Local correlations for predicting the transition process in separated flows tuned with a large experimental database

Abstract

This work provides new correlations based on local variables for characterizing the transition process developing in the case of separated flows. The goal is to improve the capability of correlation-based transition models through the use of local variables. This may indeed simplify the implementation of the correlations into modern numerical codes, especially dealing with parallel computation and unstructured meshes. A large experimental database considering about 90 different flow conditions has been used to tune the present correlations. The database accounts for the variation of the flow Reynolds number, the free-stream turbulence intensity and the adverse pressure gradient affecting the boundary layer developing over a flat plate. The parameter variation induces the formation of both short and long laminar separation bubbles. The vorticity Reynolds number, based on the second invariant of the vorticity tensor, has been used as the main local variable for the activation of the transition process. Since it is proportional to the momentum thickness Reynolds number at the separation position, the vorticity Reynolds number (local) can be used in place of its counterpart based on the momentum thickness (integral), which is the variable usually adopted for activation of transition. Then, correlations providing a critical threshold for the vorticity Reynolds number promoting transition are tuned, for both short and long bubbles. These may be used into transition models to directly provide the location where the turbulence production needs to be activated. The proposed correlations have been validated using a second database not adopted for tuning and finally tested with data concerning a separated flow case in a low-pressure turbine cascade.