Comparison of Lighthill's Analogy and Acoustic Perturbation Equations for the Prediction of HVAC Blower Noise

2.5 Noise Prediction by Analytical or Numerical Models

The prediction of noise emitted from rotating turbomachineries is a major concern in many industries. Products that directly affect the comfort of costumers are of particular interest. In this context, our aim is to develop reliable tools for the aeroacoustic noise simulation of radial blowers in heating, ventilation and air conditioning (HVAC) systems. We use a hybrid, volume discretized approach for this purpose. As the first step, the flow field is computed by applying the commercial finite volume solver Star-CCM+, where the WALE large eddy simulation (LES) turbulence model is used and constant density is assumed. This reduces the computational cost to receive a non-stationary flow field. As a consequence thereof, no acoustic information is incorporated in the flow pressure. Subsequently, aeroacoustic sources are evaluated on the flow mesh and conservatively interpolated onto a coarse finite element mesh. Finally, the in-house code CFS++ is applied for computing the sound propagation by using a non-homogeneous wave equations.
This simulation procedure is rather complex because two volume discrete meshes need to be created, which are furthermore split into a static and a rotating domain. Its major advantage is the full insight into the acoustic source distribution and propagation field. The acoustic perturbation equations (APE) are used in form of the perturbed convective wave equation (PCWE) and compared to the Lighthill analogy. The optimal simulation configuration is determined for each method by variation of the CFD mesh size and the spatial extent of the recognized sources. Furthermore, their implementation challenges and details are outlined. While both approaches yield sound pressure spectra that are in good agreement with experimental data, the PCWE has the advantage of separating flow and acoustic pressure, thus providing more information on the sound excitation processes in the blower. Its drawback is the high sensitivy to numerical noise of the flow simulation pressure. Taken together, the presented methologies can be used in radial blower development with respect to noise emission.