Acoustics Simulation of the Flow over a Door-Boot Cavity
	Noise from resonant (or Helmholtz-like) cavity flows is an issue in both aerospace and automotive industries.  Cavities are frequently present in aerodynamic structures and in many cases these cavities can't be sealed, resulting in a significant source of noise, The sound generated from the flow over these cavities can have a tiring and stressful effect on personnel and can even lead to structural fatigue. Most noise in cavity structures originates from fluid-acoustic resonance, with the energy in the external flow driving a feedback mechanism, in which pressure waves amplify the initial vortical disturbances.  Centerline schematic of the Fiat Punto boot-door cavity. The above figure illustrates an example of this type of scenario, generated by the gap in the boot-door seal on the Fiat Punto passenger car.  To simulate this problem, a 78,000 quadrilateral element mesh was first constructed of the cavity environment and a steady state RANS solution computed on this mesh.  This mesh included a large lead-in section to generate the boundary layer observed upstream (this was confirmed with experimental measurements).   Subsequently, an acoustics mesh was generated by relaxing the near-wall clustering on the RANS mesh, truncating the outer domain, extruding in the spanwise direction and interpolating the 2D RANS solution onto the 3D acoustics mesh.  The NLAS acoustics solver was then run on this acoustics mesh using a 16-processor Pentium 4 Linux cluster.                                                            Instantaneous particle paths showing recirculation within cavity. Vorticity contours showing flow structures in cavity opening. The above figure shows the development of the coherent flow structures within the cavity.  Sound pressure levels (A-weighted) are shown compared to experiment in the figures below.     Noise measured inside boot-door cavity. Result of NLAS simulation of noise inside boot-door cavity. It should be noted that the use of conventional (full-domain) LES for the simulation of this kind of phenomena would require prohibitively huge mesh resources, because of the need to resolve the source region, the incoming boundary layer and the subsequent propagation.  The current state of the art in most hybrid RANS/LES methods is not yet able to account accurately for the transport and transfer of turbulent fluctuation data from the RANS region to the refined, downstream LES zones (corresponding here to the boundary layer above the cavity opening).  On coarser meshes, the eddy viscosity models that are currently popular in hybrid RANS/LES methods have a tendency to suppress unsteadiness on small cavity gaps like that considered here, since the shear layer instabilities can be too weak to fully overcome the damping effects of the SGS viscosity.  Larger sub-grid viscosities and (unrealistic) steady inlets also have a tendency to generate structures which are spatially too well correlated.  These deficiencies limit the accuracy of both broadband and tonal noise predictions, tending to concentrate too much energy into the narrow frequency ranges corresponding to the dominant coherent motions. The generality of the NLAS tool has allowed predictions of both resolved, coherent motion and fine-scale turbulent fluctuations, justifying its use (at least for these classes of flow) as an engineering tool for the comparison of different design proposals.  The figure below shows the results of a parametric study into the effects of changes in gap opening, internal cavity volume and orientation of the flow surfaces upstream and downstream of the cavity opening. Results of NLAS simulations from various cavity designs.

Figures courtesy of Centro Richerche Fiat.