Broadband Noise Predictions on the ACAT1 Fan Stage Using Large Eddy Simulations and Analytical Models
Danny Lewis, Stéphane Moreau, Marc C. Jacob
Abstract
The present paper assesses turbulent flow through and the noise radiation by the "ACAT1" fan stage, which was experimentally tested at the Universal Fan Facility for Acoustics (UFFA) of AneCom AeroTest (ACAT) as part of the European project TurbonoiseBB. The present study focuses on a short-gap configuration, at approach condition, on a so-called Sea Level Static (SLS) working line. Its main objective is to analyze and predict the broadband noise resulting from the impact of the fan wakes onto the Outlet Guide Vane (OGV), also called Rotor-Stator Interaction (RSI) noise. This is achieved via a Large Eddy Simulation (LES) of the full fan-OGV stage that is overall in very good agreement with the test performance parameters and with the interstage hot-wire measurements. An in-depth analysis of the multiple flow features is carried out and leads to the identification of the mechanisms at stake in the broadband noise production. The noise is then estimated using two different hybrid approaches. On the one hand, the analytical approach uses an analytical model informed with flow parameters retrieved from the LES. On the other hand, a numerical approach couples the source predicted by the LES with the free-field sound propagation by the Ffowcs Williams and Hawkings (FW-H) analogy. Regarding the analytical approach, Hanson's and Posson's cascade models have been used to perform the broadband RSI noise predictions. The shape of the estimated noise spectra obtained from these models is relatively similar to those of the experimental acoustic power spectra. However, depending on the analytical model and the turbulence lengthscale estimate, some discrepancies on the absolute noise levels may appear. The FW-H analogy recovers the overall shape of the experimental spectra with an overestimation of the radiated noise that is mainly attributed to the fact that the duct propagating properties are neglected by the free-field FW-H analogy. Both the analytical and the numerical noise prediction methodologies eventually suggest that additional non-negligible noise sources might be present in both the experiment and the simulation.