Ref. Ares(2020)5667449 - 20/10/2020
Optimization of scalaBle rEaltime modeLs and functIonal testing for e-drive ConceptS
EUROPEAN COMMISSION
Horizon 2020
GV-07-2017
GA # 769506
Deliverable No.
OBELICS D3.6
Deliverable Title
Tools supporting advanced EV trade-off process
Deliverable Date
2020-09-30
Deliverable Type
REPORT
Dissemination level
Public – (PU)
Written By
Approved by
Hellal Benzaoui (RT SAS)
Matthieu Ponchant, Franck Sellier (SIE-SAS)
Nicola Tobia (CRF)
Tommaso Favilli, Luca Pugi, Lorenzo Berzi (UNIFI)
Thilo Bein (FhG-LBF)
Tomaz Katrasnik (UL)
Horst Pfluegl (AVL) – Project Coordinator
2020-08-27
2020-06-18
2020-06-26
2020-07-02
2020-10-02
2020-10-03
2020-10-12
Status
Final version
2020-10-15
Reviewed by
Change log:
No Who
0
Hellal Benzaoui
1
Hellal Benzaoui /
Matthieu Ponchant
2
Matthieu Ponchant
/ Franck Sellier
3
Nicola Tobia
4
Tommaso Favilli
5
Hellal Benzaoui
6
Matthieu Ponchant
7
Hellal Benzaoui
8
Matthieu Ponchant
9
Matthieu Ponchant
Description
skeleton
Skeleton refinement
Date
12/04/2020
12/06/2020
SIE-SAS contribution
18/06/2020
CRF contribution
UNIFI contribution
Volvo contribution
Concatenation and refinement
Minor updates – UC1.1 contribution
Report finalization
Final version after reviewer comments integration
26/06/2020
02/07/2020
27/08/2020
28/08/2020
28/08/2020
04/09/2020
05/10/2020
GA # 769506
D3.6– Tools supporting advanced EV trade-off process - PU
2 / 37
Contents
Contents .................................................................................................................................................................... 3
Figures ................................................................................................................................................................... 4
Tables..................................................................................................................................................................... 4
Publishable Executive Summary ................................................................................................................................ 6
1
Introduction....................................................................................................................................................... 7
2
Tools for early phase concept analysis (RT-SAS, VALEO/SIE-SAS, UNIFI) .......................................................... 8
2.1
2.1.1
Design optimization process ................................................................................................................. 8
2.1.2
Configuration and automation capabilities ........................................................................................ 10
2.1.3
Trade-off process ................................................................................................................................ 13
2.1.4
Process execution on a case study ..................................................................................................... 14
2.2
VALEO/SIE-SAS – UC1.2: 48V electric vehicle configurator .................................................................... 19
2.2.1
Configuration and automation capabilities ........................................................................................ 19
2.2.2
Standardization ................................................................................................................................... 20
2.2.3
Trade-off process ................................................................................................................................ 20
2.3
3
RT-SAS (VOLVO) – UC1.1: Electric powertrain optimization tool ............................................................. 8
UNIFI/SIE-NV – UC2.2: Braking system model integration methodology ............................................... 23
2.3.1
Configuration and automation capabilities ........................................................................................ 23
2.3.2
Standardization ................................................................................................................................... 23
2.3.3
Trade-off process ................................................................................................................................ 26
2.3.4
Discussions and Conclusions ............................................................................................................... 27
Tools supporting virtual system integration studies (SIE-SAS, CRF) ................................................................ 29
3.1
CRF/SIE-SAS – UC2.5: Integrated tool for virtual integration ................................................................. 29
3.1.1
Configuration and automation capabilities ........................................................................................ 29
3.1.2
Standardized processes implementation and user interfaces............................................................ 30
3.1.3
Process execution validation on case studies ..................................................................................... 31
4
Conclusions...................................................................................................................................................... 33
5
Abbreviations and definition ........................................................................................................................... 34
6
Risk Register .................................................................................................................................................... 35
6.1
Risk register............................................................................................................................................. 35
7
References ....................................................................................................................................................... 36
8
Acknowledgement........................................................................................................................................... 37
GA # 769506
D3.6– Tools supporting advanced EV trade-off process - PU
3 / 37
Figures
Figure 0-1: OBELICS model-based development concept to reduce development and testing efforts. ................... 6
Figure 2-1:
Multi-layers system design methodology......................................................................................... 9
Figure 2-2:
UC1.1 optimization design process .................................................................................................. 9
Figure 2-3:
VOLVO’s in-house optimization platform main user interface....................................................... 10
Figure 2-4:
Automated process for alternative EM (PMSM) design exploration ............................................. 11
Figure 2-5:
Generic architecture adapted for vehicle sub-system models implementation ............................ 11
Figure 2-6:
Powertrain control process description for multiple powertrain variants evaluation ................... 12
Figure 2-7:
Example of EM scaling (diameter, length, number of turns …- Example of performances MAP for
vehicle level simulation of electric machine designs from this design space.......................................................... 15
Figure 2-8:
Optimal electric machine torque distribution (EMx2) and gear selection (2-speed transmission) 15
Figure 2-9:
Component sizing optimization with backward vehicle simulation with a) grid search and using b)
optimization algorithm ............................................................................................................................................ 17
Figure 2-10: Vehicle speed considered for backward calculation ((ACEA Urban delivery cycle - 16T vehicle
application), EM best candidate and EM operating points with powertrain layout 1 ............................................ 17
Figure 2-11: Vehicle speed considered for backward calculation ((ACEA Urban delivery cycle – 27 T vehicle
application); EM best candidate, EM1 and EM2 operating points with powertrain layout 2 ................................. 17
Figure 2-12: Multi-objectives for EM component sizing optimization and detailed results .............................. 18
Figure 2-13:
Vehicle speed comparison (ACEA Urban delivery cycle) and performance constraints verification
(acceleration, gradability and startability) for best EM design............................................................................... 18
Figure 2-14:
EM sizing optimization - Backward approach with single objective optimization (energy
consumption) versus Forward approach with multi-objective optimization (vehicle performance, energy
consumption and component cost)......................................................................................................................... 19
Figure 2-15: 48V EV architecture variants .......................................................................................................... 19
Figure 2-16: HEEDS optimization process ........................................................................................................... 20
Figure 2-17: Simcenter Amesim simulator variants in HEEDS ............................................................................ 21
Figure 2-18: HEEDS study results ........................................................................................................................ 21
Figure 2-19: Electric machines operating point: baseline and optimized control .............................................. 21
Figure 2-20: model including Air conditioning and auxiliary power management............................................. 22
Figure 2-21: Siemens SimRod fully instrumented (left); SimRod braking system scheme (right) ...................... 23
Figure 2-8: Pressure sensor (left), Pedal wire potentiometers (middle) and piston rod with strain gauges .......... 24
Figure 2-23: Sensors set up for the UC2.2 .......................................................................................................... 24
Figure 2-24: Force imposed model overview ..................................................................................................... 25
Figure 2-25: Displacement imposed model overview ........................................................................................ 25
Figure 2-26: Confidence interval rear axle.......................................................................................................... 26
Figure 2-27: Force-imposed (left) and the Displacement -imposed (right) models comparison for the
30_kph_high test ..................................................................................................................................................... 27
Figure 3-1:
State of the art of virtual integration in CRF .................................................................................. 29
Figure 3-2:
Integrated tool for virtual integration in CRF ................................................................................. 30
Figure 3-3:
Vehicle subsystem integration in Simcenter Amesim .................................................................... 30
Figure 3-4:
1D/3D coupling interface integration ............................................................................................. 31
Figure 3-5:
Coupling strategy between 1D and 3D models .............................................................................. 31
Figure 3-6:
Simulation time for different scenarios .......................................................................................... 32
Tables
Table 2-1 Possible approaches related to component sizing optimization ............................................................. 10
Table 2-2 Example of powertrain layouts considered in UC1.1 for commercial electric vehicle application ......... 12
Table 2-3 Powertrain component loss model parameterization............................................................................. 13
Table 2-4 Powertrain layout and configuration in focus ......................................................................................... 14
Table 2-5 Design parameter considered for PMSM electric machine scaling ......................................................... 14
GA # 769506
D3.6– Tools supporting advanced EV trade-off process - PU
4 / 37
Table 2-6: VOLVO FE and TESLA Semi-truck electric machines optimal torque split for baseline and design variants
................................................................................................................................................................................. 16
Table 2-7: Table VOLV FE and TESLA Semi-truck transmission optimal control for baseline and design variants .. 16
GA # 769506
D3.6– Tools supporting advanced EV trade-off process - PU
5 / 37
Publishable Executive Summary
This document presents different tools developed either for early design phase or for virtual integration studies.
Each tool is designed for a specific purpose from electric vehicle architecture design exploration to full complete
electric vehicle integration analysis. These tools answer different requirements presented in the OBELICS grant
agreement, allowing reduction of time and effort in development process as highlighted in the left part of the Vcycle (requirement/specification definition & System architecture selection) in one hand and right part in second
hand (test and verification, verification software system integration) in Figure 0-1:
• Configuration and automation capabilities in the tools
• Standardized process implementation and user interface
• Process execution validation
Figure 0-1: OBELICS model-based development concept to reduce development and testing efforts.
Some tools and methodology focus on the trade-off process optimization by considering different electric vehicle
powertrain architectures and analysis the relevancies of each of them. Processes have been standardized to
generate specification of all electric subsystem like the battery and the E-motor. Other presented tool is dedicated
to generic braking model integration, but not only the plant model but also the control dedicated with a focus on
energy saving and methodologies. Another methodology focuses on standardized virtual integration allowing 1D
and 3D model co-simulation on desktop and HPC. This method is being standardized/automated through python
script and allow deeper trade-off analysis with complete subsystem interaction impact.
Finally, these tools developed by simulation expert are accessible to non-expert thanks to standardized process
and user interface leading to design phase time reduction which is key objective of this project.
GA # 769506
D3.6– Tools supporting advanced EV trade-off process - PU
6 / 37
1
Introduction
In the context of offering flexible vehicle line-up virtual integration by allowing execution of integrated sizing and
calibration processes reducing costs and delay while ensuring optimal vehicle performance, tools supporting
advanced EV trade-off process have been developed, especially in the task 3.3 dealing with these topics. Indeed
outputs from others WP3’s tasks have used to support the development of these tools, like toy models [1] from
task 3.2 or control strategies [2] from task 3.4 or electric component models from WP2 [3], [4] & [5]. Furthermore,
these tools have been used and validated through WP6 demonstrators, where results and analysis are detailed.
Two kind of tools have been developed in the task 3.3:
• Tools for early phase concept analysis
• Tools supporting virtual system integration
Both tools are complementary, because first ones deal with early design phase, where subsystem are not specified
yet (purpose of these tools) and latter ones deal with integration phase, where all subsystems are designed and
need to be validated at the full vehicle system level.
The objectives of early phase concept analysis are the reduction of the gap, that may occur between simulation
expert (generally in one domain or subsystem) and vehicle architects. Indeed, by supplying high level of
accessibility non expert people can evaluate multiple vehicle configuration for different realistic conditions while
balancing vehicle attributes and generate specification and requirement to simulation expert in each domain. Such
tools will be presented for either the full vehicle level or component levels.
The objectives of tool supporting virtual system integration is a combination of several functions ensuring
consistent and relevant vehicle level model integration to realize simulation for different realistic condition while
validating component/subsystem design with interaction of all other subsystems. These functions cover the
following capabilities:
• to run simulation in different generation from desktop to HPC and to ensure stable co-simulation
• to offer standardized processes implementation through scripting like python script
• to optimize process execution especially with the management of models in co-simulation.
GA # 769506
D3.6– Tools supporting advanced EV trade-off process - PU
7 / 37
2
Tools for early phase concept analysis (RT-SAS, VALEO/SIE-SAS, UNIFI)
In this chapter, several methods for early concept phase will be described and have been illustrated in different
use cases.
In this concept phase, generally few information are available and only main requirements are already set, like
vehicle range target and vehicle weight target. Nevertheless, it is important to have as soon as possible good
insight of the vehicle to properly generate specification of all subsystems, especially the electric powertrain. That
is why auxiliaries and thermal management must be integrated even in this early phase as highlighted in Figure
0-1.
Furthermore, automatic configuration to explore as much as possible potential architecture is a key point to reach
more and more constraint targets from regulation. Then standardized process and modeling tool is also important
to reduce the design and development phase allowing the maximum user with common simulation tools.
2.1
RT-SAS (VOLVO) – UC1.1: Electric powertrain optimization tool
The optimization of an electric powertrain aims to provide a system design that can meet all performance and
packaging requirements while minimizing the objectives e.g., the added cost and/or energy consumption (range)
of the vehicle. Moreover, the optimization process itself must have the following characteristics:
• It must use components and system models with a sufficient level of detail to capture all relevant
interactions between the different components,
• It must provide results that can serve as a starting point for a final design,
• It should require a short set-up time and be computationally efficient to execute in order to reduce
development times and associated costs,
• It must be flexible, allowing the introduction of new technologies, components, topologies or system
layouts as they emerge and finally,
• If cost minimization is a major objective of the optimization, then accurate cost models for the different
components based on a similar set of assumptions needs to be implemented in order to ensure that the
cost-tradeoffs between the components are being correctly captured.
The overall powertrain design and optimization process proposed in UC1.1 is semi-automated in order to enable
a faster exploration of the design space (powertrain layout, component design and sizing) and optimization
process execution. The user still needs to set up the analysis and configure the different tools. When a powertrain
model is required for new topology performance analysis, it needs to be developed by the user. Nevertheless,
with the availability of high-level virtual integration tools, this powertrain model can be developed faster either
by sharing standardized model interfaces, increasing re-usability of models or by using parametrization scripts.
Moreover, a general optimization framework is also considered enabling its deployment for solving a wide range
of optimization problems through the development process.
2.1.1
Design optimization process
Design optimization of an electric powertrain system, as for hybrid electric powertrain, can be formulated as a
multi-objective optimization problem that spreads over multiple levels (topology, technology, component sizing
and control). The design optimization process implemented in UC1.1 is strongly inspired by the methodology and
the optimization framework as described in for hybrid electric vehicle design optimization [6]. This complete
design process together with its different (nested) design levels is depicted in Figure 2-1.
GA # 769506
D3.6– Tools supporting advanced EV trade-off process - PU
8 / 37
Figure 2-1:
Multi-layers system design methodology
The topology, technology and sizing of components are layers related to the physical system. The control layer is
dependent on the physical system, yet it will not change its physical parameters (e.g., the battery size, electric
machine type or gear ratios). These physical system parameters will act as bounds with which the control
algorithm must cope. In addition, the BEV topology will define the variables of the control algorithm (i.e., their
number and type). This inter-dependence (coupling) between the plant design layers and the control algorithm
supports the statement that the performance, which is obtained from optimal per-layer design, is influenced by
the design of other layers.
Figure 2-2 describes the design optimization process considered in UC1.1. This process includes the following
design layers: topology, component, sizing and control. This design methodology differs from other methods by
its very general framework and by a systematic analysis of the powertrain system.
Figure 2-2:
UC1.1 optimization design process
Two approaches are possible in the powertrain component sizing optimization layer. The first approach consists
of a grid search; it creates and run a batch of parameter combinations (experiments). It can create full factorial
batch (all combinations) or reduced parameter combinations. The second approach performs optimization tasks
in order to minimize/maximize user cost function(s) under certain constraints. Moreover, both backward and
forward approaches are considered for vehicle drive cycle simulation. Table 2-1 summarizes all the approaches
developed for component sizing optimization problem.
GA # 769506
D3.6– Tools supporting advanced EV trade-off process - PU
9 / 37
Table 2-1 Possible approaches related to component sizing optimization
Optimization methods
Grid search
Optimization algorithm
2.1.2
Vehicle model for powertrain performance assessment
Backward model
Backward model
Forward (and scalable) model
Forward (and scalable) model
Configuration and automation capabilities
2.1.2.1 Optimization tool
The application of VOLVO’s in-house optimization platform caters in supporting the workflow and the optimization
problems defined in UC1.1 due to its automation capabilities. The existence of both continuous and discrete design
variables defined in UC1.1 requires the extension of its features to be able to handle mixed integer problems. This
platform is generic and highly configurable, enabling its deployment for solving a wide range of optimization
problems. In step 1, it is used to determine the torque and power requirements at the wheel, while in step 2 it is
used to optimize powertrain component design and sizing towards energy efficiency, performances evaluation or
powertrain cost. In the later steps, it is can be used to calibrate the controller parameters. For all these steps once
the process is configured, the optimization is run without the provision of user input, in order to save time, the
process can be split in multicore mode running parallel loops simultaneously. There is also a possibility to interrupt
the optimization process in between in order to view the progressive result and can be continued from the stopped
step.
Figure 2-3:
VOLVO’s in-house optimization platform main user interface
2.1.2.2 PMSM configuration tool
An advanced model for PMSM multi-domain performance simulation has been developed in WP2 (details are
reported in deliverable D2.2). This model is based on finite element modeling methods for accurate prediction of
magnetic, electric and thermal behaviors. Scalability technics have been developed in order to reutilize data from
this finite element multi-domain model (by far the most time consuming) to be able to evaluate the effect of
changes in electric machine design parameters on the electric machine performances characteristics and therefore
impacts on powertrain efficiency on drive cycle operations. Figure Figure 2-4 summarizes the different steps
involved in this workflow where execution has been automatized with the help of dedicated parameterization
scripts.
GA # 769506
D3.6– Tools supporting advanced EV trade-off process - PU
10 / 37
Figure 2-4:
Automated process for alternative EM (PMSM) design exploration
Alternative electric machines design can be generated from a reference design by varying the design parameters
such as active length, diameter, inverter maximum current or number of turns. For each alternative design;
scalable electric machine performance models are automatically generated (map-based model, multi-domain
performance model) for electric machine sizing optimization with vehicle level system simulation.
2.1.2.3 Powertrain high level integration tool
To support various powertrain topologies exploration and multiple powertrain configuration analysis in early
project phases, new requirements have been considered for the development of a more flexible modeling
environment leading towards a faster vehicle/powertrain system modeling process execution. This flexible vehicle
and powertrain modeling framework is based on the existing in-house virtual integration environment (GSP)
presented in deliverable D3.1, mainly dedicated today for detailed system integration analysis in the design,
verification and validation phases. For the development of this flexible powertrain modeling framework, several
requirements are considered from the existing in-house simulation tool GSP to keep a continuous simulation
environment between early & design phase for vehicle modeling & data management, among which the most
important are:
•
•
•
Simulink as simulation integration platform
Vehicle/powertrain model development according to VMA standard (Vehicle Modular Architecture)
Generic architecture and interfaces for vehicle sub-system models, as illustrated in the figure, to enable
plug-and-play operation of the simulation models.
Figure 2-5:
Generic architecture adapted for vehicle sub-system models implementation
To increase tool flexibility and enable faster vehicle modeling process, new requirements and guidelines are
considered to support the development of generic vehicle subsystem model enabling multiple configurations of
key powertrain component (electric machine and transmission systems). In addition to battery, electric machine
or inverter systems, transmission systems are one of the key systems for new electric powertrain concepts
development. While transmission design for conventional vehicles is rather fixed or rarely evolved, there is much
more flexibility in transmission design when it comes to battery electric vehicles. Therefore, development of
GA # 769506
D3.6– Tools supporting advanced EV trade-off process - PU
11 / 37
flexible transmission models is necessary to support the powertrain system sizing studies, in particular the
transmission pre-design analysis with a possibility of large design space exploration.
2.1.2.4 Powertrain control process for multiple variant analysis
The development of the powertrain control strategies plays an important role throughout the vehicle
development process to enable efficient operation of the electric vehicle in the different stages of the
development. Control is especially important in early project phases to allow a fair comparison of different electric
powertrain concepts for robust concept selection and tasks which is already difficult and has become even further
complex with the development of innovative powertrain concepts combining multiple electric machines and
multiple speed transmissions systems. Therefore, application of optimal control techniques is essential for the
performance assessment of new powertrain concepts for a fair comparison. In addition, development of generic
control strategy models for vehicle system simulation is important for faster execution of control development
process in early phase. In order to support multiple powertrain concepts analysis with multiple e-component
variants integration, application of optimization-based control design methodologies in the early stage of the
powertrain control development process can help to cover many powertrain topologies and component
configurations. With such approach, it is possible to automatize powertrain control process as illustrated in Figure
2-6 with a focus in this document on the powertrain layouts depicted in Table 2-2. The powertrain layouts can
consist of multiple electric machines and transmission systems depending on the powertrain expected torque and
power requirements at the wheels.
Figure 2-6:
Powertrain control process description for multiple powertrain variants evaluation
Table 2-2 Example of powertrain layouts considered in UC1.1 for commercial electric vehicle application
The blocks C1, C2 and C3 can correspond to the following powertrain components:
• C1 can represent a simple reduction gear
• C2 can represent a single or multi speed transmission with or without neutral position (neutral position
means that electric machine can be disconnected from the driveline)
• C3 can represent a simple reduction gear or a differential system.
GA # 769506
D3.6– Tools supporting advanced EV trade-off process - PU
12 / 37
For each component, an approximation of the losses can be formulated using an affine dependency relation
between the component input and output power:
𝑇𝑜𝑢𝑡 ∙ 𝑤𝑜𝑢𝑡 = 𝐶𝑖 ∙ 𝑇𝑖𝑛∙ 𝑤𝑖𝑛 − 𝑃𝑖𝑑𝑙𝑒,𝐶𝑖 (𝑤𝑖𝑛 ),
𝑇𝑖𝑛∙ 𝑤𝑖𝑛 > 0
𝑇𝑖𝑛 ∙ 𝑤𝑖𝑛 = 𝐶𝑖 ∙ 𝑇𝑜𝑢𝑡∙ 𝑤𝑜𝑢𝑡 − 𝑃𝑖𝑑𝑙𝑒,𝐶𝑖 (𝑤𝑖𝑛 ),
𝑇𝑖𝑛∙ 𝑤𝑖𝑛 < 0
Where 𝑃𝑖𝑑𝑙𝑒 is the power that the component Ci needs to idle at an input-shaft speed 𝑤𝑖𝑛 . This equation is valid
when the vehicle is in traction mode. If 𝑇𝑖𝑛∙ 𝑤𝑖𝑛 < 0 a similar equation can be formulated to describe the losses
for the component Ci that affects the regenerative torque
Parameters related to C1, C2 and C3 components are listed in Table 2-3.
Table 2-3 Powertrain component loss model parameterization
Parameter
C1
C2
C3
Number of ratios
ratio
Efficiency
1
rc1
c1
αc1
βc1
D
N
[rc2_1 rc2_2 … rc2_n]
[c1_1 c2_2 … c2_n]
[αc1_1 αc2_2 …αc2_n αneutral]
[βc2_1 βc2_2 …βc2_n βneutral]
D
1
rc3
1
α c3
βc3
D
Losses
A quasi-static model for the complete powertrain system can be established in the following form by combining
these different component loss equations
𝑛
∑ 𝑟𝑖 ∙ 𝑖 ∙ 𝑇𝐸𝑀𝑖 = 𝑓(𝑇𝑤ℎ𝑒𝑒𝑙 , 𝑁𝑤ℎ𝑒𝑒𝑙 , 𝑝𝑐1 , 𝑝𝑐2 , 𝑝𝑐3 )
1
This model makes it possible to link the powertrain control variables (𝑇𝐸𝑀1, 𝑇𝐸𝑀2,…,𝑇𝐸𝑀𝑛 : torque request related
to electric machine 1, 2, …and n; transmission gear selection) to the powertrain torque demand 𝑇𝑤ℎ𝑒𝑒𝑙 . Brute force
search method is applied in order to define the optimal motor torque and gear shifting controls to optimize vehicle
energy consumption by selecting the best torque split and transmission ratios from a discretized set of
possibilities. Brute-force search, also known as generate and test, is a very general Problem-solving technique and
algorithmic paradigm that consists of systematically enumerating all possible candidates for the solution and
checking whether each candidate satisfies the problem's statement.
2.1.3
Trade-off process
The trade-off process can be performed considering different design objectives or vehicle attributes. In UC1.1, the
following design objectives are considered:
•
•
•
Vehicle performances evaluated with real conditions of operations (drive cycles)
Energy consumption (range)
Component cost
Design optimization can be done considering a single objective (energy consumption) or multiple objectives for
trade-off analysis. For component cost objective evaluation, only cost of the electric machine is considered here
with a simple cost model (linear relation between cost and electric machine power rate). Considering the total
cost of ownership (energy cost, all powertrain component cost, maintenance cost …) as objective in the trade-off
process could deliver more relevant results. This was not considered in this study due to the complexity of the
evaluation of such objective (uncertainties on input, lack of economical knowledge …).
GA # 769506
D3.6– Tools supporting advanced EV trade-off process - PU
13 / 37
2.1.4
Process execution on a case study
In this section, the proposed process and methodology are used in the design and optimization of an electric
powertrain for multi-purpose commercial vehicle application (distribution, refuse …). Two powertrain layouts are
in focus in session and shown in Table 2-4.
Table 2-4 Powertrain layout and configuration in focus
Current product
Corresponding layout
Powertrain configuration
•
•
•
•
EM1
C1: Not considered
C2: 2 speed transmission
C3: rear axle
•
•
•
EM1 = EM2
C1: Simple gear (left)
C2: 2 speed transmission
(left)
C3: rear axle
•
The execution of the design process described in figure will mainly focus on the two following optimization layers:
• Component sizing optimization,
• Powertrain control optimization.
The main objectives involved is the optimization of the sizing of a PMSM electric machine by taking into
consideration efficiency, cost and performance also considering new assumptions for the powertrain design
optimization: targeted electric vehicle with new characteristics, vehicle operating on new drive cycles; ASTERICS
cycle and some of ACEA cycles are discussed in this session. The main steps in these two nested layers are:
• powertrain component population (with a focus only on PMSM electric machine),
• Powertrain control optimization,
• Component sizing optimization.
2.1.4.1 EM design generation
EM design candidate are generated from for example two permanent magnet synchronous machine (PMSM)
reference designs by varying the following electric machine design parameters: active length, diameter, inverter
maximum current and number of turns. A “short” (respectively “large”) EM catalogue is obtained from PMSM
reference design 1 and PMSM reference design 2; respectively with the following design parametric values:
Table 2-5 Design parameter considered for PMSM electric machine scaling
PMSM Reference design 1
PMSM Reference design 2
5
9
3
5
19
1425
278
7
5
18
5670
1836
Active length
Diameter
Inverter maximum current
Number of turns
Number of combinations
Number of feasible designs generated
GA # 769506
D3.6– Tools supporting advanced EV trade-off process - PU
14 / 37
From the scaling done in Table 2-5, 278 alternative designs among all possible combinations (1425) have been
generated (not all combinations are possible). Similarly, 1836 alternative electric machines among 5670 possible
combinations are obtained considering PMSM reference design 2. Figure 2-7 illustrates the scaling considered for
the PMSM reference design 2 and performance assessment for some of these generated designs.
Figure 2-7: Example of EM scaling (diameter, length, number of turns …- Example of performances MAP for vehicle level
simulation of electric machine designs from this design space
2.1.4.2 Powertrain control process evaluation
The powertrain control process is evaluated considering powertrain layouts 1 and 2. For powertrain layout 1,
torque distribution and 2-speed gear-shifting maps are calculated considering all electric machine design variants
defined in Table 2-5 (1836 EM variants); for powertrain layout 2, torque distribution and gear-shifting maps are
calculated considering alternative design variants VOLVO FE and Tesla semi-trucks (clutchable electric machines).
• Powertrain control with multiple EM design variants
When in the early phase of designing an electrical platform, proper management on the powertrain like the correct
Electric machine design selection and its involvement in the sizing of the powertrain is critical. With the list of
multiple combination of design parametric EM design variants available, an efficient and fast powertrain control
can be achieved with regarding component design variant selection as illustrated in Figure 2-8 for different electric
machine design variants.
Figure 2-8:
Optimal electric machine torque distribution (EMx2) and gear selection (2-speed transmission)
GA # 769506
D3.6– Tools supporting advanced EV trade-off process - PU
15 / 37
• Powertrain control with multiple transmission design variants
Table 2-6 and Table 2-7 illustrates the electric machine optimal torque and gear selection (gear 1, gear 2, …,
neutral) that can be implement for VOLVO FE and Tesla semi-truck vehicle performance and efficiency simulation
for baseline design and design variants.
Table 2-6: VOLVO FE and TESLA Semi-truck electric machines optimal torque split for baseline and design variants
Table 2-7: Table VOLV FE and TESLA Semi-truck transmission optimal control for baseline and design variants
2.1.4.3 Component sizing optimization
Consider the effect of component sizes on the optimization of BEV design, a baseline is selected for the electric
machine. The sizes of corresponding component are varied during the early design process using the varied scaling
factors and baseline parameters. Defining components that fulfil basic powertrain performances requirements
and elaborating the optimum combination of component characteristics to achieve the highest powertrain
efficiency. In this case, study the objective is to assess the area of evaluating the different variants of machine
types from a fixed catalogue of pre-calculated list, and optimizing the selected parameterization and a multidomain performance assessment in order to achieve the target of optimizing the efficiency of the electric
powertrain concept on complete vehicle level requirements. The problem of design optimization of this E-Machine
component can be tackled in two different approach, by using the backward approach and the forward approach.
• Based on backward vehicle simulation
With integrating the in-house optimization tool, we can initiate the initial backward approach, which includes the
grid-search algorithm, and the optimization-based algorithm where a differential algorithm is used to undergo a
design optimization for the appropriate component selection method. This backward approach utilizes the brute
force calculation script in order to assess the optimum feasibility of EM design with respect the different Vehicle
application and the drive cycle. This will give in some key feedback input to the forward approach with the control
map files generated.
GA # 769506
D3.6– Tools supporting advanced EV trade-off process - PU
16 / 37
Figure 2-9:
algorithm
Component sizing optimization with backward vehicle simulation with a) grid search and using b) optimization
Figure 2-10: Vehicle speed considered for backward calculation ((ACEA Urban delivery cycle - 16T vehicle application), EM best
candidate and EM operating points with powertrain layout 1
Figure 2-11: Vehicle speed considered for backward calculation ((ACEA Urban delivery cycle – 27 T vehicle application); EM best
candidate, EM1 and EM2 operating points with powertrain layout 2
GA # 769506
D3.6– Tools supporting advanced EV trade-off process - PU
17 / 37
• Based on forward vehicle simulation
With the forward approach, based on the iterated control map files generated using the backward optimization,
and the data settings of the improved target vehicle spec and model, we find the performance analysis of this
vehicle powertrain system with the multiple EM catalogue and obtain the optimized result for various road
applications. The main benefit of this approach is the use of a more realistic vehicle that includes keys powertrain
component limit among which impact of torque interruption during gear shifting or electric machine thermal
limitation. The forward optimization will also suggest us the study of the performance constraint where in, we will
be able to assess the gradeability, startability and acceleration evaluation of the target truck with the feasible EM
candidate. This will help us narrow down on the more feasible passed designs.
Figure 2-12: Multi-objectives for EM component sizing optimization and detailed results
Figure 2-13: Vehicle speed comparison (ACEA Urban delivery cycle) and performance constraints verification (acceleration,
gradability and startability) for best EM design
GA # 769506
D3.6– Tools supporting advanced EV trade-off process - PU
18 / 37
Figure 2-14: EM sizing optimization - Backward approach with single objective optimization (energy consumption) versus
Forward approach with multi-objective optimization (vehicle performance, energy consumption and component cost)
2.2
2.2.1
VALEO/SIE-SAS – UC1.2: 48V electric vehicle configurator
Configuration and automation capabilities
UC1.2 focuses on 48V electric vehicle simulations. Two vehicle concepts have been evaluated, a 2 axles drive and
a 4-wheels drive, leading to the development of two vehicle simulators using Simcenter Amesim. Standard
components have been used to model complete vehicles.
Figure 2-15: 48V EV architecture variants
Parameter optimization can be done using Amesim built-in parameter batches capabilities, using scripting (Python,
Matlab) or by using external optimization tool. In this use case the baseline models have been parameterized using
the Simcenter Heeds [7] tool.
GA # 769506
D3.6– Tools supporting advanced EV trade-off process - PU
19 / 37
2.2.2
Standardization
The two vehicle configurations share a similar structure and use the same sub-models for vehicles, driver,
transmission, e-motors and batteries. E-Motors used in the two simulators are strictly identical. Battery packs are
using the same pre-calibrated cell models, and only differs the packs configurations (numbers of parallel modules).
The standardized interfaces developed in WP3 and described in [8] have been used. This enables to adjust the
modeling level of each subsystem depending on the simulation focus (scalability), as well as to easily add additional
subsystems to the baseline models. For instance, map-based quasi-static e-motor models are used for energy
consumption on driving cycle evaluations but are replaced by electromagnetic thermal model in Simulink using
co-simulation to check the influence of dynamics effect on energy consumption.
2.2.3
Trade-off process
The trade-off process is performed in three steps:
• 1st step: optimization of the transmissions and battery pack sizing
• 2nd step: optimization of e-motor control strategies
• 3rd step: optimization of auxiliary power management
1st Step description: Multi-objective optimization using Heeds
The Heeds-Simcenter Amesim portal allows to modify parameter values (including initial values of state variables)
and extract variable values of the Amesim model. The portal launches Amesim with customized python code to
update and/or extract details of the model. The main vehicle parameters optimized in this process are the
transmission parameters (front and wheel gear ratio, wheels dimensions) and the battery pack configuration.
Figure 2-16: HEEDS optimization process
The optimization objectives are the following, using three variants of the Amesim simulator:
• to maximize the powertrain efficiency on a WLTP driving cycle,
• to maximize acceleration while ensuring gradability test.
Multiple instances of the Amesim models can run simultaneously on the same computer. Indeed, the number of
instances to run concurrently is limited by the number of cores of the machine, allowing HEEDS to complete
design explorations significantly faster than running sequentially. Nevertheless, some calculation on HPC for
bigger design of exploration is also possible.
GA # 769506
D3.6– Tools supporting advanced EV trade-off process - PU
20 / 37
Figure 2-17: Simcenter Amesim simulator variants in HEEDS
The Hybrid Adaptive Method SHERPA has been chosen for the optimizer and the analysis has been launched on
500 iterations. Heeds post-processing capabilities enables to analyze the different design performed and to
identify relationships between variables and/or responses using correlation plots. From the feasible designs, the
best compromise has been selected and is use as input to the optimization of the e-motor control strategy.
Figure 2-18: HEEDS study results
2nd Step description: optimization of e-motor control strategies
An optimal control of the electric machines has been generated using the Simcenter Amesim Hybrid Optimization
Tool (HOT). HOT is based on optimal control and specifically on the method known as Pontryagin's Minimum
Principle (PMP). With this control, the torque distribution between the machine is optimized, and the electric
machines are operating on a larger range and for operating points with higher efficiencies. The regenerative
braking is optimized as well.
Figure 2-19: Electric machines operating point: baseline and optimized control
The potential gains of this control have been evaluated on 3 different cycles (WLTC class3, JC08 and specific city
cycle), and import efficiency improvements can be achieved especially in urban conditions.
3rd Step: auxiliary power management
GA # 769506
D3.6– Tools supporting advanced EV trade-off process - PU
21 / 37
An auxiliary power management system has been introduced to optimize the trade-off between the average
consumption and the driver thermal comfort. The auxiliary power management system has been developed by
UNIFI in Simulink and is working in cosimulation with the Amesim multi-domain vehicle model. According to
battery conditions (SoC) and cabin temperature, a strategy (Eco mode) is activated to reduce some none
prioritized auxiliary consumptions (e.g. fan, blower, compressor).
Figure 2-20: model including Air conditioning and auxiliary power management
With this three-steps optimization methodology significant improvement of the performances and efficiency of
the vehicle has been obtained with respect to the initial baseline design. All results will be detailed in deliverable
D6.1 (Design of new e-drive concepts, optimal system sizing based on high Level virtual system integration tools
and simulation report on assessment of virtual simulation methods).
GA # 769506
D3.6– Tools supporting advanced EV trade-off process - PU
22 / 37
2.3
UNIFI/SIE-NV – UC2.2: Braking system model integration methodology
This work presents an investigation of braking system characteristics, brake system performance and component
design parameters that influence brake pedal force and displacement characteristics. It includes detailed studies
of individual brake system component design parameters, operation, and the linear and non-linear characteristics
of internal components through experimental study and simulation modelling. Each brake system component was
modelled individually before combining them into the whole brake system in order to identify the parameters and
the characteristics of the internal components that influence prediction capabilities and allows real-time
simulations. Within this study, two type models were created and compared. Moreover, an assessment is carried,
in order to enable further implementation in a repeatable way.
The proposed approach aims at the definition of a standardized methodology allowing to define precise and
reliable model of EV (so-called Toy model) by accounting in a more accurate and simplified way braking actuator
non-linear model, to be easily adopted in the early design stage of several plant layouts. Calibration of the models
is based on a systematic testing procedure, allowing flexible transition from MiL to HiL.
This supporting tool, useful for components sizing and investigation, has been applied to a reference UC, as
described in the following pages. Conclusion on the effectiveness of the proposed methodology is done.
2.3.1
Configuration and automation capabilities
2.3.1.1 Challenges and Aims of the activities
OBELICS set the goal of applying the developed models to various case studies. In our work, we will focus on the
UC2.2, i.e. Siemens SimRod Figure 2-21. The SimRod brake scheme is also represented. The plant is made by two
TT pipelines, one for each master cylinder.
Figure 2-21: Siemens SimRod fully instrumented (left); SimRod braking system scheme (right)
The main goal for these activities is to develop a general brake model, real-time capable and able to assess the
synergy within newly electric braking strategy. The test procedures carried out in previous work [1] [9] [2]. In the
first part of this work a general description of the performed experimental activities, based on the previous model
developed in Simulink environment [10] [11], is carried out and then a review of the entire model is made with a
general assessment on the test procedures.
The resulting model consist of a hydraulic, thermal and wear behavior [11]. An in-depth analysis related to
hysteresis is assessed. Previously proposed model was implemented in Simulink. The goal was the development
of a standalone model able to receive braking demand and predict the braking torques, allowing to Model In the
Loop e Hardware In the Loop simulations (MIL & HIL). Since a scalable model is needed to fit most case studies,
the hydraulic plant was developed according to a functional decomposition method.
2.3.2
2.3.2.1
Standardization
Experimental Validation of Brake models: Hydraulic, Thermal and Wear
2.3.2.1.1 Hydraulic Model
Test procedure related to the hydraulic model is split up into two main parts:
• Calibration test: during this phase, the vehicle should be in standstill condition.
GA # 769506
D3.6– Tools supporting advanced EV trade-off process - PU
23 / 37
• Validation test: we proceed with a validation run to evaluate vehicle behavior in real operation.
Data used comes from a setup carried out in previous works [1] [9] [2] [10] [11]. Understanding how
measurements and experimental test were carried out will be helpful to assess if calibration test procedures are
adequate to determine braking system characteristics. The signals needed for model calibration are:
• Pressure of hydraulic plant: measured using 2 piezometric transducers, one for each axle. Since UC2.2 has
a TT scheme, is correct to assume the same pressure for the right and the left calipers.
• Pedal displacement or piston rod displacement: since the UC2.2 doesn’t have enough space to install a
larger variety of displacement sensors, 2 wire displacement sensors have been applied.
• Piston rod strain: to measure the applied force on the pedal, 2 single-grid strain gauges were installed.
This particular configuration allows us to compensate for the bending moves during load application.
Figure 2-22: Pressure sensor (left), Pedal wire potentiometers (middle) and piston rod with strain gauges
2.3.2.1.2 Thermal and Wear Models
In order to better represent the phenomenology of our brake plant two important factors must be considered
• Thermal behavior
• Wear behavior
The first one will lead to a drop in the friction factor and a reduction in braking force while the other will cause
severe environmental burdens related to PM emissions. Let’s check some of the possible causes and which
components are affected by these phenomena [11].
Figure 2-23: Sensors set up for the UC2.2
GA # 769506
D3.6– Tools supporting advanced EV trade-off process - PU
24 / 37
Let’s take an overview of the experimental activities conducted for the thermal and the wear model. Installed
sensors are: Rubbing thermocouples, Pad thermocouples, IR sensors, Standard thermocouples, Pressure sensors
and Torque sensors (Figure 2-23).
2.3.2.2 Model-based approach: The Braking System
In this work, brake models in Amesim environment were proposed and then validated by experimental results.
In modern vehicles, electronic control unit and mechanics automated actuations are constantly increasing. This
means that testing entire systems in real vehicle start to be time-consuming lead the development process into
time-to-market delay. In this scenario of uncertainty, related to EVs switch, ADAS offer growth and customer
habits all the O&M, car manufacturers need to optimize the testing process switching from a real-world evaluation
into accurate laboratory test campaign. Therefore Hardware-In-the-Loop simulations take the field both for the
electronics and for the mechanical components [12].
Models can be validated early on through simulation and verified continuously as the component models are
refined with additional implementation detail.
Force Imposed Model: The first part of the study is related to carry out a force-imposed model for the UC2.2 within
the OBELICS project. In the previous work, a simplified hydraulic model was done, and reduced parameters needed
to describe the system were estimated. The new aim for the activity is to improve the model adding new features
and allowing better parameter estimation, maintaining the real-time capabilities (Figure 2-24).
Figure 2-24: Force imposed model overview
Displacement Imposed Model: In this section, we will carry out a simple displacement-imposed brake model in
order to predict the pressure giving master cylinder travel as an input. This will help in the implementation of the
model in the Virtual Reality environment. Indeed, measuring the displacement is easier than install on the system
any device capable to estimate the force with acceptable accuracy.
The model carried out share most of the part with the previous force-imposed model since one of the main
purposes within the OBELICS project was to develop a model-based brake system able to switch easily between
different configurations. This results in our case in a force-imposed model with a feedforward on displacement.
Figure 2-25: Displacement imposed model overview
2.3.2.3 Hysteresis behavior in brake plant
Hysteresis is a nonlinear phenomenon exhibited in various science and engineering field. Hysteretic systems often
present a quasi-static response where input and output lead to a cycle and not a line as in the linear model
assumption. This behavior is firstly due to the physics of the problem and for instance, is not that easy to properly
GA # 769506
D3.6– Tools supporting advanced EV trade-off process - PU
25 / 37
understand all the physical law for every engineering application. This is an arduous task and most of the resulting
models are too complex to be used. In this scenario, several mathematical models have been proposed in order
to describe the hysteresis phenomena. One of the most used is the so-called Bouc-Wen model. Mechanical
hysteresis in brake plant is a well-known effect. It will lead to delay in system reactions reducing both the pedal
feedback and the control accuracy in safety systems like ABS and ESC.
Any investigation related to this application it’s a complex task because conventional hydraulic brake system
presents many hysteresis elements, e.g. vacuum booster, pipelines, brake cylinders, seals and friction pads [13].
In this chapter, an in-depth study for the hysteretic behavior is carried out and suggestion for new test procedures
are assessed.
For the experimental validation, we exclude a priori the thermal effects since the tests are static and carried out
at a constant temperature. We exclude also the elasticity of the pipeline since the hydraulic system for the UC2.2
it’s equipped with stiff tube reducing the lung effect below zero.
The next step consists in removing the loops from the cycles, so we implement a sort of conditional sampling
which avoids choosing points within the loops and then interpolate the sample vector with given equations. Then,
estimate a 95% confidence interval or rather the interval where our hysteresis cycle will be with a 95% probability.
What we can clearly see in Figure 2-26 is that the confidence interval plotted for the downhill curve seems to be
thinner respect to the other two. This means that most of the backward travels come out with lower variability
respect to the forward travel and it’s attributable only to how the procedure was carried out. Related to these
observations it’s important to point out that tests were made with no speed control, leaving the experimenter
free to impose the pedal velocity according to his capabilities. During forwarding travel tester is not capable to
maintain a constant velocity while during the backward travel helped by the spring the experimenter is able to
maintain the pedal rate within a lower spread. This also means that hysteresis cycles are affected by a different
pedal rate that we are not able to characterize since in a single run more than one velocity can be detected.
Figure 2-26: Confidence interval rear axle
2.3.3
Trade-off process
2.3.3.1 Force-imposed and Displacement-imposed Models Comparison
A global comparison between force-imposed model and displacement-imposed model is done, considering the
results for at least one maneuver for each test set.
GA # 769506
D3.6– Tools supporting advanced EV trade-off process - PU
26 / 37
Figure 2-27: Force-imposed (left) and the Displacement -imposed (right) models comparison for the 30_kph_high test
In Figure 2-27 we can see a comparison between the displacement-imposed model and the force-imposed model.
Regarding the near-zero pressure phase in the force-imposed model, we can see an error in the prediction. This is
probably due to the fact that the strain measured during moving test is not that accurate like in the static
calibration and this result in lower quality force and, therefore, in error in pressure prediction. Results of the
displacement-imposed model for the same test is shown, as we can see there’s a difference in the curve shape
since the predicted one is scaled starting from the displacement one.
A possible explanation for this uncertainty in the pressure prediction consist of two overlapping phenomena:
• When the piston stop within near-zero velocity static friction is established and this results in a friction
force. However, in a brake system, the friction coefficient is not the only things that cause a change in
friction force. Indeed, the normal load applied to the gasket depends a lot from the pressure magnitude.
This behavior’s amenable to the seal shape that causes an increasing normal load on the seal introducing
nonlinearity within the friction force estimation. When the master cylinder piston is moving, pressure and
force showed a good correlation. Once displacement becomes stationary and the static friction is
established, the relationship between force and pressure starts to diverge and is not recovered until the
difference reaches a certain value that causes the piston to move.
• Another phenomenon is the delay between the displacement force and pressure. Firstly, the force at
master cylinder start to rise and, once the force wins the preload value, the piston at the master cylinder
start to move, however, it’s only when pads come into contact with the disc that pressure will increase.
This means that between pressure and displacement there will be always a delay.
This means that displacement-imposed model will be able to well approximate the pressure shape during the
transient phase even with a little delay due to the aforementioned phenomenon. However, these contributions
are reasonable for most of the performed tests.
In the following list, main observations are resumed
• Looking at the global assessment for the RMSE, it is clear that both models are good enough to allow
pressure prediction within the UC2.2 braking system.
• To improve displacement-imposed model an approximate measurement for the pad disc clearance can
be measured in further campaign test.
• A good pressure prediction is possible even neglecting the Coulomb friction within various components.
• Related to the test procedure, calibration could be revisited in order to assess leakage entity within
various conditions. Single-step test results more suitable for our purpose and carrying out with one step
maneuver at different displacement will be better to evaluate leakage parameter.
• Regarding the instrumentation, at least one more pressure sensor needs to be installed in order to carry
out a proper validation for the master cylinder and asses pipelines elasticity.
• Strain measurement can be affected by chassis motion within various driving conditions.
2.3.4
Discussions and Conclusions
In this work, two complete brake plant models were carried out, a force-imposed model and a displacementimposed model with feedforward displacement. In the overall model, all the contributions were considered,
GA # 769506
D3.6– Tools supporting advanced EV trade-off process - PU
27 / 37
hydraulic, thermal and wear. Model validation for the hydraulic part has been carried out within different
manoeuvres and within various moving, tests ensuring a statistical overview of the model accuracy. Moreover, a
comparison between the two models is made in order to assess the pros and cons of both.
Above all, the most important thing that comes out from this study is a generalizable approach that allows the
experimenter to move from raw dataset to a model coherent with the experimental results in a reduced number
of steps. Within this purpose, a critical analysis of the test procedures and the measuring instruments is done
considering all the limitations that these two aspects introduce.
Lastly, we focused on the hysterical behaviour in the hydraulic circuit of which a detailed study is reported. Further
improvements related to this aspect are closely related to the calibration test procedures, that needs to be
reorganized in order to accomplish the goal.
Trying to draw out overall conclusions
• Force imposed model and displacement-imposed model shown satisfying results within various moving
tests procedures even with some limits due to nonlinearity that are not considered within this thesis.
• New methods able to characterize hysteresis behaviour have been carried out. This part has not been yet
completed, however interesting conclusion on how the actual experimental data should be used and
improved using other datasets from new activities, has been made.
Further improvements for this work will be carry out new calibration test procedures, to see if a proper hysteresis
characterization is possible, perform an in-depth study on real-time capabilities and optimizing within this purpose
and use a hardware platform able to translate input from user to input to software in order to enable the model
to virtual reality environment. Finally, this methodology allows standardized process as well as process execution
validation in early stage of the EV design development concept phase.
GA # 769506
D3.6– Tools supporting advanced EV trade-off process - PU
28 / 37
3
Tools supporting virtual system integration studies (SIE-SAS, CRF)
In this chapter, several methods for integration will be described and have been illustrated in different use cases.
In system integration stage (complete vehicle modeling), subsystems generally come from different department.
As example, electric powertrain and thermal comfort could be developed in parallel, as already described in
previous deliverable [1]. Nevertheless, final validation by combining these subsystems and studying their
interaction is mandatory to tends towards a closer behavior in terms of energy consumption.
Another challenge in integration process is the reliability in terms of connection. Indeed, the definition of inputs
and outputs must be addressed early in design phase [8] to ensure consistent subsystem model development. So
dedicated tool and method should be applied to ensure numerical stability and correct connection to reach
consistent system integration studies as highlighted in Figure 0-1.
3.1
3.1.1
CRF/SIE-SAS – UC2.5: Integrated tool for virtual integration
Configuration and automation capabilities
Until now, with conventional powertrain, most part of components that have an impact from the energetic point
of view have been modelled separately. In fact, from energy management point of view, cabin comfort auxiliary
systems are not considered because the normative does not require their activation, and the cooling systems
neither because the cycles are performed in a temperature condition (20-23°C) in which they do not affect
significantly the powertrain performances.
Therefore, all subsystems are studied separately with minor integration between them (only some boundary
conditions), as illustrated in Figure 3-1. This previous approach led to estimation of consumption, without
considering realistic behavior or by combining consumption and thermal comfort.
Nowadays, an assessment of the vehicle energetic behavior in a more realistic condition than the one required for
the CO2 homologation is becoming very important in order to avoid a too large gap between the vehicle
consumption declared and the one observed by the customers. This item is of course important for vehicle with
conventional powertrain, but it is even more to be monitored for electric vehicles in which the autonomy can have
an important variation depending on the environment conditions in which the car works, the auxiliary’s systems
activation and the driving style.
Figure 3-1:
State of the art of virtual integration in CRF
A more realistic mission of the car means new complex requirements for the simulation tools that should be more
accurate in several different conditions with a higher fidelity of the models.
Therefore, during this project, an enhanced modelling approach has been developed (described in D3.4), with
complete integration of subsystem in order to integrate all subsystems in only one simulation tool. In this way, all
GA # 769506
D3.6– Tools supporting advanced EV trade-off process - PU
29 / 37
subsystems interact with others by sharing variable boundary conditions along more realistic driving cycle,
especially for the thermal comfort, as illustrated in Figure 3-2.
Basically, 1D model, corresponding to the complete vehicle modelling from electric powertrain to all thermal
subsystem, has been connected with 3D phenomenon like external flow, flow under the hood and internal flow in
cabin, calculated using CFD on HPC.
Figure 3-2:
3.1.2
Integrated tool for virtual integration in CRF
Standardized processes implementation and user interfaces
By using standardized interface, integration can be easily performed, avoiding lot of wasting time in integration
process. This approach is also applied for al level of modeling during the design cycle from functional model to
mapped-based or detailed one. With this approach, the connection between system is identical and only physics
inside subsystem model is changed.
Furthermore, the reduction of simulation tool by focusing on multi-physics tool like Simcenter Amesim™ [14]
allows easier and faster integration for most of subsystem in a vehicle, as illustrated in Figure 3-3. Another
important advantage is the numerical stability, because some information is generally lost in cosimulation
environment due to the communication interval, which could lead to an energy misbalance.
Figure 3-3:
Vehicle subsystem integration in Simcenter Amesim
Nevertheless, some physical phenomenon could not be caught by 1D simulation, like 3D flow or 3D magnetics
field calculation. In this case, some numerical “bridges” have been developed to connect these different levels of
GA # 769506
D3.6– Tools supporting advanced EV trade-off process - PU
30 / 37
modeling, especially between Simcenter Amesim™ 1D system modeling and Simcenter STAR-CCM+™ 3D flow
modeling, as illustrated in Figure 3-4.
Figure 3-4:
1D/3D coupling interface integration
All models are easily piloted thanks to a python script that allows to launch 1D model locally and send 3D model
on HPC before retrieving results.
3.1.3
Process execution validation on case studies
A dedicated coupling strategy has been developed in this use case to exchange data in a smarter way, by triggering
the call of the 3D simulation and by storing previous results, as illustrated in Figure 3-5.
Some control signals have been defined, and their variation produces a CFD call (blue points in Figure 3-5). When
an already run CFD calculation is needed, stored data are recovered and therefore a new CFD call is avoided (red
points in Figure 3-5), with a consequent reduction in computational time.
Figure 3-5:
Coupling strategy between 1D and 3D models
GA # 769506
D3.6– Tools supporting advanced EV trade-off process - PU
31 / 37
This methodology allows interaction of all subsystems for a better understanding and study capability, but also
allows running 3D simulation in driving cycle context in a limited time, as highlighted in Figure 3-6. We can
observe that, thanks to this smart coupling, computational time becomes lower than one day for each analyzed
scenario.
Figure 3-6:
Simulation time for different scenarios
All results will be detailed in deliverable D6.1 (Design of new e-drive concepts, optimal system sizing based on
high Level virtual system integration tools and simulation report on assessment of virtual simulation methods).
Additional results are also available in different publications [15], [16] & [17].
GA # 769506
D3.6– Tools supporting advanced EV trade-off process - PU
32 / 37
4
Conclusions
This document presents different tools developed by different partners (RT-SAS, VALEO, SIE-SAS, UNIFI, SIE-NV &
CRF) either for early design phase or for virtual integration studies. All these tools are fully in line with OBELICS
objectives to reduce time and effort in EV concept development. All detailed results and conclusion obtained
thanks to these tools are described in deliverable D6.1.
Each tool is designed for a specific purpose from electric vehicle architecture design exploration to full complete
electric vehicle integration analysis. These tools answer different requirements presented in the OBELICS grant
agreement:
• Configuration and automation capabilities in the tools developed by RT-SAS and methods proposed by
SIE-SAS
• Standardized process implementation and user interface in the tools developed by RT-SAS and UNIFI and
CRF
• Process execution validation in the tools developed by RT-SAS and CRF
Tools developed by RT-SAS and methodology developed by SIE-SAS & VALEO focus on the trade-off process
optimization by considering different electric vehicle powertrain architectures and analysis the relevancies of each
of them. Processes have standardized to generate specification of all electric subsystem like the battery and the
E-motor. UNIFI’s tool is dedicated to generic braking model integration, but not only the plant model but also the
control dedicated with a focus on energy saving and methodologies. SIE-SAS and CRF focus on standardized
methodology for virtual integration allowing 1D and 3D model co-simulation on desktop and HPC. This method is
being standardized through python script and allow deeper trade-off analysis with complete subsystem
interaction impact.
Finally, these tools developed by simulation expert are accessible to non-expert thanks to standardized process
and user interface leading to design phase time reduction which is key objective of this project.
GA # 769506
D3.6– Tools supporting advanced EV trade-off process - PU
33 / 37
5
Abbreviations and definition
ABS
AC
ACEA
BBC
BMS
CA
CAD
CCL
CFD
DC
DCL
EBD
ECU
EM
ESC
ESP
EV
FMI
FMU
GSP
GSPDB
HiL
HMI
HPC
HV
HVAC
ICE
IWM
IM
ITEA
LV
MCU
MiL
NN
PI
PMSM
RSM
RT-SA
SiL
UIC
VMA
VCU
WRSM
AntiBlockierSystem
Alternative current
European Automobile Manufacturers' Association
Brake Blending Controller
Battery Management System
Control Allocation
Computer Assisted Design
Charge Current Limit
Computational Fluid Dynamics
Direct Current
Discharge Current Limit
Electronic Brakeforce Distribution
Engine Control Unit
Electric Motor
Electronic Stability Control
Electronic Stability Program
Electric Vehicle
Functional Mock-up Interface
Functional Mock-up Unit
Global Simulation Platform
Global Simulation Platform DataBase
Hardware In the Loop
Human-Machine Interface
High Performance Computing
High Voltage
Heating, Ventilation & Air Conditioning
Internal Combustion Engine
In-Wheel Motor
Induction machine
Information Technology for European Advancement
Low Voltage
Motor Control Unit
Model In the Loop
Neural Network
Proportional Integral
Permanent Magnet Synchronous Machines
Response Surface Methodology
Renault Trucks SA
Software In the Loop
Union International of Chemins de Fer
Vehicle Modular Architecture
Vehicle Control Unit
Wound Rotor Synchronous Motor
GA # 769506
D3.6– Tools supporting advanced EV trade-off process - PU
34 / 37
6
6.1
Risk Register
Risk register
Mention here the risks that are linked to this deliverable. See the list of risks on the OBELICS sharepoint:
https://projects.avl.com/16/0142/Documents/02_Deliverables/OBELICS_Deliverables%20listTIMELINE_27092017.xlsx?Web=1
If a new risk occurred, please introduce in the table below, and mention;
“With reference to the critical risks and mitigation actions this deliverable is not linked to any open risk. See the
monitoring file of the WPLB
https://projects.avl.com/16/0142/Documents/02_Deliverables/OBELICS_Deliverables%20listTIMELINE_27092017.xlsx?Web=1
New identified risks that occurred are listed in the table below.
Risk No.
What is the risk
Probability
of risk
occurrence1
Effect of
risk2
Solutions to overcome the risk
WPx.x
Describe here the risks!! And please
refer to the section of the text in the
document dealing with this.
Indicate
the level
Indicate
the level
Give a description how to
overcome the risk / give here
possible solution(s)
1
Probability risk will occur: 1 = high, 2 = medium, 3 = Low
2
Effect when risk occurs: 1 = high, 2 = medium, 3 = Low
GA # 769506
D3.6– Tools supporting advanced EV trade-off process - PU
35 / 37
7
References
[1] M. Ponchant, L. Pugi, R. Kluppel, H. Benzaoui and R. Mothier, "D3.2 Toy model," OBELICS, Lyon, 2018.
[2] T. F. J. P. Matthieu Ponchant, "D3.4: "Control calibration tools for real driving operations"," OBELICS, 2019.
[3] T. Katrasnik, I. Mele, K. Zelic, F. Sellier, R. N. Da Fonseca, H. Teichmann and Y. Firouz, "D2.1: "Innovative
battery modeling techniques"," OBELICS, 2019.
[4] F. Sellier, D. Roiu, A. Primon, S. C. Kaeck, H. Benzaoui, D. Miljavec, K. Yamamoto, M. El-Baghdadi and M.
Ranieri, "D2.2: "Innovative E-motor modelling techniques"," OBELICS, 2019.
[5] M. Baghdadi, F. Sellier, R. Estrada Vazquez, S. S. Guduguntla, A. Primon, K. Yamamoto and E.-H. Ourami,
"D2.3: "Innovative modelling techniques"," OBELICS, 2019.
[6] E. Silvas, Integrated optimal design for hybrid electric vehicles, Eindhoven: TU-Eindhoven, 2015.
[7] Siemens,
"Simcenter
HEEDS,"
Siemens,
[Online].
Available:
https://www.plm.automation.siemens.com/global/en/products/simulation-test/designexploration.html.
[8] M. Ponchant, A. barella, G. Stettinger and H. Benzoui, "D3.1 Standardized Model Integration," OBELICS, Lyon,
2017.
[9] T. F. N. T. F. S. Benzaoui Hellal, "D3.3: "Methods and tools for EV optimization"," OBELICS, 2019.
[10] L. B. e. al., "Brake Blending Strategy on Electric Vehicle Co-simulation Between MATLAB Simulink ® and
Simcenter Amesim™," in 5th International forum on Research and Technology for Society and Industry
(RTSI), Florence, 2019.
[11] D. e. al., "Modeling and Identification of an Electric Vehicle Braking System: Thermal and Tribology
Phenomena Assessment," in WCX2020, Detroit, 2020.
[12] A. Bergmann, "Benefits and Drawbacks of Model-based Design," in KMUTNB: IJAST, 2014.
[13] L. Z. Z. Y. D. Meng, "A dynamic model for brake pedal feel analysis in passenger cars," in Proceedings of the
Institution of Mechanical Engineers, Part D: Journal of Automobile Engineering, 2016.
[14] SISW, Simcenter Amesim Reference guide, leuven: Siemens Industry Software NV, 2019.
[15] N. Tobia and M. Ponchant, "Methodology Applied to Couple 1D & 3D Models for Electric Vehicles Thermal
Management Design," in JSAE, Yokohama, May 2019.
[16] N. Tobia and M. Ponchant, "Methodology applied to couple 1D & 3D models on HPC in context of electric
vehicle Fiat 500e thermal management design," in 32nd Electric vehicle Symphosium, Lyon, May 2019.
[17] M. Ponchant and N. Tobia, "Methodology Applied to Couple 1D & 3D Models for Electric Vehicles Thermal
Management Design; Use of High Fidelity Models for HPC Smart Coupling," in Siemens Simcenter
Conference, Amsterdam, 2019.
[18] G. M. T. Grigoratos, "Brake wear particle emissions: a review," in Environ Sci Pollut Res, 2015.
[19] J. Wahlström, "A comparison of measured and simulated friction, wear, and particle emission of disc brakes,"
in Tribology International, 2015.
GA # 769506
D3.6– Tools supporting advanced EV trade-off process - PU
36 / 37
8
Acknowledgement
The author(s) would like to thank the partners in the project for their valuable comments on previous drafts and
for performing the review.
Project partners:
Partner
Partner organisation name
no.
1
AVL List GmbH
2
Centro Richerche Fiat SCpA
3
FORD Otomotiv Sanayi Anonim sirketi
4
Renault Trucks SAS
5
AVL Software and Functions GmbH
6
Robert Bosch GmbH
7
SIEMENS INDUSTRY SOFTWARE NV
8
SIEMENS Industry Software SAS
9
Uniresearch BV
10
Valeo Equipements Electroniques Moteurs
11
Commissariat à l’Energie Atomique et aux Energies Alternatives
12
LBF Fraunhofer
13
FH Joanneum Gesellschaft M.B.H.
14
National Institute of Chemistry
15
University Ljubljana
16
University Florence
17
University of Surrey
18
Das Virtuelle Fahrzeug Forschungsgesellschaft mbH
19
Vrije Universiteit Brussel
Short Name
AVL
CRF
FO
RT-SAS
AVL-SFR
Bosch
SIE-NV
SIE-SAS
UNR
Valeo
CEA
FhG-LBF
FHJ
NIC
UL
UNIFI
US
VIF
VUB
Copyright ©, all rights reserved. This document or any part thereof may not be made public or disclosed, copied or otherwise
reproduced or used in any form or by any means, without prior permission in writing from the OBELICS Consortium. Neither OBELICS
Consortium nor any of its members, their officers, employees or agents shall be liable or responsible, in negligence or otherwise, for
any loss, damage or expense whatever sustained by any person as a result of the use, in any manner or form, of any knowledge,
information or data contained in this document, or due to any inaccuracy, omission or error therein contained.
All Intellectual Property Rights, know-how and information provided by and/or arising from this document, such as designs,
documentation, as well as preparatory material in that regard, is and shall remain the exclusive property of the OBELICS Consortium
and any of its members or its licensors. Nothing contained in this document shall give, or shall be construed as giving, any right, title,
ownership, interest, license or any other right in or to any IP, know-how and information.
This project has received funding from the European Union’s Horizon 2020 research and innovation programme under grant
agreement No 769506.
The information and views set out in this publication does not necessarily reflect the official opinion of the European Commission.
Neither the European Union institutions and bodies nor any person acting on their behalf, may be held responsible for the use which
may be made of the information contained therein.
GA # 769506
D3.6– Tools supporting advanced EV trade-off process - PU
37 / 37
Mathieu Sarrazin