Unsteady coherent surface-pressure fluctuations from time-averaged flow data with given two-point statistics

Sound generation due to vibrations induced by hydrodynamic pressure fluctuations is a major concern in many industrial sectors such as automotive and aerospace. Accurate measurements of the wavenumber-frequency spectrum are necessary for its characterization and prediction. Often these measurements lack of spatial or temporal accuracy, particularly in the low frequency and wavenumber range. Accurate unsteady numerical simulations can overcome these limitations. However, the high computational costs restrict their applications to lowand mediumReynolds number flows. As a consequence, particularly in the design phase, the reconstruction of unsteady surface-pressure fluctuations from time-averaged computational and experimental data for attached flows can allow reduction of time and costs. As a matter of fact, in this phase, analytical or semi-analytical models are often used. They use a singlepoint power spectrum, that can be modeled from time-averaged flow quantities, an analytic expression of the correlation length, such as the Corcos’ model, and a known Green’s function. However, in real-life configurations, a tailored Green’s function is not always known and an integral method is used. In this case, unsteady surface-pressure signals with correct phase difference and known two-point statistics are needed. This paper proposes a methodology, based on the multisensor eigenvalue decomposition, to reconstruct unsteady surface-pressure signals from time-averaged flow data with correlated one and two-point statistics. Results show that the methodology is able to mix correctly the signals. The output signals generate a reliable synthetic wavenumber-frequency spectrum without affecting the energy content of the input signals.