Preprints of the 19th World Congress

The International Federation of Automatic Control

Cape Town, South Africa. August 24-29, 2014

Electric Drive Vehicle Development and Evaluation Using System Simulation

A. Rousseau*, S. Halbach*, L. Michaels*, N. Shidore*, Na. Kim*, N. Kim*. D. Karbowski*
M. Kropinski**

*Argonne National Laboratory, Argonne, IL 60439

USA (Tel: 630.252.7261; e-mail: [email protected]).
**General Motors, 3300 General Motors Road, MI 48380 USA (e-mail:
[email protected])

Abstract: To reduce development time and introduce technologies faster to the market, many companies
have been moving to Model-based System Engineering (MBSE). In MBSE, the development process
centers around a multi-physics model of the complete system being developed, from requirements to
design, implementation and test. Engineers can avoid a generation of system design processes based on
hand coding, and use graphical models to design, analyze, and implement the software that determines
system performance and behavior. This paper describes the process implemented in Autonomie, a Plug-
and-Play Software Environment, to design and evaluate electric drive powertrain and component
technologies in a multi-physics environment. We will discuss best practices and provide examples of the
different steps of the V-diagram including model-in-the-loop, software-in-the-loop and component-in-
the-loop simulation.
Keywords: Simulation, System Engineering, Control, Electric Drive Vehicles, Model Based System
Engineering.

Project management requirements also exist, to ensure that all

1. INTRODUCTION
the files and data needed to create a complete simulation are
Building hardware is expensive, time consuming, and organized, maintained, and provided as a complete package.
limiting with respect to comprehending and accounting for
variation in a system design. Traditional design processes in The Autonomie software [2, 3] provides an integrated
the automotive industry often delay control system design environment and set of processes for managing,
until late in the product development process, in some cases interconnecting, and integrating dynamic models of vehicle
requiring several costly hardware iterations. To reduce costs components and subsystems, to build and execute multi-
and improve time to market, OEMs are turning towards physics system simulations. These simulations are then used
Model-based System Engineering by placing greater for many different types of analyses, from trade-off studies of
emphasis on modeling and simulation early in the alternative propulsion system architectures to detailed control
development process. MBSE is the formalized application of system design. Autonomie’s Plug-and-Play architecture
modeling to support system requirements, design, analysis, facilitates the use of component and subsystem models of
verification and validation activities, beginning in the selectable fidelity, permitting users to focus in detail on the
conceptual design phase and continuing throughout area of interest they want to investigate.
development and later life cycle phases [1]. To fully realize
the benefits of math-based design, flexible and reusable 2. MODEL BASED SYSTEM ENGINEERING
multi-physics models must be created.
Model-Based System Engineering [4] provides an efficient
But greater reliance on modeling and simulation does come at methodology that includes four key elements in the
some cost. Even if institutional inertia can be overcome, new development process: modeling a plant (from first principles
processes must be put in place to facilitate communication or system identification), synthesizing and analyzing a
between the model creators and consumers and to handle an controller for the plant, simulating the plant and controller
increase in the number of files, which can become quite together, and programming/deploying the controller. MBSE
significant and overwhelming. supports all these multiple phases and provides a common
framework for communication throughout the entire design
Model management introduces a set of requirements that are process.
quite similar to software management requirements. There is
a need to maintain version control for the models, as well as The MBSE paradigm is significantly different from
to implement a system-level model assembly capability, traditional methodology. Rather than using complex
comparable to the need to interface, compile, and link the workflows and extensive software code, designers can
code modules that comprise a complete software build. formulate advanced functional characteristics by using

Copyright © 2014 IFAC 7886

19th IFAC World Congress
Cape Town, South Africa. August 24-29, 2014

continuous-time and discrete-time computational building 3.1 Configuration Definition

blocks. These models and associated simulation support tools
The relative position of systems and their interconnections
can provide rapid prototyping, virtual functional verification,
are defined by a set of configuration files. For example, a
software testing, and hardware/software validation. In MBSE,
conventional vehicle can be represented by multiple
models are at the core of the development process, from
configurations. One configuration may have a torque
requirements development, through design, implementation,
converter and a gearbox. Another may have those two
and testing. The control algorithm model is an executable
systems grouped together under a single transmission system.
specification that is continually refined and elaborated
Both configurations are valid for a conventional vehicle, and
throughout the development process. MBSE allows one to
the selection between the two of them depends on the
improve efficiency by:
application. Figure 1 shows an example of a completed
- Using a common design environment across configuration in Simulink for generic component systems.
project teams
- Linking designs directly to requirements
- Integrating testing with design
- Investigating the effects of variation on system
performance and robustness
- Refining algorithms through multi-physics
simulation
- Automatically generating embedded code
- Developing and reusing test suites
- Automatically generating documentation
Figure 1: Configuration Example for Component (MIL)
- Reusing designs to deploy systems across
multiple processors and hardware targets
3.2 Selection through the GUI
Different steps are supported by a variety of approaches from The GUI manages the system definition complexity and
Model-in-the-Loop (MIL), to Software-in-the-Loop (SIL), makes it possible for novice users, as well as experts, to
Hardware-in-the-Loop (HIL), Rapid-Control-Prototyping quickly find the files they need, to stitch them together into a
(RCP), or Component-in-the-Loop (CIL). Each process is cohesive simulation, to execute standard work processes, and
used to address different stages of the development. to perform database management functions without becoming
Automotive companies have widely implemented these overwhelmed. The first step in defining a model is to select
processes from code generation [5], to RCP [6] and HIL [7]. its configuration. As the user selects various systems in the
GUI, several configuration options may be available in the
3. PLUG-AND-PLAY ARCHITECTURE TO SUPPORT GUI and are selected through Drag-and-Drop. Once the
MBSE configurations are selected, experts select their models and
initialization files to define a system.
While most companies currently use MBSE, there is
currently no single tool that allows them to share and reuse 3.2 Building the Entire Model Automatically
models from one phase to the next. Autonomie [3] is a
software package designed in collaboration with General Autonomie uses a unique approach that allows users to build
Motors (GM) to support the ideal use of modeling and a complete system model on the basis of configurations and
simulation for math-based automotive control system design. systems selected from the GUI. This gives users the
While Autonomie does not meet all the needs of the flexibility of saving and versioning models independently
development process, it provides a framework to link and without having to manually connect each model’s inputs and
integrate multiple tools (i.e. requirements, plant modeling, outputs. Users select the correct files in a user interface, and
optimization, database management…) into an integrated the automatic building algorithm uses metadata associated
workflow. Autonomie supports the assembly and use of with the models to create the correct connections. This GUI
models from design to simulation to analysis with complete also uses the metadata to facilitate the other necessary
plug-and-play capabilities. Models in the standard format functions, such as compatibility checks and file selection.
create building blocks, which are assembled at runtime into a Figure 2 shows how each selected system from the GUI is
simulation model of a vehicle, system, subsystem, or built and connected to represent the final model. It also
component. All parts of the Autonomie graphical user allows component experts (e.g., engine experts) to focus on
interface (GUI) are designed to be flexible to support the areas they are familiar with, while being able to evaluate
architectures, systems, components, and processes not yet their component both as a standalone system as well as
envisioned. This allows the software to be molded to within a vehicle context. Systems can be exported and
individual uses, so it can grow as requirements and technical transferred that way, or the systems can be shared through the
knowledge expand [2]. This flexibility also allows for database. Autonomie includes as many pre-built systems as
integration of legacy code, including models, controller code, possible to give users a “jump start” into creating vehicles
processes, and post-processing equations. from scratch.

7887
19th IFAC World Congress
Cape Town, South Africa. August 24-29, 2014

Once the states are selected, the next step is to define the
component operating points under different conditions.
Particular focus should be paid to thermal effects as shown in
Figure 4.

Figure 2: A Powertrain Is Assembled by System Experts on

the basis of Selected Configuration
The unique architecture of Autonomie allows users to
perform multi-physics system simulation through the
different steps of the V-diagram using a single tool. The
following sections provide examples of the tool’s
Figure 4 Prius HEV Engine Operating Conditions for
capabilities.
Different Temperatures
4. MODEL IN THE LOOP EXAMPLE
The complete vehicle model can then be validated ensuring
Model-in-the-loop (MIL) can be used to address a very wide that each individual component behaves properly compared
range of questions. As mentioned previously, pure simulation to test data. Figure 5 shows an example for the 2010 Toyota
is usually used toward the beginning of the process, to allow Prius HEV model comparing fuel consumption, battery state-
engineers to study the performance of the system and design of-charge and component temperature. Outstanding
the control algorithm(s). Due to the short simulation time, a differences are attributed to test driver uncertainties which
very large number of simulations can be quickly performed. influence engine ON events and consequently battery SOC.
This section provides an overview of some of the main
applications that can be addressed using Autonomie for MIL.

4.1 Vehicle Model Validation

Before performing any analysis, it is important to validate the
component and sub-system models [8, 9] to provide
confidence in the results. Autonomie allows users to import
and analyze component and vehicle test data within the same
environment as the models, thus facilitating the analysis and
validation. Most of the vehicle test data used to validate the
models comes from Argonne’s Advanced Powertrain
Research Facility (APRF). For hybrid and plug-in hybrid
electric vehicles, one of the critical steps of the analysis
consists of understanding the vehicle energy management.
Indeed, it is important to understand the different states of the
system (i.e. engine ON/OFF, shifting, regenerative
braking…) and the selection criteria from a control point of
view. For example, the engine turn ON conditions of the
Prius HEV 2010 are shown in Figure 3 (i.e. the engine will be Figure 5. Prius HEV Model Validation with Test Data
turned ON above a wheel torque demand of 400Nm at a
wheel speed of 20 rad/s). 4.2 Impact of Powertrain Comparison
1600
Electric drive vehicles offer increased flexibility from a
1400
powertrain configuration point of view with hundreds of
possible combinations. As a result, it is critical for any multi-
(Nm)

1200
demand(rad/s)

1000
physics system simulation tool to handle a wide number of
torque demand

800
powertrain configurations to understand the advantages and
drawbacks of different options under various driving and
Wheeltorque

600

400
thermal conditions [10]. Autonomie currently includes close
Wheel

200
to 200 different powertrain configurations.
0

-200
Figure 6 shows a comparison between the GM Voltec
powertrain configuration and a series hybrid configuration for
0 10 20 30 40 50 60 70
Wheel speed
Wheel speed (Nm)
(rad/s)

Figure 3. Prius HEV Engine Turn ON Conditions Used for a PHEV application. Such analysis provides researchers with
Vehicle Level Control insight on component sizing, performance and energy

7888
19th IFAC World Congress
Cape Town, South Africa. August 24-29, 2014

management. In the example shown below, the sizing results 5. SOFTWARE IN THE LOOP EXAMPLE –
showed that the GM Voltec powertrain requires less PRODUCTION CONTROL DEVELOPMENT (GM)
component power to meet the vehicle technical specifications
Once high level analysis is performed during the MIL phase,
than does a series hybrid as a result of the many component
higher fidelity models and controls can be used to develop
efficiencies between the engine and the wheel. In addition,
and test production controls. During that phase, it is critical
simulations were performed to characterize the impact on
that the tool be able to handle a wide range of modeling
component operating conditions and fuel consumption during
languages and model complexity. Autonomie is currently
urban and highway driving. Using the series mode in the GM
being used by GM in several critical projects to accelerate the
Voltec implies that a relatively larger motor is required to
implementation of the next generation of advanced
address the vehicle’s power requirement. In the GM Voltec
technologies for engine, transmission, and hybrid
power-split mode, the motor is used to assist the engine; in
applications (including the next generation Chevrolet Volt
the plug-in series hybrid, the motor works as a generator. It
and Chevrolet Malibu with eAssist). The methodology used
was determined that the GM Voltec powertrain consumed up
for virtual ECU software development is that of system
to 5% less energy during all driving condition modes than did
simulation, which calls for the entire system to be modeled,
a pure series hybrid configuration on the UDDS cycle.
including plants, sensor/actuators, controller hardware, and
algorithm/application software.

Although plant models may be available, or can be readily

developed, the control algorithm models are not all available,
and the only representation of the algorithm functionality
may be contained in existing software. In these cases, a full
system simulation can be performed only when a complete
software build has been created and loaded into the target
embedded controller. Then the control algorithm software
EV1 EV1 EV1 can be tested in a hardware-in-the-loop (HIL) system, where
the actual controller and its application software is connected
Charge Depleting mode (SOCinit = 80%) Charge Sustaining mode (SOCinit = 30%) to a real-time computer system that runs a simulation of the
plant model(s) representing the physical system with which
Figure 6. GM Voltec PHEV and Series PHEV Comparison
the controller will eventually be used. This is only possible
on the UDDS driving Cycle
when controller hardware and associated software are
available, and is quite difficult and costly when multiple
4.4 Impact of Vehicle Energy Management through GIS
controllers are involved.
Another critical benefit of system simulation is the ability to
quickly and efficiently develop and evaluate the impact of System simulation has a number of challenging requirements,
advanced vehicle level energy management algorithms. In the including:
case of plant models that link to expert tools (i.e. AMESim, • The need to support flexible modelling,
GTPower…), Autonomie supports integration of third party • The ability to interconnect many complex models
tools for specific use cases. Since optimization based control involving hundreds of input and output signals,
methods for Plug-in Hybrid Electric Vehicles require the • Representation of all control system functionality,
knowledge of an entire driving cycle and an elevation profile when many control function models are not
to obtain the optimal performance over fixed driving route, available
linking with additional tools such as HERE for traffic • Model sharing facility, where models are made
information is quickly becoming of upmost importance [12, available to a large user community
13]. Figure 7 shows an example of route based control
including a geographical information system (GIS). The challenges listed above were solved with a combination
of the Autonomie software developed at Argonne National
Laboratory and a unique software-in-the-loop (SIL)
capability created at GM Powertrain Engineering [14]. Since
the adoption of Autonomie, the Electric Vehicle Controls
group at GM has been using this methodology for virtual
controls development. GM has created a distributed work
environment that leverages the plug-and-play capabilities of
Autonomie such that plant models, sensor/actuator models,
control algorithm models, and compiled production software
components of the system that are developed by different
people throughout the organization may be seamlessly
Figure 7: Route Based Control Implementation using integrated into a system simulation. This facility enables the
Geographical Information System use of models having varying degrees of fidelity that are
implemented in a number of commercial simulation tools.

7889
19th IFAC World Congress
Cape Town, South Africa. August 24-29, 2014

Figure 8 shows the controller model structure and includes: 6. COMPONENT IN THE LOOP EXAMPLE
• A block for the real-time operating system (RTOS)
Figure 10 shows a block diagram representation of BIL.
that schedules the execution of the tasks in the
software and algorithm models
• A Controller Area Network (CAN) communication
block to supply serial data signals to the controller
software
• An algorithm model (AM) block that contains the
AMs under development
• An hardware I/O (HWIO) block that contains
behavioral models of the HWIO functions in the
controller

Figure 10: Battery in the Loop Block Diagram

Battery hardware is subjected to power demands from a

virtual vehicle in Autonomie [15]. A high-voltage DC power
RTOS SIL Algo HWIO supply (the ABC-170 CE) sinks and sources power from the
battery to emulate battery utilization on standard
dynamometer cycles or real-world driving. Real-time
feedback from the battery (SOC, voltage, temperature,
CAN current limits) is used by the virtual vehicle energy
management controller as a part of its strategy. To control
hardware or receive feedback, specific logic needs to be
introduced. This provides a convenient point for the user to
Figure 8: Configuration Selected for Software-in-the-Loop implement a testing plan or to enforce checks to ensure safe
operation of the hardware. Figure 11 describes the generic
A formal process has been established to allow users to configuration setup used for hardware/software interactions.
archive and share models and complete system simulations. Note that the configuration selected for CXIL application is
Plant models are reused for HIL as well as system simulation, radically different than for both MIL and SIL.
minimizing rework and redundant development. Migration
from one controller software release to the next is easily
accommodated by the automated features of the Software-in-
the-Loop process and the Autonomie interconnect capability.
The open architecture of the system simulation facility has
permitted integration of GM internal tools, such as the Global
Automated Test Tool (GLATT) and the Test Automation
Suite (TAS). Figure 9 illustrates an example of the
automated insertion of a control algorithm model into the
simulation.
Figure 11: Configuration for Component-in-the-Loop
The Autonomie simulation of a midsize pre-transmission
parallel vehicle with a 41-A·h, 10-kWh JCS battery in
electric operation was compared to the same vehicle with an
actual physical battery (same specification) by using BIL.
The aim of the experiment was to validate the battery
simulation model. Proper validation is only possible if the
battery model can be seamlessly replaced by an actual
battery, with the rest of the vehicle model remaining
untouched. Autonomie provides the flexibility for such kinds
of experiments. After the simulation, Autonomie was
configured for CIL, as described in the section above. Figure
12 compares the Autonomie simulation model voltage (red)
with the actual battery voltage (blue).
Figure 9. An Algorithm Model Inserted Into a Baseline
System Simulation

7890
19th IFAC World Congress
Cape Town, South Africa. August 24-29, 2014

simulated
ACKNOWLEDGEMENTS
This work was supported by DOE’s Vehicle Technology
Office under the direction of Lee Slezak and David
Anderson. The submitted manuscript has been created by
UChicago Argonne, LLC, Operator of Argonne National
Laboratory ("Argonne"). Argonne, a U.S. Department of
Energy Office of Science laboratory, is operated under
measured Contract No. DE-AC02-06CH11357. The U.S. Government
retains for itself, and others acting on its behalf, a paid-up
nonexclusive, irrevocable worldwide license in said article to
reproduce, prepare derivative works, distribute copies to the
Figure 12: Battery Voltage Comparison public, and perform publicly and display publicly, by or on
behalf of the Government.
7. CONCLUSIONS
To reduce costs, the automotive industry has been embracing
Model-based System Engineering for modeling, simulation, REFERENCES
testing, and analysis. This paper proposes an ideal modeling 1. INCOSE SE Vision 2020 (INCOSE-TP-2004-004-02), Sept 2007
process, wherein experts produce libraries of high-quality 2. Argonne National Laboratory, Autonomie Documentation, 2013.
models in varying levels of fidelity for use throughout an 3. Argonne National Laboratory, Autonomie, www.autonomie.net, 2013
organization and across the automotive industry. These 4. Paul Smith, The MathWorks, “Best Practices for Establishing a Model-
models connect seamlessly for maximum reusability and Based Design Culture,” SAE 2007-01-0777, SAE World Congress.
flexibility, making collaboration quick and easy. The models 5. General Motors, “Automatic Code Generation Process,” MathWorks
developed by these experts can be used from the beginning to
Automotive Conference, June 2005.
the end of the design process, from high-level configuration
6. Oguz Dagci, General Motors, “Real Time Interface Blockset
sorting studies, to code testing with production software
Development in Matlab/Simulink for On-Target Rapid Prototyping,”
(such as SIL, HIL, or RCP), and, finally, to solving
SAE 2006-01-0169.
production problems. Each system (e.g., plant or control)
7. Sergey Semenov, Ford Motor Company, “Automation of Hardware-in-
can be either represented by a set of equations or by the
actual hardware. To quickly evaluate new technologies in the-Loop and In-the-Vehicle Testing and Validation for Hybrid Electric
different environments, a generic process was developed to Vehicles at Ford,” SAE 2006-01-1448.
replace any part of the system through the graphical interface. 8. N. Kim, M. Duoba, N. Kim, and A. Rousseau, "Validating Volt PHEV
This generic process allows companies to increase their Model with Dynamometer Test Data Using Autonomie," SAE Int. J.
productivity and bring technologies to the market faster. The Passeng. Cars - Mech. Syst. 6(2):985-992, 2013,
tools and methods described in this paper have been adopted 9. N. Kim, A.Rousseau, E. Rask, “Vehicle-level control analysis of 2010
by a large number of companies. For example, the software- Toyota Prius based on test data”, Proceedings of the Institution of
in-the-loop process has been adopted by the General Motors Mechanical Engineers, Part D: Journal of Automobile Engineering
Electric Vehicle Controls group to enable virtual ECU Volume 226 Issue 11, November 2012
software development via system simulation. Control 10. N. Kim, J. Kwon, A. Rousseau, “Comparison of Powertrain
algorithm development in the context of a complete system Configuration Options for Plug-in HEVs from a Fuel Economy
simulation provides many advantages that result in high- Perspective”, SAE 2012-01-1027, SAE World Congress, Detroit
quality robust control systems. Engineers are able to account 11. B. Walton, A. Rousseau, “Fuel Efficiency Benefit for Electrified
for interactions among the many functions, subsystems, and Vehicles from Advanced Spark-ignition Engine Technologies”,
controllers that comprise modern automotive controls, and to EVS27, Nov 2013, Barcelona
study the effects of variation on the system's robustness. 12. N. Kim and A. Rousseau, "Sufficient conditions of optimal control
based on Pontryagin’s minimum principle for use in hybrid electric
The main features of the software that empower the algorithm vehicles," IMechE Part D: J. Automobile Engineering, vol. 226, no. 9,
developer are:
Sept. 2012, pp. 1160-1170.
• Automatically build a complete system simulation 13. D Lee, Suk Won Cha, A Rousseau, N Kim, D Karbowski, “Optimal
from the constituent subsystem models
Control Strategy for PHEVs using Prediction of Future Driving
• Provide a development environment that supports
Schedule”, EVS26, May 2012, Los Angeles
MIL, SIL, HIL, and virtual rapid prototyping
14. L. Michaels, et al. (Argonne National Laboratory), M. Kropinski, et al.
• Establish a set of tools and processes for enterprise-
(General Motors), “Model-Based Systems Engineering and Control
wide collaboration
System Development via Virtual Hardware-in-the-Loop Simulation,
• Support customization of architectures, models,
SAE 2010-01-2325, SAE Convergence Conference, October, 2010
processes, and analysis
15. N. Shidore, T. Bohn, “Evaluation of cold temperature performance of
• Allow simulation at any level – single component ->
the JCS-VL41M PHEV battery using Battery HIL”, SAE 2008-01-
subsystem -> Powertrain -> complete vehicle
1333, April 2008, Detroit
• Manage models, data, processes, results, and
controller code from research to production

7891