Driving Innovation in

Vehicle Dynamics
High performance physical modeling and simulation increases
efficiency and productivity in vehicle design

A Cyb ern et Grou p Co m p an y

 Driving Innovation
High performance physical modeling and simulation
increases efficiency and productivity in vehicle design

Vehicle manufacturers are constantly faced and electronic controllers, where the control
with the challenge of balancing fuel efficiency software determines the dynamic behavior of
and safety against the demands for greater the system. In newer vehicles, it is common to
performance. This, along with the growing find 50 controllers or more.
need for reducing design and prototyping
The proliferation of in-vehicle controllers
costs, has led to the implementation of
means that there is a lot more software going
many innovations; not only in features of the
into the design of a vehicle. This, of course,
vehicle components, but also in the design
presents its own problems: embedded
and development process of those features.
software is expensive to develop, and bug-free
Through the use of model-based design and
software even more so. In addition, in order to
virtual prototyping, automotive manufacturers
have been able to reduce design and
prototyping costs significantly while fulfilling
the demands of governments and markets.
Authors
Driven by these needs, a wide range of
computer-aided modeling tools has emerged,
Paul Goossens
covering all aspects of the dynamic behavior
V.P. of Application Engineering,
of vehicles. However, these tools tend
Maplesoft to be very application-specific and use
computationally intensive numerical methods,
Dr. John McPhee such as finite difference and finite elements.
Professor, Engineering, University of While these are useful for the “off-line” stages
Waterloo, and NSERC/Toyota/Maple- in the design process, where time-to-result is
The proliferation of in-vehicle controllers means there
soft Industrial Research Chair not too critical, there is an increasing demand is a lot more software going into the design of a vehicle
for faster simulations to the point where high- which, of course, presents its own problems…
Dr. Chad Schmitke fidelity models need to be run in real-time for
Director of MapleSim Development, hardware-in-the-loop (HIL) testing. reduce time to market, component hardware is
Maplesoft often developed in parallel with the controller,
This need is becoming increasingly apparent
and manufacturers face the challenge of
Dr. Janos Pinter as manufacturers replace traditional passive
developing and testing the control software
components by active mechatronic systems
President, PCS Inc. without access to the real plant. The demand
in order to reduce weight and costs, improve
for better understanding of the dynamics, and
design efficiency, and control dynamic
the ability to produce high-fidelity physical
stability. These sub-systems are typically
models of vehicle systems that can be used in
made up of electro-mechanical components
HIL systems to test prototype controllers, has
reached a critical point for many companies.
Engineers in these companies frequently
need to go back to basics and develop
system models from first principles, deriving
the equations that describe the system (often
as systems of differential equations and/
or algebraic differential equations), solving
them, and then implementing the model in an
engineering simulation tool.
The development of the system (or “plant”)
model can be a time-consuming stage, often
taking up to 80% of the total project time,
especially if the wrong tools are used. Even
the use of tools like MATLAB® and Simulink®
requires significant manual manipulation of
System modularization drives the need for standard the math to get it into a form that they can
dynamic components and control systems across
vehicle platforms. Software determines system behavior
use. All of this is time-consuming, costly, and
and how the components interact with each other. error-prone.

2
 Vehicle Dynamics Application:
Kinematic Modeling, Simulation, and Optimization of a
Double-Wishbone Suspension
The dynamic stability of a vehicle,
as well as fuel efficiency and tire
wear, is directly affected by the
kinematic behavior of the suspension
Automatic Model Derivation mechanism. A great deal of effort is
This is one of the reasons why MapleSim , TM
needed to achieve the best possible
a high performance physical modeling movement of the wheel camber,
and simulation tool, is gaining popularity caster, and toe-in as the suspension
around the world. The appeal of MapleSim responds to a bump in the road. This is particularly complex
is that mathematical equations representing to achieve in a double-wishbone suspension because there
a model are automatically generated, are ten joints that influence the geometry of the system, known Hard Points - positions of joints
and wheel center
saving weeks, sometimes even months, on as the “Hard Points”.
complex projects. These equations are also The challenge is to compute the spatial coordinates of the
simplified automatically, yielding concise, hard points that will achieve the desired kinematics for the
computationally efficient models and high- vehicle. Using MapleSim, it is very easy to graphically define
speed simulations of sophisticated systems. the topology of the mechanism using blocks and lines from
which the equations that define camber, caster, and toe-in
will be automatically derived.
Once the model has been defined, the engineer can enter
the numeric values for the x,y,z coordinates of each hard
point. MapleSim symbolically derives the differential
equations that define the system, and simplifies these
equations through index reduction and algebraic
substitution. MapleSim then solves this resulting system
of ODEs to produce the camber, caster, and toe-in
over a given vertical
displacement applied
to the wheel (in this
case, a sine wave of
amplitude 100 mm to represent a bump and rebound).
These are the results for a given set of hard-point values,
defined prior to performing the calculations. Normally, these
Even tools like MATLAB and Simulink require
® ®

significant manual manipulation of the math to get it into are compared with standard curves to ensure they comply
a form that they can use. with the desired kinematics for the vehicle. On a first pass this is
very unlikely and the engineer will need to adjust the hard points
to give the desired curve, which can be very painstaking and
time-consuming.
By using MapleSim to develop the model, the
engineer arrives at an error-free mathematical A more efficient method is to import the standard curves as data sets and use the Global
solution in a fraction of the time it would take Optimization Toolbox for Maple. Using the imported curves as the objective functions, you
with more traditional approaches. can identify which hard points can be moved (very often, there are only a few that can be
moved because of other design constraints) and by how much.
For example, something as simple as a
DC motor requires the solution of a system The Global Optimization toolbox will then find the “best-fit” coordinates that will produce the
of two differential equations that need to kinematics closest to the desired curves.
be rearranged into a form that can then Once you have optimized the model you can then export it to Simulink® as an S-Function
be constructed using signal flow blocks block using the MapleSim Connector, so you have a very high fidelity component for
(integrators, gains etc.). With MapleSim, simulating the vehicle dynamics, particularly in real-time for hardware-in-the-loop testing in
you don’t have to worry about deriving the lab, but with close-to-road conditions.
the math. You simply re-create the electric
circuit diagram and add the mechanical
components (in this case, the rotor) on
the screen using compact and intuitive
components, then simulate.

3
 Powerful Math Simulation
MapleSim’s graphical user environment makes the burden of developing the math model
Behind MapleSim lies the power of Maple , TM

much lighter. For example, if you are modeling a multibody electro-mechanical system,
the world’s strongest math engine. MapleSim MapleSim allows you to define the system graphically and then generate the equations of
builds on Maple’s extensive numeric and motion from that.
symbolic computational abilities to perform
high speed simulations of very complex The concept is very simple: the MapleSim
models. This means that Maple can be Interface allows you to define the
used to access and view the equations system topology as a block diagram,
underneath a system model and perform where blocks representing the system
advanced mathematical analysis using components – linkages, flexible beams,
those equations. Extensive advanced pneumatic tires, electrical circuits etc. –
functionalities are available, ranging are connected to others to represent how
from optimization routines to a rich set of they relate to each other (essentially, the
control analysis tools. Maple itself is a rich “topology” of the system). The blocks
programming language, giving you complete have a number of parameters to set
control over your simulation runs. (mass, moments of inertia, resistance,
inductance, joint constraints, etc.) and Equations of motion are automatically generated from
Documents created in Maple are live. Any you can easily define points of interest by the graphically defined system.
changes made to a parameter or the model attaching probes to the model diagram.
results in dependencies being automatically
updated throughout the document. It is easy When the model is simulated, MapleSim automatically generates and simplifies the
to share and re-use design documents, governing equations that describe the behavior of the system.
in particular by attaching them to a single
The system model defined in MapleSim is tightly integrated with the Maple Technical
MapleSim model. Its Document Folder lets
Document interface. Within Maple, you can extract the equations underlying the model,
you keep your simulation models together
and perform detailed analysis using, among other things, Maple’s DynamicSystems
with all relevant design documents and
tools, a library of routines for analyzing linear systems (Bode, Root Locus, controllability,
custom component designs.
observability, etc.). These and other analysis tools can be used to optimize your design
parameters within the Maple environment. Optimized parameters can then be pushed back
and updated live into the MapleSim
model.
With the MapleSim Connector, the model
is easily evaluated in a form that can
be rapidly imported into Simulink®. The
MapleSim Connector is one of a range
of products from Maplesoft that aid in
the development and deployment of
dynamic models of vehicle systems.
By taking advantage of Maple’s built-in
code generation feature, as well as index
reduction and symbolic simplification
techniques, it produces an optimized
code version of the model, and
MapleSim allows you to define the system topology as a
block diagram, from which the equations of motion are automatically turns it into a Simulink®
automatically generated. S-Function (user code) block.

“The concept is very simple: MapleSim

allows you to define the system
graphically and then generate the
equations of motion from that.”

4
Real-Time Simulation
Once you’re satisfied with the model, you can use the MapleSim Connector to generate
a sub-system block for Simulink® with a click of a button. And, because the mathematical
model is in a highly compact form, it is possible to run the system model in real-time.
Benchmarks have shown much higher execution speeds with MapleSim-generated multi-
body mechanism models than with comparable models developed in other systems (such
as Adams and SimMechanics ). This allows the engineer to develop much higher-fidelity
TM TM

models for HIL testing.

As an example, in a recent design project, engineers were able to develop a full-vehicle
model of a Chevrolet Equinox including pneumatic tires (22 degrees-of-freedom, 36 state-
space variables) with a handful of blocks, then on to a dSPACE Simulator via the
MapleSim Connector.
Even with a modest-performance simulator (PowerPC 1 GH3), it was
possible to achieve update rates of 63 μs. With similar benchmark models,
execution speed has been up to 20 times faster than the same ADAMS
model and 16 times faster than SimMechanics . TM

Design Optimization
One of the major benefits of using MapleSim return an erroneous, unfeasible, or sub-optimal
for dynamic systems modeling is that it is very solution. For example, in models that could
easy to run through many “what-if” scenarios possess multiple optima, most traditional
and refine your design parameters. For numerical methods will find only the local
many people, this is as simple as adjusting optimum closest to an initial “guess” of the
parameter values and observing the results solution. One can visualize the model objective
until the acceptable results are achieved. Such as a hilly surface on which an automated
tweaking-by-hand approaches can easily search method attempts to find the “lowest
become time consuming to the point that valley”. A global scope search approach is
the time gains in the initial model development needed to achieve this.
are lost.
By using MapleSim’s live document templates,
There are a number of fully automated you can work with the mathematical model Model-fitting, or parameter matching, is a common use
optimization techniques available as off- of your system within the Maple environment of optimization techniques. It allows you to fit model
the-shelf tools. These allow you to present parameters to data in much more complex models than
itself, and take advantage of sophisticated with traditional regression techniques.
the model, the target behaviors (as model solvers and tools.
objectives) and the engineering, economic,
environmental (or perhaps other) model For example, the Maple Global Optimization Optimization Toolbox to determine these
constraints, and then the optimization method Toolbox combines a state-of-the-art parameters automatically. This is a relatively
will return the “best possible” solution. Note, solver suite with the intuitive modeling simple example to illustrate a point, but the
however, that some optimizers are better than environment of Maple. This allows you to principles can be applied to problems with
others at handling complicated nonlinearities. set up an optimization problem with little or many variables and constraints.
Therefore, care should be taken in defining the no knowledge of the underlying numerical
One of the many areas where the Maple
problem, and then selecting the method for the techniques. You simply set up your model
Global Optimization Toolbox is used
task at hand. This in itself requires specialized variables, their bounds, provide your target
extensively is in model fitting or calibration.
knowledge. function and constraints. The Maple Global
Very often, engineers will have a good
Optimization Toolbox, with its seamlessly
Used “blindly”, optimization solvers may either theoretical model of a system; however, they
integrated global and local scope search
not return a solution or, just as bad, they could do not have the necessary parameter values
algorithms, does the rest for you.
to make the model behave the same as the
For example, let’s consider a vehicle real system. In situations like this, you can
suspension system and assume that we need capture experimental data from the real system
to find the values for the spring stiffness and to determine the target output of the model
damper parameters that provide a desired and then use optimization techniques to find
ride. From the basic second-order model, we the necessary parameter values to match
can determine the time response of the system that output as closely as possible. Since you
using Maple’s DynamicSystems tools, but are typically dealing with highly non-linear
we would have to use initial parameters that systems, often with dynamic responses, the
will probably not give the desired response. use of global optimization for this type of
We could manually adjust the parameters application is necessary to obtain the “best
Using global optimization, you can rapidly arrive at and, with some effort, approach the solution fit”. This is a more general, flexible, and more
parameters that minimize the difference between the
response of a system and the desired response without
iteratively. To do this more accurately and successful model-fitting approach than using
manually iterating to a solution. efficiently, we can use the Maple Global traditional statistical regression techniques.

5
 Next-Generation Tools that Break Model Validation
Through the Barriers to Innovation and Simplification
Early in the process, control design
The automotive design engineer is well engineers have found that they can develop
engineers need to analyze their systems
served with software tools that provide dynamic system models that are much more
and often have to work with the equations
access to highly sophisticated computational numerically efficient than models developed
that describe them. When a system is
techniques for modeling and analyzing using more traditional tools. This means that
defined using a Simulink® block diagram,
vehicle dynamic behavior long before a incorporating dynamic vehicle models into
the related equations have to be derived
prototype is taken on the road. Many of the hardware-in-the-loop and driver-in-the-loop by hand, which is time consuming and
products that provide this capability do so in simulators is achievable at much greater error-prone. Added to this, there is a
a highly dedicated form that can often limit degrees of fidelity, further reducing the risk growing requirement for the ability to
the scope of the application. For instance, of design error and minimizing costs further assess the mathematical integrity of
traditional approaches to modeling require down the line. a Simulink® model, especially when a
the engineer to perform time-consuming,
Maple complements this and other existing component model has been provided
error-prone manual derivations of the
system equations which require advanced tools by providing a highly flexible, easy-to- by another group or an external supplier.
mathematical knowledge. Others account use environment in which the engineer can Again, there is little option but to derive
only for the multibody aspects of a vehicle explore ideas and concepts way beyond the these by hand. Until now…
with no regard for the drivetrain. inherent limits of more traditional tools.
The range of engineering products available By offering a powerful, flexible, easy-to-use
from Maplesoft offers a unique alternative. environment, along with an array of add-on Simulink® Model
MapleSim, with its next-generation physical products, application-specific user interfaces,
modeling techniques, makes it easier to analytical libraries and import/export
build and understand models of systems. capabilities, Maplesoft is able to provide the Import into Maple:
Model equations are automatically generated vehicle design engineer with tools that make Over 100 Equations
and simplified from physical system the job of defining systems, producing and
representations, yielding concise models validating systems models, and subsequently
and high-speed simulations of sophisticated deploying them to the tool chain, much
Mathematical Simplification:
Reduced to 11 Equations
systems. easier and quicker. This integrated approach
Furthermore, through the use of state-of-the- contributes significantly to the overall
art symbolic computation techniques, along efficiency and effectiveness of the vehicle
with the tools described in this article, many design process.

Export to Simulink®: 3 x Speed

up, No Loss of Fidelity

BlockImporter™ allows the engineer to

import a Simulink® model and convert it to
mathematical equations within the Maple
environment. This makes validation,
analysis, or simply gaining a better
understanding of the model much easier
– either by inspection of the equations
or by plugging in data and checking the
response.
Furthermore, Maple’s analytical capability
can be used to optimize and simplify
the model to provide a more concise
representation that runs significantly
faster than the original model without
losing fidelity. Used in conjunction with
the MapleSim Connector to create an
S-Function to bring the model back in
Simulink®, this approach can typically
deliver a significant increase in execution
speed, which can make the difference
between a feasible HIL model, or not.

6
Author’s Biographies Illustrative References
Pintér, J.D. (1996) Global Optimization in
Action. (Continuous and Lipschitz Optimization:
Paul Goossens, V.P. of Application Engineering, Maplesoft Algorithms, Implementations and Applications.)
Kluwer Academic Publishers, Dordrecht. Now
distributed by Springer Science and Business
Paul directs the Applications Engineering Group that supports Maplesoft’s
Media, New York.
new line of engineering modeling products. A mechanical engineer by
background, Paul has over 20 years of experience in both engineering and Pintér, J.D. (2006) Global Optimization with Maple:
An Introduction with Illustrative Examples. An
software business management. Past positions included senior management
electronic book published by Pintér Consulting
positions for companies in engineering modeling solutions and high- Services, Halifax, NS and Maplesoft, Waterloo,
performance real-time simulations. ON.

Pintér, J. D. (Ed.) (2006) Global Optimization:

Scientific and Engineering Case Studies. Springer
Dr. John McPhee, Professor, Engineering, University of Waterloo, Science and Business Media, New York.
and NSERC/Toyota/Maplesoft Industrial Research Chair
Autobeat Daily: Improving the Reliability of
Dr. John McPhee is a Professor in Systems Design Engineering, and the Networked Electronics, An interview with Kevin
NSERC/Toyota/Maplesoft Industrial Research Chair in Mathematics-based Kott, President dSPACE North America. http://
www.dspaceinc.com/shared/data/pdf/press/
Modeling and Design. Dr. McPhee’s main area of research is multibody system autobeat_viewpoint_dspace_060426.pdf
dynamics, with principal application to the analysis and design of vehicles,
mechatronic devices, and biomechanical systems. He has won many awards, Blundell, M. and Harty, D.: The Multibody
Systems Approach to Vehicle Dynamics. Elsevier
including a Premier’s Research Excellence Award and the I.W. Smith Award Butterworth-Heinemann, 2004
from the Canadian Society of Mechanical Engineers. Dr. McPhee completed
his term in 2009 as the Executive Director of the Waterloo Centre for Automotive He, Y. and McPhee, J.: A Review of Automated
Research, spending a sabbatical year at the Toyota Technical Center in Ann Design Synthesis Approaches for Virtual
Development of Ground Vehicle Suspensions,
Arbor, Michigan. He holds a PhD in Mechanical Engineering from the University SAE Congress 2007
of Waterloo, Canada.
K. Morency, J. McPhee, and C. Schmitke, Using
Graph Theory and Symbolic Computing to
Dr. Chad Schmitke, Director, MapleSim Development, Maplesoft Generate Efficient Models for Vehicle Dynamics,
Proc. CSME Forum, Alberta, 19 pages, 2006.
Dr. Schmitke holds a Ph.D. in Systems Design Engineering from the
Reimpell, J., Stoll, H. and Betzler, J: The
University of Waterloo, where he serves as an adjunct professor. His Automotive Chassis, Engineering Principles,
research focuses on the symbolic modeling of multibody mechatronic Second Edition, SAE International
systems. Before joining Maplesoft, Chad worked as Chief Developer at
Bhatt, R.M. and Krovi V.N.: Product Review:
MotionPro, and at the Canadian Space Agency as a post-doctoral fellow, DynaFlexPro for Maple, IEEE Control Systems
developing tools for modeling robotic manipulators. Magazine, Volume 26 Issue 6, December 2006.

Dr. János D. Pintér, President, PCS Inc.

János Pintér, Ph.D., D.Sc. has authored and edited several books and
published more than 180 other peer-reviewed papers and other technical
documents. He has run PCS Inc. since 1994, and he is also an adjunct
professor at Dalhousie University, in Halifax, Nova Scotia, Canada.
Dr. Pintér’s R&D, consulting, and teaching activities are related to the
application of advanced optimization technology across a range of
business, engineering, and scientific decision problems.

7
 Maplesoft

Reduce Development Risk. Create Better Products. Get to Market Faster.

Maplesoft Engineering Solutions provide you with the expertise and tools you need to meet your project
requirements quickly and effectively. Offering experts in a variety of engineering fields, extensive experience
in model-based design, and the superior system-level modeling and analysis tools MapleSim and Maple,
Maplesoft can help you reduce development risk and bring high-quality products to market faster.

Products Services
Develop system-level models with MapleSim With expertise in a variety of engineering fields
and Maple, and dramatically reduce model and extensive experience in model-based design,
development time, gain greater insight into Maplesoft experts are available to help you solve
system behavior, and produce fast, high-fidelity your engineering design problems.
simulations.

MAPLESOFT
SERVICES
The Essential Tool for Mathematics and Modeling

Talk to a Maplesoft Engineering Solutions expert to

learn how we can help you with your design projects.
www.maplesoft.com

A C yb er n et Gr o u p C o mp an y

www.maplesoft.com | [email protected]
Toll-free: (US & Canada) 1-800-267-6583 | Direct:1-519-747-2373
© Maplesoft, a division of Waterloo Maple Inc., 2014. Maplesoft, Maple, and MapleSim are trademarks of Waterloo Maple Inc. MATLAB, Simulink, and SimMechanics
are trademarks or registered trademarks of The Mathworks, Inc. Adams is a trademark of the MSC Software Corporation in the United States and/or other countries.
All other trademarks are the property of their respective owners.