energies
Article
High Performance Electric Vehicle Powertrain Modeling,
Simulation and Validation
Feyijimi Adegbohun 1 , Annette von Jouanne 1, *, Ben Phillips 2 , Emmanuel Agamloh 1 and Alex Yokochi 2






Citation: Adegbohun, F.; von
Jouanne, A.; Phillips, B.; Agamloh, E.;
Yokochi, A. High Performance
Electric Vehicle Powertrain Modeling,
Simulation and Validation. Energies
2021, 14, 1493. https://doi.org/
10.3390/en14051493
Academic Editor: Hrvoje Pandzic
Received: 16 February 2021
Accepted: 1 March 2021
Published: 9 March 2021
Publisher’s Note: MDPI stays neutral
with regard to jurisdictional claims in
published maps and institutional affil-
iations.
Copyright: © 2021 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
Energies 2021, 14, 1493. https://doi.org/10.3390/en14051493
1 Department of Electrical and Computer Engineering, Baylor University, Waco, TX 76798, USA;
jimi_adegbohun@baylor.edu (F.A.); Emmanuel_Agamloh@baylor.edu (E.A.)
2 Department of Mechanical Engineering, Baylor University, Waco, TX 76798, USA;
ben_phillips@baylor.edu (B.P.); Alex_Yokochi@baylor.edu (A.Y.)
* Correspondence: annette_vonjouanne@baylor.edu
Abstract: Accurate electric vehicle (EV) powertrain modeling, simulation and validation is paramount
for critical design and control decisions in high performance vehicle designs. Described in this paper
is a methodology for the design and development of EV powertrain through modeling, simulation
and validation on a real-world vehicle system with detailed analysis of the results. Although simula-
tion of EV powertrains in software simulation environments plays a significant role in the design
and development of EVs, validating these models on the real-world vehicle systems plays an equally
important role in improving the overall vehicle reliability, safety and performance. This modeling
approach leverages the use of MATLAB/Simulink software for the modeling and simulation of
an EV powertrain, augmented by simultaneously validating the modeling results on a real-world
vehicle which is performance tested on a chassis dynamometer. The combination of these modeling
techniques and real-world validation demonstrates a methodology for a cost effective means of
rapidly developing and validating high performance EV powertrains, filling the literature gaps in
how these modeling methodologies can be carried out in a research framework.
Keywords: electric vehicle; chassis dynamometer; drive cycle; modeling
1. Introduction
The electric vehicle (EV) market continues to grow, with over seven million EVs on the
road worldwide [1]. The International Energy Agency (IEA) forecasts that these numbers
will exceed 125 million by 2030 [2]. EVs offer increased efficiency and energy savings,
reduced emissions (especially when the electricity is being generated from renewable
resources), higher performance and a greater diversity of fuel choices for transportation.
For these reasons, the world is also seeing an increased trend in high performance EVs,
and in converting traditional internal combustion engine (ICE) vehicles to EVs. The goals
of high performance EVs include addressing the challenges of cost, range anxiety, charging
time and infrastructure, battery state of health, and impacts of vehicle to grid (V2G)
operations, while increasing transportation safety [3]. According to the U.S. Department of
Transportation (DoT), 94% of all serious car accidents occur due to human error [4]. Thus,
the opportunities that high performance EVs provide for increased levels of autonomous
EV operation further enable increased transportation safety, while also adding levels of
convenience and independence for the world’s aging population [5].
Understanding the operation of the EV under a variety of driving and environmental
conditions is critical to optimizing vehicle performance, vehicle health and vehicle safety.
Some of the conditions that could affect vehicle performance such as temperature, road
conditions, road grade/elevation, aggressive/conservative driving, etc. need to be vali-
dated for vehicle reliability and performance estimation [6]. To estimate these metrics and
make the critical design and control decisions needed during vehicle design and vehicle
validation, the EV powertrain must be modeled accurately, and simulated and analyzed in
https://www.mdpi.com/journal/energies
Energies 2021, 14, 1493 2 of 22

a scalable fashion [7–10]. Furthermore, these models must be flexible and robust enough to
be tuned based on the real-world data to provide improved vehicle serviceability once the
vehicle is in the field.
Vehicle modeling occurs in different forms, such as mathematical models, steady state
models, multi-physics domain physical modeling, dynamics and transient modeling [11],
with significant advancements in recent years. Vehicle modeling can also occur at different
levels of fidelity and precision, such as system level modeling, component modeling or
sub-system level modeling. The selection of the model type and modeling tools is indicative
of the kinds of questions that the model is expected to answer and the forms of analyses
that are intended to be performed on the modeling results. However, the reliance on
simulation-based platforms alone for the validation of vehicle models is insufficient, and
the development and testing of vehicle prototypes for public road testing could be unsafe,
impractical and sometimes very costly. Therefore, a balance between simulation and some
level of real-world testing in a scalable fashion is necessary in the development of high
performance EV powertrains.
MATLAB/Simulink software is a tool capable of modeling complete EV powertrains
of different levels of fidelity and detail and has become an invaluable modeling platform.
This software features a variety of shipped sample models for simulation of pure battery
electric as well as hybrid electric vehicles of different configurations and types [12–15].
The MATLAB/Simulink platform supports many add-ons which have been used in vehi-
cle modeling, such as SimPowerSystems and SimDriveline [15], Advisor [13], Simscape,
Powertrain Blockset, etc. Simulink supports an equation-based modeling approach, a data-
driven modeling approach, as well as a physical modeling approach for vehicle modeling.
Simulink also supports code generation for hardware testing and deployment, testing and
analysis frameworks for test case management and report generation. MATLAB/Simulink
models have been widely studied in the literature [10–15], however, methodologies for
validating these models within the real-world environment in a research setting has not
been addressed adequately in existing literature.
In this paper, the authors describe the modeling and simulation steps for high perfor-
mance EVs by defining key vehicle specifications and then developing an equation-based
model of a battery EV in MATLAB/Simulink software. The modeling goal is to test
for the efficiency and performance of the EV and the results of the model are validated
by real-world tests on the target EV platform. In Section 2, a summary of vehicle mo-
tion dynamics is presented and the modeling methodology is described where a generic
equation-based model that ships with the MATLAB/Simulink software [16] is significantly
modified based on the specifications and initial testing conducted on the target vehicle
platform. In Section 3, the procedure and the detailed results of the chassis dynamometer
performance testing conducted on the modeled vehicle is described. Section 4 presents the
results of the model and detailed analysis. Section 5 concludes the paper including the
primary findings and statistical results.

2. Equation-Based Modeling Summary

For equation-based modeling, the dynamics of vehicle motion represented in drive
cycles and the vehicle systems must be formulated and then implemented in the form
of subsystems, blocks and signals within Simulink. The next sections summarize the
equations that govern vehicle motion and vehicle systems needed for modeling.

2.1. Dynamics of Vehicle Motion

Newton’s second law states that the acceleration of an object is proportional to the net
force exerted on it. In other words, an object accelerates when the net force exerted on it is
nonzero. Similarly, a vehicle moves due the force of the propulsion unit (the powertrain)
overcoming the gravitational force of the vehicle exerted on the roadway, the air resistance
and the tire rolling resistance. The acceleration and speed at which the vehicle moves
are dependent on the power delivered to the wheel by the powertrain, the curb mass
 Energies 2021, 14, 1493 3 of 33
Energies 2021, 14, 1493 3 of 22

moves are dependent on the power delivered to the wheel by the powertrain, the curb
of the of
mass vehicle (including
the vehicle all of all
(including theofcomponents and passengers
the components on board),
and passengers the condition
on board), the condi-
of the roadway and the aerodynamics of the vehicle on the roadway.
tion of the roadway and the aerodynamics of the vehicle on the roadway. When When a vehicle is
a vehicle
accelerating, the vehicle is subject to the forces described in Equation (1) [17]:
is accelerating, the vehicle is subject to the forces described in Equation (1) [17]:
dv 𝑑𝑣
k m𝑘m 𝑚 xT = =
FTR𝐹 − −
FRL𝐹 (1)
(1)
dt 𝑑𝑡
whereFTR
where 𝐹 represents
representsthe the tractive
tractive force
force which
which thethe electric
electric machine
machine delivers,
delivers, 𝐹 repre-
FRL represents
sents road load forces, m represents vehicle mass, 𝑣
road load forces, m represents vehicle mass, v xT is the velocity in the tangential tangential
is the velocity in the direction
direction
of of the kroadway,
the roadway, m accounts 𝑘 foraccounts for the inertia
the rotational rotational inertia coefficient
coefficient of thevehicle’s
of the onboard onboard
vehicle’smass,
rotating rotating k m m dvand
andmass, dt
xT 𝑘 𝑚
ccounts for ccounts
forces for
that forces
propel that
the propel
vehicle the vehicle
forward. F forward.
RL is the
𝐹 is the summation
summation of the gravitational force FgxT acting
of the gravitational forceon𝐹 the acting
vehicleon in the downward
vehicle in thedirection,
down-
rolling resistancerolling
ward direction, of the resistance
vehicles tires Frollvehicles
of the , and the aerodynamic
tires 𝐹 , and the drag FAD as described
aerodynamic drag 𝐹in
Equations
as described (2)–(5) [17]:
in Equations (2)–(5) [17]:
FRL = FgxT + Froll + FAD (2)
𝐹 = 𝐹 + 𝐹 + 𝐹 (2)
FgxT = m × g × sinβ (3)
𝐹 = 𝑚 × 𝑔 × 𝑠𝑖𝑛𝛽  (3)
Froll = m × g × cosβ Co + C1 v xT 2 for v xT > 0 (4)
𝐹 = 𝑚 × 𝑔 × 𝑐𝑜𝑠𝛽(𝐶 + 𝐶 𝑣 ) for 𝑣 > 0 (4)
1
FAD𝐹 == ρC 𝜌𝐶D A𝐴 F(𝑣(v xT ++ 𝑣v0))2 forfor v xT𝑣 > >0 0 (5)
(5)
2
whereC𝐶o and
where andC1𝐶areare coefficients
coefficients of of rolling
rolling resistance,
resistance, thethe grade
grade in in degrees
degrees is is given
given byby β, 𝛽,
g
is𝑔 the
is the acceleration
acceleration duedue to gravity,
to gravity, air density
air density is ρ,isC𝜌,
D is𝐶 the is coefficient
the coefficient A F is𝐴theis
of drag,
of drag,
frontal
the frontal and vand
area,area, 𝑣 is
0 is the measured head wind
the measured head velocity
wind velocityahead ahead
of the vehicle. Equation
of the vehicle. (5)
Equa-
defines
tion (5)the aerodynamic
defines force relationship
the aerodynamic to the aforementioned
force relationship constants and
to the aforementioned variables.
constants and
Figure 1 is Figure
variables. referred1 to
is as a free body
referred to as diagram
a free body or point
diagrammassor diagram and itdiagram
point mass describesandtheit
forces thatthe
describes act forces
on thethat
EV onactaon roadway
the EV [18].
on a roadway [18].

Figure1.1.Free
Figure Freebody
bodydiagram
diagramof
offorces
forcesacting
actingon
onan
anEV.
EV.Adapted
Adaptedfrom
from[18],
[18],IEEE:
IEEE:2020.
2020.

The
The tire-road
tire-road interaction
interaction of of the
the vehicle
vehicle andand the
the traction
traction properties
properties ofof the
thetire-road
tire-road
interface
interface are fundamental to the dynamics of a vehicle. The traction torque fromfrom
are fundamental to the dynamics of a vehicle. The traction torque the
the pro-
propulsion
pulsion system is converted into a traction force through the interaction between the
system is converted into a traction force through the interaction between the
pneumatic
pneumatic tiretire and
and the
the road
road surface
surface atat the
the tire-road
tire-road interface
interface [17]
[17] illustrated
illustrated in
in Figure
Figure 2.2.
The
Thespeed
speedofoftire
tireand
andthethelongitudinal
longitudinal speed
speedof of
thethe
vehicle vary
vehicle in magnitude
vary in magnitudeandand
direction
direc-
and
tionare
anda are
function of theofforces
a function acting
the forces on the
acting ontire,
the including the vertical
tire, including loadload
the vertical forceforce
that that
the
vehicle body exerts on the wheel, Fz , and the longitudinal forces exerted on the tire at the
the vehicle body exerts on the wheel, 𝐹 , and the longitudinal forces exerted on the tire at
tire-road interface, Fx . The rolling speed of the tire is related to the tire angular velocity
the tire-road interface,𝐹 . The rolling speed of the tire is related to the tire angular velocity
and wheel radius rw and is given by Equation (6) [16]:
and wheel radius 𝑟 and is given by Equation (6) [16]:
𝑣 = r=w Ω
vtire 𝑟 Ω (6)
(6)
 Energies 2021, 14, 1493 4 of 22
Energies 2021, 14, 1493 4 of 33

Figure
Figure2.2.Speed
Speedand
andforces
forcesin
inthe
thetire-road
tire-roadcontact
contact patch
patch area.
area. Adapted
Adapted from
from [16],
[16], MathWorks:
MathWorks: 2020.
2020.
The ratio of the longitudinal velocities of the vehicle Vx and the tire Vtire , is defined as
The ratio
the wheel slipofs given
the longitudinal
by velocities of the vehicle 𝑉 and the tire 𝑉 , is defined
as the wheel slip s given by Vx
s = 1− (7)
Vtire
𝑉
𝑠 =1− (7)
𝑉 by
During braking, the slip of the vehicle is given
During braking, the slip of the vehicle is given
V by
s = 1 − tire (8)
V𝑉x
𝑠 =1− (8)
𝑉
2.2. Drive Cycles for Modeling and Testing
In practice, the environmental protection agency (EPA) sets the standards for allowable
2.2. Drive Cycles
emmisions andfor Modelingrequirements
efficiency and Testing for automakers and original equipment manu-
In practice,
facturers (OEMs).the Aenvironmental
critical part of protection
the design agency (EPA) sets
philosophy the vehicle
in any standards for allow-
development
able emmisions
program and efficiency
is meeting requirements
these stringent efficiency forrequirements.
automakers and For original equipment
ICE vehicles, the EPA man-
has
historically(OEMs).
ufacturers leaned on the Single
A critical partCycle Test
of the (SCT)philosophy
design in determiningin anyemmisions and efficiency
vehicle development
ratings of
program is new vehicles
meeting theseand enforcing
stringent the emmision
efficiency regulations.
requirements. For ICESCTs, however,
vehicles, the EPA arehas
not
suitable forleaned
historically characterizing
on the Singlethe effects
Cycle Testof ambient
(SCT) intemperatures and other ancillary
determining emmisions and efficiencyloads
that anofEV
ratings newmight be subject
vehicles to in thethe
and enforcing real world. A
emmision new testing
regulations. procedure
SCTs, however, referred
are not to
as the Multi-Cycle
suitable Test has
for characterizing thetherefore
effects ofbeen adopted
ambient in the efficiency
temperatures and other testing conducted
ancillary loads
by EPA
that an EV which
mightreduces
be subjectthe to
testing
in thetimereal for an EV
world. by over
A new 75%,
testing from over
procedure 18 h totojust
referred as
over 4 h [19]. Given the goal of the model described in this
the Multi-Cycle Test has therefore been adopted in the efficiency testing conducted work, which is to test bythe
efficiency
EPA whichofreduces
the target
the EV platform,
testing time fortheanauthors
EV byhave over devised
75%, from a shorter
over 18form
h to ofjustthe MCT
over 4
hwhich still captures
[19]. Given the goaltheof basic principles
the model of theintests
described thisused
work, bywhich
the EPA.is toThis
testtest
the developed
efficiency
ofincludes
the targeta combination
EV platform,ofthe theauthors
US06, which is an aggressive
have devised a shorterdrive
form cycle,
of theand
MCT the Highway
which still
captures the basic principles of the tests used by the EPA. This test developed includes of
Fuel Economy Test (HWFET). The US06 cycle is an 8-mile route with an average speed a
48.4 mph, with
combination maximum
of the US06, which speedsis anofaggressive
80.3 mph and drivetotal duration
cycle, and theof 596 s. The
Highway Fuel HWFET
Econ-
covers
omy Testa (HWFET).
distance ofThe 10.26
US06miles,
cyclewith
is anan average
8-mile routespeed
with ofan48.3 mphspeed
average and aofduration
48.4 mph, of
765 s providing a total MCT test duration of ~22 min and 18.26
with maximum speeds of 80.3 mph and total duration of 596 s. The HWFET covers a dis- miles covered in distance.
Figure
tance of 310.26
is a plot
miles,ofwith
the combined
an averageMCT speeddrive cycle
of 48.3 mph used
andina duration
this workoffor 765efficiency
s providing test
modeling and validation.
a total MCT test duration of ~22 min and 18.26 miles covered in distance. Figure 3 is a plot
of the combined MCT drive cycle used in this work for efficiency test modeling and vali-
dation.
Energies 2021, 14, 1493 5 of 22
Energies 2021, 14, 1493 5 of 33

Figure 3. Multi-Cycle Test (MCT) drive cycle, reducing the testing time of EVs. The total time and
Figure Multi-Cycle
distance3.for Test~22
this MCT test was (MCT) drive
min and 18.26cycle, reducing
miles (29.3 km). the testing time of EVs. The total time and
distance for this MCT test was ~22 min and 18.26 miles (29.3 km).
2.3. EV Powertrain Modeling and Simulation
2.3. EV Powertrainmodeling
EV powertrain Modelingandand Simulation
simulation begins with the determination of key design
specifications as detailed
EV powertrain in Table 1,and
modeling adapted from the spec
simulation sheetwith
begins of a Chevrolet Bolt EV.
the determination of key design
These specs are then used in an equation-based model of an EV using MATLAB/Simulink,
specifications as detailed in Table 1, adapted from the spec sheet of a Chevrolet
and the modeling results are then validated using a real-world vehicle performance tested
Bolt EV.
These specsdynamometer.
on a chassis are then used in an equation-based model of an EV using MATLAB/Simulink,
and the modeling results are then validated using a real-world vehicle performance tested
Table 1. Vehicle Specification for Chevrolet Bolt EV.
on a chassis dynamometer.
Description Value
Initial1.Acceleration
Table (0–60 mph) for Chevrolet Bolt 7.5
Vehicle Specification EV.s
Curb weight 1616.15 kg
Description
Motor power 200 Value
hp/150 kW
Motor torque 266 lb.ft/360 Nm
Initial Acceleration (0–60 mph) 7.5 s
Final drive ratio 7.05:1
Curb weight
Energy efficiency 300 1616.15
Wh/milekg
Battery
Motor capacity
power 53 kWh
200 hp/150 kW
Top speed 93 mph
Motor torque 266 lb.ft/360 Nm
Aerodynamic drag coefficient 0.308
Final drive ratio 7.05:1
2.4. Glider Model
Energy efficiency 300 Wh/mile
The equation-based modeling approach in Simulink involves the representation of
Battery
each capacity
vehicle 53 kWhblocks connected to each
component or subsystem as a series of equation
other through signals that are calculated and updated at each time step. Figure 4 is an
Top speed 93 mph
overview of the equation-based model in Simulink to meet the EV specifications previ-
Aerodynamic
ously described. drag coefficient
The glider 0.308model of the vehicle dynam-
model represents the point mass
ics previously described. It sums up the forces acting on vehicle body represented by a
point mass, as described in Equations (1)–(5). Table 2 is a summary of parameters used
2.4.
for Glider Model of vehicle dynamics and determining the vehicle speed at each
the calculation
The equation-based modeling approach in Simulink involves the representation of
each vehicle component or subsystem as a series of equation blocks connected to each
other through signals that are calculated and updated at each time step. Figure 4 is an
overview of the equation-based model in Simulink to meet the EV specifications previously
described. The glider model represents the point mass model of the vehicle dynamics
previously described. It sums up the forces acting on vehicle body represented by a point
mass, as described in Equations (1)–(5). Table 2 is a summary of parameters used for
the calculation of vehicle dynamics and determining the vehicle speed at each timestep.
Figure 5 is a closer look at the glider model subsystem within the Simulink model where
Equations (1)–(5) are implemented by taking in an input of tractive force and subtracting
the summation of the road load forces to get the inertial force, which is converted into an
Energies 2021, 14, 1493 6 of 33

timestep. Figure 5 is a closer look at the glider model subsystem within the Simulink
Energies 2021, 14, 1493 6 of 22
model where Equations (1)–(5) are implemented by taking in an input of tractive force
and subtracting the summation of the road load forces to get the inertial force, which is
converted into an acceleration based on the vehicle mass and then integrated over time to
form the output
acceleration basedofonthethe
model, vehicle
vehicle speed,
mass and thenwhich is thenover
integrated fed time
backtotoform
the driver control
the output of
subsystem
the in Figure
model, vehicle 6. Figure
speed, which7 isisthen
the energy
fed backand power
to the analysis
driver controlofsubsystem
glider model subsys-
in Figure 6.
tem. Figure
Figure 8 isenergy
7 is the a blockand
diagram
powerwhich represents
analysis of gliderthe calculations
model subsystem.for determining
Figure 8 is avalues
block
such as distance,
diagram tractive energy
which represents consumed,
the calculations for tractive power,
determining velocity
values such braking energy,
as distance, etc.
tractive
energy consumed,
Note that simulationtractive
blockpower,
inputsvelocity braking
are indicated energy,
with etc.and
yellow Note that outputs
block simulation
areblock
indi-
inputs are indicated
cated with red. with yellow and block outputs are indicated with red.

Figure 4. Simulink
Figure 4. Simulink equation-based
equation-based EV
EV model.
model. Adapted
Adapted from
from [16],
[16], MathWorks:
MathWorks:2020.
2020.

Energies 2021, 14, 1493 Table 2. Glider Model Parameters Adapted from [16], MathWorks: 2020. 7 of 33

Parameter Unit Description Value

ρ Table 2. Glider kg/m3Parameters
Model Air[16],
Adapted from Density
MathWorks: 2020. 1.23
Cd
Parameter -
Unit Drag coefficient
Description 0.38
Value
A𝜌f kg/m
m2 Airfrontal
Vehicle DensityArea 1.23
2.1
V𝐶 m/s- Drag coefficient
Vehicle Speed 0.38
-
𝐴 m2 Vehicle frontal Area 2.1
a m/s Vehicle acceleration -
𝑉 m/s Vehicle Speed -
m𝑎i kg
m/s Vehicle
Vehicleinertial mass
acceleration 1678.30
-
𝑚
m kg
kg Vehicle inertial
Vehicle Massmass 1678.30
1616.15
𝑚
g kg
m/s2 Vehicle Mass
Gravity 1616.15
9.81
𝑔 m/s Gravity 9.81
θ Degrees Road angle 0
𝜃 Degrees Road angle 0
𝐶rr
C -- Rolling
Rolling resistance
resistance coefficient
coefficient 0.01
0.01

Figure 5.
Figure 5. Vehicle
Vehicle body
body glider
glider model
model Adapted
Adapted from
from[16],
[16],MathWorks:
MathWorks:2020.
2020.
 Energies 2021, 14, 1493 8 of 33

Energies 2021, 14, 1493 7 of 22

Energies 2021, 14, 1493 9 of 33

Figure 6. Vehicle
Figure 6. Vehicle glider
glider model subsystem Adapted
Adapted from
from [16],
[16], MathWorks:
MathWorks: 2020.
2020.

Energies 2021, 14, 1493 10 of 33

Figure7.7.Energy
Figure Energyand
andpower
poweranalysis
analysisofofthe
theglider
glidermodel
modelsubsystem
subsystemAdapted
Adaptedfrom
from[16],
[16],MathWorks:
MathWorks:2020.
2020.

Figure8.8.Driver
Figure Drivercontrol
controlsystem
systemblock
blockAdapted
Adaptedfrom
from[16],
[16],MathWorks:
MathWorks:2020.
2020.

2.5. Driver Model

The inputs to the driver control block are the drive cycle reference speed and the
vehicle speed feedback from the glider model block. The error between the two speeds is
fed back into the PID controller and the output is a driver command of either an acceler-
ator pedal position (APP%) or a brake pedal position (BPP%).
The modeling approach of the driver subsystem mimics a real-world driver control
operation, where the driver is the PID in this case and he/she observes the vehicle speed
at any given time and responds with a force on either the accelerator or brake pedal to
Energies 2021, 14, 1493 8 of 22

2.5. Driver Model

The inputs to the driver control block are the drive cycle reference speed and the
vehicle speed feedback from the glider model block. The error between the two speeds is
fed back into the PID controller and the output is a driver command of either an accelerator
pedal position (APP%) or a brake pedal position (BPP%).
The modeling approach of the driver subsystem mimics a real-world driver control
operation, where the driver is the PID in this case and he/she observes the vehicle speed at
any given time and responds with a force on either the accelerator or brake pedal to bring
the vehicle to a desired speed.

2.6. Brake System

The brake system takes an input of vehicle speed and BPP%. The BPP% signal is used
to determine the amount of braking force that the driver/PID is intending to apply to the
vehicle and distributes this force into a regenerative braking force output and frictional
Energies 2021, 14, 1493 brake output, based on the limitation of the motor and systems regen power. The vehicle
11 of 33
speed input is used to determine if the vehicle speed is above a certain speed threshold
needed for regen braking to be applied. Figure 9 is a detailed description of the vehicle
braking system block.

Figure 9.
Figure 9. Braking system block
Braking system block Adapted
Adapted from
from [16],
[16], MathWorks:
MathWorks: 2020.
2020.

2.7. Motor Model

Electric motors
motorsare
arecritical
criticalcomponents
componentsofof thethedrivetrains
drivetrains of EVs
of EVs[20].[20].
The The
electric mo-
electric
tor model is based on the power loss equation for electric motors described
motor model is based on the power loss equation for electric motors described in Figure in Figure 10,
where
10, wherethethe
motor
motoroutput
outputpower
power is is
defined
definedbybythethemotor
motorinput
inputpower
power minusminus the motor
losses. Table
Table33describes
describesthe
theparameter
parameter definitions
definitions forfor calculating
calculating the the motor
motor loss loss model.
model. The
The
motormotor
model model receives
receives an acceleration
an acceleration pedalpedal
position position
signalsignal
APP%APP%input and input and motor
motor speed
feedback,
speed these values
feedback, are then
these values areused
theninused
the motor torque torque
in the motor limiter limiter
sub-system described
sub-system de-
in Figure
scribed in11 to determine
Figure the maxthe
11 to determine torque output.output.
max torque Similarly, the regen
Similarly, thelimiter subsystem
regen limiter sub-
determines
system the max the
determines allowable regen torque
max allowable regenwhich
torque is awhich
specified
is a by a factorby
specified of the maximum
a factor of the
allowable torque
maximum of the
allowable motor.
torque Themotor.
of the net torque is then
The net torqueused to calculate
is then used tothe outputthe
calculate power
out-
using
put the motor
power usingloss
the model as described
motor loss model asby Equations
described by(9)–(11)
Equations[16].
(9)–(11) [16].

Pmot = Tmot ωmot (9)

Ploss = k c T 2 + k i ω + k ω ω 3 + C (10)
Pin = τ × ω + Ploss (11)
 𝑃 = 𝑇 𝜔 (9)
𝑃 = 𝑘 𝑇 +𝑘 𝜔+𝑘 𝜔 +𝐶 (10)
Energies 2021, 14, 1493 9 of 22
𝑃 = 𝜏×𝜔+ 𝑃 (11)

Energies 2021, 14, 1493 13 of 33

Figure 10. Motor loss model. Adapted from [16], MathWorks: 2020.
Figure 10. Motor loss model. Adapted from [16], MathWorks: 2020.
Table 3. Motor model parameters. Adapted from [16], MathWorks: 2020.

Parameter Table 3. Motor model

Unitsparameters. Adapted from [16], MathWorks: 2020.
Description Value
T
Parameter Nm
Units Maximum motor torque
Description 450
Value
𝑇
ω Nm
rad/s Maximum motor
Motor base torque
speed 450
834
𝜔
k rad/s
s Motor base speed
Motor loss constant 834
0.12
c Kgsm2
𝑘 Motor loss constant 0.12
ki KgJm Motor loss constant 0.01
𝑘 J Motor loss constant 0.01
kw Kg m2 Motor loss constant 1.2 × 10−−55
𝑘 Kg m Motor loss constant 1.2 × 10
C𝐶 W
W Motor
Motor loss
loss constant
constant 600
600

Figure 11. Electric

Figure 11. Electric motor
motor model
model with
with motor
motor loss
loss model
model calculations.
calculations. Adapted
Adapted from
from [16],
[16], MathWorks:
MathWorks: 2020.
2020.

2.8. Battery Model

2.8. Battery Model
The battery model is developed based off a constant voltage source, Voc , in series with
The battery model is developed based off a constant voltage source, 𝑉 , in series
an internal resistance, Rint = 0.1 ohms, as described by Figure 12. This model relies on a
with an internal resistance, 𝑅 = 0.1 ohms, as described by Figure 12. This model relies
power loss calculation to determine the battery output current as described by Equations
on a power loss calculation to determine the battery output current as described by Equa-
(12)–(15). Figure 13 describes the battery model and loss model calculation in the Simulink
tions (12)–(15). Figure 13 describes the battery model and loss model calculation in the
Electric Vehicle model [16]. The state of charge (SOC) of the battery is determined by
Simulink Electric Vehicle model [16]. The state of charge (SOC) of the battery is deter-
mined by integrating the power over time and comparing to the energy capacity variable
set at the initialization of the simulation.
𝑃 =𝑃 +𝑃 (12)
𝑃 = 𝐼𝑉 (13)
 Energies 2021, 14, 1493 10 of 22

integrating the power over time and comparing to the energy capacity variable set at the
initialization of the simulation.

Pideal = Pactual + Ploss (12)

Pideal = IVoc (13)

2
Energies 2021, 14, 1493 Pactual = IVoc − I Rint 14 of(14)
33

Ploss = I 2 Rint (15)

Energies 2021, 14, 1493 15 of 33

Figure12.
Figure Batterymodel
12.Battery modelwith
withinternal
internalresistance.
resistance.Adapted
Adaptedfrom
from[16],
[16],MathWorks:
MathWorks:2020.
2020.

Figure 13.
Figure 13. Battery
Battery model
model with
with SOC
SOC and
and power
power loss
loss calculations.
calculations. Adapted
Adapted from
from [16],
[16], MathWorks:
MathWorks: 2020.
2020.

2.9. Driveline Model

The driveline model describes the driveline losses that are present in the the EV
EV due
due to
to
gear reduction. The electric motor spins at higher speeds than the wheel and, as a result,
gear
gear reduction
reductionisisused
usedtotoreduce
reducethe speed
the speedat at
thethe
wheel while
wheel multiplying
while the the
multiplying torque at the
torque at
wheel. ThisThis
the wheel. geargear
reduction is notislossless
reduction and isand
not lossless given
is given byin𝑇the Equation
by Tloss (16), where
in the Equation (16),
where 𝐺 is the gear ratio of the rear reduction assembly, 𝑟 is the radius of the wheel,
𝐹 in net tractive force at the wheel and 𝐹 is the braking force. Figure 14 is the Simulink
model of the driveline and accompanying loss calculations.
(𝑇 −𝑇 )𝐺
𝐹 = − 𝐹 (16)
Energies 2021, 14, 1493 11 of 22

G is the gear ratio of the rear reduction assembly, rw is the radius of the wheel, Ftr in net
tractive force at the wheel and FBr is the braking force. Figure 14 is the Simulink model of
the driveline and accompanying loss calculations.

( Tmotor − Tloss ) G
Energies 2021, 14, 1493 Ftr = − FBr 16(16)
of 33
rw

Figure 14.Driveline
Figure14. Drivelineloss
lossmodel
modeland
andcalculations
calculationsAdapted
Adaptedfrom
from[16],
[16],MathWorks:
MathWorks:2020.
2020.

3.
3. EV
EV Chassis
Chassis Dynamometer
DynamometerTesting Testing
EVs are characterized on a chassis
EVs are characterized on a chassis dyno to determine
dyno their their
to determine driving performance
driving [19,21–23].
performance [19],
The vehicle
[21–23]. Theperformance testing typically
vehicle performance involves
testing typically drivingdriving
involves the vehicle through
the vehicle an EPA
through an
drive cycle, such as the MCT drive cycle in Figure 3. Prior to dyno testing
EPA drive cycle, such as the MCT drive cycle in Figure 3. Prior to dyno testing of an EV, of an EV, the
dyno mustmust
the dyno be calibrated to accurately
be calibrated represent
to accurately the road
represent loadload
the road forces thatthat
forces are exerted on
are exerted
aonvehicle. This involves road load coast down tests, where the vehicle is
a vehicle. This involves road load coast down tests, where the vehicle is driven on a driven on a flat
straight roadroad
flat straight and and
accelerating the vehicle
accelerating to ~60
the vehicle mph,
to ~60 thenthen
mph, shifting the the
shifting vehicle to neutral
vehicle to neu-
gear
tral gear to remove the regenerative braking capabilities and allowing the vehicledown
to remove the regenerative braking capabilities and allowing the vehicle coast coast
naturally to determine the road load forces acting on vehicle, which include
down naturally to determine the road load forces acting on vehicle, which include aero- aerodynamic
drag and rolling resistance. One such test was conducted in [18] and validated by putting
dynamic drag and rolling resistance. One such test was conducted in [18] and validated
the vehicle through similar drive cycles on the road and the dyno, collecting data through
by putting the vehicle through similar drive cycles on the road and the dyno, collecting
the vehicle on-board diagnostic (OBD) port and then comparing the power and energy
data through the vehicle on-board diagnostic (OBD) port and then comparing the power
of the vehicle over the drive cycles. Figure 15 is a speed plot of the road and dyno tests
and energy of the vehicle over the drive cycles. Figure 15 is a speed plot of the road and
including the difference in speed between both tests, showing a close correlation between
dyno tests including the difference in speed between both tests, showing a close correla-
speed over the entire drive cycle. Figure 16 is a plot of the power and energy for both
tion between speed over the entire drive cycle. Figure 16 is a plot of the power and energy
the road and dynamometer tests. Although the dynamic behavior of the vehicle during
for both the road and dynamometer tests. Although the dynamic behavior of the vehicle
the road and dyno tests varied, they followed a similar trend over time and the energy
during the road and dyno tests varied, they followed a similar trend over time and the
consumption was comparable.
energy consumption was comparable.
Energies 2021, 14, 1493 12 of 22
Energies 2021, 14, 1493 17 of 33

Energies 2021, 14, 1493 18 of 33

Figure
Figure 15.
15. Road
Road and
and dynamometer
dynamometer test
test vehicle
vehicle speed and difference
speed and difference Adapted
Adapted from
from [18],
[18], IEEE:
IEEE: 2020.
2020.

Figure 16. Road and dynamometer test vehicle power (left axis) and energy (right axis). Adapted from [18], IEEE: 2020.
Figure 16. Road and dynamometer test vehicle power (left axis) and energy (right axis). Adapted from [18], IEEE: 2020.

Oncethe
Once the dynamometer
dynamometer was wascalibrated,
calibrated,the
themodified
modifiedMCT MCTtest described
test describedin Figure 3 3
in Figure
was conducted on the vehicle using the dynamometer and the vehicle OBD to
was conducted on the vehicle using the dynamometer and the vehicle OBD to instrument instrument
thevehicle
the vehicleand
and collect
collect data
data for
for calculating
calculatingthe
theenergy
energyefficiency ofof
efficiency thethe
vehicle. Three
vehicle. runs
Three runs
of the same MCT test were carried out on the vehicle and described in Figure 17. The
of the same MCT test were carried out on the vehicle and described in Figure 17. The
instantaneous power and energy of each cycle was recorded during the drive cycles and
summarized in Figure 18. Finally, a vehicle acceleration performance test, known as the
wide-open throttle (WOT) test, was conducted on the vehicle to determine the vehicle
initial acceleration capability described in Figure 19. The dynamometer tests resulted in
an energy efficiency average of 297 Wh/mile and an initial acceleration of 7.5 s from 0–60
mph (96.44 kph).
Energies 2021, 14, 1493 13 of 22

Energies 2021, 14, 1493
instantaneous power and energy of each cycle was recorded during the drive cycles and
summarized in Figure 18. Finally, a vehicle acceleration performance test, known as the
wide-open throttle (WOT) test, was conducted on the vehicle to determine the vehicle 19 of 33
initial acceleration capability described in Figure 19. The dynamometer tests resulted in an
energy efficiency average of 297 Wh/mile and an initial acceleration of 7.5 s from 0–60 mph
(96.44 kph).

Energies 2021, 14, 1493 20 of 33

Figure
Figure 17.
17. MCT
MCT drive
drive cycle
cycle vehicle
vehicle speed
speed on
on chassis
chassis dynamometer
dynamometer Adapted
Adapted from
from [18],
[18], IEEE:
IEEE: 2020.
2020.

Figure 18. MCT drive cycle vehicle power (left axis) and energy (right axis) on chassis dynamometer Adapted from [18],
Figure 18. MCT drive cycle vehicle power (left axis) and energy (right axis) on chassis dynamometer Adapted from [18],
IEEE: 2020.
IEEE: 2020.
 Energies 2021, 14, 1493 21 of 33

Energies 2021, 14, 1493 14 of 22

(a) (b)
Figure
Figure 19.19.Wide-open
Wide-openthrottle
throttle(WOT)
(WOT)test
test(a)
(a)motor
motor torque
torque envelope
envelope (b)
(b) vehicle
vehiclespeed
speedAdapted
Adaptedfrom
from[18],
[18],IEEE: 2020.
IEEE: 2020.

TheWOT
The WOT test test showed
showed thethe vehicle
vehiclewaswascapable
capableofofattaining
attaininga atop
topofof
speed
speed of of
93 93
mph,
mph,
while almost instantly reaching the maximum torque treshold of
while almost instantly reaching the maximum torque treshold of 360 Nm as indicated 360 Nm as indicated in
inFigure
Figure 19.19.
In In
Summary,
Summary, thetheEV EV
chassis dynamometer
chassis dynamometer results serveserve
results as a strong basis for
as a strong basis
validating
for validatingthetheresults of aof
results vehicle model
a vehicle developed
model and simulated
developed and simulatedwithin a software
within en-
a software
vironment through
environment throughananiterative
iterativedesign
designprocess.
process.The
Thecontrolled
controlled test environment
test environment of of
using
using
a chassis dyno to emulate road load forces allows for ease and repeatability
a chassis dyno to emulate road load forces allows for ease and repeatability of testing cycles of testing
cycles during
during the vehicle
the vehicle development
development and vaidation
and vaidation stages.
stages. The real-world
The real-world driving
driving use use
cases
cases emulated on the dyno can then be validated in the sofware simulation
emulated on the dyno can then be validated in the sofware simulation environment and environment
andversa.
vice vice versa.
AlthoughAlthough
slightslight
errorerror
and and deviations
deviations can can be observed
be observed in dyno
in dyno vs. vs.
roadroad
tests
tests as indicated in Figures 15 and 16, these errors are negligible in most
as indicated in Figures 15 and 16, these errors are negligible in most cases, however, it is cases, however,
it is difficult
difficult and sometimes
and sometimes unsafeunsafe to perform
to perform some ofsome of the rigourous
the rigourous vehicle vehicle
driving driving
scenerios
scenerios needed for testing high performace vehicles on public
needed for testing high performace vehicles on public roads. The next section roads. The next section
presents
presents detailed results and analysis of the dyno test results and the simulations results.
detailed results and analysis of the dyno test results and the simulations results.

4.4.Validation
Validationof
ofEV
EV Model
Model Simulation
Simulation Results
Resultsand
andAnalysis
Analysis
Given the high repeatability of the chassis dynamometertesting,
Given the high repeatability of the chassis dynamometer thethe
testing, EVEVmodel
modelde-de-
veloped above was then validated against the dyno work, with careful
veloped above was then validated against the dyno work, with careful comparison ofcomparison of thethe
observations, including speed, distance, energy expended and power between
observations, including speed, distance, energy expended and power between modeling, modeling,
chassis dynamometer,
chassis dynamometer, and andon-road
on-roadmeasurements.
measurements. It isItimportant to note
is important that due
to note thattodue
the to
advanced dynamics controls on the real vehicle in comparison to the simplified
the advanced dynamics controls on the real vehicle in comparison to the simplified PID PID con-
trols modeled, the dynamic responses of the modeled vehicle and the real vehicle, such as
controls modeled, the dynamic responses of the modeled vehicle and the real vehicle, such
motor torque and power, are expected to vary to a certain degree. However, the steady
as motor torque and power, are expected to vary to a certain degree. However, the steady
state values, such as speed, distance and energy, do show a strong correlation.
state values, such as speed, distance and energy, do show a strong correlation.
Figure 20 is a plot of the vehicle speed during the simulation overlayed on the drive
Figure 20 is a plot of the vehicle speed during the simulation overlayed on the drive
cycle speed, which was recorded via the OBD port during dyno testing. From the plot in
cycle speed, which was recorded via the OBD port during dyno testing. From the plot in
Figure 20, we observe that the modeled vehicle is capable of following the drive cycle
Figure 20, we observe that the modeled vehicle is capable of following the drive cycle input
input closely within a very small margin of error while meeting the speed and torque
closely within a very small margin of error while meeting the speed and torque response
response requirements. Figure 21 is a plot of the battery SOC over the drive cycle period.
requirements. Figure 21 is a plot of the battery SOC over the drive cycle period. The model
The model was initialized to 95% SOC at the beginning of the model and ended at 85% at
was initialized to 95% SOC at the beginning of the model and ended at 85% at the end of
the end of the simulation. Recall that the modeled vehicle battery total capacity 53 kWh,
the simulation. Recall that the modeled vehicle battery total capacity 53 kWh, indicating
indicating that ~10% of the battery total capacity was depleted during the drive cycle.
that ~10% of the battery total capacity was depleted during the drive cycle.
Energies 2021, 14, 1493 15 of 22
Energies 2021, 14, 1493 22 of 33

Energies 2021, 14, 1493 23 of 33

Figure 20. Modeled vehicle speed (mph) over time (secs) compared to dyno drive cycle speed.
Figure 20. Modeled vehicle speed (mph) over time (secs) compared to dyno drive cycle speed.

Figure 21. Modeled battery SOC over time.

Figure 21. Modeled battery SOC over time.

Figure22
Figure 22isisaaplot
plotcomparison
comparisonofofthe
thebattery
batteryenergy
energyoverovertime
timebetween
betweenthe
thedyno
dynotest
test
andvehicle
and vehiclesimulation,
simulation,which
whichisisderived
derivedbybyintegrating
integratingthe thebattery
batterypower
powerover
overtime,
time,asas
seenininthe
seen themodel
modelblock
blockdiagrams
diagramsininFigure
Figure13.
13.Figure
Figure2323isisaaplot
plotcomparison
comparisonofofthe
thevehicle
vehicle
distance over time, which is an integration of the vehicle speed
distance over time, which is an integration of the vehicle speed over time. over time.
 Energies 2021, 14, 1493 24 of 33
Energies 2021, 14, 1493 16 of 22

Energies 2021, 14, 1493 25 of 33

Figure 22. Battery energy over time.

Figure 22. Battery energy over time.

Figure 23. Vehicle distance covered over time.

Figure 23. Vehicle distance covered over time.
4.1.Energy
4.1. Energy Efficiency
Efficiency
The energy
The energy efficiency
efficiencyofof
the modeled
the modeled EV EV
can can
be derived by theby
be derived dividing the battery
the dividing the battery
energy output by the vehicle distance covered. The total energy delivered by the battery
energy output by the vehicle distance covered. The total energy delivered by the battery
model was 5.249 kWh, while the distance traveled was 17.85 miles, which resulted in an
model was 5.249 kWh, while the distance traveled was 17.85 miles, which resulted in an
energy efficiency of 294 Wh/mile compared to the dynamometer tests results, which had
energy efficiency
an overall of 294 Wh/mile
energy efficiency average compared to the
of 297 Wh/mile. dynamometer
The testserror
root-mean-square results, which had
(RMSE)
an
between the modeled EV results and the dyno results was calculated for a final value of (RMSE)
overall energy efficiency average of 297 Wh/mile. The root-mean-square error
between the modeled
0.2 for measured energyEVconsumption
results andover
the dyno results
time. The was
small calculated
error forina the
deviations final value of 0.2
model
for
canmeasured energy
be attributed consumption
to negligible over
transient time.
errors Thethe
within small error deviations in the model can
model.
be attributed to negligible transient errors within the model.
4.2. Motor Torque and Power
Figure 24 is a plot of the vehicle power over the vehicle speed range, indicating the
power regions the motor operated over the drive cycle. From the plot, we can determine
the vehicle model rated speed is around 40 mph, and then, between 40 mph to 70 mph,
Energies 2021, 14, 1493 17 of 22

4.2. Motor Torque and Power

Figure 24 is a plot of the vehicle power over the vehicle speed range, indicating the
power regions the motor operated over the drive cycle. From the plot, we can determine
the vehicle model rated speed is around 40 mph, and then, between 40 mph to 70 mph, the
vehicle is in the constant power region, with a concentration of motor operating points in
this region. The diagonal line in the power plot is attributed to the initial acceleration of
the vehicle from 0 mph to ~40 mph at the very beginning of the drive cycle and towards
the 600 s mark. Figure 25 is a scatter plot of motor torque over the vehicle speed change
from the simulated drive cycle and the dyno test. From these plots, we draw similarities
between the two plots, where the maximum torque (simulation) and the peak torque (dyno)
occur between 0 and 40 mph speed range, the constant torque region, and then the gradual
ramp down in torque as the speed increases. Similar to the observations in Figure 24, there
is a concentration in operating point between 40 mph to 70 mph in both plots, which is
consistent with the drive cycle speeds in Figure 20. There is however a significant difference
Energies 2021, 14, 1493 in the dynamic behavior of the dyno motor torque compared to the simulation that26can of 33
be attributed to the sophisticated motor control schemes implemented in the real vehicle
compared to the simplified PID controller that is modeled. These sophisticated controls are
implemented
responsiveness in the real-world
of the vehicle. vehicle to allow
For example, in for
theease
dynooftorque
drivability
plots,and
weresponsiveness
observed a con-
oftinuity
the vehicle.
in the torque values at all times without sudden jumps or dropscontinuity
For example, in the dyno torque plots, we observed a in theas
in torque value
torque values at all times without sudden jumps or drops in torque value as speed
speed changes compared to the simulated torque -speed plots. This is due to the effective changes
compared
and careful to avoidance
the simulated torque
of torque -speed
ripples byplots. This controller
the motor is due to the effective and
implemented on careful
the real-
avoidance of torque ripples by the motor controller implemented on the
world vehicle. The simulated vehicle model’s torque controls calculate the maximum real-world vehicle.
The simulated
torque based on vehicle model’s
current vehicletorque
speedcontrols calculate the
and the requested maximum
APP% at each torque based
time step. on
A more
current vehicle speed and the requested APP% at each time step. A more advanced
advanced model will take into consideration the dynamic response of the vehicle and sim- model
will take into consideration the dynamic response of the vehicle and simulate a smoother
ulate a smoother transition in torque applied, similar to the dyno vehicle response. The
transition in torque applied, similar to the dyno vehicle response. The expected torque
expected torque envelope regions of the torque plots captures the expected torque-speed
envelope regions of the torque plots captures the expected torque-speed envelope however,
envelope however, showing an attainment of maximum traction and regen torque at
showing an attainment of maximum traction and regen torque at lower speeds and reduced
lower speeds and reduced torque at higher speeds, as indicated in the torque over time
torque at higher speeds, as indicated in the torque over time plots in Figure 26 as well.
plots in Figure 26 as well.

Figure 24. Motor power vs. speed for MCT drive cycle simulation.
Figure 24. Motor power vs. speed for MCT drive cycle simulation.
Energies 2021, 14, 1493 18 of 22
Figure 24. Motor power vs. speed for MCT drive cycle simulation.

Energies 2021, 14, 1493 28 of 33

(a) Simulation (b) Dyno
Figure 25. Motor torque vs. speed (a) simulation (b) dyno test.

Figure 26. Motor torque vs. time plot.

Figure 26. Motor torque vs. time plot.
4.3. Loss Model
4.3. Loss Model
The battery energy at the battery terminal is 5.229 kWh as indicated in Figure 22 and
The battery energy at the battery terminal is 5.229 kWh as indicated in Figure 22 and
is consistent with the energy consumption during dyno testing. The total tractive energy
is consistent with the energy consumption during dyno testing. The total tractive energy
consumption of the vehicle over the drive cycle is measured at 3.18 kWh as indicated in
consumption of the vehicle over the drive cycle is measured at 3.18 kWh as indicated
Figure 27, while the motor energy losses are measured at 0.71 kWh as indicated in Figure
in28,
Figure 27, while the motor energy losses are measured at 0.71 kWh as indicated in
and the driveline losses From Figure 29 are measured at 0.5 kWh, indicating that the
Figure 28, and
total energy thetodriveline
used overcomelosses From
the road loadFigure
forces29 are measured
acting at 0.5
on the vehicle, FRLkWh, indicating
, is 0.839 kWh.
Although Figure 22 validated the total energy comsumption correlation of the modelFRL
that the total energy used to overcome the road load forces acting on the vehicle, to , is
0.839
the dyno, the figures below allow us to estimate, experiment and validate the efficiencythe
kWh. Although Figure 22 validated the total energy comsumption correlation of
model
of the to the dyno,
real-world the figures
vehicle below
compared allow
to the us tomodel.
vehicle estimate,
The experiment
vehicle model and canvalidate
be opti-the
efficiency of the real-world
mized iteratively to reducevehicle compared
these losses to the
through vehicle
either model.
reduction in The
motor vehicle
lossesmodel
by ad-can
justing the motors operational limits or modifying the driveline gear selection.
Energies 2021, 14, 1493 19 of 22

Energies 2021, 14, 1493 be optimized iteratively to reduce these losses through either reduction in motor losses
29 ofby
33
adjusting the motors operational limits or modifying the driveline gear selection.

Energies 2021, 14, 1493 30 of 33

Figure 27. Vehicle tractive energy over time.

Figure 27. Vehicle tractive energy over time.

Figure 28. Motor energy losses over time.

Figure 28. Motor energy losses over time.
Energies 2021, 14, 1493 31 of 33
Energies 2021, 14, 1493 20 of 22

Figure 29. Driveline energy losses over time.

Figure 29. Driveline energy losses over time.

5. Conclusions
5. Conclusions
This paper
This paper highlights
highlights the
the importance
importance of of high
high performance
performance EV EV modeling,
modeling, simulation,
simulation,
and validation due to the revolutionization of the transportation industry by
and validation due to the revolutionization of the transportation industry by electric
electric mo-
mo-
bility. The
bility. The objective
objective ofof this
this work
work waswas to to demonstrate
demonstrate the the methodologies
methodologies for for developing
developing
high performance EV models in a software software simulation
simulation environment
environment and and validating
validating thethe
withreal-world
models with real-worldhardware
hardwareinina repeatable
a repeatable and
and scalable
scalable fashion,
fashion, thereby
thereby enabling
enabling the
the rapid
rapid development
development of EVoftestbeds.
EV testbeds.
A briefAreview
brief review of EV modeling
of EV modeling techniques techniques
and softwareand
software toolsets were introduced. A summary of EV motion
toolsets were introduced. A summary of EV motion dynamics was described mathemati- dynamics was described
mathematically
cally and an equationand an equation
based based
EV model EVimplemented
was model was implemented based on the models
based on the mathematical mathe-
matical models
described. described.
The design The design
philosophy of EVphilosophy
OEMs withofregard EV OEMs with efficiency
to vehicle regard to drive
vehicle ef-
cycle
ficiency
tests drive cycle
mandated by thetests
EPA mandated
were brieflyby the EPA were
described. briefly
Finally, thedescribed. Finally,
results of the the results
EV model were
of the EV by
validated model were validated
comparing by comparing
them to the them to
results of vehicle the results tests
performance of vehicle
of the performance
specified EV
on a chassis
tests dynamometer.
of the specified EV onThe energydynamometer.
a chassis efficiency results The of energy
the modeled high results
efficiency performance
of the
EV compared to the real-world vehicle showed close correlation
modeled high performance EV compared to the real-world vehicle showed close correla- with a root-mean-square
error (RMSE)
tion with value of 0.2. Theerror
a root-mean-square vehicle speedvalue
(RMSE) and vehicle
of 0.2. distance
The vehicletraveled
speedofandthe vehicle
model
also resulted
distance in negligible
traveled of the modelerror values.
also Theinmodeled
resulted negligible motor
errorcontrol
values.strategies,
The modeled however,
motor
did not capture
control thehowever,
strategies, dynamic did and not
transient
capture behavior of the real-world
the dynamic and transient vehicle adequately.
behavior of the
Although
real-worldthe equation-based
vehicle adequately. model provided
Although a valuable means
the equation-based of gaining
model providedan understand-
a valuable
ing
meansof the how the an
of gaining modeled vehicle will
understanding perform
of the how under certain vehicle
the modeled drivingwill
conditions
perform from an
under
energy efficiency standpoint, this modeling approach is not suitable
certain driving conditions from an energy efficiency standpoint, this modeling approach for detailed analysis of
vehicle dynamics and evaluation of control strategies. Future work will
is not suitable for detailed analysis of vehicle dynamics and evaluation of control strate- consider a physical
modeling
gies. Future approach
work will which enables
consider the development
a physical modelingofapproach
an advanced whichmotor model
enables thesuch as
devel-
data
opmentdriven/map based motor
of an advanced motor model
models, andasadvanced
such motor controller
data driven/map based motor schemes
models,suchand
as
six-step control or space vector modulation (SVM) controls which
advanced motor controller schemes such as six-step control or space vector modulation should result in more
detailed and accurate
(SVM) controls whichtransient and dynamic
should result in more responses
detailed and compared
accurate totransient
the real-world vehicle.
and dynamic
A
responses compared to the real-world vehicle. A physical modeling approach would that
physical modeling approach would also include more advanced battery models also
consider
include morethe thermal
advanced relationship between
battery models thatindividual
consider the cellsthermal
within relationship
a pack and ambient
between
temperature
individual cellsconditions,
within aas welland
pack as degradation mechanisms
ambient temperature that affect
conditions, asstate
well ofas power (SOP)
degradation
and state of health (SOH) during drive cycles simulations.
Energies 2021, 14, 1493 21 of 22

Author Contributions: Conceptualization, F.A.; methodology, F.A. and A.v.J.; validation, F.A., A.v.J.,
A.Y., B.P. and E.A.; formal analysis, F.A.; investigation, F.A., A.v.J. and A.Y.; data curation, F.A.;
writing—original draft preparation, F.A.; writing—review and editing, F.A., A.v.J., A.Y., B.P. and
E.A.; supervision, A.v.J., A.Y.; project administration, A.v.J. All authors have read and agreed to the
published version of the manuscript.
Funding: This research received no external funding.
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement: Not applicable.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. BNEF. EV Sales 2020; Bloomberg New Energy Finance: New York, NY, USA, 2020.
2. IEA. Global EV Outlook 2019; International Energy Agency: Paris, France, 2019.
3. Kumar, R.R.; Alok, K. Adoption of electric vehicle: A literature review and prospects for sustainability. J. Clean. Prod. 2020, 253,
119911. [CrossRef]
4. Traffic Safety Facts. 2016 Fatal Motor Vehicle Crashes: Overview; Research Note DOT HS 812 456; U.S. Department of Transportation,
NHTSA’s National Center for Statistics and Analysis: Washington, DC, USA, 2017.
5. Rowley, J.; Liu, A.; Sandry, S.; Gross, J.; Salvador, M.; Anton, C.; Fleming, C. Examining the driverless future: An analysis of
human-caused vehicle accidents and development of an autonomous vehicle communication testbed. In Proceedings of the IEEE
Systems and Information Engineering Design Symposium (SIEDS), Charlottesville, VA, USA, 27 April 2018.
6. Kricke, C.; Hagel, S. A hybrid electric vehicle simulation model for component design and energy management optimization. In
Proceedings of the FISITA World Automotive Congress, Paris, France, 27 September–1 October 1998.
7. Auert, B.; Cheny, C.; Raison, B.; Berthon, A. Software tool for the simulation of the electro- mechanical behavior of a hybrid
vehicle. In Proceedings of the Electric Vehicle Symposium 15, Brussels, Belgium, 29 September–3 October 1998.
8. Onoda, S.; Emadi, A. PSIM-Based Modeling of Automotive Power Systems: Conventional, Electric, and Hybrid Electric Vehicles.
IEEE Trans. Veh. Technol. 2004, 53, 390–400. [CrossRef]
9. Amrhein, M.; Krein, P.T. Dynamic Simulation for Analysis of Hybrid Electric Vehicle System and Subsystem Interactions,
Including Power Electronics. IEEE Trans. Veh. Technol. 2005, 54, 825–836. [CrossRef]
10. Baisden, A.C.; Emadi, A. An ADVISOR based mode of a battery and an ultra-capacitor energy source for hybrid electric vehicles.
IEEE Trans. Veh. Technol. 2004, 53, 199–205. [CrossRef]
11. Gao, D.W.; Mi, C.; Emadi, A. Modeling and Simulation of Electric and Hybrid Vehicles. Proc. IEEE 2007, 95, 729–745. [CrossRef]
12. Alam Chowdhury, M.S.; Al Mamun, K.A.; Rahman, A.M. Modelling and simulation of power system of battery, solar and fuel
cell powered Hybrid Electric vehicle. In Proceedings of the 2016 3rd International Conference on Electrical Engineering and
Information Communication Technology, Dhaka, Bangladesh, 22–24 September 2016; pp. 1–6. [CrossRef]
13. Yaich, M.; Hachicha, M.R.; Ghariani, M. Modeling and simulation of electric and hybrid vehicles for recreational vehicle.
In Proceedings of the 2015 16th International Conference on Sciences and Techniques of Automatic Control and Computer
Engineering (STA), Monastir, Tunisia, 21–23 December 2015; pp. 181–187. [CrossRef]
14. Zhou, J.; Shen, X.; Liu, D. Modeling and simulation for electric vehicle powertrain controls. In Proceedings of the 2014 IEEE
Conference and Expo Transportation Electrification Asia-Pacific (ITEC Asia-Pacific), Beijing, China, 31 August–3 September 2014;
pp. 1–4. [CrossRef]
15. Tremblay, O.; Dessaint, L.; Dekkiche, A. A Generic Battery Model for the Dynamic Simulation of Hybrid Electric Vehicles. In
Proceedings of the 2007 IEEE Vehicle Power and Propulsion Conference, Arlington, TX, USA, 9–12 September 2007; pp. 284–289.
[CrossRef]
16. MathWorks Student Competitions Team 2021. MATLAB and Simulink Racing Lounge: Vehicle Modeling. GitHub. Available
online: https://github.com/mathworks/vehicle-modeling/releases/tag/v4.1.1 (accessed on 10 January 2021).
17. Husain, I. Electric and Hybrid Vehicles: Design Fundamentals, 2nd ed.; CRC Press LLC: Boca Raton, FL, USA, 2010.
18. Von Jouanne, A.; Adegbohun, J.; Collin, R.; Stephens, M.; Thayil, B.; Li, C.; Agamloh, E.; Yokochi, A. Electric Vehicle (EV) Chassis
Dynamometer Testing. In Proceedings of the 2020 IEEE Energy Conversion Congress and Exposition (ECCE), Detroit, MI, USA,
11–15 October 2020; pp. 897–904. [CrossRef]
19. SAE International. J1634 in Battery Electric Vehicle Energy Consumption and Range Test Procedure; SAE International: Warrendale,
PA, USA, 2017.
20. Agamloh, E.; Von Jouanne, A.; Yokochi, A. An Overview of Electric Machine Trends in Modern Electric Vehicles. Machines 2020, 8,
20. [CrossRef]
21. Schauer, J.J.; Kleeman, M.J.; Cass, G.R.; Simoneit, B.R.T. Measurement of Emissions from Air Pollution Sources. 2. C1through
C30Organic Compounds from Medium Duty Diesel Trucks. Environ. Sci. Technol. 1999, 33, 1578–1587. [CrossRef]
Energies 2021, 14, 1493 22 of 22

22. Zielinska, B.; Sagebiel, J.; McDonald, J.D.; Whitney, K.; Lawson, D.R. Emission Rates and Comparative Chemical Composition
from Selected In-Use Diesel and Gasoline-Fueled Vehicles. J. Air Waste Manag. Assoc. 2004, 54, 1138–1150. [CrossRef] [PubMed]
23. Kogelschatz, U. Dielectric-barrier Discharges: Their History, Discharge Physics, and Industrial Applications. Plasma Chem. Plasma
Proc. 2003, 23, 1–46. [CrossRef]