EVS28 International Electric Vehicle Symposium and Exhibition 1
Goyang, Korea, May 3-6, 2015
An Accessible Predesign Calculation Tool to Support EV
Components Definition
Topic: Electric Vehicle, Presentation preference: Lecture session
Roche, Sabrià, Mammetti
Marina Roche Arroyos, Applus IDIADA group, [email protected]
Dídac Sabrià, Applus IDIADA group, [email protected]
Marco Mammetti, Applus IDIADA group, [email protected]
Abstract
The new freedoms in design that electric powertrains provide lead to a wide variety of configurations to
consider when developing an electric vehicle (EV) from scratch. Furthermore, the strong relation of the
battery size with vehicle weight, range and performances leads to a set of interrelated dependencies that
can result in many design loops to fulfil the targets and regulations simultaneously.
The paper presents a tool that integrates the main relations regarding vehicle targets, market and
regulations constraints and plots them as restrictions for vehicle development. As a result, the tool depicts
a set of feasible vehicle configurations that could fulfil the targets. Furthermore, to better assist selection,
it also provides a sensitivity analysis of the performances and the user can introduce a cost function
depending on vehicle weight and battery size. The tool is aimed at providing an overview of the possible
solutions and guide for component selection in the early predesign phase in which vehicle characteristics
and even powertrain architecture are unknown.
Finally, the tool is evaluated by modelling one of its solutions for passenger car for three different
architectures in the simulation software vemSim. Furthermore, for one of the architectures, two control
strategies were simulated, leading to a total of four simulations. The results of the simulations are
compared to the solution of the predesign tool to evaluate the level of fidelity and the deviations in the final
result that can appear depending on the architecture, components and control strategy.
Keywords: design, energy consumption, definition, simulation, performances
1 Introduction
Nowadays, transportation faces higher energy
costs and continuously increasing restrictive
emission targets, aiming at 95 g/km CO2 by the
year 2020. This implies the necessity of a
significant change in road vehicle propulsion
technologies. This necessity is especially patent
in dense urban areas with high traffic volumes,
polluted atmosphere and high noise levels. Fully
electric vehicles (EVs) offer the potential to be
locally emission free and recover energy through
regeneration while meeting the individual mobility
demand of passenger cars as well as fleet vehicles.
In the recent years IDIADA has participated in
several EV research and development projects,
EVS28 International Electric Vehicle Symposium and Exhibition 2
namely ELVA, Electric Race Car, More Zero,
IMPROVE and Puma Mind (European
Cofounded projects), eTruck and VeLoW
(locally cofounded projects), and iShare,
iTorque, eValuate and vemSim (internally
funded projects) as well projects for clients.
Based on the knowledge obtained in these
projects, IDIADA has developed an interactive
tool aimed at assisting EV design in the very first
development steps.
The main advantage of this tool is that it
interconnects calculations that typically involve
different departments in a loop; for example, the
target range and battery chemistry influence the
battery size, which affects the package and
vehicle weight, and thus the consumption and
range in an iterative loop. The tool was intended
to provide the user a first insight from a design
point of view and sufficient orientative
information on the vehicle requirements to fulfil
the performance targets regardless of the
powertrain architecture. A first approximation of
a vehicle solution which fulfils all these targets
and the market and regulations constraints is key
to speeding up the development process and
select the best powertrain architecture for
packaging, performances and cost.
The paper is structured in two main parts: the
first one describes the interactive tool that
models the constraints for the vehicle
development and shows a result example.
Afterwards, the same example vehicle is further
designed for three different architectures by
modelling each configuration in the in-hose
developed simulation software vemSim. The
required vehicle specifications are re-calculated
in more detail for each architecture of EV.
Finally, the results from the pre-design tool and
the detailed simulation models are compared to
show that the pre-design tool can provide a good
guideline for component selection when the
package and the architecture is still to be
decided.
2 Pre-design Tool
The pre-design tool was aimed at defining the
problem of EV development from an integrated
point of view. Commonly, a powertrain
definition with the eye on consumption and
performance cannot meet the dynamic targets or
may not be feasible for packaging. This situation
leads to a loop of iterations among departments
that modify the vehicle characteristics and affect
other department targets until a compromise is
reached.
The aim of the tool, despite its simplicity, is to
provide sufficient information on the vehicle
requirements to fulfil different targets
simultaneously in order to reduce the iterations
among departments. This solution can be used as a
specification to select the components and start the
packaging saving many design loops.
The process to find a solution is focused on the
design and package point of view: what volume
should I keep for the battery? What is my weight
target? How much the battery will weight? Where
are the powertrain components located? The main
targets and regulations to be fulfilled were
translated to weight and volume targets, leading to
a solution in terms of these two variables that
fulfils the requirements of different departments
and regulations simultaneously.
The pre-design tool was developed in Matlab.
Figure 1 shows the tool GUI in which the user can
select the inputs from sliders or also type in the
value. The dynamic GUI is programmed so the
default, maximum and minimum values for each
field are refreshed automatically when the user
changes the vehicle segment. The result of the
calculation is instantaneously refreshed as the user
moves a slider or modifies an input value. The
main parameters that affect vehicle longitudinal
dynamics [4] and targets are present in the
interface:
- 𝑖: Vehicle segment
- battery type: default chemistries [1].
- εbat: battery vol. energy density (kWh/l)
- ρbat: battery density (kg/l)
- πbat: battery vol. power density (kW/l)
- η: average powertrain efficiency (%)
- Af: aerodynamic frontal area (m2)
- Cx: aerodynamic drag coefficient
- PAux: auxiliaries power consumption (W)
- f: rolling resistance [5] (kg/ton)
- x: motor torque characteristic (see Figure 3)
- Reg: recovered energy in deceleration (%)
- Irot: rotating parts equivalent inertia (%)
- mload: playload mass (kg) (if applies)
- R1−3 : vehicle range targets in up to three
cycles (loaded or uloaded).
- vmax: target vehicle speed (m/s).
- ta: target 0-100 km/h acceleration (s)
- tan(α): gradeability target (%)
EVS28 International Electric Vehicle Symposium and Exhibition 3
Figure 1: Interface with highlighted inputs.
As all the main parameters are interrelated (range
- battery capacity - volume and weight - vehicle
weight – consumption and power required, etc),
the constraints due to the targets, regulations and
segment’s characteristics were translated to
battery volume and vehicle weight requirements
[6]. The principle of linear programming was
used to apply linear inequation constraints. The
purpose is to check whether there is a set of
solutions that fulfil all the requirements and, if
not, which limitations should be modified. If
there is a solution, the feasible solutions region is
a convex polyhedron, which is defined by
intersection of half spaces defined by linear
inequations representing the constraints as shown
in Figure 2. In this case, there is a range of
vehicle mass and battery volume combinations
that could lead to feasible solutions.
Figure 2: Internal tool result with numbered
constraints.
2.1 Deterministic constraints
In this section, the inequations associated to each
constraint in Figure 2 are explained. It can be
observed that the half spaces defined for the
constraints are coloured for better understanding:
the restrictions regarding weight (due to market
practices, structural feasibility or regulations) are
coloured in dark blue, the constraints coming from
performance targets or power limitations in
regulations are coloured in red. Finally, the
restrictions caused by the range targets are
coloured in cyan.
The possibility to dynamically modify the inputs
and obtain a picture of the feasible solutions is
helpful to determining the most restrictive
requirements and for decision-making in vehicle
development. It must be noted that the restrictions
applied by the inputs are deterministic, and all of
them can be expressed in the form of inequations.
The equations describing the 13 constraints are
defined in the following paragraphs in terms of
vehicle mass (𝒎) and/or battery volume (𝑽). The
list of all the abbreviations used can be consulted
at the end of the paper.
1.1.1 Vehicle segment constraints The following equations describe the constraints
due to restrictions coming from the market typical
values for each segment, structural feasibility and
weight restrictions that appear in regulations.
Maximum battery volume: it is restricted by the
possible available space per segment and it must
fulfil the inequation (1).
𝑽 < 𝑉𝑚𝑎𝑥,𝑖 (1)
Minimum battery volume: it must fulfil a
consistent minimum defined per segment as
specified in ineq. (2).
𝑽 > 𝑉𝑚𝑖𝑛,𝑖 (2)
Maximum battery mass: for each vehicle mass
there is a limit battery mass that allows structural
feasibility. The minimum mass requirements
without battery (𝑚𝑊/𝑂) for each segment lead to a
battery volume limit that can be calculated with the
battery density as shown in ineq. (3).
𝑽 <(𝒎−𝑚𝑊/𝑂,𝑚𝑖𝑛,𝑖)
𝜌𝑏𝑎𝑡(3)
Maximum vehicle mass: it must be consistent
with the vehicle segment mass as in ineq. (4).
𝒎 < 𝑚𝑚𝑎𝑥,𝑖 (4) For light duty vehicles, if a payload is selected, the
vehicle mass is also limited by the maximum
Gross Vehicle Weight of 3500 kg as specified in
the regulations [7], and must also fulfil ineq. (5).
EVS28 International Electric Vehicle Symposium and Exhibition 4
𝒎 < 𝑚𝑖𝑛(𝑚𝑚𝑎𝑥,𝑖 , (3500 −𝑚𝑙𝑜𝑎𝑑))(5)
Minimum vehicle mass: it is defined to allow
structural feasibility and leads to ineq. (6).
𝒎 > 𝑚𝑚𝑖𝑛,𝑖 (6)
Quadricycle mass limit: the vehicle mass
without battery for quadricycles is limited in
regulations [8] to 350 kg for light quadricycles
and 400 kg for heavy quadricycles and thus, the
battery volume must fulfil ineq. (7).
𝑽 >(𝒎−𝑚𝑊/𝑂,𝑚𝑎𝑥,𝑖)
𝜌𝑏𝑎𝑡(7)
1.1.2 Power constraints
The power constraints [4] are defined by the
battery power delivery limitations and the vehicle
power requirements to fulfil the targets. The
auxiliary variables defined in eq. (8-13) are
required for the calculations.
𝐹2 =1
2· 𝜌𝑎𝑖𝑟 · 𝐶𝑥 · 𝐴𝑓(8)
𝑘 = 1 + 𝐼𝑟𝑜𝑡 (9)
𝑚𝑒𝑥𝑡𝑟𝑎 = 𝑚𝑑𝑟𝑖𝑣𝑒𝑟 +𝑚𝑙𝑜𝑎𝑑 (10)
𝑚𝑇 = 𝒎+𝑚𝑒𝑥𝑡𝑟𝑎(11)
𝐹0 = 𝑓 · (𝑚𝑇(𝒎)) · 𝑔(12)
𝑚𝑒𝑞 = 𝒎 · 𝑘 +𝑚𝑒𝑥𝑡𝑟𝑎(13)
Maximum speed: the power required to reach
the maximum speed at a minimum reference
grade leads to a power demand depending on the
vehicle mass and payload as shown in eq. (14)
and thus, to a battery volume requirement as
specified in ineq. (15).
𝑃𝑣 = 𝐹2 · 𝑣max3 +𝑚𝑇(𝒎) · 𝑔 · 𝑣𝑚𝑎𝑥 ·
(𝑓 · cos(𝛼𝑟𝑒𝑓) + sin(𝛼𝑟𝑒𝑓))(14)
𝑽 > (𝑃𝑣(𝒎)
𝜂𝑝𝑒𝑎𝑘+𝑃𝐴𝑢𝑥)
𝜋𝑏𝑎𝑡(15)
Gradeability: the gradeability target at a
minimum reference speed mainly affects the
torque requirement which is calculated a
posteriori and does not have an effect at this
stage of the calculations. Even though, the power
required from the battery to achieve the
gradeability depends on the vehicle mass and is
calculated following eq. (16) and entails the
battery volume requirement in ineq. (17).
𝑃𝛼 = 𝐹2 · 𝑣𝑟𝑒𝑓3 +𝑚𝑇(𝒎) · 𝑔 · 𝑣𝑟𝑒𝑓 ·
(𝑓 · cos(𝛼) + sin(𝛼))(16)
𝑽 > (𝑃𝛼(𝒎)
𝜂𝑝𝑒𝑎𝑘+𝑃𝐴𝑢𝑥)
𝜋𝑏𝑎𝑡(17)
Acceleration time 0-100 km/h: the acceleration
time to 100 km/h or to the speed that the vehicle
allows implies a power demand that depends on
the motor torque characteristic. In an electric
motor the base speed is known as the speed in
which the motor switches from a constant torque
operation to a constant power operation, and leads
to the relation specified in eq. (18) known as motor
characteristic [9].
𝑥 = 𝑣𝑚𝑎𝑥
𝑣𝑏(18)
For a better understanding of this magnitude,
Figure 3 depicts the results of a simulation in
which motors with different motor characteristic
provide the same 0-100 km/h acceleration time.
The relevance of 𝑥 is stated since motors with very
different torque can provide the same acceleration
performance.
Figure 3: Different possible motor characteristic curves
for same 0-100 km/h acceleration time.
For combustion vehicles, acceleration usually has
to be simulated in time steps, but this approach is
not suitable for an instantaneous calculation. For
EVs, as the maximum torque curve can be
simplified in constant torque region and a
decreasing torque region governed by a constant
power limitation, the calculation can be divided in
two steps: one ruled by a constant torque equation
and one ruled by a constant power equation. This
hypothesis allowed the calculation of the power
and torque requirement at vehicle level with a
simplified equation obtained through integration of
Newton’s second law under the assumption of
average drag power as specified in [9].
Consequently, the power needed to reach the
acceleration target is calculated through the eq.
(19-22).
𝐸 = (𝑣𝑏
′2+𝑣𝑎2)
2·𝑡𝑎(19)
𝐹 =2
5· 𝐹2 · 𝑣𝑎
3(20)
𝐺 = 2
3· 𝑣𝑎(21)
𝑃𝑎 = 𝐸 · 𝑚𝑒𝑞(𝒎)+ 𝐹 + 𝐺 · 𝐹0(𝒎)(22)
EVS28 International Electric Vehicle Symposium and Exhibition 5
Where the values of 𝑣𝑎 and 𝑣𝑏′ used depend on
the vehicle characteristics as specified in eq. (23-
24). 𝑣𝑎 is the target speed to which the vehicle
must accelerate and it is 100 km/h unless the
vehicle is limited to a lower speed, and 𝑣′𝑏 is the
base speed unless it is higher than 𝑣𝑎.
𝑣𝑎 = min(𝑣𝑚𝑎𝑥, 100)(23)
𝑣𝑏𝑎𝑠𝑒′ = min(𝑣𝑎, 𝑣𝑏)(24)
Finally, the battery volume constraint due to
acceleration target can be calculated with the
battery characteristics through ineq. (25).
𝑽 > (𝑃𝑎(𝒎)
𝜂𝑝𝑒𝑎𝑘+𝑃𝐴𝑢𝑥)
𝜋𝑏𝑎𝑡(25)
Quadricycle power limitation: the
quadricycles maximum continuous power is
defined in regulations [8] as 4 kW for light
quadricycles and 15 kW for heavy quadricycles.
This power limitation delimits a mass that, if
exceeded, the performance targets cannot be
reached. For each power requirement (speed,
gradeability and acceleration) this limit mass is
calculated in eq. (26-28) and leads to the
restriction established in ineq. (29).
𝑚𝑚𝑎𝑥,𝑣 =𝑃𝑚𝑎𝑥,𝑖−𝐹2·𝑣𝑚𝑎𝑥
3
𝑣𝑚𝑎𝑥·(𝑓·cos(𝛼𝑟𝑒𝑓)+sin(𝛼𝑟𝑒𝑓))·𝑔−
𝑚𝑒𝑥𝑡𝑟𝑎(26)
𝑚𝑚𝑎𝑥,𝛼 =𝑃𝑚𝑎𝑥,𝑖−𝐹2·𝑣𝑟𝑒𝑓
3
𝑣𝑟𝑒𝑓3 ·(𝑓·cos(𝛼)+sin(𝛼))·𝑔
−
𝑚𝑒𝑥𝑡𝑟𝑎(27)
𝑚𝑚𝑎𝑥,𝑎 =𝑃𝑚𝑎𝑥,𝑖−𝐹−(𝐸+𝐺·𝑓·𝑔)·𝑚𝑒𝑥𝑡𝑟𝑎
(𝐸·𝑘+𝐺·𝑓·𝑔)(28)
𝒎 < min(𝑚𝑚𝑎𝑥,𝑣, 𝑚𝑚𝑎𝑥,𝛼 ,𝑚𝑚𝑎𝑥,𝑎)(29)
1.1.3 Energy constraints
Range in cycle 1: The vehicle must be
dimensioned to provide a consumption [7, 10]
with which the target range in the specified
regulations or real driving cycle can be satisfied.
The connection among battery volume and
vehicle weight is very noticeable in terms of
energy constraints, because a bigger battery to
increase the range also increases the weight and
thus, consumption.
To properly simulate an accurate consumption, a
step-by-step calculation with a vehicle model is
required. However, this simulation cannot run in
the pre-design tool interface instantaneously and
requires a set of inputs that are not available at
the first development phase to provide a level of accuracy that is also not required.
Therefore, the tool was aimed at overcoming this
obstacle during pre-design. One of the most
remarkable contributions of the tool was the
parameterization of the consumption in a specific
cycle with a single formula that requires just
simple vehicle characteristic inputs. This formula
is based on a backward-looking vehicle model in
which the main consumption sources were
classified in inertia (also considering regeneration),
rolling resistance and aerodynamic resistance
forces. The main assumption that the user has to
make is the average working efficiency of the
powertrain (inverter + motor + transmission), to
which reference information is provided.
Assuming the average efficiency, each
consumption source was divided into two
coefficients: one that depended only on the cycle
selected and one that depended just on the vehicle
characteristics as specified in eq. (30), where 𝐵 is
the energy consumption and 𝑅 the regeneration
factor.
𝐵 =1
· (𝐹0(𝒎) · 𝐶1 + 𝐹2 · 𝐶2 +𝑚𝑒𝑞(𝒎) · 𝐶3) +
· 𝑅 · (𝐹0(𝒎) · 𝐶4 + 𝐹2 · 𝐶5 +𝑚𝑒𝑞(𝒎) · 𝐶6) +
𝑃𝐴𝑢𝑥 · 𝐶7 (30)
Thus, the cycle dependent coefficients are defined
in eq. (31-40) where 𝜒represents a step function.
Since the maximum vehicle speed can be decided
by the user, the cycle’s speed profiles are also
modified consequently [7] under the speed limit to
calculate the cycle coefficients.
𝐶1 = ∫𝑣(𝑡) · 𝜒1(𝑡) · 𝑑𝑡(31)
𝐶2 = ∫𝑣(𝑡)3 · 𝜒1(𝑡) · 𝑑𝑡 (32)
𝐶3 = ∫𝑣(𝑡) ·𝑑𝑣(𝑡)
𝑑𝑡· 𝜒1(𝑡) · 𝑑𝑡 (33)
where𝜒1(𝑡) = {1,
𝑑𝑣(𝑡)
𝑑𝑡≥ 0
0,𝑑𝑣(𝑡)
𝑑𝑡< 0
(34)
𝐶4 = ∫𝑣(𝑡) · 𝜒2(𝑡) · 𝑑𝑡 (35)
𝐶5 = ∫𝑣(𝑡)3 · 𝜒2(𝑡) · 𝑑𝑡 (36)
𝐶6 = ∫𝑣(𝑡) ·𝑑𝑣(𝑡)
𝑑𝑡· 𝜒2(𝑡) · 𝑑𝑡 (37)
where𝜒2(𝑡) = {0,
𝑑𝑣(𝑡)
𝑑𝑡≥ 0
1,𝑑𝑣(𝑡)
𝑑𝑡< 0
(38)
𝐶7 = ∫𝑑𝑡(39)
𝐷 = ∫𝑣(𝑡) · 𝑑𝑡 (40)
The simplification stated in eq. (30-39) allows
quick characterization of the relative importance of
the three main consumption sources in different
cycles. Through the simplified formula, the
consumption is modelled as a function of the
EVS28 International Electric Vehicle Symposium and Exhibition 6
vehicle mass, and with the target range, the
battery volume requirement can be related to the
vehicle mass as specified in ineq. (41).
𝑽 >𝑅1
𝐵(𝒎)
𝐷·𝜀𝑏𝑎𝑡·𝑆𝑂𝐶𝑙𝑖𝑚
(41)
Range in cycle 2: this extra range requirement
allows enter a different target for another cycle.
Range in cycle 3: this extra range requirement
allows enter a different target for another cycle.
2.2 Non-deterministic constraints
The constraints specified in the previous
paragraphs can be expressed through physical
equations and the limit the possible feasible
solutions. However, how to select an optimal
solution within the viable region is not ruled by
deterministic equations. The cost objective
function commonly used in linear programming
is not that straight forward for this situation, as it
depends on discrete and non-linear
characteristics such as the powertrain
architecture, the components selected, the usage
or not of high-technology lightweight materials,
the development costs and the vehicle
manufacturing volume and procedure. The
weighting of all these factors in the final cost to
make a final decision require a proper market
study, interaction among departments, internal
company knowledge and direct contact with
TIERs for components quotations.
For better assisting the decision making inside
the viable region, the pre-design tool allows
introducing two cost functions, one that is
dependent on the battery capacity and one that is
dependent of the mass of the vehicle without the
battery, and represents the cost of the rest of the
material. The total cost function is then defined
by equation 42. The specific coefficients that rule
each cost function have to be decided based on
the know-how of the user, the production volume
and the market. Figure 4 shows an example of
application of a cost function.
𝑪 = 𝑓(𝑽 · 𝜀𝑏𝑎𝑡) + 𝑓(𝒎− 𝑽 · 𝜌𝑏𝑎𝑡)(42)
2.3 Sensitivity analysis
In some cases and depending the background of
the user, it is difficult to know in advance which
modifications will improve consumption and,
furthermore, which will affect consumption
more. For this purpose, once the user selects a
point inside the feasible region, a sensitivity analysis appears to show the magnitude in which
different parameters affect consumption for the
first selected cycle as shown in Figure 5. The
sensitivity analysis is a good summary to compare
the influence of very different parameters on
consumption, and to evaluate the most cost-
effective measure to reduce it.
2.4 Pre-design Tool Application Example
In this section, the pre-design tool was used to
estimate the main characteristics of a C segment
passenger car. The main characteristics of the
vehicle are presented in Table 1.
Table 1: Main characteristics of the vehicle
Aerodynamic drag coefficient (Cx) 0.28
Frontal area (m2) 2.19
Tire drag coefficient 0.007
Auxiliaries consumption (W) 450
Battery vol. energy density (Wh/l) 190
Powertrain average efficiency 75%
Amount of recovered energy 80%
The main targets of the vehicle are the ones
presented in Table 2.
Table 2: Main targets of the vehicle
WLTP Range (km) 200
Max Speed (km/h) 140
0-100 km/h (s) 10
The function of the cost of the battery and the cost
of the mass of the vehicle without the battery were
determined based on two studies [11, 12]. As per
discussed in these studies, the battery was
considered to have a linear cost increase with the
capacity and the weight of the vehicle was
considered to have a non-linear trend. The
lightweight design implies less amount of
materials but also more development costs and
more expensive materials, namely aluminium or
carbon fibre. However, an increase of the mass
over due to a low investment in design also causes
a cost increase due to the need to use more
kilograms of steel. Therefore, the function of the
mass-dependent cost has a local minima. Figure 4
shows the feasible region obtained for a C-segment
vehicle with the characteristics of Table 2 and the
targets of Table 2. It can be observed that the
battery size for this case is determined by the
target range (cyan area) and not by the power
required to fulfil the targets (red area), and that the
size requirement increases with the increase of
vehicle weight, because a higher vehicle weight
implies more consumption and, therefore, a bigger
battery for achieving the same range.
EVS28 International Electric Vehicle Symposium and Exhibition 7
Figure 4 Feasible for a C-segment vehicle and cost
function
Figure 4 also shows the cost function for
batteries from 100 to 200 l as a function of the
battery size and the vehicle mass. It can be
observed that the optimal is to select the
minimum battery that allows to achieve the range
(line limiting with the cyan area) and that the
cost function has a local minimum around 1400-
1500 kg. A lower mass would include the cost of
the materials, and a higher mass would increase
the size of the battery and the amount of material
required.
With this information, the pre-design vehicle
selected was a 1500 kg vehicle, which requires
approximately a 140 l battery to fulfil the range
and 84 kW of power. Figure 5 shows the
sensitivity analysis performed to this vehicle.
Figure 4: Sensitivity analysis in WLTP
It is remarkable, that for EV, an increase of 5%
on the average powertrain efficiency has more
than 5% positive impact on the total consumption.
This phenomenon occurs because when the
powertrain is more efficient, not only the vehicle
consumes less in traction mode (5% less) but also
regenerates more when braking (5% more),
leading to an impact higher than the relative
improvement.
3 Vehicle configurations With the pre-design configuration hint, it was
possible to implement the idea in three different
powertrain architectures and to simulate more
accurate consumption and component
requirements for each case. The simulations were
performed with vemSim, a vehicle longitudinal
dynamics simulation software developed at
IDIADA which is implemented in
Matlab/Simulink environment. The common
characteristics for the simulated configurations
were the ones shown in Table 3.
Table 3: Common characteristics for the simulation
Base vehicle weight 1250 kg
Motor power 85 kW
Motor torque curve characteristic 3
The three powertrain architectures considered
were:
- Longitudinal with one motor, reducer and
differential (Figure 5).
- Two motors connected directly to the front
wheels with two reducers (Figure 7)
- Four motors to wheel with four reducers
with the possibility to adjust the torque split
between the front and rear axle (Figure 8)
The main unknown system is the battery, which
will be different for each case to fulfil the same
range target because of the different weight and
average efficiency of each vehicle configuration.
The internal tool was useful to provide an accurate
hint of the size, but the accurate simulation is used
to check the viability of the predesign for the
specific powertrain architecture.
The motor torque curve characteristic was taken of
3 for all the motors and the motors were
considered of the same power with an accumulated
power of 85 kW for the cases with more than one
motor. The simulations were performed
considering a characteristic electric motor
efficiency map that was re-scaled to fit for the
different-sized motors of the three powertrain
architecture cases.
The weight of the powertrain components is also
different for each configuration, depending on the
amount and size of the motors, reducers and
EVS28 International Electric Vehicle Symposium and Exhibition 8
differential. The mass values were taken from an
analysis of the available components for electric
vehicles currently in the market.
3.1 First configuration
The first case to study was the conventional
architecture for a vehicle, a car with a
longitudinal motor that provides traction to front
wheels but in this case, the IC engine was
changed for an electric motor. To transmit the
power to the wheels, the vehicle used a reducer
and a differential. The configuration simulated is
shown in Figure 5.
Figure 5: Configuration 1 example
Figure 6 shows the efficiency points in which the
motor works for this configuration. It can be
observed that the cycle does not require to use
the maximum power or torque that the vehicle
can provide. Therefore, the torque required is
low compare with the torque available by the
motor and it works very often in low efficiency
points.
Figure 6: Operative points distribution for one
longitudinal motor
3.2 Second configuration
The second case to study was a different
architecture that may be implanted in electric
vehicle thanks to the new possibilities of the
electric traction. The concept of one motor for
traction power can be substituted by the concept of
one motor for every traction wheel. For this case, a
front wheel drive (FWD), car was simulated with
two motor in the front of the car. A reducer was
used to adjust the required torque to the wheels.
Figure 7 shows the simulated vehicle model.
Figure 7: Configuration 2 example
Both motors have half the power of the motor for
the first configuration, but the torque split among
them was 50% for each motor, the operative points
distribution obtained was very similar to the one
depicted in Figure 6, the motors worked very often
at low efficiency points.
3.3 Third configuration
The third case was the same concept of the second
case but with four tractive wheels, simulating an
all-wheel drive (AWD) vehicle. Figure 8 shows the
architecture simulated. This AWD configuration
allows implementing a more efficient operating
strategy of the motors to work in the most efficient
operative points without compromising the
performance. For that, a controller was designed to
calculate in which case is more efficient to work
with two motors or four motors depending on the
motors efficiency map.
EVS28 International Electric Vehicle Symposium and Exhibition 9
Figure 6: Configuration 3 example
The results that would be obtained without
operating strategy by just dividing the target
torque for the four motors would be very similar
to the ones obtained for the first and second
configuration, the motor would mainly work at
low efficiency points.
However, if an operating strategy is implemented
to calculate when it is better to operate only with
two motors and when to split the torque between
front and rear motors, high average efficiency
improvements can be achieved as shown in
Figure 9.
Figure 9: Operative points distribution for AWD
configuration for the front motors (blue) and rear
motors (yellow)
3.4 Simulation results
With the powertrain defined, was possible to
make an approach of the battery capacity
required for each specific case and define the
final performance of the vehicle. The objective
was to design a vehicle which was able to
achieve the range of 200 km following the
WLTP cycle and compare the results obtained
for different architectures with the results obtained
with the pre-design tool before the architecture
was selected.
Figure 10 shows the final weight for each
configuration. The use of four motors implies a
significant increase of the powertrain weight due
to the two extra motors and transmissions. The two
motors configuration implies only a small increase
in weight respect to the longitudinal configuration
because the differential is removed. A difference in
weight can also be observed for the AWD
configuration depending on the operating strategy
of the motors, because the battery size can be
reduced is such strategy is implemented. It can be
observed that the pre-design tool hint was close for
the longitudinal motor configuration but that is
was not that accurate for the AWD configuration.
The predesign tool is aimed at giving a fist hint on
the vehicle characteristics, but many different
implementations can appear when further
developing the configuration for different
powertrain architectures.
Figure 10: Total weight for each configuration
Figure 11 shows that a configuration with four
motors allows including an operating strategy to
optimize the powertrain average efficiency, which
in this case increases from 79% to 82%. The
average efficiency of the powertrain is also
improve with the two-motor configuration,
because the differential and the pinion crown used
to change the direction of the torque delivery from
longitudinal to transversal is no longer required.
EVS28 International Electric Vehicle Symposium and Exhibition 10
Figure 7: Efficiency of the powertrain for each
configuration
Figure 13 shows the kilometric consumption,
which is an effect of the vehicle weight, the
powertrain average efficiency among other
sources. Figure 13 shows that, for this case, even
though the AWD configuration implies a higher
vehicle mass, the kilometric consumption and
thus, the battery size, can be reduced if an
efficiency optimization strategy is implemented.
Figure 8: Consumption for each configuration.
Finally, Figure 15 shows the battery capacity
required for each configuration. The configuration
with four motors to wheel and an optimized
operating strategy requires less battery capacity
than the other configurations, what could
compensate the extra-cost required by the more
complex powertrain.
Figure 9: Battery capacity for each configuration
Conclusion In this paper, a pre-design tool to assist the electric
vehicle development process was presented. The
advantage of the tool is its capability to represent
the different development constraints that are
affected by regulations and decision-making of
different departments in one figure that can be
instantaneously refreshed in an interactive
interface.
The tool makes use of the equations that are used
to perform longitudinal dynamic simulations.
However, these equations were studied to obtain
simple approximations to avoid performing step-
by-step simulations and produce instantaneous
results. The main advantage of this tool is that it
interconnects calculations that typically involve
different departments in a loop to reduce the
iterations in the pre-design phase. The tool was
intended to provide the user a first insight from a
design point of view and sufficient information on
the vehicle requirements to fulfil the performance
targets regardless of the powertrain architecture.
The main equations and assumptions introduced in
the tool to obtain these targets are explained in this
paper. Finally, the usage of the tool is illustrated
through one example that is then further developed
in three different powertrain architectures and
simulated in the software vemSim to assess the
0,7
0,72
0,74
0,76
0,78
0,8
0,82
0,84P
ow
ert
rain
eff
icie
ncy
(%
)
112114116118120122124126
Co
nsu
mp
toin
Wh
/km
25
25,5
26
26,5
27
27,5
28
Bat
tery
cap
acit
y (k
Wh
)
EVS28 International Electric Vehicle Symposium and Exhibition 11
level of accuracy of the estimation that the pre-
design tool provides.
Acknowledgments
This study was developed within the framework
of an internal project in Applus IDIADA. The
author would like to express his gratitude to the
vemSim software team
References 1. ELVA Project Webpage, http://www.elva-
project.eu/
2. Dávila, A., Romero, E., Roche, M., Mammetti,
M. et al., "The ELVA Project's EV Design
Support Tool," SAE Technical Paper 2014-01-
1967, 2014, doi:10.4271/2014-01-1967.
3. Dávila, A., Romero, E., Gutiérrez, J., Mammetti,
et al., “ELVA Project – Innovative Approaches
for Electric Vehicle Design,” FISITA 2014 World
Automotive Congress, Maastricht, The
Netherlands, F2014-MVC-032.
4. BOSCH; Automotive Handbook, 2nd edition,
1986
5. UN/ECE Regulation No. 1222/2009, “Labelling
of tyres with respect to fuel efficiency and other
essential parameters”.
6. Sánchez Ruelas, J.G., Stechert, C., Vietor, T.,
Schindler, T., “Requirements Management and
Uncertainty Analysis for Future Vehicle
Architectures,” FISITA 2012 World Automotive
Congress, Beijing, China, F2012-E02-006.
7. Regulation No 83 of the Economic Commission
for Europe of the United Nations (UN/ECE),
“Uniform provisions concerning the approval of
vehicles with regard to the emission of pollutants
according to engine fuel requirements”.
8. Directive 2002/24/EC relating to the type
approval of two or three wheel motor vehicles
9. Ehsani, M., Gao, Y., Gay, S. E., Texas A&M
University, Ali Emadi, Illinois Institute of
Technology, “Modern Electric, Hybrid Electric,
and Fuel Cell Vehicles, Fundamentals, Theory,
and Design,” 2005 by CRC Press LLC. p.108.
10. ISO 2416:1992; Passenger cars – Mass
distribution; 1992
11. Redelbach, M., Klötzke, M., Friedrich, H. E.,
“Impact of lightweight design on energy
consumption and cost effectiveness of alternative
powertrain concepts”, EEVC, Brussels 2012.
12. Mckinsey & Company, “Lightweight, heavy
impact”.
Authors
Ms Marina Roche holds Mechanical
and Industrial Engineering degrees
both with honours. In 2012 she joined
IDIADA for research in vehicle
modelling, powertrain NVH, Wavelet
post-processing and software
development. Previously researched
on modelling the combustion of
gaseous fuels in IC engine in the
University of Zaragoza.
Mr Dídac Sabrià holds an Industrial
Engineering degree and a diploma in
Automotive Engineering from the
Universitat Politècnica de Catalunya
(UPC). He joined IDIADA in 2014 as
powertrain integration engineer for
research and development in vehicle
modelling and software development.
He has also researched on two-
wheeler fuel cell vehicles in UPC.
Mr Mammetti holds PhD in
Mechanics and an Aeronautical
degree. Since 2009 he worked in
IDIADA as powertrain integration
Product Manager. Recent activities
include powertrain conversions for
consumption reduction or to full
electric. In 2006 he was responsible of
powertrain integration in Maserati
and, before he worked in Ferrari as
responsible for engine performance
and calibration and innovation.
Abbreviations 𝛼 Gradeability target (º)
𝛼𝑟𝑒𝑓 Reference grade for speed target (º)
𝜒1(𝑡) True when positive or null acceleration in a
cycle
𝜒2(𝑡) True when negative acceleration in a cycle
𝜀𝑏𝑎𝑡 Battery volumetric energy density (kWh/l)
𝜂𝑝𝑒𝑎𝑘 Peak powertrain efficiency (%)
𝜋𝑏𝑎𝑡 Battery volumetric power density (kW/l)
𝜌𝑎𝑖𝑟 Air density (kg/m3)
𝜌𝑏𝑎𝑡 Battery density (kg/l)
𝐴𝑓 Aerodynamic frontal area (m2)
𝐵 Consumption in a specific cycle (kWh)
𝐶1−7 Coefficients to compute the contribution of
different consumption sources in a cycle
𝐶𝑥 Aerodynamic drag coefficient
𝐷 Distance overtook during a cycle (m)
𝐸 Coefficient to compute the inertia
contribution to the acceleration power
requirements (kW/kg)
𝐹 Coefficient to compute the drag contribution
to the acceleration power requirements (kW)
EVS28 International Electric Vehicle Symposium and Exhibition 12
𝐹0 Coast Down coefficient to compute the
constant resistance forces (N)
𝐹2 Coast Down coefficient to compute the
square-speed dependent resistance forces
(N/(m2/s
2))
𝐺 Coefficient to compute the constant
resistance forces contribution to
acceleration power (m/s)
𝑓 Tire rolling resistance (kg/t)
𝑔 Gravity constant (m/s2)
𝐼𝑟𝑜𝑡 Relation of rotating parts equivalent inertia
to vehicle mass (%)
𝑖 Segment selected
𝑘 Vehicle inertia coefficient
𝒎 Vehicle mass (kg)
𝑚𝑑𝑟𝑖𝑣𝑒𝑟 Driver mass (kg)
𝑚𝑒𝑞 Equivalent total mass considering inertia
(kg)
𝑚𝑒𝑥𝑡𝑟𝑎 Extra mass applied (driver + load) (kg)
𝑚𝑙𝑜𝑎𝑑 Playload mass (kg)
𝑚𝑚𝑎𝑥,𝑖 Upper vehicle mass limit per segment (kg)
𝑚𝑚𝑖𝑛,𝑖 Lower vehicle mass limit per segment (kg)
𝑚𝑚𝑎𝑥,𝑎 Quadricycles: acceleration target mass
limit (kg)
𝑚𝑚𝑎𝑥,𝛼 Quadricycles: gradeability target mass
limit (kg)
𝑚𝑚𝑎𝑥,𝑣 Quadricycles: speed target mass limit (kg)
𝑚𝑇 Total mass (vehicle + driver + load) (kg)
𝑚𝑊/𝑂,𝑚𝑖𝑛,𝑖 Lower limit to vehicle mass without
battery (kg)
𝑚𝑊/𝑂,𝑚𝑖𝑛,𝑖 Quadricycles: upper limit to vehicle mass
without battery (kg)
𝑃𝐴𝑢𝑥 Auxiliaries power consumption (kW)
𝑃𝑎 Acceleration power requirement (kW)
𝑃𝛼 Gradeability power requirement (kW)
𝑃𝑚𝑎𝑥,𝑖 Quadricycles: upper power limit
𝑃𝑣 Speed power requirement (kW)
𝑅1−3 Vehicle range targets for three cycles (m)
𝑅 Relation of recovered to recoverable
energy in deceleration (%)
𝑆𝑂𝐶𝑙𝑖𝑚 Battery State Of Charge lower limit (%)
𝑡 Time (s)
𝑡𝑎 Acceleration time target (s)
𝑽 Battery volume (l)
𝑉𝑚𝑎𝑥,𝑖 Upper limit to battery volume (l)
𝑉𝑚𝑖𝑛,𝑖 Lower limit to battery volume (l)
𝑣(𝑡) Speed profile in a cycle (m/s)
𝑣𝑎 Speed to achieve in acceleration target
(m/s)
𝑣𝑏 Transition speed from constant torque to
constant power in an electric motor (m/s)
𝑣𝑏′ Corrected base speed for acceleration (m/s)
𝑣𝑚𝑎𝑥 Target vehicle speed (m/s)
𝑣𝑟𝑒𝑓 Reference speed for gradeability target
(m/s)
𝑥 Motor torque curve characteristic