher the rolling resistance (rolling resistance coefficient), which is an attribute of tires, the more energy
is required to drive the vehicle. Although the rolling resistance depends also on certain road characteristics
[9], we will focus on tire factors here (vehicle-dependent).
To better quantify this impact on energy efficiency for different types of tires, the EU has introduced
a tire label with a classification regarding the rolling resistance (see table 2.2). When driving with the
best tires (class A) compared to driving with the worst tires (class G), it is assumed that the energy
consumption decreases by about 7.5 %. The automotive industry further estimates that reducing a tire’s
rolling resistance by 10 % results in an improvement between 1 to 2 % of the overall consumption [89].
Due to abrasion of the tread it is also assumed that tire resistance gradually decreases by about 20 %
during its lifetime, but of course, energy consumption competes with safety aspects in this regard [89].
In fact it is not just production-specific characteristics of tires, but also the air pressure inside the tires
that affects rolling friction and energy efficiency. With 20 % lower tire pressure, for instance, energy
consumption might increase significantly by up to 5 % [89].
In addition it should be noted also that tire pressure itself is indirectly dependent on other indirect factors
like ambient temperatures which affect air temperatures inside the tire. A rough estimation shows that for
every 2.8C increase in air temperature, the tire pressure will rise by about 1 % [89]. Hence, a tire filled
to 2.2 bar at 25 C will have only about 2.0 bar at 0 C.
Even further, one might also consider the coherence between altitude and atmospheric pressure which also
affects tire pressure (relativity measure). When the tire pressure at sea level is 1.0 bar, it may decrease to
about 0.94 bar with a higher elevation of 600 meters [89].
When energy consumptions are computed inside the vehicle, it is also important to include the right
tire diameter, as the driving distance is typically calculated by the number of tire rotations. Hence, to
deliver accurate (and comparable) values for an amount of energy consumed per distance, this should be
considered as well.
class rolling resistance (CR) in kg/t
6.5
6.6 - 7.7
7.8 - 9.0
-
9.1 - 10.5
10.6 - 12.0
12.1
A
B
C
D
E
F
G
Figure 2.2: Classification of rolling resistances for the EU efficiency label on passenger vehicle tires.
- 33 -
Chapter 2 - Related Work
Drivetrain Components
As described in the introduction (chapter 1.2), there are several factors that can minimize the efficiency
of a BEV drivetrain. Some losses can be contributed to the motor, the inverter, or to the transmission, but
it’s also the suspension that can influence energy consumption and driving range of BEVs.
Battery
It may seem obvious that the capacity and state of charge (SoC) of the battery dictates the remaining
driving range of BEVs, but it should be noted that not all of the stored energy is actually available for
driving. Experimental tests have shown, for instance, that just 19 kWh could be used by a BEV with a
24 kWh battery pack (taking battery and inverter losses into consideration) [30]. Most often this is an
intended behavior of the battery management system (BMS) which limits the real capacity and prevents
deep discharges in order to avoid negative impacts on battery life.
2.2.3 Personal Factors
It has been shown that human behavior is essential as a range-limiting factor of BEVs. When looking at
environmental coherences, the driver may help indirectly to reduce impacts by choosing the right time (e.g.
influence of traffic, temperature or unfavorable atmospheric conditions), but also the right driving route
(e.g. influence of topography/elevation, road condition, or also traffic). For vehicle-dependent impacts,
the driver has an indirect influence by maintaining optimal tire pressure and taking care of internals like
the battery or drivetrain.
But of course there are also factors that are in direct control of humans at the moment they are operating
the vehicle. These personal factors range from driving style (e.g. speed, rate of acceleration, stop-and-go,
distance to other vehicles) to settings of energy consuming components and systems (e.g. climate control,
heating systems, radio, and many more). Hence, it is not hard to imagine that individual actions can have
a tremendous impact on energy consumption and BEV driving range. Therefore we focused our attention
on analyzing such factors and reasoning about potentials for optimizations.
Individual Driving Style
Driving styles may be characterized as aggressive usage, normal (average) usage, and energy-efficient
”eco-driving”. To analyze the impacts of different styles, it is important to consider only realistic (real-
world) behavior in order to show meaningful potentials because indeed if someone really tries to drive
inefficient on purpose, he/she will be able to limit BEV range even greater.
Researchers conclude that the main factor that impacts range is driving style and individual user char-
acteristics [35, 31]. For conventional ICE vehicles, this has been analyzed and proven for some time
now. Energy-efficient driving training in the Netherlands, for instance, was proven to result in average
improvements of about 10 % [92]. Others indicate fuel savings in the range of 6 to 15 %, or even up to
20 % [8, 92]. And even better, some studies claim reduced fuel consumptions of 50 % [26], or up to 58 %
[47].
- 34 -
Chapter 2 - Related Work
For battery-electric vehicles (BEVs) the variance is similar: driving aggressively, without taking advan-
tage of regenerative braking systems, is claimed to reduce fuel efficiency by more than 30 % [8]. On a
test course another study found out that aggressive driving (155.4 Wh/km) consumes about 18 % more
energy than normal usage (131 Wh/km). And on the other hand, eco-driving (104.7 Wh/km) may easily
allow to drive 20 % more efficiently (see also table 2.1) [30]. Finally, a comprehensive survey concluded
that an average saving potential of 26.8 % (comparing normal to efficient BEV driving) is possible in a
realistic scenario. And in addition, they note that driving more energy-efficiently does not equate with
longer trip duration (no significance) [47].
At first, it is necessary to highlight driving speed as a key factor on the overall energy consumption. As
depicted in Figure 2.3, three kinds of forces/counterforces are relevant when moving the vehicle: rolling
resistance, internal friction, and air drag. But the proportional distribution of the mechanical power
(available at the engine) is strongly dependent on the speed of the vehicle [9]. With up to 60 km/h air
drag is comparatively low and rolling resistance is the predominant factor. Then, at higher speeds, energy
consumption starts to grow relatively fast due to the required power to overcome aerodynamic force which
grows with the square of speed [31, 39].
0 20 40 60 80 100 120 140 160
0
20
40
60
80
100
speed (km/h)
energy (%)
rolling resistance (Frr)
internal friction (Fv)
air drag (Fd)
Figure 2.3: The energy required to move a typical passenger car (at constant speed without slope) is
distributed as a function of the vehicle speed. The main forces are rolling resistance (Frr ), internal
friction (Fv) and air drag (Fd) which becomes the predominant factor at higher driving speeds [9].
Of course, this does not imply that lower driving speeds are better in every situation. In particular we have
to consider a certain base load that constantly draws energy from the battery (e.g. on-board computer,
electric control systems, and auxiliary loads regulated by the driver). Thus, the longer the vehicle is
driven, the more energy is consumed. This is best demonstrated with energy-hungry components like
passenger cabin heating, as indicated in Figure 2.4 [57, 58]. It shows the huge impact of heating at lower
- 35 -
Chapter 2 - Related Work
driving speeds, which means that the optimal speed (in order to maximize driving range) gets pushed
upwards with higher energy consumptions of auxiliary loads. One could argue also that slower driving
might be more relevant with pleasant outdoor temperatures (usually rather spring and autumn seasons)
[57]. Probably a good compromise to save time but sill conserve energy would be an optimum driving
speed between 60 to 90 km/h [19].
0 10 20 30 40 50 60 70 80 90 100 11 0 120
0
20
40
60
80
100
120
140
speed (km/h)
driving range (km)
range-optimal speed
4.2 kW heating
2 kW heating
no heating
Figure 2.4: The estimated driving range at different speed levels also depends on auxiliary loads. The
optimal speed to maximize range increases from disabled heating to medium heating (2 kW) up to
maximum heating (4.2 kW) settings [57].
Besides the driving speed itself, individual driving styles are characterized by stop-and-go occurrences.
In order to reduce unnecessary energy consumption, it is recommended to anticipate traffic flow and
signals, maintain constant speed (with cruise control when appropriate), and avoid sudden acceleration
or deceleration. It is estimated (at least for ICE vehicles) that one second of high-powered driving may
produce almost the same amount of carbon monoxide emissions as 30 minutes of normal driving [6]. But
in terms of EVs, it is not only the rate of acceleration that may drastically reduce driving range, rather it is
also heavy braking which leads to the use of the mechanical brakes instead of regenerative braking [31].
On average, EVs are estimated to achieve a regeneration of braking energy in the range of 30 % (Nissan
Leaf) or even 50 % (Tesla Roadster) [46]. However, the optimal use of regenerative braking, which is
depending on both the accelerator pedal and the brake pedal, requires a certain change in driving style
(e.g. the so called ”one foot driving”) [92].
In fact, a study on intelligent driver feedback systems [39] concluded that the elimination of stop-and-go
driving can yield the greatest benefit in order to reduce energy consumption.
Finally, regarding characteristics of driving styles, there is also the aspect of interactions with other road
users. It is known that the distance to a vehicle driving in front can influence air drag and energy consump-
- 36 -
Chapter 2 - Related Work
tion due to the slipstream. Although it might increase the range of BEVs when driving very close to a
leading vehicle, it cannot be recommended because of safety aspects. Nevertheless, this might be another
opportunity for autonomous driving, because due to the interconnection with other vehicles and almost
instantaneous ”reaction” times, the formation of narrow road convoys could be safe and energy-efficient
at the same time.
Occupants and Payload
Moving more weight inside the vehicle requires more energy, as described in the section of vehicle-
dependent factors. We can assume, however, that the additional consumption impact of occupants is quite
low compared to the total weight of the vehicle (empty weight). Simulations for a mid-size passenger car
(ICE) have shown that a weight reduction of 5 % (about 33kg) might lead to efficiency-improvements of
about 1 % on a mixed drive cycle [17]. But for EVs with regenerative braking such improvements are
expected to be even lower. Although thinking of occupant weights might be pointless for the driver, it is
a good idea to avoid unnecessary payload.
Auxiliary Load Settings
There are multiple in-car systems and components that can draw energy from an EV’s battery while be-
ing in control of the driver. Based on rough estimations and specifications of modern BEVs, we can
assume that most of them account for relatively little average power consumption [80, 82, 28]: navigation
or entertainment systems (radio) with about 0.05 kW, LED headlights with about 0.18 kW, fog lights
with about 0.1 kW, electric window regulator with about 0.1 kW, windscreen wipers with about 0.15 kW,
and electric power steering with about 0.1 kW. In addition to these rather small loads, we will focus on
the much higher consumptions of heating, ventilation, and air conditioning (HVAC). This includes heat-
ing/cooling of the passenger cabin (comfort aspect), as well as defogging and deicing the windows (safety
aspects) [46]. Although the driver can typically adjust these systems, some BEVs also include thermal
management systems for battery heating/cooling [96] which cannot be controlled and are therefore not
considered as personal factors.
An assessment of electric transportation in Switzerland [96] assumes that the consumption for BEV air
conditioning (cooling) in Central Europe is about 1 kWh/100km which corresponds to an average power
consumption of about 1.5 kW (required for four months per year). Heating, on the other hand, will
consume 3 kWh/100km which corresponds to an average thermal energy requirement of 3kW (for six
months per year). Firstly, this high energy demand can be attributed to the technical specifications of
the heating systems. While some EVs are still relying on electric resistance heaters, which are known
to be very inefficient, some others (e.g. Renault ZOE) are using a more energy-efficient bidirectional
heat pump system which is capable of heating and cooling in one. Secondly, most vehicles have a bad
thermal insulation, such that their energy demand is about the same as for a small single-family house
[96]. Although this was less relevant for ICE vehicles, in which the heat loss of engines was more than
sufficient for ”free” cabin heating, a better insulation might be essential for EVs. One interesting concept
to overcome this issue is the thermoelectric variable temperature seat (VTS) by Amerigon [18]. Instead
of wasting energy to heat/cool interior surfaces and the surrounding space inside the vehicle, this seat
- 37 -