• separated flow
• optimization
• experimentation
• diffuser
• CFD
• bluff bodies
• base drag
• flow control

The present work gives a contribution to the experimental and numerical characterization of separated flows and to the set-up of strategies for their control. The manuscript is divided in two parts.

In the first part the performance of a passive control method aimed at improving the performance of two-dimensional diffusers is evaluated in laminar and turbulent flow conditions. The passive method is based on the introduction of single or multiple contoured cavities in the diffuser solid walls and its effectiveness is investigated by means of numerical simulations carried out using a commercial code, previously validated through comparison with available numerical and experimental results. As for the laminar regime, the flow in symmetric plane diffusers having an area ratio of 2 and different divergence angles is considered. The Reynolds number based on the inlet quantities (half width and velocity on the centreline) is Re = 500. Without the control, different flow patterns are present in the diffuser depending on the divergence angle, i.e. fully-attached flow or an asymmetric zone of separated flow that reattaches before the end of the diffuser. The location and geometry of the cavities are numerically optimized to maximize the pressure recovery in the diffuser. In all cases, the introduction of the optimal cavities leads to a strong increase in the pressure recovery and, when present, to a significant reduction of the main flow separation zone. The flow separates at the cavity upstream edge and rapidly reattaches, forming a small closed recirculation region within and immediately downstream of the cavities. These recirculation zones lead to both a favourable local modification of the virtual shape of the diffuser and to a reduction of the dissipation in the near-wall region due to the relaxation of the no-slip condition. A classical optimization of the diffuser shape is also carried out; if the number of degrees of freedom is large enough, the presence of small local recirculations is again found in the optimized configuration. As for the turbulent flow regime, an asymmetric plane diffuser having an area ratio of 4.7 and a divergence angle of 10 degrees is chosen. The Reynolds number is Re = 20000, based on the inlet cross-section of the diffuser, where fully developed turbulent conditions are present. This is a test case for which reference experimental and numerical data are available, and the adopted RANS model is chosen and validated by comparison against these reference data. Also in this case, the introduction of one and two optimized-contour cavities leads to a strong reduction of the separation extent and to an increase in pressure recovery. The same physical mechanisms as in the laminar regime are responsible for the enhanced diffuser performance and the proposed control is shown to be robust to small variations of the cavity configuration both at low and high Reynolds numbers.

In the second part of the present work the main findings of an experimental and numerical research activity whose goal is characterizing and reducing the base drag of bluff bodies are presented. We consider the flow around an axisymmetric body with a sharp-edged base perpendicular to its axis. Wind-tunnel tests and Variational MultiScale Large Eddy Simulations are carried out at a Reynolds number Re=5.5x10^5, based on the body length and the freestream velocity, corresponding to Re_d=9.6x10^4, based on the body diameter. Direct Numerical Simulations are performed at a Reynolds numbers roughly two orders of magnitude lower, i.e. Re_d=1500. The results of experiments, VMS-LES and DNS simulations show that a decrease of the base suctions - and thus of the base drag - is directly proportional to the increase of the thickness of the separating boundary layers, with quantitative differences according to the experimental and numerical flow conditions. However, in all cases the data collapse on a single straight line when the base pressure is plotted against the length of the mean recirculation region behind the body, which, in turn, is connected with the location of the incipient instability of the detaching shear layers. It is shown that the location of this instability can be moved downstream, and thus base drag can be reduced, by increasing the thickness of the separating boundary layer. The results of the present analysis may also be useful to devise further strategies for pressure drag reduction.