Research Paper

An Integrated Systems Engineering Method

for Conceptual Design of Vehicle Handling
Dynamics
Mandar Hazare∗ and Paul Venhovens

Department of Automotive Engineering, Clemson University, 4 Research Dr, Greenville, SC 29607

Received 1 September 2014; Revised 11 September 2015; Accepted 19 December 2015, after one or more revisions
Published online in Wiley Online Library (wileyonlinelibrary.com).
DOI 10.1002/sys.21333

ABSTRACT
The handling characteristics of road vehicles are important attributes that define a major part of its unique
selling proposition. The engineering process of tuning vehicle handling dynamics is a challenging process
due to the multitude of competing design requirements/parameters and requires a well-defined system-
atic approach. The primary objective of this publication is to describe an integrated system engineering
method for the conceptual design of vehicle handling dynamics early in the product development process.
A systems engineering-based simulation framework is described that connects subjective, customer-
relevant handling expectations to higher-level, objective vehicle engineering targets and consequently
breaks these targets down into subsystem-level requirements and component-level design specifications.
A case study supporting this proposed system engineering method for vehicle handling design is also
described in detail. The integrated systems engineering approach guides the engineering development
process and provides insight into the compromises involved in vehicle handling layout, ultimately saving
product development time and costs and helping to achieve a higher level of product maturity early in the
design phase. C⃝ 2016 Wiley Periodicals, Inc. Syst Eng 18: 659–675, 2015

Key words: Systems Engineering; Vehicle Handling Design

1. INTRODUCTION (relevant to the consumers), and reflects brand’s strategic

The handling characteristics of road vehicles are important at-
tributes that define a major part of the vehicle’s unique selling
propositions. For vehicle manufacturers with specific focus
on driving dynamics (emphasizing for example fun/joy of
driving), on-road vehicle handling is an important attribute of
the strategic brand “DNA.” The vehicle brand specific DNA
(shown in Fig. 1) represents the different vehicle attributes
∗ Author to whom all correspondence should be addressed (e-mail: mhaz-
are@g.clemson.edu).
Systems Engineering Vol. 18, No. 6, 2015
C⃝ 2016 Wiley Periodicals, Inc.
preferences regarding relative importance of the attributes.
Vehicle handling can be broadly defined as the interaction
between driver, vehicle and environment, which takes place
during transportation of people and goods [Bergman, 1969].
It is an indicator of driver’s ease of controlling the motions
of the vehicle while driving. The roots of vehicle handling
design theory can be traced to mid 1950s, when the first
comprehensive understanding of both the theory and practice
of the linear handling response of the automobile was
introduced [Olley, 1946–1947, Segel, 1993; Crolla, 1995].
Since then, the field of vehicle handling dynamics design
has come a long way, with a plethora of research activities
in almost all areas of vehicle handling including, theories

659
660 HAZARE AND VENHOVENS
Figure 1. Brand specific DNA or vehicle attributes.
explaining nonlinear limit handling behavior [Ono et al.,
1998], use of computer simulations [Allen et al., 1999],
complex multibody models [Pacejka, 2005; Blundell and
Harty, 2014], specialized handling measurement devices,
vehicle characterization test rigs [Garrott et al., 1988],
application of active control systems [Statistical Analysis of
the Effectiveness of Electronic Stability Control (ESC) 2007;
Browalski et al., 2010], etc. It should be pointed out that
even with these significant advances in the field of vehicle
handling and objectification of the handling characteristics,
subjective testing of the vehicle by trained test drivers still
dominates the final chassis setup and sign-off process.
The domain of vehicle handling engineering is focused on
developing and applying methods to qualify (and quantify)
the directional behavior of the vehicle during different driv-
ing maneuvers. Studying handling dynamics involves under-
standing the controllability and stability of the vehicle, and
is also closely associated with driver’s subjective interaction
with the vehicle. The engineering process of tuning the vehi-
cle handling dynamics is a challenging task due to the mul-
titude of (competing) design requirements and parameters.
Some of the key challenges involved in the process are listed
below:
• It is difficult to understand the voice of the customer
(i.e., interpret the driver’s expectations/perception of the
vehicle’s handling behavior) and translate these expecta-
tions into meaningful objective metrics to be used in the
vehicle product development process.
• It is challenging to find a unified chassis setup that meets
the expectations of various segments of drivers interested
in the same vehicle. Drivers of different age group, such
as young enthusiast and aging drivers, might have dif-
ferent expectations regarding handling behavior for one-
and-the-same vehicles.
• Different aspects of vehicle handling—such as steady-
state handling, transient handling, straight-line stability,
parking, and emergency handling—are often in conflict
with one another.
• The vehicle’s handling properties by themselves as part
of the overall brand specific DNA (Fig. 1), are often in
conflict with competing properties aspects, such as ride
comfort, acoustic comfort, or passive safety. The vehi-
cle brand specific DNA spider/radar diagram (shown in
Systems Engineering DOI 10.1002/sys
Fig. 1) allows for a visual representation of important
vehicle attributes and also show the relative importance
of the attributes.
• Trial and error approaches to find the best compromise
can lead to suboptimal solutions, resulting in increased
product cost and weight and the prolongation of product
development time.
To address the challenges above, it is important to establish
a well-defined and systematic approach toward handling dy-
namics design to ensure that the final product meets customer
expectations. Most vehicle manufacturers follow their own set
of unique, proprietary methods to design a vehicle’s handling
characteristics. Commonly, current best practice design meth-
ods [Kinstle, Hassler, and Johnson, 2009]:
• Benchmark competitor vehicles to develop vehicle-level
targets and component design specifications. Compet-
itive benchmarking can lead to products with perfor-
mance levels that go beyond customers’ expectations
and lead to unnecessary engineering effort, higher prod-
uct costs and weight, and product performance that
might not be perceived by the end-user.
• Rely heavily on the application of physical prototypes
during the initial development phase. Early-stage vehi-
cle prototypes are very expensive, not infinitely tunable
(such as the setup of static and dynamic body-in-white
stiffness), and frequently of insufficient build quality to
address competing vehicle properties like acoustics.
• Focus on physical testing of prototypes. Repetitive phys-
ical testing can be time-consuming and costly.
• Practice “trial-and-error” and “test-and-tune” philoso-
phies, which can be time-consuming if not supported
by sufficient system behavior knowledge. This approach
can also lead to an oversight of conflicting design objec-
tives, resulting in a sub-optimal final setup.
• Lead to expensive design changes late in the develop-
ment process, especially if the interaction with and de-
pendency of the handling on other vehicle properties and
design parameters is not properly understood.
The objective of this research is to address the challenges
described in the previous section by developing a method
for conceptual vehicle/chassis layout based on a systems
engineering approach. A system engineering process provides
 SYSTEMS ENGINEERING METHOD FOR VEHICLE HANDLING DESIGN
Figure 2. Systems engineering process “V” diagram.
a comprehensive, sequential, and top-down approach for
the successful realization of complex systems. According
to International Council of Systems Engineering (INCOSE)
[International Council on Systems Engineering Systems
Engineering Handbook, 2006], systems engineering seeks to
focus on the definition of customer needs and requirements
early in the development process and subsequently proceed
with design synthesis and systems validation. The systems en-
gineering process inputs focus primarily on the stakeholder’s
(i.e., the customer, legislator, manufacturer, or retailer) needs,
objectives, expectations, requirements, project target visions,
and business objectives. These process inputs are used to
derive system-level, functional, and performance targets,
which are then realized by the systematic development of
subsystem-level requirements and component-level specifi-
cations. The systems engineering approach is often described
with a “V” diagram, shown in Figure 2. The left-hand side
of the “V” diagram deals with decomposition and definition
of requirements, the bottom with product design using the
defined requirements, and the right with the integration and
verification of the requirements through testing.
Successful implementation of a system engineering
method for the design and development of complexly
engineered systems is in itself challenging. For example,
it is often not straightforward to systematically decompose
and define the requirements and specifications for the
different design and development levels due to “build-in”
system trade-offs and competing properties. Furthermore, the
implementation method must be concurrent and consistent
(concurrent here implies that the individual tasks at different
levels are carried out separately in parallel, and consistent
implies that the key interactions among different design tasks
are identified, observed, and enforced until the concurrent
design process results in a final product [Kim et al., 2003]).
One approach to assuring concurrency and consistency is
the use of a decomposition-based target cascading method-
ology [Kim et al., 2003]. In a decomposition-based target
cascading methodology, the entire complex problem is par-
titioned or decomposed into smaller and simpler problems. A
661
decomposition-based approach helps to better understand and
explore the compromises and trade-offs involved between the
different subsystems and hence provides valuable insights for
the system engineer. Once the system is decomposed, the tar-
gets, or specifications for top levels, are identified first. These
targets are propagated, or cascaded, systematically to the rest
of the system (i.e., the subsystems and smaller components.
The actual design tasks are executed locally at subsystem and
component level, and interaction with the rest of the system
is revisited only when a target cannot be met. This often
leads to an iterative target cascading process. When the design
decisions can be modeled analytically, the process can be
formalized as a multilevel optimization problem referred to
as Analytical Target Cascading (ATC) [Kim et al., 2003, Li,
Lu, and Michalek, 2008].
In this publication, a simulation based design method for
conceptual design of vehicle handling characteristics is de-
scribed using decomposition based target cascading process
and systems engineering principles.
2. PROPOSED METHOD FOR DESIGN
OF VEHICLE HANDLING DYNAMICS
Systems engineering approach toward product design always
begins with understanding the customer’s expectations from
a particular product, to make sure that the final product meets
the end-user’s expectations as best as possible. Equally im-
portant for the product developer is to understand the com-
pany’s brand essence and how the branded products need to
be designed for and aligned with the brand strategy in mind
to ensure consistency in the message and product experience.
Translating these general ideas to vehicle handling design
targets requires a target setting process that is in alignment
with the brand essence (or DNA) and a target realization and
tracking process that assures that higher-level specific cus-
tomer expectations are met during the vehicle development
process.
Systems Engineering DOI 10.1002/sys
662 HAZARE AND VENHOVENS
Figure 3. Systems engineering method for conceptual design of vehicle handling dynamics.
Systems engineering recommends a systematic top-down
approach toward product design where the customer require-
ments are cascaded step-by-step from overall high-level vehi-
cle targets into subsystem level requirements and component
level specifications. The targets, requirements and specifi-
cations must be validated at each step during the product
engineering and build phase as part of multiple design review
processes. For the specific application of conceptual design
of vehicle handling dynamics, the five steps involved in this
process are:
Step 1: Define driving maneuvers and qualitative metrics
of vehicle handling based on correlative anal-
yses between customers’ handling expectations,
brand DNA targets, and objective metrics.
Step 2: Quantify handling metrics.
Step 3: Develop a set of knowledge-based lower-order
models as the basis for engineering design opti-
mization.
Step 4: Develop and apply a multiobjective, multisce-
nario optimization framework to drive the prod-
uct design.
Step 5: Validate and verify the recommended design con-
figurations to ensure customer satisfaction.
Figure 3 schematically shows the application of system
engineering process for conceptual design of vehicle handling
dynamics.
Step 1: Define driving maneuvers and qualitative metrics
of vehicle handling based on correlative analyses
Systems Engineering DOI 10.1002/sys
between customer’s handling expectations brand
DNA targets, and objective metrics.
The first step is to understand the customer’s relevant ve-
hicle handling expectations. The average consumer often de-
scribes handling highly subjectively with attributes such as
“fun to drive,” “sporty,” or “safe”. Translation of these subjec-
tive attributes into the engineering domain is a big challenge
in itself. Customers with different lifestyles and backgrounds
(i.e., age groups, income levels, and hobbies) might have
very different expectations with respect to vehicle handling
behavior, which makes the qualification and quantification of
customers’ vehicle handling requirements even more difficult.
A possible approach to understanding customer’s han-
dling expectations is based on the use of marketing research
and clinics to better understand the product characteristics and
features desirable for customers. Clinics, driving events, and
marketing surveys aimed at understanding end-user prefer-
ences and expectations of vehicle handling can be used as the
first step in the vehicle dynamics development program.
Customer handling expectations should be defined and
quantified with respect to the various domains of vehicle han-
dling. The vehicle handling domains are formulated consider-
ing the handling performance requirements of a driver during
different scenarios of vehicle operation. Step 1 requires the
development of statistically relevant correlations between
customer expectations and qualitative objective metrics,
which can then be used by chassis engineers for the develop-
ment of vehicle handling targets used in the product design
phase. Table I shows the qualitative overview of vehicle-
 SYSTEMS ENGINEERING METHOD FOR VEHICLE HANDLING DESIGN 663

Table I. Vehicle Handling Domains and Objective Metrics

Handling Domains Description of Handling Domains Handling Objective Metrics

Steady-State Handling Scenarios of constant speed; constant steer angles with vehicle Understeer Gradient, Yaw Rate
turning along a constant radius of curvature Gain, Roll Gain, Side Slip Angle
Gain
Transient Handling Scenarios of changing yaw velocity, side-slip velocity, and path Yaw Rate Time Constant, Lateral
curvature; represents vehicle’s response during dynamic situations Acceleration Phase Lag, Yaw
(i.e., turn-entry and turn-exit); evaluated with metrics: agility, Rate Damping Ratio, Roll Angle
responsiveness and damping Overshoot, Roll Angle Response
Time
Steering Feedback Steering system response, described in terms of steering-wheel torque Steering Torque Feel (Torque vs.
(Off-Center) feedback, of vehicle during normal driving scenarios Angle, Torque vs. Lateral
Acceleration Gradient)
On-Center Steering Steering system response during straight-line driving at highway Steering Torque Time Lag (vs.
speeds steering angle) at Low Lateral
Accelerations
Emergency (Limit) Vehicle’s response during critical maneuvers such as obstacle Yaw Stability, Roll Stability
Handling avoidance
Parking Ease of vehicle maneuverability during low-speed, high-steer angle Static Parking Torque, Turn Circle
maneuvers Diameter, Lock-to-Lock Steering
Turns
Coupled Dynamics Vehicle’s directional stability in scenarios where cornering is coupled Yaw Rate Increment
with other dynamic motions such as braking or acceleration
Road Adaptability Handling behavior of vehicle on different road surfaces (i.e., rough Yaw Rate Increment
roads, bumps, or low friction surfaces).
Straight-Line Stability Pull and drift behavior of vehicle (i.e., tendency of vehicle to deviate Pitch Gradient, Straight-Line
from intended path during straight-line cruising, acceleration, and Stability Index
braking scenarios); vehicle’s response during acceleration and
braking on split-mu surfaces.
Disturbance Sensitivity Vehicle’s straight-line performance in presence of external Yaw Moment Sensitivity
environmental disturbances such as winds, road crown, and road
roughness.

In this research, all handling domains described in

Table I contribute to the total objective vehicle handling
DNA of a vehicle (see Fig. 4).The handling DNA represents
one arm of the overall vehicle DNA (shown in Fig. 1) and
provides a more detailed representation of vehicle handling
characteristics (and the relative importance of the different
vehicle handling domains). It is important to note that there
are dependencies and trade-offs among the different domains
of vehicle handling. For example, designing a vehicle for
stability during emergency handling usually results in an
understeered vehicle that customers may perceive as less
agile and sporty during normal driving scenarios. Making
a vehicle more agile, and thus oversteered, will result in a
vehicle setup that is more nervous with regard to straight-line
stability. Making a vehicle easier to turn using a low steering
ratio leads to high steering effort and torque during parking.

Figure 4. Typical vehicle handling DNA. Step 2: Quantify handling metrics.

Once the various vehicle-handling scenarios and objective

handling domains associated with different, everyday driving metrics associated with different everyday driving tasks are
tasks (scenario). As part of the systems engineering process, identified, the next step is to quantify objective metrics with
each scenario should be quantified with objective metrics that realistic numbers. These metrics can then be used in the prod-
captures the driver’s assessment of vehicle handling behavior. uct development process.

Systems Engineering DOI 10.1002/sys

664 HAZARE AND VENHOVENS
Figure 5. Target cascading flow diagram.
There are three traditionally different approaches for quan-
tifying objective handling related metrics:
1. Physical testing of vehicles using highly trained pro-
fessional test engineers (Bergman [1969, 1978], Weir
and DiMarco [1978], Mimuro et al. [1990], King et al.
[2002], Xia and Willis [1995], Chen [1997], Lincke,
Richter, and Schmidt [1973], and Good [1977]). The
subjective feedback of trained test engineers using qual-
itative engineering descriptions such as “progressive
handling,” “predictable behavior,” “cornering traction,”
“overall grip,” “direct steering response,” etc., corre-
lates with objective responses measured by sensors in-
stalled on the vehicle. This approach assumes that the
nonexpert target customer’s expectations can be suffi-
ciently described by expert test engineers’ subjective
judgments. This approach, although most widely used
within the industry, can lead to over-engineered prod-
ucts tuned for expert professional test engineers/drivers
instead of “normal” end-users.
2. Simulation-based strategies (McRuer and Klein [1978],
Venhovens and Hazare [2012], Horiuchi and Yuhara
[2000], and Ishio, Ichikawa, Kano and Abe [2008]).
This approach relies on simulation methods and re-
quires synthesized vehicle dynamics and driver models
to predict and assess the driver’s perception of vehi-
cle handling dynamics. A simulated driver model is
used in conjunction with a vehicle dynamics model,
and the adaptive parameters of the driver model are
used as indicators of the vehicle’s handling quality.
According to McRuer and Klein [1978], drivers adapt
their driving control in such a way that they maintain
Systems Engineering DOI 10.1002/sys
near-constant closed-loop driver-vehicle system perfor-
mance. The gains and lead time-constants are indicators
of the driver’s perception of handling quality. A range
of gains exists that are perceived as enjoyable by the
driver. Gains that are too high (perceived as “twitchy”
by customers) or too low (“sluggish”) lead to degraded
perception of handling quality. Although this method
seems very attractive, uncertainty associated with the
human driver simulation model has restricted this
approach.
3. Clinics and analytic research. This approach isolates
the driver’s preferences and perceptions of particu-
lar vehicles and brands by conduction-driven events
with nonexpert target consumers to help manufactur-
ers understand customer expectations and preferences.
Drivers’ preferences are then correlated with objective
metrics using statistical tools (i.e., regression and cor-
relation techniques). These objective metrics can be de-
rived from vehicle responses measured during physical
testing of the vehicles or by the use of simulated vehicle
dynamics models.
The quantification of vehicle handling objectives with re-
alistic targets requires understanding both the brand essence
and brand DNA targets. The brand image closely relates to the
customer’s perception of a particular brand in comparison to
the other brands; for example, the customer might perceive a
certain brand to be “sportier” or more “comfortable” than an-
other brand. Brand DNA targets provide a way to realistically
account for trade-offs between different conflicting attributes
(i.e., sportiness vs. comfort) relevant to the customer while
emphasizing certain attribute more than others.
 SYSTEMS ENGINEERING METHOD FOR VEHICLE HANDLING DESIGN 665

Table II. Handling Performance Metrics Vehicle A and B

Objective Handling
Handling Domains Metrics Unit Vehicle A (FWD) Vehicle B (RWD)
Steady-State Handling (v = 80 km/h) Understeer Gradient deg/g 0.727 1.197
Yaw Rate Gain 1/sec 0.491 0.307
Side-Slip Angle Gain deg/g −1.29 −1.34
Roll Angle Gain deg/g 2.39 3.57
Transient Handling (v = 80 km/h) Yaw Rate Time Constant ms 113 105
Yaw Rate Damping Ratio − 0.994 0.918
Lateral Acceleration deg −44.1 −43.4
Phase Lag
Roll Angle Overshoot % 1.9 8.0
Steering Feedback (v = 80 km/h) Steering Torque Gain (per Nm/deg 0.391 0.308
Steering Angle)
Steering Torque Feel (per Nm/g 20.1 25.4
Lateral Acceleration)
On-Center (v = 80 km/h) Steering Torque Time Lag ms 86 63
(@ 0.2 Hz)
Parking Lock-to-Lock Steering − 2.3 3.2
Rotations
Turn Circle Diameter m 10.52 10.47
Parking Static Torque Nm 13.2 10.3
Disturbance Sensitivity (v = 80 km/h) Yaw Moment Sensitivity deg/kN m-sec 3.327 2.301
Coupled Dynamics (v = 80 km/h) Yaw Rate Increment % −1.47 3.54
(Acceleration out of
Turn)
Road Adaptability (v = 80 km/h) Yaw Rate Increment % −0.85 −1.95
(Cornering on Rough
Roads)
Straight-Line Stability Straight-Line Stability Nm/N 1.197 1.518
Index
Pitch Gradient deg/g 1.99 1.79
Emergency Handling (Roll Stability) Static Stability Factor − 1.416 1.499
NHTSA Stars − 4 5

Step 3: Develop a Set of Knowledge-Based Lower-Order
Models as the Basis for Engineering Design Op-
timization.
The next step is to develop suitable vehicle handling mod-
els (physics-based, knowledge-based), which can capture and
connect the different vehicle handling scenarios and metrics
(described in Step 1). These “hybrid” lower-order models
with appropriate amount of complexity must be able to sim-
ulate vehicle behavior at each level (vehicle, subsystem and
component). Since these models will be used in an iterative
optimization framework (Step 4) it is important to ensure
that the models are easy to characterize, computationally in-
expensive, transparent and insightful for the chassis design-
ers.
Note that there are several highly complex vehicle dy-
namics models available (as part of commercially available
software packages). These commercially available handling
packages/models are often not suitable during the vehicle
concept development phase. Firstly, the detailed vehicle han-
dling software packages (such as multibody dynamics simu-
lation tools) requires building elaborate models with detailed
component specifications that are generally not available dur-
ing the initial stages of the conceptual vehicle design, sec-
ondly, having highly nonlinear and complex models make
the simulation (and numerical optimization) process very
computationally expensive, and thirdly, some commercially
available simulation tools are of a black-box nature which
means that the chassis engineer has no clear insight into the
descriptive language of the model make-up and models of
such nature often do not allow access to model parameters
by 3rd party optimization routines.
Step 4: Develop and Apply a MultiObjective, Multi-
Scenario Optimization Framework to Drive the
Product Design.
Systematically following Steps 1 and 2 will lead toward
the development of quantifiable engineering metrics that cor-
relate to customers’ vehicle handling expectations. As de-
scribed in Step 1, to comprehensively describe customers’
vehicle handling requirements, a variety of scenarios and
corresponding objective metrics are needed. These objective

Systems Engineering DOI 10.1002/sys

666 HAZARE AND VENHOVENS
Figure 6. Vehicle handling DNA performance spider diagram, higher value on handling spider diagram is better (indicates improvement).
handling metrics can be analytically computed using vehicle
handling models developed in Step 3.
Step 4 requires the availability of a multiscenario,
multiobjective optimization framework to balance competing
customer relevant vehicle handling requirements; that will
drive the product design development and optimization.
This optimization framework should account for the
interaction between various aspects of vehicle handling and
supports developing chassis subsystem- and component-
level design specifications with respect to realistic design
constraints. The framework can be applied in two consecutive
steps: in the first step, objective vehicle-level handling targets
derived from customer handling expectations are translated
into subsystem-level engineering requirements and balanced
against various competing design objectives using an
optimization method. In thesecond step, the subsystem-level
requirements are translated into component-level design
specifications where an optimization algorithm searches for
the best set of design parameters.
In this research, the chassis design problem is decom-
posed into a meaningful subsystem-level—such as sus-
pension, steering and tires—and component-level—such as
kinematics and bushing compliances—design problem. The
desired vehicle-level targets are cascaded systematically to
lower levels (i.e., subsystems) and components are rebalanced
upwards based on lower-level designs. Analytical models
for subsystems and components are identified and tied to-
Systems Engineering DOI 10.1002/sys
gether in the optimization framework. An iterative optimiza-
tion scheme has been established, which aims at reducing
the discrepancy between targets (from higher levels) and re-
sponses (from lower levels) in order to achieve a consistent,
optimized chassis design solution with respect to constraints
at the subsystem and component levels. Figure 5 shows the
overall target cascading flow diagram proposed.
Genetic Algorithms (GA) [Deb, 2001], a type of stochastic
optimization method is used at each level. As the vehicle
handling design process is complex and possibly multimodal
in nature, the use of stochastic optimization approaches will
ensure that the final optimal solution is not restricted to a
local minimum as with traditional gradient-based optimiza-
tion methods. Note that it is possible to use other stochastic
optimization methods for solving the vehicle handling design
problem. For example, Haghiac, Haque, and Fadel [2004]
used GA to solve a more simplified vehicle handling design
problem and compares the results obtained from using Ge-
netic Algorithms against other optimization methods such
as Monte Carlo [Scarlat et al., 2002; Chakravartula, Haque,
and Fadel, 2005] and Simulated Annealing [Chandramoulli,
2002]. The results and conclusions from their research indi-
cate that the GA approach has some distinct advantages over
the other methods.
The optimization framework has been implemented using
a decomposition-based Analytical Target Cascading (ATC)
methodology [Kim et al., 2003, Li et al., 2008]. ATC is an
 SYSTEMS ENGINEERING METHOD FOR VEHICLE HANDLING DESIGN
Table III. Customer Relevant Subjective Handling Attributes
and Objective Vehicle Handling Metrics
Customer Relevant
Subjective Handling
Attributes Objective Handling Metrics
Sporty 5 Metrics to be Higher Yaw Rate Gain,
Maximized and 8 Higher Steering Torque Gain,
Metrics to be Higher Steering Torque Feel,
Minimized Higher Damping Ratio,
Higher Yaw Rate Increment
(Acceleration-in-Turn).
Lower Understeer Gradient,
Lower Side-Slip Gain,
Lower Roll Gain,
Lower Pitch Gradient,
Lower On-Center Lag,
Lower Yaw Rate Time Constant,
Lower Lateral Acceleration Phase Lag,
Lower Roll Angle Overshoot.
Fun-to-Drive 2 Metrics Higher Yaw Rate Damping Ratio,
to be Maximized Higher Steering Torque Feel.
and 6 Metrics to be
Minimized
Low Side-Slip Gain,
Lower Roll Gain,
Lower Pitch Gradient,
Lower On-Center Lag,
Lower Time Constant,
Lower Lateral Acceleration Phase Lag.
Safety 2 Metrics to be High Understeer Gradient (Yaw
Maximized and 4 Stability),
Metrics to be Higher Static Stability Factor (Roll
Minimized Stability).
Lower Yaw Rate Increment
(Accelerating in Turn),
Lower Yaw Rate Increment (Rough
Road Cornering),
Lower Roll Angle Overshoot,
Lower Straight-Line Stability Index
(i.e., Higher Straight-Line Stability).
Comfort 2 Metrics to Higher Yaw Rate Time Constant,
be Maximized and 6 Higher On-Center Delay.
Metrics to be
Minimized
Lower Yaw Moment Disturbance
(i.e., Higher Side Wind Stability),
Lower Steering Torque Gain,
Lower Steering Torque Feel,
Lower Parking Torque,
Lower Lock-to-Lock Steering Turns,
Lower Turn Circle Diameter.
667
effective hierarchical multilevel optimization-based design
methodology; it applies a decomposition approach wherein
the overall system is split into subsystems, which are then
solved separately and coordinated via target-response consis-
tency constraints [Li et al., 2008].
In the most general form, ATC problem can be represented
as follows ([notations and formulations adapted from Li et al.
[2008] and Tosserams, Etman, and Rooda [2006])
∑
N ∑ ∑
N ∑
𝑚𝑖𝑛 f i j (x i j ) + 𝜋(ti j − ri j ) (1)
x ...,x 𝑁𝑀
11 i=1 J =𝜀i i=1 J =𝜀i
For Augumented Lagrangian ATC Formulation:
𝜋(ti j − ri j ) = 𝜆iTj (ti j − ri j ) + ||w i j ∙ (ti j − ri j )||22
s.t. gi j (x i j ) = 0
h i j (x i j ) = 0
(2)
ti j − ri j = 0
w her e x i j = [x i j , t(i+1)k ∀k ∈ Ci j ]
ri j = ai j (x i j )∀ j ∈ 𝜀i , i = 1, . … , N
Here, the system is decomposed into N levels with M
elements each. The subscript i j represents thejth element
of the system in the ith level. The variable f i j represents
the scalar objective function, and gi j ≤ 0, h i j = 0 are
the inequality and equality constraints respectively. Local
variables of element j are denoted by xi j . The variable
ri j is the response of element j calculated by analysis
model ai j . ∈i is the set of elements at level i, and Ci j is
the set of children of element j. ti j represents target vari-
able created for each shared variable. 𝜋 denotes the con-
sistency constraint relaxation function. In the case of the
Augmented Lagrangian (AL) method, applied to ATC for-
mulation, the consistency constraint function, 𝜋, is formu-
lated as a combination of the quadratic penalty function,
w =∥ w i j⋅(ti j −ri j ) ∥22 , w = [w i j , ∀ i, j] and the Lagrangian
function, 𝜆iTj (ti j − ri j ), 𝜆 = 𝜆i j , ∀ i, j ].
In this work, ten different scenarios of vehicle handling
dynamics are considered (see Table I). The vehicle handling
objective metrics from each scenario are grouped under four
customer-specific brand attributes derived from market data
(i.e., Sporty, Fun to Drive, Safety, and Comfort). The fitness
function at the first layer (vehicle-level) of the optimization
framework is based on customer-relevant brand attributes.
The design variables used in the first layer of the optimization
framework can be grouped into three separate subsystems—
tires, suspension, and steering.
In the second layer of the optimization framework, the sus-
pension subsystems are analyzed using separate kinematics
and compliance modules. The suspension kinematic charac-
teristics are represented by eight design variables namely, sus-
pension roll camber, roll steer, roll center height, mechanical
trail, scrub radius, king pin inclination, caster angle, and anti-
dive geometry. The suspension compliance characteristics are
represented with three design variables: namely, suspension
lateral force compliance steer, lateral force camber compli-
ance, and aligning moment compliance steer. The suspen-
sion kinematic characteristics are a function of suspension
geometry, and the suspension compliance characteristics are
a function of the suspension geometry and bushing stiffness.
Systems Engineering DOI 10.1002/sys
668 HAZARE AND VENHOVENS
Figure 7. Analytical target cascading flow diagram for design of vehicle handling dynamics.
In the proposed ATC optimization framework, a suspension
geometry model and a suspension compliance model work
separately to achieve the desired kinematics and compliance
targets set at the first layer of the optimization framework.
Note that both suspension kinematics and compliance mod-
els use suspension pickup points—three-dimensional spatial
coordinates—as common design variables and hence are rep-
resented as linking variables in the ATC framework.
Step 5: Validate and Verify the Recommended Design
Configurations to ensure Customer Satisfaction.
In a typical system engineering process, the derived
subsystem-level and component-level specifications are vali-
dated by physically building and testing systems at each level.
(Refer to the right-hand side of the “V” diagram shown in Fig.
2.) The upward process ensures that the final product meets
all desired vehicle level targets and end-user expectations.
For the conceptual design of vehicle handling dynamics
described in this research, it is recommended that the valida-
tion process be performed virtually - potentially with higher-
order, higher-quality simulation models. The component de-
sign specifications derived using the optimization framework
(Step 4) can be used to characterize commercially available
higher-order simulation tools, which usually have a higher
degree of correlation with real vehicle behavior. For exam-
ple, the suspension pick-up points generated from the opti-
mization algorithm can be used to characterize a multibody
simulation model, which can then be used to simulate the
Systems Engineering DOI 10.1002/sys
kinematics and compliance behavior validate suspension sub-
system targets, such as the compliance steer of an entire axle.
The simulation models can also be used to simulate different
vehicle handling scenarios to ensure that vehicle-level targets
derived during Step 2 are achieved. Real world testing with
physical prototypes can be used to complete the validation
process.
3. CASE STUDY
In this section, a case study based on the proposed systems
engineering framework for the systematic design of vehicle
handling characteristics will be carried out. The case study
comprises of two vehicles: vehicle A (compact FWD hatch-
back) and vehicle B (sporty RWD coupe).
In our case study, a vehicle customer satisfaction survey
from AutoPacific [2013] is used to understand customer’s
vehicle satisfaction with attributes that relate to vehicle han-
dling. The survey (2013 New Vehicle Satisfactory Survey)
consists of around 56,000 responses from consumers who
purchased a new vehicle within 6 months of filling out the
questionnaire. The survey captures new vehicle owners’ sat-
isfaction with their purchases with respect to different ve-
hicle attributes. From the general survey responses, four
key attributes—Sporty, Fun to Drive, Safety, and Comfort—
related to vehicle handling behavior were selected to correlate
the customer’s satisfaction to the various scenarios and objec-
tive metrics of vehicle handling.
 SYSTEMS ENGINEERING METHOD FOR VEHICLE HANDLING DESIGN
Figure 8. Vehicle handling DNA performance spider diagram, higher value on handling spider diagram is Better (indicates improvement).
From the survey, it was found that customers perceived
vehicle B to be more Sporty, Fun-to-Drive, and slightly more
Comfortable than vehicle A. The customers also perceived
vehicle A to be much Safer than vehicle B. In this scenario,
the challenge for a chassis engineer would be to develop a new
chassis setup solution for vehicle B, such that it improves its
Safety behavior without excessively sacrificing its Sporty, Fun
to Drive and Comfort attributes.
3.1. Objectives and Scope of the Case Study
To limit the scope of this case study, it is assumed that the
manufacturer (or the chassis engineer) would like to redesign
the chassis setup for vehicle B such that the new concept (say,
optimized-vehicle B), handles similar to vehicle A. It is also
assumed that the vehicle manufacturer can only redesign the
steering system, front and rear tires, and front-axle suspension
characteristics of vehicle B. Thus, the objective of this case
study is to illustrate a systematic chassis design process that
will attempt to give vehicle B the handling characteristics—or
the DNA—of vehicle A within realistic design constraints.
3.2. Quantification of Vehicle Handling
Characteristics
The handling requirements are grouped and categorized with
respect to the various scenarios of vehicle handling. These
669
include: steady-state handling, transient handling, on-center
handling, emergency handling, parking, steering feedback,
handling on different road surfaces, coupled dynamic corner-
ing, disturbance sensitivity, and straight-line stability.
Table II illustrates the differences in the handling char-
acteristics of vehicle A (compact FWD hatchback) and ve-
hicle B (sporty RWD coupe), described with respect to the
ten scenarios of vehicle handling. The four key customer-
relevant vehicle handling attributes (i.e., Sporty, Fun to Drive,
Safety, and Comfort) were correlated with the various han-
dling objective metrics shown in Table II. Table III shows
the higher-level relationship between the customer-relevant
handling attributes and the objective handling metrics used
for this case study. Figure 6 shows a relative performance
spider diagram comparing the handling performance DNA
(expressed in terms of customers’ subjective expectations) for
vehicles A and B.
Figure 6 is based on the numerical values of objective
metrics shown in Table II and empirical associations shown
in Table III. These correlations have been developed such that
a higher value of a metric on the spider diagram represents
an improvement of the attribute. The first step in developing
these correlations is to calculate the normalized values of the
vehicle handling objective metrics. The normalized values
are calculated by dividing the handling objective metrics by
values for a reference vehicle. In this case a best-in-class ve-
hicle was considered as the reference. The next step involves
either addition or subtraction of these normalized metrics,
Systems Engineering DOI 10.1002/sys
670 HAZARE AND VENHOVENS

Table IV. Handling Performance Metrics for Vehicle A, B and Optimized Vehicle B

Handling Objective Handling

Domains Metrics Units Vehicle A Vehicle B Optimized - Vehicle B
Steady-State Handling (v = 80 km/h) Understeer Gradient deg/g 0.727 1.197 1.171
Yaw Rate Gain 1/sec 0.491 0.307 0.352
Side-Slip Angle Gain deg/g −1.29 -1.34 −0.77
Roll Angle Gain deg/g 2.39 3.57 2.32
Transient Handling (v = 80 km/h) Yaw Rate Time Constant ms 113 105 85
Yaw Rate Damping Ratio 0.994 0.918 0.983
Lateral Acceleration deg −44.1 −43.4 −30.0
Phase Lag
Roll Angle Overshoot % 1.9 8.0 1.9
Steering Feedback (v = 80 km/h) Steering Torque Gain (per Nm/deg 0.391 0.308 0.327
Steering Angle)
Steering Torque Feel (per Nm/g 20.1 25.4 23.5
Lateral Acc.)
On−Center (v = 80 km/h) Steering Torque Time Lag ms 86 63 36
(@ 0.2 Hz)
Parking Lock−to−Lock Steering 2.3 3.2 2.8
Rotations
Turn Circle Diameter m 10.52 10.47 10.50
Parking Static Torque Nm 13.2 10.3 11.5
Disturbance Sensitivity (v = 80 km/h) Yaw Moment Sensitivity Deg/kN m−sec 3.327 2.301 1.937
Coupled Dynamics (v = 80 km/h) Yaw Rate Increment % −1.47 3.54 2.62
(Acceleration out of
Turn)
Road Adaptability (v = 80 km/h) Yaw Rate Increment % −0.85 −1.95 −0.83
(Cornering on Rough
Roads)
Straight−Line Stability Straight−Line Stability Nm/N 1.197 1.518 1.471
Index
Pitch Gradient Deg/g 1.99 1.79 2.12
Emergency Handling (Roll Stability) Static Stability Factor − 1.416 1.499 1.499
NHTSA Stars − 4 5 5

depending upon whether they need to be maximized or
minimized, to calculate the relevant handling attribute based
on associations shown in Table III. Figure 6 clearly indicates
that the vehicle B is Sportier than vehicle A; on the other hand,
vehicle A is perceived as much Safer than vehicle B. Although
not very apparent from the figure, it is found that vehicle B is
slightly more Fun-to-Drive and Comfortable than vehicle B.
3.3. Vehicle Handling Models
The key idea at this step is to develop models with an appro-
priate amount of complexity and that are easy to character-
ize, computationally in-expensive, transparent, and insightful
for chassis engineers. The models applied must be effec-
tive/efficient when used in an iterative optimization frame-
work.
The vehicle handling model used for this case study is
based on a vehicle model with roll, yaw, and lateral motion as
the three Degrees of Freedom (DOF). This is coupled with a
steering system model, which adds another DOF and accounts
for steering system compliance between the road wheel and
steering wheel. The tire force model is based on Pacejka’s
Magic Formulae [Pacejka, 2005] and includes a simple tran-
sient tire side force model extension based on a first-order lag
using the tire’s relaxation length as the time constant. The in-
fluence of steering system compliance, suspension kinematics
and compliance, weight transfer due to height of the center
of gravity, roll stiffness, and centrifugal forces are included
in the tire force calculations using effective axle cornering
characteristics [Pacejka, 2005]. Such effective cornering char-
acteristics include tire properties based on Pacejka’s Magic
Formulae, incorporate tire force dependency on slip angle
and vertical load, and provide mechanisms for combined cor-
nering and braking with tire force saturation. The suspen-
sion elasto-kinematic characteristics are modeled by using
the suspension compliance matrix formulations described by
Knapczyk and Dzierżek [1995].
3.4. Optimization Framework
A multiscenario, multiobjective optimization framework nec-
essary to account for customers’ conflicting vehicle handling

Systems Engineering DOI 10.1002/sys

 SYSTEMS ENGINEERING METHOD FOR VEHICLE HANDLING DESIGN 671

Table V. Optimized Vehicle Design Variables (Suspension, Steering and Tires)

Vehicle Design Variables

Suspension Parameters

Variables Units Vehicle B Optimized −Vehicle B % Change

Roll Stiffness (Total) Nm/deg 1,641 2,522 53.7
Roll Stiffness Distribution (Front) % 65.3 78.1 19.5
Roll Stiffness (Front) Nm/deg 1,072 1,969 83.7
Roll Stiffness (Rear) Nm/deg 569 554 −2.7
Front Wheel Rate (Suspension Stiffness, Front) N/mm 26.60 39.55 48.7
Rear Wheel Rate (Suspension Stiffness, Rear) N/mm 18.40 11.09 −39.7
Roll Stiffness (sway bar, Front) Nm/deg 573 1,194 108.3
Roll Stiffness (sway bar, Rear) Nm/deg 129 336 160.6
Shock Damping (Front) N−sec/mm 2,420 1,970 −18.6
Shock Damping (Rear) N−sec/mm 1,746 1,592 −8.8
Lateral Force Compliance Steer (Front) deg/N −1.01E−04 −4.65E−05 −53.8
Lateral Force Compliance Steer (Rear) deg/N 5.90E−06 5.90E−06 0.0
Lateral Force Camber Compliance (Front) deg/N 1.18E−04 9.04E−05 −23.4
Lateral Force Camber Compliance (Rear) deg/N 6.35E−05 6.35E−05 0.0
Roll Camber (Front) deg/deg −0.620 −0.907 46.5
Roll Camber (Rear) deg/deg −0.637 −0.637 0.0
Roll Steer (Front) deg/deg 0.035 0.037 6.1
Roll Steer (Rear) deg/deg −0.044 −0.044 0.0
Roll Center Height (Front) mm 57.00 38.59 −32.3
Roll Center Height (Rear) mm 99.7 99.7 0.0
Anti−Dive Angle (Front) deg 2.00 2.09 4.3
Anti−Dive Angle (Rear) deg 14 14 0.0
Steering Parameters
Steering Ratio − 16.4 14.8 −9.9
Mechanical Trail mm 30.00 25.83 −13.9
King Pin Inclination Angle deg 12.39 9.54 −23.0
Caster Angle deg 7.50 8.53 13.8
Scrub Radius mm 34.00 −5.15 −115.2
Aligning Torque Compliance Steer (Front) deg/Nm 2.83E−04 1.33E−04 −52.8
Aligning Torque Compliance Steer (Rear) deg/Nm 3.34E−04 3.34E−04 0.0
Tire Parameters
Cornering Stiffness (Front, per tire) N/deg 1,279 1,520 18.8
Cornering Stiffness (Rear, per tire) N/deg 1,145 1,349 17.9
Cornering Stiffness Load Dependence Coefficient a3 2,203 3,076 39.6
BCD = a3 * sin (2 * atan (Fz ./ a4)) a4 11.21 13.61 21.3
Camber Stiffness (Front, per tire) N/deg 191.9 227.9 18.8
Camber Stiffness (Rear, per tire) N/deg 171.7 202.4 17.9
Relaxation Length mm 422.50 146.02 −65.4
Pneumatic Trail mm 27.70 22.33 −19.4

requirements has been developed for this case study. The op-
timization framework is implemented using a decomposition-
based, Analytical Target Cascading (ATC) [Kim et al., 2003,
Li et al., 2008] methodology. The ATC optimization frame-
work developed for this research works in a two-layer opti-
mization schedule. GA [Deb, 2001], a type of evolutionary
optimization algorithm, are used at each layer of the frame-
work.
The first layer is used to derive subsystem-level
requirements from overall vehicle level targets. These
subsystem-level requirements are passed on as targets to the
second layer of optimization and the second layer attempts to
derive component-level specifications from subsystem-level
requirements derived in the first step. The second layer
optimization utilizes component level design variables and
analysis models and attempts to minimize the difference
between the targets transferred from the vehicle level and
response generated from the component level analysis. An
iterative loop is set up with an objective to minimize the
target/response consistency constraints, that is, the targets

Systems Engineering DOI 10.1002/sys

672 HAZARE AND VENHOVENS
Figure 9. Optimized suspension geometry configuration for vehicle B and optimized-vehicle B.
Table VI. Optimized Front Suspension Bushing Stiffness
Bushing Radial Stiffness (N/mm)
Vehicle B
Link 1 (Lower A Arm (F)) 3,305
Link 2 (Lower A Arm (R)) 1,885
Link 3 (Upper A Arm (F)) 7,831
Link 4 (Upper A Arm (R)) 3,531
Link 5 (Tie Rod) 1.0E + 09
at the vehicle level are constantly rebalanced to achieve a
consistent and feasible solution.
Figure 7 shows the ATC flow diagram illustrating the op-
timization process used in this study. As described earlier in
the paper, the first layer (vehicle-level) of the optimization
framework is based on customer-relevant brand attributes.
The design variables used in the first layer of the optimization
framework can be grouped into three separate subsystems—
tires, suspension, and steering. In the second layer of the
optimization framework, the suspension subsystems are an-
alyzed using separate kinematics and compliance modules.
The suspension kinematic characteristics are represented by
eight design variables namely, suspension roll camber, roll
steer, roll center height, mechanical trail, scrub radius, king
pin inclination, caster angle, and anti-dive geometry. The
suspension compliance characteristics are represented with
three design variables: namely, suspension lateral force com-
pliance steer, lateral force camber compliance, and aligning
moment compliance steer. The suspension kinematic char-
acteristics are a function of suspension geometry, and the
suspension compliance characteristics are a function of the
suspension geometry and bushing stiffness. Note that both
Systems Engineering DOI 10.1002/sys
Optimized-Vehicle B % Change
5,126 55.1
2,935 55.7
1,801 −77.0
9,741 175.8
1.0E + 09 0
suspension kinematics and compliance models use suspen-
sion pickup points—three-dimensional spatial coordinates—
as common design variables and hence are considered as
linking variables in the ATC framework.
3.5. Results from the Optimization Process
The objective of the optimization process is to derive a new
chassis configuration for vehicle B (called “optimized-vehicle
B”) such that it has handling characteristics are similar to
those of vehicle A. Most importantly, the objective was to
improve the Safety characteristics of vehicle B without ex-
cessively sacrificing its Sporty, Fun-to-Drive and Comfort
attributes. The steering system, front and rear tires, and front
axle suspension characteristics were set as design variables in
the optimization process.
Figure 8 shows a relative performance spider diagram com-
paring the handling performance DNA, expressed in terms
of customers’ subjective expectations, for vehicle A, vehicle
B, and optimized-vehicle B. The handling spider diagram is
based on correlations described in Table III and uses a ranking
 SYSTEMS ENGINEERING METHOD FOR VEHICLE HANDLING DESIGN 673

Figure 10. Optimized suspension kinematic characteristics (bump steer, bump camber).

Figure 11. Tire cornering stiffness (N/deg) vs. normal load (N).

scheme where a higher value on the spider diagram repre-
sents improvement in the handling attribute. From Figure 8
it is observed that the optimized-vehicle B is slightly less
Sportier than vehicle B but is Sportier than vehicle A; it is
also more Fun-to-Drive than both vehicle A and vehicle B.
The optimized-vehicle B is Safer than vehicle B but still less
Safe than vehicle A. Although not very apparent in the figure,
it was also observed that the optimized-vehicle B was the least
Comfortable among the three vehicles.
Table IV shows the final vehicle handling performance
metrics for the optimized configuration obtained from this
case study. The proposed system engineering framework

Systems Engineering DOI 10.1002/sys

674 HAZARE AND VENHOVENS
ensures that the proposed vehicle-level solution can always
be realized with feasible subsystem and component-level
specifications. Table V compares the subsystem-level design
variables for the optimized vehicle configuration (optimized-
vehicle B) and the original vehicle (vehicle B). Note that only
front suspension design variables were independent for this
case study (i.e., only the front suspension of vehicle B was
redesigned).
Figure 9 shows the optimized suspension geometry config-
uration (i.e., the re-designed suspension pick-up points). In
this case study, only wheel-side points of suspension assem-
bly were optimized, that is, only the front wheel knuckle was
redesigned. Figure 10 shows the suspension kinematic curves
for the optimized suspension configuration. Table VI shows
the optimized bushing stiffness obtained from the optimiza-
tion schedule. Figure 11 shows the tire cornering stiffness
characteristics obtained from the optimized schedule. Note
that both front and rear tires are assumed to have identical
characteristics in this case study.
The results of the case study indicate that the optimized-
vehicle B can only partially attain the handling character-
istics of vehicle A. There are three key reasons explaining
this effect, firstly, the entire optimization schedule is per-
formed with realistic design constraints; for example, pack-
aging constraints, which restricts drastic changes in suspen-
sion redesign. Specifically, only wheel-side points of front
suspension assembly were optimized, that is, only the front
wheel knuckle was redesigned. Secondly, it is important to
note that not all targets can be simultaneously achieved be-
cause of the inherent interdependence between conflicting
handling requirements and objective metrics. For example,
in this case study, the optimized–vehicle B becomes less un-
dersteer in steady-state response and simultaneously becomes
more responsive in transient scenarios. It was not possible
to simultaneously make the vehicle less understeer and less
responsive, while imposing the constraint of using the same
tires on front and rear axle. Lastly, it should be noted that
the two vehicles used for this case study belong to different
vehicle platforms and have completely different architectures.
They have different geometric and inertial properties which
has some fundamental influence on the handling characteris-
tics. Also, the fact that vehicle A is FWD and vehicle B is
RWD is another major constraint which cannot be entirely
compensated using the chassis optimization process.
The case study described here demonstrates the application
of a systems engineering approach to vehicle handling design.
This approach starts with understanding the customer’s ve-
hicle handling expectations and then systematically deriving
vehicle-, subsystem- and component-level specifications to
develop a new chassis configuration that meets the customer’s
requirements under realistic design constraints.
4. SUMMARY/CONCLUSION
The handling characteristics of a road vehicle are important
attributes that define a major part of its unique selling propo-
sition. The engineering process of tuning vehicle handling
dynamics is a challenging task due to the multitude of compet-
Systems Engineering DOI 10.1002/sys
ing design requirements and parameters and requires a well-
defined systematic approach.
The primary objective of this publication is to describe
an integrated system engineering method for the conceptual
design of vehicle handling dynamics early in the product
development process. A systems engineering-based
simulation framework is described that connects subjective,
customer-relevant handling expectations to higher-level,
objective vehicle engineering targets and consequently
breaks these targets down into subsystem-level requirements
and component-level design specifications. A case study
supporting this proposed system engineering method for
vehicle handling design is described in detail.
Such an integrated systems engineering approach will
guide the engineering development process and provides in-
sight into the compromises involved in vehicle handling lay-
out, ultimately saving product development time and costs
and helping to achieve a higher level of product maturity early
in the design phase.
REFERENCES
R.W. Allen, T.J. Rosenthal, et al., Computer simulation analysis of
light vehicle lateral/directional dynamic stability, Proceedings
of the SAE 1999-01-0124, Society of Automotive Engineers,
Warrendale, PA, March 1999.
AutoPacific 2013 New Vehicle Satisfactory Survey.
W. Bergman, Considerations in determining vehicle handling re-
quirements, SAE Technical Paper 690234, 1969.
W. Bergman, Correlation between vehicle tests and subjective eval-
uation, Ford Motor Company, State of the Art of Objective Pro-
cedures for Testing Vehicle Handling, 1978.
M. Blundell and M. Harty, Multi-body systems approach to vehicle
dynamics, 2nd edition, SAE International and Elsevier, USA,
2014.
E. Browalski, S. Jogi, J. Waraniak, T. Gillespie et al., ESC per-
formance of aftermarket modified vehicles: Testing, simulation,
HIL, and the need for collaboration, SAE Int J Passeng Cars -
Electron Electr Syst 3(2) (2010), 204–214.
S. Chakravartula, I. Haque and G. Fadel, A modified Monte-Carlo
simulation approach to heavy vehicle design for good dynamic
performance in multiple scenario, Int J Vehicle Des (Heavy Ve-
hicle Systems A Special Issue) 10(1.2), (2005), 112–143.
T. Chandramoulli, Application of simulated annealing for optimiza-
tion of the lateral dynamics behavior of an automobile, M.S. the-
sis, Department of Mechanical Engineering, Clemson University,
2002.
D. Chen, Subjective and objective vehicle handling behavior, PhD
thesis, School of Mechanical Engineering, The University of
Leeds, 1997
D.A. Crolla, Vehicle dynamics-theory into practice, Automobile Di-
vision Chairman’s Address, IMechE, 1995.
K. Deb, Multi-objective optimization using evolutionary algorithms,
Wiley, London, 2001.
W.R. Garrott et al., Vehicle inertial parameters-measured values and
approximations, Proceedings of the SAE 881767, SAE Passenger
Car Meeting & Exposition, Dearborn, MI, October 1988.
M.C. Good, Sensitivity of Driver-Vehicle performance to vehicle
characteristics revealed in open-loop tests, Vehicle System Dy-
namics, Int J Vehicle Mech Mobility 6(4) (1977), 245–277.
 SYSTEMS ENGINEERING METHOD FOR VEHICLE HANDLING DESIGN 675

H. Haghiac, I. Haque, and G. Fadel An assessment of a genetic
algorithm-based approach for optimizing multi-body systems
with applications to vehicle handling performance, Int J Vehicle
Des, 36(4), (2004), 320–344.
S. Horiuchi and N. Yuhara, An analytical approach to the prediction
of handling qualities of vehicles with advanced steering control
system using multi-input driver model, ASME J Dyn Syst, Mea-
sur Contl 122(3) (2000), 490–497.
International Council on Systems Engineering Systems Engineering
Handbook, A Guide for System Life Cycle Processes and Activ-
ities, INCOSE - TP- 2003-002-03 version 3, June 2006.
J. Ishio, H. Ichikawa, Y. Kano, and M. Abe, Vehicle-handling quality
evaluation through model-based driver steering behavior, Vehicle
Syst Dyn 46(1) (2008), 549–560.
H.M. Kim, D.G. Rideout, P.Y. Papalambros, and J.L. Stein, Analyti-
cal target cascading in automotive vehicle design, Trans ASME:
J Mech Des 125 (2003), 474–480.
R.P. King, D.A. Crolla, H.A.S. Ash, and J. Whitehead, Identifica-
tion of subjective-objective vehicle handling links using neural
networks for the foresight vehicle, SAE Technical Paper 2002-
01-1126, 2002.
M. Kinstle, D. Hassler, and B. Johnson, Vehicle dynamics bench-
marking and simulation, SAE Technical Paper, 2009-01-0465,
2009.
J. Knapczyk and S. Dzierżek, Displacement and force analysis of
five-rod suspension with flexible joints, ASME J Mech Des
117(4) (1995), 532–538.
Y. Li, Z. Lu, and J.J. Michalek, Diagonal quadratic approximation
for parallelization of analytical target cascading, J Mech Des, 130
(2008), 051402-1–051402-11.
W. Lincke, B. Richter and R. Schmidt, Simulation and measurement
of driver vehicle handling performance, SAE Technical Paper
730489, 1973.
D. McRuer and R. Klein, Effect of automobile steering character-
istics on driver vehicle performance for regulation tasks, SAE
Technical Paper 780778, 1978.
Mimuro et al., “Four Parameter Evaluation Method of Lat-
eral Transient Response,” SAE Technical Paper 901734,
1990.
M. Olley, Road manners of the modern car, Proc Inst. Automob Eng
51 (1946–47), 147–182.
E. Ono, S. Hosoe, H.D. Tuan, S. Doi, Bifurcation in Vehicle Dynam-
ics and Robust Front Wheel Steering Control, IEEE Trans Contl
Syst Technol 6(3) (1998), 412–420.
H. Pacejka, Tire and vehicle dynamics, 2nd edition, SAE Interna-
tional and Butterworth Heinemann, USA, 2005.
G. Scarlat, I. Haque, G. Fadel, and J. Schuller A modified Monte-
Carlo approach to simulation-based vehicle design with multiple
objectives and multiple scenarios, SAE Technical Paper 2002-
01-1186, 2002.
L. Segel, An overview of developments in road vehicle dynam-
ics: Past, present and future, Proceedings of the IMechE Con-
ference on Vehicle ride and Handling, London, 1993, pp.
1–12.
Statistical Analysis of the Effectiveness of Electronic Stability Con-
trol (ESC) Systems – Final Report, Report No. DOT HS 810 794,
Jennifer N. Dang, July 2007.
S. Tosserams, L.F.P. Etman, and J.E. Rooda, An augmented La-
grangian relaxation for analytical target cascading using the alter-
nating directions method of multipliers, Struct Multidiscip Optim
31(3) (2006), 176–189.
P.J. Venhovens, M.A. Hazare, Setting vehicle handling engineer-
ing targets based on driver’s preferences, abilities and limita-
tions, Proceeding of the 11th International Symposium on Ad-
vanced Vehicle Control, AVEC 12, Seoul, South Korea, August
2012.
D. Weir and R.J. DiMarco, Correlation and evaluation of
driver/vehicle handling data, SAE Technical Paper 780010,
1978.
X. Xia and J.N. Willis, The effect of tire cornering stiffness on vehi-
cle linear handling performance, SAE Technical Paper 950313,
1995.

Dr. Mandar Hazare is vehicle dynamics researcher and a motorsports professional. He received his PhD in Automotive Engineering from
Clemson University International Center for Automotive Research (CU-ICAR) in 2014. His research interests include vehicle dynamics,
control systems, tire mechanics, and suspension design.

Dr. Paul Venhovens is the BMW Endowed Chair in Automotive Systems Integration at CU-ICAR and comes to CU-ICAR from BMW’s
Research and Development headquarters in Munich, Germany, where he worked in the field of systems integration since joining the company
in 1995. Dr. Venhovens received his Ph.D. in mechanical engineering in 1993 from Delft University of Technology in his home country of
the Netherlands. Before joining BMW, he conducted post-doctoral research at the University of Michigan Transportation Research Institute
(UMTRI) in Ann Arbor, Mich., in the area of simulation and control design for ITS (lane departure warning and control) applications. At CU-
ICAR his research focuses on vehicle dynamics, new vehicle architectures and automotive systems engineering tools that balance multiple,
often competing aspects, including markets, policies, brand, geometry, functions, weight, costs, and manufacturing.

Systems Engineering DOI 10.1002/sys