A Statistically Based Computational Approach for Tonal and Broadband Noise Attenuation

Uncertainty of the source model for both tonal and broadband noise, and the result on prediction of acoustic treatment attenuation, is investigated. A statistical source model is proposed. A finite element simulation for propagation in non-uniform ducts with compressible mean flow with a random source description is developed. Probability density functions for transmitted acoustic power and attenuation are determined based on as many as 100,000 trials with random distributions of input modal power and phase. For cases with a moderate to large number of input modes, transmitted power and attenuation appear to follow simple statistical distributions. A comparison is made with limited experimental data. For broadband noise represented by a random distribution of all propagating modes, attenuation statistics for mean attenuation have a relatively small standard deviation, and in all cases considered measured attenuation is within or close to the predicted distribution.