Cavity buffeting noise is the main contributor to discomfort in sky-dive tunnels. Low frequency noise, generated by self-sustained cavity shear-layer oscillations, similar to automotive sunroof-buffeting noise and side-window buffeting noise, is observed in sky-dive tunnels. Typical low frequencies are not conventionally audible to the human ear: registration of this phenomenon occurs due to the high strength of the oscillations felt on the inner-ear. Long-term exposure to high intensity low frequency noise can be experienced as fatiguing and annoying. A design Study of Streamlines Design BV has shown a fully effective design solution by simulation, which proved to be ineffective in real-life. The current study deals with validation of a different numerical set-up and the analysis of various retrofit designs for the reduction of sky-dive buffeting noise.

Existing literature has shown the promising results for the use of compressible computational fluid dynamics solvers, combined with detached eddy simulation for similar problems. The computational cost of detached eddy simulation has been found within the computational resources for this thesis, opposed to large eddy simulation or direct numerical simulation. For this thesis, the delayed detached eddy simulation model has been used. For the acquisition of validation data a Scanivalve DSA3217 pressure scanner has been used. For computational fluid dynamics and experiments, pressure has been probed at the same location in the cavity. Four retrofit designs have been proposed for analysis in this thesis: trailing edge extension, wall normal cylinder, span wise cylinder and a wing. Simulations have been conducted for one specific tunnel velocity and experiments for a full range of operating set-points.

Simulations have shown reductions of 6.0 dB and 2.6 dB for respectively the trailing edge extension and the wing in sound pressure levels of cavity buffeting noise. Furthermore an increase of 3.6 dB and 1.5 dB have been observed for respectively the span wise cylinder and wall normal cylinder. A reduction of the wake by 5.2% has been observed for the trailing edge extension, where the other designs increased the wake.
For the clear reduction in sound pressure level, introduced by the trailing edge extension, the formation of a larger scale vortex has been delayed downstream compared to the current situation. Visualising the vortices in the case of the span wise cylinder, it has been found that the formation of the large scale vortex has been triggered further upstream compared to the current situation, possibly due to interaction with cylinder vortex shedding.

Comparison of simulation and experiment data has shown errors of 11.8% in sound pressure level for the current situation and 17.6% when the trailing edge extension is employed. The reduction in experiment has been observed to be stronger than in simulation.

An evaluation experiment has shown the effectiveness of the trailing edge extension for all tunnel velocities at which high sound pressure levels have been observed. Reductions between 7.7 dB and 22.6 dB have been observed. Over the whole range of tunnel velocities, the sound pressure level has been reduced to 130.1 dB and below.

Concluding on the thesis work, the numerical model is able to predict a reduction. Following the simulation work the trailing edge extension is most effective in both reducing the noise and maintaining tunnel performance. Furthermore, the trailing edge extension has proved effectiveness for all critical tunnel velocities in experiment.