World Electric Vehicle Journal Vol.

7 - ISSN 2032-6653 - ©2015 WEVA Page WEVJ7-0447

EVS28
KINTEX, Korea, May 3-6, 2015

Electric vehicle energy consumption modelling and

prediction based on road information
Jiquan Wang1, Igo Besselink1, Henk Nijmeijer1
1
Dynamics and Control Group, Department of Mechanical Engineering, Eindhoven University of Technology,
Den Dolech 2, 5612 AZ Eindhoven, the Netherlands

Abstract
The limited driving range is considered as a significant barrier to the spread of electric vehicles. One
effective method to reduce “range anxiety” is to offer accurate information to the driver on the remaining
driving range. However, the energy consumption during driving is largely determined by driving
behaviour, road topography information and traffic situation, which are hard to predict. This paper will
discuss an accurate electric vehicle energy consumption model validated using driving tests on different
public roads, and then the model is used to predict future energy consumption based on road information.
The energy consumption model includes five parts: the road load model, the powertrain loss model, the
regenerative braking model, the auxiliary system model and the battery model. The parameters of these
models are obtained through driving tests on public road and dynamometer tests in the TU/e Automotive
Engineering Science lab. The results show that the model can calculate the energy consumption with a
maximum error of 5% based on driving speed under different circumstances. To predict the future energy
consumption, the road information is obtained from OpenStreetMap and Shuttle Radar Topography
Mission. An offline algorithm is built to predict the energy consumption for a future trip based on the road
information. The algorithm gives two energy consumption results: one is for the fastest driving speed; the
other one is for the most economic driving speed. The results show that the measured energy consumption
results for different types of road driving are all within the algorithm’s prediction scope. Therefore, the
offline algorithm can give an accurate energy consumption estimation to the driver before a trip begins.

Keywords: Electric vehicle, Energy consumption model, Road information, Energy prediction

“range anxiety”, which can be explained as the

1 Introduction drivers’ concern of not reaching the destination
during driving [1]. Range anxiety is considered as
Electric vehicles (EVs) are considered as future one of the major factors that affect the acceptance
cars to solve oil dependency and environmental of electric vehicles.
problems. However, the spread of EVs is limited Apart from a bigger battery, an accurate range
by high price, long charging time, few charging estimation system is necessary to solve range
stations and limited driving range. The limited anxiety. However, the prediction of the remaining
driving range and long charging time may lead to range is complicated, because it is dependent on

EVS28 International Electric Vehicle Symposium and Exhibition 1

World Electric Vehicle Journal Vol. 7 - ISSN 2032-6653 - ©2015 WEVA Page WEVJ7-0448

some stochastic factors such as vehicle driving tests, including highway driving, city
characteristics, driving behaviour, traffic state, driving and rural driving are shown to verify the
road topography and weather condition. Several offline algorithm.
studies have been performed to predict the
remaining driving range of EVs. They mainly 2 Energy consumption model
focus on predicting the driving speed [2-4] and
obtaining the traffic information and road The energy flow of an electric vehicle for driving
topography information [5-7]. is shown in Figure 2. The electric energy is taken
The predicted driving speed and road information from the battery: a small part is used for the
can help to improve the prediction accuracy. auxiliary system; the main energy is transformed
However, there are still two problems that need by the electric motor into mechanical energy,
to be solved. Firstly, the energy consumption is which is used to overcome the road load.
also determined by the vehicle characteristic The electric vehicle energy consumption model is
parameters, and some of the vehicle parameters built in a backward way in this research. This
are influenced by the weather condition and means that the vehicle driving speed is the model
driving behaviour. Secondly, the driving speed input and the battery energy is the model output.
may be difficult to predict accurately before a The energy consumption model can be divided
trip begins, because the driving speed is into five parts from the vehicle wheel to the
influenced by the traffic flow. battery: the road load model, the powertrain loss
In this paper, an offline algorithm is developed to model, the regenerative braking model, the
solve these two problems. The algorithm can auxiliary system model and the battery model. The
give the driver a first view on the energy auxiliary system components energy consumption
consumption for a future trip based on different can be measured, the results are reported in [9].
driving speed predictions including considering The battery model adopted is a simple battery
the weather influence. This research is based on equivalent circuit model [10]. This paper will
an electric vehicle: TU/e Lupo EL. The TU/e mainly introduce the road load model, the
Lupo EL is built from a donor vehicle VW Lupo powertrain loss model and the regenerative braking
3L by the Dynamics and Control group of model. The parameters of those models are
Eindhoven University of Technology in 2009, obtained from different tests.
where EL is the abbreviation of “Electric
Lightweight”. The TU/e Lupo EL is allowed to Auxiliary
system
drive on the public road since spring 2011 [8-9].
drive shaft
80 % Inverter
of nominal
capacity
motor reduction
battery wheel
Discharging

Figure 2: Energy flow of an electric vehicle

2.1 Road load model

The moving of a vehicle on a road is influenced by
the road load, which includes rolling resistance 𝐹𝑟 ,
aerodynamic drag force 𝐹𝑎𝑖𝑟 , road slope force 𝐹𝑔
and acceleration force 𝐹𝑖 , as shown in Figure 3.
Figure 1: Lupo 3L with the existing diesel and new
electric powertrain
This paper is organized as follows. In section 2,
an energy consumption model is built with
considering the influence of the weather
condition based on the measurement data. In
section 3, the road information is obtained from
the OpenStreetMap (OSM) and Shuttle Radar
Topography Mission (SRTM). Then the energy
consumption for a future trip is predicted based
on the road information. In section 4, three Figure 3: Outside forces acting on a driving vehicle

EVS28 International Electric Vehicle Symposium and Exhibition 2

World Electric Vehicle Journal Vol. 7 - ISSN 2032-6653 - ©2015 WEVA Page WEVJ7-0449

The vehicle’s longitudinal dynamic equation is: slope can be calculated based on the road
𝐹𝑥 = 𝐹𝑟 + 𝐹𝑎𝑖𝑟 + 𝐹𝑔 + 𝐹𝑖 information. However, the air density 𝜌 is
1 changing with the weather condition, and the
= 𝑓𝑟 ∙ 𝑚 ∙ 𝑔 + ∙ 𝜌 ∙ 𝐶𝑑 ∙ 𝐴 ∙ (𝑣 − 𝑤)2 (1) rolling resistance coefficient 𝑓𝑟 is determined by
2
+𝑚 ∙ 𝑔 ∙ sin 𝛼 + 𝑚𝑒𝑓𝑓 ∙ 𝑎𝑥 the weather and road condition. Therefore, these
two parameters are always changing at different
where 𝐹𝑥 is the propelling force, 𝐹𝑟 is the rolling driving conditions. To model the energy
resistance force, 𝐹𝑎𝑖𝑟 is the air dynamic drag consumption, the changing trend of these two
force, 𝐹𝑔 is the gravity force, 𝐹𝑖 is the parameters should be identified.
acceleration force caused by vehicle inertia, 𝑓𝑟 is
the rolling resistance coefficient, 𝑚 is the vehicle 2.1.1 Air density
mass, 𝑔 is the gravitational acceleration, 𝜌 is the The air density is a function of pressure, relative
air density, 𝐶𝑑 is the aerodynamic drag humidity and ambient temperature, while the
coefficient, 𝐴 is the vehicle frontal area, 𝑣 is the humidity has a minor influence on the air density
vehicle speed, 𝑤 is the wind speed in the vehicle at higher temperature [12]. According to the
driving direction, 𝛼 is the road slope, 𝑚𝑒𝑓𝑓 is the equations published in [12], the value of air
vehicle effective mass, 𝑎𝑥 is the vehicle density is shown in Figure 4 for the condition that
acceleration. the humidity equals 80%. As can be seen, the air
The vehicle effective mass 𝑚𝑒𝑓𝑓 is the sum of density will decrease with a temperature increase
and air pressure decrease.
the vehicle mass and the equivalent mass of the
motor and wheel inertia. Air density, 80% humidity

4 ∙ 𝐽𝑤 𝐽𝑒𝑚
1.5
1050 hPa
𝑚𝑒𝑓𝑓 = 𝑚 + 2 + 2 2 (2) 1.45 1025 hPa

𝑅𝑒 𝑟𝑑 ∙ 𝑅𝑒 1.4
1000 hPa
975 hPa
950 hPa
1.35
Air density [kg/m3]

where 𝑚 is the vehicle mass, 𝐽𝑤 is the wheel 1.3

inertia, 𝐽𝑒𝑚 is the motor inertia, 𝑅𝑒 is the tyre 1.25

effective rolling radius and 𝑟𝑑 is the gear 1.2

reduction ratio. 1.15

Table 1: Variability of vehicle longitudinal dynamics 1.1

equation parameters 1.05

-20 -10 0 10 20 30
Ambient temperature [deg.C]

Parameters Dependency Variability Figure 4: Air density as a function of ambient temperature

𝑔 and pressure [12]
𝐶𝑑 - Stable
𝐴 2.1.2 Rolling resistance coefficient
𝑚 Passengers A series of coast down tests have been performed
Low
𝜌 Weather
to obtain the rolling resistance coefficient at
Road
𝑓𝑟 different conditions. During a coast down test, the
Weather High
𝜃 Weather propulsion is removed when the vehicle reaches a
Traffic certain speed, and then the road load, consisting of
𝑎 rolling resistance, aerodynamic drag and friction
Road
𝑣
Driver Extremely high force, will slow down the vehicle until the vehicle
Weather comes to a standstill. The differential equation
𝑤
route describing the longitudinal dynamics in a coast
down test is given as
All the parameters in Equation (1) need to be 𝑚𝑒𝑓𝑓 ∙ 𝑣̇ = −𝑓𝑟 ∙ 𝑚 ∙ 𝑔
determined to build the road load model. Since 1 𝑇𝑓𝑟 (3)
the variability properties of these parameters are − ∙ 𝜌 ∙ 𝐶𝑑 ∙ 𝐴 ∙ (𝑣 − 𝑤)2 −
2 𝑅𝑒
different (see Table 1), those parameters can be
determined separately. The stable parameters can where 𝑇𝑓𝑟 is the friction moment in the gearbox
be obtained by doing tests in a lab. The and wheel bearings.
extremely high variability parameters can be The rolling resistance coefficient is dependent on
measured during the driving. The vehicle mass is the road surface, ambient temperature and driving
dependent on the passengers number. The road speed [13]. Different coast down tests are done to

EVS28 International Electric Vehicle Symposium and Exhibition 3

World Electric Vehicle Journal Vol. 7 - ISSN 2032-6653 - ©2015 WEVA Page WEVJ7-0450

obtain the relation between the rolling resistance The effect of temperature on rolling resistance is
coefficient with these factors. shown in Figure 6. A comparison of simulation
A comparison between simulation and and measurement of coast down tests on a highway
measurement of the coast down tests on different road in summer and winter is shown in Figure 6.
road types in summer are shown in Figure 5. The summer test was done on 1st August 2013, the
Through the comparison, the value of rolling ambient temperature was about 29 degree Celsius,
resistance coefficient can be obtained. The and 𝑓𝑟 was then determined to be 0.0095. The
rolling resistance on a highway road (smooth winter test was done on 11th December 2013, the
asphalt) is 0.0095 and it increases to 0.0157 on a ambient temperature was about 2.5 degree Celsius,
coarse rural road (Rijtvenweg test); the increase and 𝑓𝑟 was then determined to be 0.0132. The
is about 65%. It can be seen that the difference is increase of 𝑓𝑟 is 39% from 29 ºC to 2.5 ºC.
really significant. However, it is difficult to Based on the measurement data at different
obtain the rolling resistance coefficient on every temperatures, the relationship between the rolling
road section. This relative difference between resistance coefficient with the temperature is given
highway road and rural road will be used to in Equation 5.
2
estimate the relative relationship of rolling 𝑓𝑟0 = 1.9 × 10−6 ∙ 𝑇𝑎𝑚𝑏 − 2.1 × 10−4
(5)
resistance coefficient on other types of road, and ∙ 𝑇𝑎𝑚𝑏 + 0.013
the result is shown in Table 3.
120
2.2 Powertrain loss model
Rijtven test 1
Rijtven test 2 The vehicle powertrain components consist of a
Simulation frr=0.0157
100
Highway test 1 MES-DEA TIM 600 inverter, a MES-DEA A200-
Highway test 2
80 Simulation frr=0.0095 200W water cooled AC induction motor and a
Velocity [km/h]

Carraro reduction gear/differential (gear ratio

60
1:8.654).
40
To determine the power loss, a test was executed
on a dynamometer in the TU/e AES lab. The
20 experiment is shown in Figure 7 [14]. The
0
vehicle’s front wheels are put on the dynamometer
0 50 100
Time [s]
150 200
in the test.
Figure 5: Coast down test on highway road and rural road
Apart from the road surface, the rolling
resistance coefficient is influenced by the
weather and the driving speed. Therefore, the
rolling resistance coefficient can be expressed as
𝑓𝑟 (𝑇𝑎𝑚𝑏 , 𝑣) = 𝑓𝑟0 (𝑇𝑎𝑚𝑏 ) + 𝑓𝑟1 ∙ 𝑣 (4)

where 𝑇𝑎𝑚𝑏 is the ambient temperature, 𝑓𝑟1 is a

constant value 5.4 × 10−5, and 𝑓𝑟0 is determined
by experiments. Figure 7: Dynamometer test in AES lab
120
summer test 1
When the drum is running at a constant speed, the
100
summer test 2
simulation frr=0.0095
output of the motor is changed by changing the
winter test 1
winter test 2
inverter control signal. When the inverter control
80 simulation frr=0.0132
signal is bigger than 50, the vehicle is in traction
Velocity [km/h]

60
mode, and the vehicle is in regenerative braking
mode when the inverter control signal is smaller
40 than 50. The motor electric power and bench
20
mechanical power can be measured during the
experiment, as illustrated by Figure 8. It can be
0
0 50 100 150 200
seen from Figure 8, when the inverter control
Time [s]
signal is 50, the motor is idling, the efficiency is
Figure 6: Coast down test on highway road at different zero, but the power loss still exists because of the
time friction. Therefore, the power loss is used to

EVS28 International Electric Vehicle Symposium and Exhibition 4

World Electric Vehicle Journal Vol. 7 - ISSN 2032-6653 - ©2015 WEVA Page WEVJ7-0451

represent the powertrain energy loss instead of For the regenerative mode, the empirical equation
efficiency map. is:
𝑃𝑙𝑜𝑠𝑠 = 0.21 ∙ 𝑇 2 + 0.097 ∙ |𝑇| ∙ 𝜔 + 𝑃𝑐 (7)

where 𝑇 is the motor torque, 𝜔is the motor angular

speed, 𝑃𝑐 is the motor idling power loss.
The power loss difference between the calculation
and the measurement is depicted in Figure 10. It
can be seen that the error is smaller than 0.5 kW in
most cases. Therefore, the empirical equations are
sufficiently accurate and can be used to calculate
the powertrain loss. Based on the equations, the
powertrain efficiency map is calculated and
depicted in Figure 11. It can be seen that the
powertrain efficiency is better in the traction mode
Figure 8: The method of calculating the powertrain loss than in the regenerative braking mode at the same
motor output.
The powertrain loss at a specific speed is the
difference between the motor electric power and 6.2 km/h
drum mechanical power. After processing the 2
12.4
24.8

measurements, the relationship between the 1.5

37.2
49.7
62.1
power loss and motor mechanical torque at Power loss difference [kW] 74.5
86.9

different driving speed is shown in Figure 9. As 1 99.3

111.8
124.2
can be seen, the power loss is different for the 0.5
traction part and regenerative braking part at the
same motor mechanical power. 0

-0.5
6.2 km/h
12.4
-40 -30 -20 -10 0 10 20 30 40
10 24.8 Mechanical power [kW]
37.2
49.7
8 62.1 Figure 10: The powertrain loss difference between
Power loss [kW]

74.5
86.9
simulation and measurement
6 99.3
111.8 0.9
250 0.3
0.3
124.2
4 200 0.2
0.20.4
0.4 0.8
0.6
0.6
150
0.7
2 100 0.7 0.80.8
0.7
0.9 0.9
0.85
0.85 0.6
Torque [Nm]

50 0.92 0.92
0
-50 -40 -30 -20 -10 0 10 20 30 40 0 0.5
Motor mechanical power [kW ] 0.85 0.85
-50
0.80.8 0.4
-100
Figure 9: The relationship between the powertrain loss 0.6
0.6
0.3
-150
and the motor mechanical power at different speeds 0.10.5
0.1 0.5
0.70.7
-200
0.2
0.4
0.4
To simplify the calculation, algebraic equations -250
0.1
are used to describe the power loss. Since the 0 200 400 600 800
Angular Velocity [rad/sec]
1000

power loss for the traction mode and regenerative

Figure 11: The powertrain efficiency map
mode are different, separate empirical equations
are used to calculate the power loss. Most of the
power loss is caused by the inverter and motor, 2.3 Regenerative braking model
thus, the power loss can be calculated based on There are two kind of regenerative braking control
the motor torque and angular speed. strategies in the Lupo EL. The first one is a
The MATLAB/Simulink Parameter Estimation parallel regenerative braking control strategy, the
Tool is used to get the empirical equation other one is a one pedal driving control strategy.
structure and optimize the value of parameters.
For the traction mode, the empirical equation is: 2.3.1 Parallel regenerative braking control
𝑃𝑙𝑜𝑠𝑠 = 0.3158 ∙ 𝑇 2 + (2.2 × 10−10 ∙ 𝜔3 strategy
−1.62 × 10−7 ∙ 𝜔2 + 3.44 × 10−5 ∙ 𝜔) (6) The parallel regenerative braking control strategy
∙ 𝑇 3 + 𝑃𝑐 in the Lupo EL is based on the brake pedal travel.

EVS28 International Electric Vehicle Symposium and Exhibition 5

World Electric Vehicle Journal Vol. 7 - ISSN 2032-6653 - ©2015 WEVA Page WEVJ7-0452

The total braking force and regenerative braking 5

accelerator pedal released
force are shown in Figure 12 when the driving 4
brake pedal actuation
speed is 60 km/h. As can be seen, when the brake 3
accelerator pedal > 30%

pedal travel is below 10%, the hydraulic braking 2

Acceleration [m/s ]
2
force is zero, this is the free travel between the
1
brake disc and brake drum. And when the brake
pedal travel is bigger than 60%, the regenerative 0

braking force is reduced to zero, which is to -1

ensure the braking stability in an emergency -2

case. -3
However, there are two disadvantages in the
-4
parallel regenerative braking strategy. The first 0 20 40 60 80 100 120
Driving speed [km/h]
issue is that the braking force is not linear with
the brake pedal travel, which may not provide a Figure 13: One pedal driving control strategy
comfortable and consistent brake feel for the
driver. The second issue is that the hydraulic 2.4 Driving verification
braking system is always working during the Some driving tests on public road are used to
braking, which results in part of the energy being verify the energy consumption model. A total of
transferred into heat. Therefore, the regenerative twelve driving tests were recorded in 2013 and
braking strategy needs to be modified to improve 2014. Six recordings, from May to November in
the braking feel and energy efficiency. 2013, are driven with the parallel regenerative
60 km/h
braking algorithm. Another six recordings in 2014
10000 are driven with the one pedal driving algorithm.
Total brake force
Regenerative force
The driving tests include constant speed driving,
8000 city driving and rural road driving. The battery
output energy is measured during driving. The
Brake force [N]

6000
details of these tests and measured energy results
4000
are listed in Table 2.
Table 2: The recording of driving tests on public road
2000
DC
Speed Distance
Date energy
0 [km/h] [km]
0 20 40 60 80 100 [kWh]
Brake pedal travel [%] May 6th 80 228.9 22.6
Parallel May 7th 100 114.1 16.6
Figure 12: Parallel regenerative braking control strategy regenerative
Jun 17th 120 113.9 18.4
braking
Jun 5th City 47.7 5.1
algorithm
2.3.2 One pedal driving control strategy (2013) Jun 6th City 49.6 5.3
A one pedal driving control strategy is designed Nov 27th City 50.1 6.8
Jun 6th 95 60.2 6.7
to improve the performance of the vehicle. For One pedal Sep 2nd 115 106.8 15.6
the one pedal driving algorithm the maximum driving Jun 26th City 6.9 0.7
deceleration that can be achieved when releasing algorithm Dec 3rd City 7.9 1.0
(2014) Nov 15th Rural 17.96 1.9
the accelerator pedal is increased to 2 m/s2. This Dec 2nd Rural 17.97 2.1
allows the vehicle to be driven by the accelerator
pedal alone in most cases, and the brake pedal is The energy consumption of a driving test can be
only applied for emergency cases. The calculated using the energy consumption model
relationship between forward speed and based on the driving speed. Finally, the error of the
acceleration/deceleration in a driving test on the model can be represented by the difference
public road is shown in Figure 13. It can be seen between the calculation and measurement:
that the driving test include the city driving and 𝐸𝑠𝑖𝑚𝑢𝑙𝑎𝑡𝑖𝑜𝑛 − 𝐸𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑚𝑒𝑛𝑡
highway driving, and the brake pedal is seldom 𝑒= ∙ 100% (8)
𝐸𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑚𝑒𝑛𝑡
used during the test.
The simulation results demonstrate that the energy
consumption model can calculate the vehicle
energy consumption based on the driving speed
with an error smaller than 5% for different

EVS28 International Electric Vehicle Symposium and Exhibition 6

World Electric Vehicle Journal Vol. 7 - ISSN 2032-6653 - ©2015 WEVA Page WEVJ7-0453

circumstances. Therefore, the model is speed is an empirical value based on the road type
considered sufficiently accurate and can be used in this research. The structure of the offline
as a tool to predict the future energy consumption algorithm is depicted in Figure 15.
information to the driver.

Figure 15: The energy prediction offline algorithm

Figure 14: The difference between the simulation and structure
measurement for energy consumption model
3.1 Road information
3 Energy consumption prediction The road information for the future driving route,
To predict the energy consumption for a future including the geographic coordinates, road type,
trip, the road information, weather condition and traffic light position and speed limitation signs, can
driving speed should be known first, and then the be obtained from the OpenStreetMap (OSM).
energy consumption model can be used to OSM is a collaboration project to create a free
calculate the required amount of energy. The editable map of the world that provides
road information, including route length, road geographical data to anyone [16]. It is freely
slope, road direction and road type, can be available and has no technical restrictions in terms
obtained from OSM and SRTM. The ambient of processing the data.
temperature, wind speed and direction can be The height information of the future route can be
obtained from the weather report. The driving obtained from the Shuttle Radar Topography
speed is determined by the road type, traffic Mission (SRTM) based on the geographic
condition and driver behaviour. The driving coordinates. SRTM is an international research
behaviour varies from driver to driver and even effort that obtained digital elevation models on a
may be different for the same driver at different near-global scale from 56 degrees south to
time and situations. Therefore, the driving speed 60 degrees north latitude, which comprises almost
cannot be predicted accurately before a trip 80% of Earth’s total landmass. The data has been
begins. However, the maximum driving speed released at two horizontal resolutions: 3 arc-
and most economic driving speed can be seconds (90 m) globally, and 1 arc-second (30 m)
obtained based on the road information to predict for the Unites States [17-18].
the energy consumption.
The offline algorithm is designed to give a first 3.2 Driving speed
estimation on the energy consumption to the Four factors can be obtained from OSM to
driver before a trip begins. In the algorithm, two determine the vehicle driving speed: road type,
kinds of driving speed are determined: one is for speed limit sign, road curvature and traffic lights.
the fastest driving, while the other one is for the The maximum driving speed and most economic
most economic driving. Two energy driving speed are a combination effect of these
consumption results are provided correspondent four factors.
to the predicted driving speed. The fastest driving
speed is determined by the speed limit based on 3.2.1 Road type
road type. The most economic speed of the Lupo
Normally, there is a legal allowed driving speed
EL is about 25 km/h for constant speed driving
scope for each type of road. Therefore, the first
[15], but this value will decrease while
estimation of the maximum and most economic
considering the influence of the frequently start-
driving speed is determined by the road type. The
and-stop driving in city route. Therefore, the
relative rolling resistance on different road types
most economic speed is set as the minimum
have been estimated based on the coast down test
driving speed on the road. The minimum driving
results in Section 2.1.2.

EVS28 International Electric Vehicle Symposium and Exhibition 7

World Electric Vehicle Journal Vol. 7 - ISSN 2032-6653 - ©2015 WEVA Page WEVJ7-0454

2
𝑣𝑚𝑎𝑥
= 𝑎𝑦𝑚𝑎𝑥 (9)
Table 3: Road type information based on OSM 𝑅

Type
Maximum Minimum Relative where 𝑅 is the corner radius, 𝑣𝑚𝑎𝑥 is the maximum
Road type speed speed rolling
number
[km/h] [km/h] resistance driving speed and 𝑎𝑦𝑚𝑎𝑥 is the maximum lateral
footway acceleration.
pedestrian
living street
1 15 5 1.25 The maximum lateral acceleration can reach 8 m/s2
residential 2 30 10 1.15 on a standard vehicle during driving [21],
service 3 30 10 1.25 however, the maximum lateral acceleration is
track 4 30 10 1.40 smaller than 4 m/s2 at most cases according to the
unclassified 5 50 20 1.20
tertiary 6 50 20 1.10 Lupo EL driving test measurement. Therefore, the
secondary 7 70 30 1.05 maximum lateral acceleration is considered as
primary 8 80 50 1.00 4 m/s2 in this research. The road curvature can be
trunk 9 100 70 1.00
motorway 10 120 80 1.00
calculated based on the coordinates obtained from
OSM.
The maximum allowed speed is 130 km/h on
about half of motorways in the Netherlands, 3.2.4 Traffic light
while the value is 120 km/h on the other half The influence of traffic lights has to be considered,
[19]. However, normally the driver doesn’t drive because the vehicle may have to stop and wait in
at the maximum speed for long distance on front of a traffic light. However, whether the traffic
motorway. Therefore, the maximum speed is set light is red or nor, cannot be predicted offline. To
to 120 km/h for motorway driving. The results obtain the average influence of traffic lights, half
are shown in Table 3. of traffic lights are assumed to be red, and the
waiting time is estimated to be 20 s.
3.2.2 Speed limit sign The traffic lights are tagged to road nodes in OSM.
The vehicle driving speed is also limited by the The position of the traffic lights can be obtained
speed limit sign. The road speed limit sign is once the future travel route has been determined.
used in most countries to set the maximum (or
minimum in some case) speed. Speed limit sign 3.2.5 Acceleration
is normally indicated as a traffic sign in the After combining the influence of the road type,
street, as Figure 16 [20]. The speed limit sign is speed limit sign, road curvature and traffic lights,
added as nodes information in OSM road the target speed for a specific route can be
information database, which can be investigated obtained. An illustration is shown in Figure 17. As
once the driving route is settled down. can be seen, the target speed isn’t continuous,
which cannot be realised by the vehicle. To obtain
a realistic driving speed, the discontinuities
between two target speed values should be
connected by the vehicle acceleration and
deceleration.

Figure 16: Speed limit sign in a street

3.2.3 Road curvature

The vehicle maximum lateral acceleration will
limit the driving speed if the steering radius is
small in a corner. The relation between the Figure 17: Speed information from OSM
maximum driving speed and maximum lateral The maximum deceleration caused by regenerative
acceleration is given as braking and acceleration of Lupo EL are
determined by the motor setting, the value can be

EVS28 International Electric Vehicle Symposium and Exhibition 8

World Electric Vehicle Journal Vol. 7 - ISSN 2032-6653 - ©2015 WEVA Page WEVJ7-0455

calculated by the vehicle longitudinal dynamic The acceleration is used to connect the target speed
equation. The equation is given as into the realistic driving speed. The result is
𝑃𝑚,𝑚𝑒𝑐ℎ 𝐹𝑟 + 𝐹𝑎𝑖𝑟 + 𝐹𝑔 illustrated by the blue curve in Figure 17.
𝑎𝑚𝑎𝑥 = − (10)
𝑚∙𝑣 𝑚
4 Results
The maximum motor electric power is set to
50 kW for traction and -24 kW for regenerative Several illustrative tests are shown in this section
braking. The motor efficiency is set to a constant to verify the offline algorithm, including the
value 80% in this calculation. Therefore, when highway driving, rural driving and city driving.
the motor efficiency is considered, the maximum The highway driving test illustrates how the offline
motor mechanical power 𝑃𝑚,𝑚𝑒𝑐ℎ is 40 kW for algorithm works. City driving and rural driving are
performed by four different drivers on the same
traction and -30 kW for regenerative braking.
routes. The results show that the algorithm can
The acceleration/deceleration is determined by
give an accurate energy prediction scope for
the motor power setting at high driving speed,
different drives.
while it is limited by the tire-road friction
coefficient at low speed driving. In this
algorithm, the maximum acceleration and 4.1 Highway driving
deceleration at low speed is determined by the A highway driving test was done to verify the
driving tests measurement. The relationship energy prediction algorithm on 2nd September
between the acceleration/deceleration and driving 2014 from Eindhoven to Nijmegen. The route of
speed in a driving test is shown in Figure 18. As the driving is shown in Figure 19.
can be seen, the maximum acceleration is set to Wind direction
3 m/s2 and the maximum deceleration is set to 51.85

2 m/s2 at low speed driving. 51.8

51.75

5 51.7
Latitude [deg]

measurement
51.65
4 measurement
maximum acceleration
51.6
3 maximum deceleration

51.55
2
Acceleration [m/s ]
2

51.5
1

51.45
0

5.4 5.45 5.5 5.55 5.6 5.65 5.7 5.75 5.8 5.85 5.9
-1 Longitude [deg]

-2
Figure 19: Driving route from Eindhoven to Nijmegen
-3

-4
0 20 40 60 80 100 120 140
To determine the driving speed, road types along
Speed [km/h]
the route have to be obtained from OSM first, and
Figure 18: The relation between the speed and the results are shown in Figure 20. Most part of the
acceleration/deceleration in a driving test route is highway road, but other types of road are
also included.
The acceleration during driving in this algorithm motorway
is determined by the difference between the trunk
current driving speed and target speed. If the
primary
difference is bigger than 10 km/h, the maximum
secondary
acceleration is adopted, or else, the acceleration
Road type

tertiary
linearly decreases from the maximum
uncalssified
acceleration to zero with the speed difference
decrease. The equation is track

𝑎𝑥,𝑚𝑎𝑥 , ∆𝑣 ≥ 10𝑘𝑚/ℎ service

𝑎𝑥 = { 𝑎𝑥,𝑚𝑎𝑥 (11) residential

∙ ∆𝑣, ∆𝑣 < 10𝑘𝑚/ℎ

0 10 20 30 40 50
Distance [km]
10
Figure 20: Road type along the highway driving route
where 𝑎𝑥 is the current acceleration, 𝑎𝑥,𝑚𝑎𝑥 is The height information along the driving route can
the maximum acceleration at the current driving be obtained from SRTM, and the result is shown in
speed and ∆𝑣 is the difference between the Figure 21. This again illustrate that the
current driving speed and target speed. Netherlands is a very level country.

EVS28 International Electric Vehicle Symposium and Exhibition 9

World Electric Vehicle Journal Vol. 7 - ISSN 2032-6653 - ©2015 WEVA Page WEVJ7-0456

9
maximum speed
30 8 minimum speed
90 km/h
7 100km/h

Energy consumption [kWh]

25 110 km/h
6
Measurement
5
20
4
Height [m]

3
15
2

10 1

0
5
-1
0 10 20 30 40 50
Driving distance [km]
0
0 10 20 30 40 50
Driving distance [km] Figure 23: The simulation and measurement energy result
along the highway driving route
Figure 21: The height information along the highway
driving route
4.2 City road driving
The speed information of simulation and
A driving route in the centre of Eindhoven is
measurement is shown in Figure 22. As can be shown in Figure 24. The road type of the city
seen, the highway driving speed can range from driving route is shown in Figure 25. Four drivers
80 km/h to 120 km/h. Normally, the driver have driven on this route, and the energy
needn’t to follow the traffic flow on highway consumption results are recorded. The maximum
road; he can choose his own driving speed. This and minimum predicted energy consumption
will obviously affect the energy consumption results for this route are shown in Figure 26. As
results. To provide more energy consumption can be seen the measured energy consumption for
prediction information to the driver, more energy four drivers are all between the minimum and
consumption results are calculated from 80 km/h maximum predicted result, which can confirm the
to 120 km/h, the interval calculation speed is accuracy of the approach. There are lots of traffic
10 km/h. The prediction and the measured lights in the city route, thus the vehicle have to
energy consumption are shown in Figure 23. start-and-stop frequently. This is the reason why
According to the measurement, most of the there are so many oscillations in the maximum
highway driving speed is between 110 km/h and driving speed energy consumption prediction
120 km/h, the measured energy consumption is result.
also between the 110 km/h and 120 km/h driving
speed energy prediction. This proves the 51.45
accuracy of the algorithm on highway driving. 51.448

51.446
140
Latitude [deg]

51.444
120
51.442

100 51.44
Driving speed [km/h]

80
51.438

51.436
60
51.434
maximum speed
5.465 5.47 5.475 5.48 5.485 5.49
40 minimum speed
90 km/h Longitude [deg]
100km/h
20 110 km/h
Measurement
Figure 24: A driving route in Eindhoven centre
0 secondary
0 10 20 30 40 50
Driving distance [km]

tertiary
Figure 22: The measurement and simulation driving
speed along the highway driving route unclassified
Road type

track

service

residential
0 1 2 3 4 5 6
Distance [km]

Figure 25: The road type in city centre driving

EVS28 International Electric Vehicle Symposium and Exhibition 10

World Electric Vehicle Journal Vol. 7 - ISSN 2032-6653 - ©2015 WEVA Page WEVJ7-0457

0.9 2.5
maximum speed maximum speed
0.8
minimum speed minimum speed
driver 1 driver 1
0.7 2
driver 2 driver 2
Energy consumption [kWh]

driver 3

Energy consumption [kWh]

0.6 driver 3
driver 4 driver 4
0.5
1.5
0.4

0.3
1
0.2

0.1
0.5
0

-0.1
0 1 2 3 4 5 6 0
Driving distance [km] 0 5 10 15 20
Driving distance [km]

Figure 26: The energy consumption in city centre driving

Figure 29: The energy consumption in rural driving route
4.3 Rural road driving
A rural driving route in Eindhoven nearby is
5 Conclusions
shown in Figure 27. The road type along this The purpose of this work is to develop an offline
route can be seen in Figure 28. Four drivers have algorithm for an electric vehicle range estimator,
driven on this route and energy consumption which can give a first impression to the driver on
results are recorded. The energy prediction the possible energy consumption scope for a
results are shown in Figure 29. It is again shown forthcoming trip.
that the energy consumption results are different An energy consumption model is built based on
for different drivers, but these four energy measurements to predict the energy consumption.
consumption results are all between the The model can calculate the energy consumption
maximum and minimum energy prediction with a maximum error of 5% for different
results. circumstances, which is accurate enough to be
used in the offline algorithm. However, the rolling
resistance coefficient varies significant on different
51.5
road types, which need to be investigated further in
51.49 future research.
The model is used to predict the future energy
Latitude [deg]

51.48

51.47
consumption based on the road information
obtained from OSM and SRTM. The highway
51.46
driving, city driving and rural driving results show
51.45 that the offline algorithm can provide an accurate
51.44 energy consumption prediction scope for a future
5.43 5.44 5.45 5.46 5.47 5.48 5.49 5.5 5.51
Longitude [deg] trip, which is suitable for different drivers.
Figure 27: A rural driving route in Eindhoven
Additionally, the algorithm gives more prediction
results based on different highway driving speeds,
primary
which can give a more precise advice when the
secondary driver is driving on highway road.
The future work is to build an online algorithm for
tertiary
the range estimator. The purpose of the online
Road type

unclassified algorithm is to adjust the energy prediction result

during driving based on the driving behaviour.
track

service Acknowledgments
residential The funding of PhD project of Jiquan Wang is
0 5 10 15 20
Distance [km] provided by China Scholarship Council (CSC).
Figure 28: The road type in rural driving route Thanks to Vital van Reeven for providing the
MATLAB tool to obtain data from OpenStreetMap.
The authors would like to thank everyone involved
in this project for their technical support and
advice.

EVS28 International Electric Vehicle Symposium and Exhibition 11

World Electric Vehicle Journal Vol. 7 - ISSN 2032-6653 - ©2015 WEVA Page WEVJ7-0458

References [15] I.J.M. Besselink, et al., Evaluating the TU/e Lupo

EL BEV performance, EVS27, 2013, Barcelona.
[1] Nilsson M., Electric Vehicle: The phenomenon
of range anxiety, Lindholmen Science Park, [16] OpenStreetMap, http://en.wikipedia.org/wiki/
Sweden, June 2011. OpenStreetMap, accessed on 2014-12-17.

[2] Hongwen He, et al., A method for identification [17] Shuttle Radar Topography Mission,
of driving patterns in hybrid electric vehicles http://en.wikipedia.org/wiki/shuttle_Radar_Topogr
based on a LVQ neutral network, Energies, ISSN aphy_Mission, accessed on 2014-12-17.
1996-1073, 5 (2012), 3363-3380. [18] SRTM, http://wiki.openstreetmap.org/wiki/ SRTM,
[3] Valera J.J, et al., Driving cycle and road grade accessed on 2014-12-17.
on-board predictions for the optimal energy [19] Speed limits in the Netherlands,
management in EV-PHEVs, EVS27, 2013, http://en.wikipedia.org/wiki/Speed_limits_in_the_
Barcelona. Netherlands, accessed on 2015-01-21.
[4] Frank A., et al., Drive cycle prediction and [20] Speed limit, http://en.wikipedia.org/wiki/
energy management optimization for hybrid Speed_limit, accessed on 2014-12-17.
hydraulic vehicles, IEEE Transaction on
vehicular technology, ISSN 0018-9545, [21] Gilles Reymond, et al., Role of lateral acceleration
62(2013), 3581-3592. in curve driving: driver model and experiments on
a real vehicle and a driving simulator, Human
[5] Ravi Shankar, et al., Method for estimating the Factors, ISSN 0018-7208, 43(3): 483-495.
energy consumption of electric vehicles and
plug-in hybrid electric vehicles under real-world
driving conditions, IET Intelligent Transport Authors
System, ISSN 1751-956X, 7(2013), 138-150. Jiquan Wang is a PhD at the
[6] Peter Ondruska, et al., Probabilistic attainability Eindhoven University of Technology,
maps: efficiently predicting driver-specific Department of Mechanical
electric vehicle range, IEEE Intelligent vehicles engineering, Dynamics and Control.
symposium (IV), 2014, Dearborn, Michigan, Current research activities include
USA. electric vehicle energy modelling and
driver support building.
[7] Christophe Moure, et al., Range estimator for
electric vehicles, EVS27, Barcelona, 2013. Dr. Ir. Igo Besselink is an assistant
professor at the Eindhoven University
[8] I.J.M. Besselink, et al., Design of an efficient,
of Technology, department of
low weight battery electric vehicle based on a
Mechanical engineering, Dynamics
VW Lupo 3L, EVS25, Shenzhen China, 2010.
and Control. Research activities
[9] P.F. van Oorschot, et al. Realization and control include tyre modelling, vehicle
of the Lupo EL electric vehicle, EVS26, dynamics and electric vehicles.
California USA, 2012.
Prof. Dr. Henk Nijmeijer is a full
[10] Jiquan Wang, et al. Evaluating and modeling the
professor at the Eindhoven University
energy consumption of the TU/e Lupo EL BEV,
of Technology, department of
Fisita 2014, Maastricht, the Netherlands.
Mechanical engineering, Dynamics
[11] Javier A. Oliva, et al., A model-based approach and Control. Current research
for predicting the remaining driving range in activities include non-linear dynamics
electric vehicles, Annual Conference of the and control.
Prognostics and Health Management Society
2013, New Orleans, USA.
[12] A. Picard, et al., Revised formula for the density
of moist air (CIPM-2007), Metrologia, ISSN
0026-1394, 45(2008) 149-155.
[13] Tony Sandberg, Heavy truck modeling for fuel
consumption simulations and measurements,
Linkoping University, thesis No. 924, Sweden.
[14] Paul Kokke, et al., TU/e Lupo EL powertrain
efficiency experiments, EEVC 2014, Brussels,
Belgium.

EVS28 International Electric Vehicle Symposium and Exhibition 12