Bayesian optimization of the acoustic response of short holes

The acoustic response of short circular holes with bias flow passing through them is relevant to many industrial applications, such as acoustic liners and perforated plates. Recent theoretical and numerical studies suggest that this acoustic response is sensitive to small geometry modifications. Additionally, flows through holes are often used to generate mean flow cooling.

In the current study, we consider a hole flow with a temperature gradient, and numerically optimize the hole geometry to maximise its acoustic absorption using Bayesian optimisation. The two-step numerical approach of Guzmán-Iñigo et al. (2022) is used, in which the incompressible Reynolds-Averaged Navier-Stokes (RANS) and energy equation are firstly solved to compute the mean flow through the hole in the presence of temperature gradient. The second step involves solving the flow perturbations superimposed on the mean hole bias flow via the linearised Navier-Stokes equations (LNSE). These two steps (including meshing) are repeated for every geometry optimization iteration.

The geometrical modification proposed uses rounded fillets on both the upstream and downstream edges of the hole, with the fillet radii as the optimisation parameters. The effect on a hole whose length and diameter have similar values, and which initially generates acoustic energy via “whistling” is demonstrated. The output of this study offers a route for designing short holes whose acoustic damping properties are maximized.