| In addition to its longstanding commercial
success with vehicle models such as the Civic and Accord, Honda is systematically
praised for its commitment to achieving high comfort, safety and performance
standards. Among the factors that contribute to vehicle comfort, the control
of road noise is of key importance. On several of its recently developed
vehicles, Honda succeeded in reducing road noise by using a new hybrid
modeling approach. Together with LMS Engineering, the Japanese carmaker
implemented a hybrid simulation process, capable of quickly and accurately
modeling road noise at frequencies up to 300 Hz. The approach consists
of coupling a test-based model of the trimmed vehicle body with a Finite
Element (FE) model of the suspension system. The obtained hybrid full-vehicle
model enables Honda to evaluate more suspension design alternatives earlier
in the development, and to come up with more effective countermeasures
for improved road noise performance.
Whether enjoying an enriched discussion, or a piece of light music during
a tranquil ride, the comfort of a quiet interior makes a big difference.
A potentially disturbing factor is road noise that propagates through
the vehicle’s mechanical structures and connections. As suspension
assembly parts play a critical role in most of these propagation paths,
it is very difficult to control and reduce road noise. The prediction
of road noise levels prior to the physical prototype refinement stage
has long been extremely challenging. The most common approach to early
suspension design is to rely on rules of thumb derived from previous experience,
such as the resonant frequency of the suspension links being above a given
frequency, or the mount stiffness being within a certain range. The main
problem with this approach is that there is no way to depend on these
rules to predict whether a design modification will increase or reduce
road noise.
Developing FE models of the entire trimmed body and all suspension components
is another way to evaluate road noise performance. A disadvantage of using
purely FE-based models of fully trimmed vehicles is the large modeling
effort they require. The scale of this effort is hardly justifiable when
the focus is limited to just the vehicle suspension. Another disadvantage
of simulations using these FE models is that their prediction accuracy
decreases as frequencies go up.
Hybrid Test-CAE approach increases speed and accuracy
In order to evaluate its accuracy and usability, the hybrid modeling and
simulation approach has been applied on an existing Honda model. The LMS
Engineering Services team started by creating an FE-based model of the
individual suspension system components, including the suspension links,
shock absorber and sub frame. The models were validated through comparison
with tests ran on their respective physical counterparts. Specific tests
were performed on the shock absorber to identify their dynamic properties,
such as bending modes, and the effects resulting from stiffness and damping.
The results of these tests were used to fine-tune the FE model of the
shock absorber. The model of the complete suspension model was obtained
by coupling the various components and integrating the connections through
the definition of bushing characteristics. The complete suspension model
was validated using a dedicated test where the suspension was fixed to
rigid boundaries. This set up ensures a realistic pre-load condition of
all the bushings, which is crucial for the refinement of the linear bushing
rates used in the FE model. Then, the structural and vibro-acoustic FRFs
of the car body were measured at all suspension-body attachment points.
The advantage is that these measurements can be performed on any car body,
for example a trimmed body prototype or a body of a previously released
vehicle.
To connect the calculated FRFs of the suspension system and the measured
FRFs of the car body, the FRF-Based Sub structuring (FBS) method was used.
This approach defines the FRFs of a combined assembly through the FRFs
of its constituting components. The dynamic forces retrieved from operational
vehicle tests were then applied through indirect force identification
at the level of the wheel spindle. The model proved to be very accurate
in simulating the measured vibration and road noise, throughout the entire
frequency range of interest, being between 20 and 300 Hertz.
Empowering efficient design modifications
After building the full-vehicle model, LMS and Honda engineers evaluated
the effect of various design modifications, for which physical test results
on road noise were available. They implemented changes to the suspension
system model by either modifying the geometry of the appropriate component
FE model or by updating the coupling definition. They calculated the FRFs
of the modified suspension system model and coupled the FRFs with those
measured on the trimmed car body. By modifying the suspension FE model
and recoupling it to the test-based vehicle body model, they systematically
evaluated the road noise performance of a series of other modifications.
One of the Honda engineers involved in the project commented, “The
hybrid FBS model was found to be able to predict the road noise effect
of large as well as more local modifications, such as changing the geometry
of individual components or adding local masses. LMS engineers also used
this model to calculate the structural and vibro-acoustic transfer functions,
and compare them to the FRFs measured on the complete vehicle, in order
to validate and further improve the model.”
NVH tire model
Indirect force identification at the wheel spindle is a very powerful
procedure and is very often applied to investigate NVH (Noise, Vibration
and Harshness) problems. One disadvantage, though, is that the forces
obtained depend to a certain extent on the type of tire used. In addition,
when modifying the suspension model to predict modifications, it is possible
that these forces change, and as a result reduce the accuracy of predictions.
More recently, LMS has developed an improved method in which an experimental
modal model of the tire is coupled to the wheel spindle.
This model can be loaded by a forced displacement, which is directly related
to the road surface geometry. This approach makes the model completely
independent from the type of tire used and modifications made to the suspension
system. The tire model can also be corrected to take into account the
effect of tire rotation on its dynamic properties. At present, the use
of this model is limited to noise evaluation on smooth roads, which are
characterized by low displacements.
From validation to mainstream development
Compared to more traditional development sequences, the application of
hybrid suspension/body modeling to reduce road noise is advantageous in
more than one way. This hybrid modeling and simulation technique has already
been used by Honda in a number of its vehicle development programmes.
Integrating an FE suspension model in a test-derived body model has enabled
Honda engineers to simulate a higher number of suspension design alterations
at earlier stages in the development.
The road noise performance of the vehicle can be increased by gradually
improving the geometry of the sub frame and suspension parts, such as
the arms, wheels and knuckles. Implementing and assessing these numerous
and profound changes would not be possible later on in the development,
when a physical vehicle prototype becomes available. Modifying parts at
this late stage is typically very expensive, time-consuming or simply
not feasible.
|
|
