Using numerical acoustics to diagnose noise problems

Numerical acoustics has traditionally been relegated to a prediction only role. However, recent work has shown that numerical acoustics techniques can be used to diagnose noise problems. The starting point for these techniques is the acoustic transfer vector (ATV). First of all, ATV's can be used to conduct contribution analyses which can assess which parts of a machine are the predominant noise sources. As an example, the sound power contribution and radiation efficiency from parts of a running diesel engine are presented in this paper. Additionally, ATV's can be used to reliably reconstruct the vibration on a machine surface. This procedure, commonly called inverse numerical acoustics (INA), utilizes measured sound pressures along with ATV's to reconstruct the surface velocity. The procedure is demonstrated on an engine cover for which the reconstructed vibration had excellent agreement with experimental results. As a second example, the particle velocity on the open end of a duct was reconstructed. Again, there was excellent agreement with the experimental results.