Integrated CFD-Acoustic Approach to the Simulation of Tonal and Broadband Noise Generated by Axial and Centrifugal Fans

2.5 Noise Prediction by Analytical or Numerical Models

An innovative computational approach for the simulation of noise generated by axial and centrifugal fans is presented, integrating mesh generation, CFD simulation and acoustic propagation analysis. The simulation chain includes methods for the efficient and accurate simulation of tonal and broadband noise, providing affordable alternatives to more traditional but expensive unsteady approaches.
The simulation of tonal noise is based on the application of the Non-Linear Harmonic (NLH) method for the noise sources related to blade passing frequencies and their harmonics, as well as propagation in the fan duct (including sound absorbing material) and in the near-field, followed by a propagation analysis to the far-field based on either a Boundary Element Method (BEM) or a Ffowcs Williams - Hawkings (FW-H) formulation. With this approach the simulation time is one to two orders of magnitude faster than with full-unsteady traditional multistage approaches with URANS models.
Typical computational approaches for broadband noise are based on time domain methods (e.g. LES, DES) requiring large computational efforts. The cost-efficient approach proposed by NUMECA exploits the Adaptive Spectral Reconstruction (ASR) method, based on a stochastic reconstruction of the turbulent noise sources. The method relies on a steady RANS solution allowing the application for preliminary design assessment and optimization. The frequency content of the turbulent field and the spatial cross-correlation of the turbulence scales are reconstructed. The noise sources (Lamb vectors) are extracted from the reconstructed turbulence and propagated in the near-field with a Finite Element Method (FEM) solving the Pierce-Howe Convective Wave Equation in the frequency domain. The far-field radiation is finally computed by a BEM solver.
The above mentioned computational chain is implemented in NUMECA's Flow Integrated Environment (FINE) suite, providing wizard automation and best practices. In the frame of this study, the suite is applied to the simulation of the noise generated by two different axial fans, respectively the Advanced Noise Control Fan (ANCF) operated by NASA Glenn Research Center and the NPU-fan, currently exploited in the IMAGE project, operated in the Turbomachinery Aerodynamics and Acoustics Laboratory (TAAL) of the Northwestern Polytechnical University of Xi'an, China. A centrifugal fan will be also considered. For all the fans, tonal and broadband noise predictions are compared with experimental data in order to demonstrate the level of accuracy and the computational performance. The results currently available indicate an accurate prediction of tonal noise and a reasonable estimation of the broadband noise level, in combination with computational efforts suitable in an industrial setting.