As automotive manufacturers compete
to produce ever-quieter vehicles, an improved way of analysing structural
noise promises an important breakthrough in reducing the time of prototype
refinement. In the DIANA project, a specialist company focused on
noise and vibration analysis and worked with a technical school, a
motor industry research centre and several vehicle manufacturers to
transform a textbook procedure into a practical engineering tool.
The result is so successful that it is rapidly becoming a standard
procedure during the vehicle refinement process. The software developed
by the project partners has matured into an established commercial
product and has been the foundation for further technologies that
address the complex world of vibroacoustic modelling.
When this project started in 1992, automotive
engineers had no practical way of modelling structure-borne noise
inside a vehicle. Five years later, 'transfer path analysis' has
become a standard tool used by many of Europe's vehicle manufacturers.
This new method quickly pinpoints the locations on the vehicle where
an optimal design change should be made. This not only helps engineers
build quieter cars - it reduces by several weeks the time it takes
to refine the prototype.
The engine, driveline and suspension system all generate vibrations
that are transmitted through the structure of a vehicle. These vibrations
eventually excite the large body panels, such as the floor pan,
which then radiate noise into the interior. LMS International, a
Belgian company specialising in noise and vibration analysis, knew
that it is not difficult to set up a mathematical model to describe
the experimental data. To be of use, however, the measurement procedure
needed further refinement to make it applicable to an industrial
environment, and some of the mathematics needed to be strengthened.
Realising the potential value of a practical way to analyse structure-borne
noise, LMS initiated the DIANA project to overcome these difficulties.
As partners the company invited the Fachhochschule Bielefeld in
Germany, the Motor Industry Research Association (MIRA) in the UK,
and vehicle manufacturers including Renault and Fiat. Ford Germany
sponsored the project by providing test vehicles.
How vibration is transmitted
A typical vehicle has a relatively small number of connection points
through which vibrations from the sources are transmitted to the
body structure. Most of these connections are mounts: rubber components
designed to isolate vibration. Some of these mounts are relatively
soft, others, such as the suspension bushings, can be quite stiff
in comparison to the body stiffness. Transfer path analysis is used
to answer the following questions:
• Which inputs contribute to the problem of a particular noise at
a specific speed or frequency?
• Does a particular contribution appear because of high injected
force levels, or because it is being transmitted too efficiently?
• How do the above questions relate to the overall design of the
vehicle? (Factors include body stiffness at connection points, mount
stiffness, source and body resonance.)
An existing theoretical background called transfer path analysis
(TPA), provided the partners with a mathematical framework. TPA
describes the total interior noise as a vector sum of individual
contributions from a given set of force inputs entering the body
over a known set of bridges - typically the engine, exhaust and
suspension mounts. The method requires two pieces of information:
knowledge of the operating forces at the body side of the mount,
and a measurement of the vibro-acoustic transfer functions between
that point and the target receiver. A ranking of the transfer paths
then becomes possible.
Transfer functions and forces
To measure the vibro-acoustic transfer functions the vehicle is
set up in the test lab. Known forces are injected into the structure
at each attachment point in turn using either a shaker or impulse
hammer. The acoustic response is then measured by a microphone placed
near the driver's ear. It is then quite straightforward to calculate
the transfer function for each attachment point. An alternative
transfer function measurement procedure was also developed. This
used the principal of reciprocity, whereby a sound source is placed
inside the cabin instead of a microphone and all attachment points
are set up in parallel with vibration pick-ups. In this way, all
the transfer functions are measured simultaneously.
Operating forces are more difficult to measure because force transducers
would have to be physically built into the vehicle's structure.
This is not feasible because force transducers are very bulky and
cannot be placed in the confined areas around the mounts. This was
the limiting factor for the technique until the DIANA project solved
One approach is to measure the vibration amplitude of the mount,
which can be done in a moving vehicle using small acceleration sensors.
It is then possible to calculate the forces that were needed to
cause the acceleration using a knowledge of the mount stiffness.
There was a problem that had to be overcome - mounts are very non-linear
devices which meant that special techniques had to be developed
to enable the stiffness value to be measured in the laboratory.
The Fachhochschule Bielefeld and MIRA developed the laboratory equipment
necessary to measure the dynamic stiffness of rubber mounts. They
investigated the influence of different parameters, such as temperature
and static preload on the stiffness values. Both parameters are
important as they vary to a large extent, depending on the driving
The Bielefeld test rig allows mounts to be tested in many ways not
previously possible in the laboratory, such as with multi-axial
preloads and measurements of rotational stiffness.
To measure the operation vibration and noise data under different
engine and road speeds, a vehicle is first fitted with accelerometers
attached to each mount. Data are then measured during a drive on
a test track, or in the laboratory using, a chassis dynamometer.
Combining the vibration data with the laboratory measurements of
mount stiffness gives the forces acting on each mount.
The partners also developed another approach to force determination
which could avoid having to know the mount characteristics. Instead,
forces can be back-calculated from measured operating vibration
data and local body dynamic stiffness at each input connection.
Local body stiffness is quite easily measured in the laboratory.
In practice the researchers found that both methods of force measurement
were important to the success of the project.
Refinements and adaptations
Vibrations originating in the engine and transmission system are
all coherent, that is, they are all related to each other. On the
other hand, vibrations caused by the contact between the wheels
and the road transmitted through the suspensions are random. This
makes the analysis of road noise transfer paths more difficult.
The basic TPA approach is not directly applicable in this situation,
so the team adapted a supplementary technique called 'Principal
Component Analysis' (PCA) to enable road noise analysis. PCA breaks
down non-coherent vibrations into coherent sets of vibrations and
assigns the latter to 'virtual' sources that can be treated in a
similar way to coherent sources.
Companies can apply both TPA and PCA at several different levels.
The simplest is to treat large components, such as the engine, as
self-contained units. An engineer planning to re-use an existing
engine design in a new vehicle can measure the engine's vibration
characteristics in an existing vehicle and assume that it will behave
identically in its new environment. Combined with calculated transfer
functions for the new body, the resulting noise level can then be
predicted. Similarly, an alternator supplier can model and predict
the vibration levels that are transmitted to and from the engine
for different alternator configurations. A design optimisation cycle
can then be applied.
The success of the DIANA project lay in its adaptation of a well-known
textbook mathematical analysis to the rigorous demands of an industrial
application; solving the technical details of force identification
and incoherent excitation; solving the practical issues of stationary
and mobile measurements; and making sure that the massive amounts
of data that would be generated during the testing phase could be
easily handled, visualised and interpreted.
LMS has used the knowledge gained during the project to develop
and launch commercial software packages for TPA and PCA. Interest
in both the software and in LMS's consulting activities in this
area has been high, with orders from many European car manufacturers.
The original consortium members are now using the techniques to
help them to develop better and quieter vehicles.