Sound Quality Assessment in Motorcycles

1
Sound Quality Assessment in Motorcycles

• Guide on measured data needed to make efficient
  use of acoustic simulations

2
Noise Sources

• A typical motorcycle noise spectra will comprise
  noise sources from the following main areas
• Snorkel noise (doesnt include structural
  vibration)
• Intake Exhaust systems
• Mechanical noise transmitted by the structures
  into the air through exterior walls
• Engine noise (including engine covers)
• Skin noise from other structures silencer
  walls, large flat panels on air intakes, etc.
• Rattle friction noise from rotating components
  chains, valve train arrangement, drive gears,
  etc.
• Tyre noise

3
Sound Quality

• Sound quality from main noise source areas
• Snorkel noise It brings the character of the
  motorbike as intakeexhaust systems are the means
  to take the engine breathing to the exterior.
  Therefore, these sources are normally associated
  to a boomy or throaty sound. However, if the
  engine generates high frequencies these will also
  be output through the exhaust unless some
  internal wadding is provided to attenuate them.
• Mechanical noise It tends to add mainly high
  frequencies (metal sound) which typically turns
  into unpleasant perception by the human ear.
• Having listened to all the sources it appears as
  if the exhaust noise is the one to be maintained
  dominant in order to achieve good sound
  quality. Intake systems normally resonate below
  500Hz at specific frequencies. Mechanical noise
  adds an undesirable harshness to the character
  of the sound. Therefore, it is desirable to make
  intake and mechanical noise as quiet as possible
  in order to leave more room for the exhaust
  note to shape the quality using the engine
  breathing and silencer internals.

Exhaust Noise
Mechanical Noise
Intake Noise
4
Noise Levels

• Current legislation sets the drive-by noise level
  to 80dBA, which means that we cannot make the
  exhaust as loud as we like. Furthermore, if the
  mechanical noise contribution is near the 80dBA
  limit the exhaust noise level will be below the
  mechanical noise level with the consequent
  detriment in the sound quality.
• We can lump the overall dBA noise level into two
  main sources mechanical and snorkel. Lets
  take an example where the drive-by noise level is
  82dBA
• If the mechanical acoustic power gives 78dBA then
  the snorkel one must be 79.8dBA in order to give
  the overall 82dBA. Assuming the powertrain parts
  are already designed and no obvious mechanical
  noisy component can be modified, as it is often
  the case, we are left with the only choice of
  redesigning the exhaust internals (we are
  assuming a quiet intake system for now).
• In order to achieve 80dBA without touching the
  mechanical components we will need to attenuate
  the exhaust by 2.33dBA. So the motorbike is legal
  but the mechanical noise is well above the
  exhaust level (double the acoustic power
  corresponds to 3dB). This reverts in bad sound
  quality.

5
Noise Simulations

• There are about 3 different approaches to noise
  simulations nowadays
• 1D engine performance codes (WAVE, LES, etc)
• These are transient codes that understand about
  flow mass and shock waves. However, they do not
  generate realistic mid-high frequency content.
  Therefore they are not ideal for sound quality
  where high frequencies are important to identify.
• 3D finite and boundary acoustic elements
• These can model complex 3D models and understand
  high frequencies and absorptive materials.
  However, these analyses are normally solved using
  linear solvers which tend to smooth out
  results. Nevertheless they are very cost
  effective and fast tools for design
  optimisations.
• Coupled CFD-acoustics
• This method understands both mass flow and
  acoustics in a 3D environment. The main drawback
  is that it is very expensive to set up and run
  leading to long development time scales.
• Results from a noise simulation will be as good
  as the input data available. Typically for
  exhaust systems it is preferred to have measured
  pressures before the silencer inlet.

6
Noise Excitations

• There are about 3 possible noise inputs for an
  exhaust noise simulation
• White noise
• This is the most simplistic input that does not
  require any measurements. It consists basically
  of exciting all frequencies with the same
  magnitude and phase. It is useful for general
  first noise optimisation but it is not accurate
  for a final sound quality assessment.
• Predicted pressures from 1D engine simulation
  codes
• This is mainly useful for low frequency (below
  1500Hz) and although it gives an indication on
  sound quality it will not be very accurate if the
  actual engine generates high frequencies which
  may influence importantly in the sound quality.
• Measured pressures on a rolling road
• This is the preferred input data. If the
  measurements are made accurately it is possible
  to obtain reliable information up to 10000Hz.
  Therefore, given that this data is available
  prior to planning the acoustic analysis it will
  be of invaluable help to focus on the noise
  frequency ranges of interest. This data will also
  indicate whether future high frequency analysis
  introducing internal wadding materials will be
  needed.

7
Desirable Measured Data

• Existing engine
• Drive-by noise recordings featuring the baseline
  silencers for the following tests
• Naked motorbike (official drive-by noise test)
• All mechanical parts lagged as much as possible
  to record only snorkel contribution
• Intakeexhaust ducts silenced as much as possible
  to obtain mechanical noise contribution
• Drive-by noise test other prototypes if available
• Measure pressures in a rolling road while running
  the engine at the speeds expected during the
  drive-by test. Sample measured data to give
  frequencies up to 10000Hz.
• In order to try to correlate predicted results
  the following tests are recommended
• Insertion loss measurement on a silencer on a rig
  using a loudspeaker excitation
• Insertion loss measurement with a running engine
  in a semi-anechoic room with the exhaust outlet
  isolated from the room to obtain noise recordings
  of the exhaust snorkel only. Pressures should
  also be measured before the silencers.

8
Desirable Measured Data

• Non-existing engine
• Use existing measured data from similar engine
  configurations.
• If this is not the case use 1D engine predictions
  and run acoustic analysis at low frequency only
  (2000Hz).
• Then wait until prototype engine is available and
  obtain pressure measurements to look for any high
  frequency content that may contribute importantly
  to the noise levels and/or sound quality.
• Post-processing of measured data
• Once the drive-by noise recordings are obtained
  then a study of these is needed to assess the
  different contributions between mechanical and
  snorkel noise. This will lead to conclusions
  indicating whether there is chance of good sound
  quality by only modifying the intake/exhaust
  systems.
• The above study will also indicate if high
  frequencies are predominant in the sound quality.
• Pressures measured in the rolling road before the
  exhaust system will indicate whether a high
  frequency acoustic analysis is needed from the
  start. It will also justify the use of active
  components (glass fiber, rock wool, etc) in the
  inside of the silencer walls. Furthermore, this
  measured data will be used as the acoustic
  excitation for the simulation work.

9
Acoustic Analysis Plan for Sound Quality

• The following steps are recommended in order to
  make efficient use of acoustic simulation tools
• 1.- Make drive-by recordings on existing
  prototype.
• 2.- Record pressures on a rolling road before the
  exhaust system.
• 3.- Post-process all measured data and draw
  conclusions.
• 4.- Plan the analysis work (low frequency or high
  frequency?).
• 5.- If the analysis requires a high frequency
  study and the inclusion of active absorptive
  components, these should be tested to calculate
  the acoustic properties to be incorporated in the
  acoustic simulation.
• 6.- It is recommended that, given that a
  motorbike exists, the above simulation work is
  supported by at least one prototype half way
  along the simulation work. This is to assess the
  design so far in terms of noise levels and sound
  quality as part of the whole motorbike.