Numerical investigations on selective detection of characteristic flow field patterns in a turbulent boundary layer flow

• Applicant: Professor Dr.-Ing. Ulrich Rist
• Subject Area: Fluid Mechanics
• Term: since 2020
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 446803563

The aim of the present application is to carry out investigations on the detection of pre-determined, transient events in a flow close to the wall by means of a sensor grid. The basic idea is to consider a sensor grid as a “coincidence detector” which “fires” exactly when a certain event passes the grid. This idea is derived from biology, where predators bite in the dark when their flow field sensors report a worthwhile prey. We believe that this approach can give new impulses to flow control. If, for example, an actuator is to be used that is tuned to certain events, it is extremely important to reliably detect such events and trigger the actuator at the right moment.Structured arrangements of wall shear stress sensors (in the form of thin fiber sensors) allow new types of spatial and temporal signal detection. Especially if they are arranged in a suitable combination and one does not consider the individual signals but their correlations. Based on fundamental investigations into the detection of flow events in a turbulent flat-plate boundary layer, in which the central question is which events in the flow field can be detected via “footprints” on the wall, basic investigations into the arrangement of discrete sensors on a wall over which a fluid flows are to be carried out. This is followed by correlation investigations between the individual signals and the previously detected structures in order to find the most suitable sensor arrangements as well as the most suitable post processing of the individual signals into an event signal. The investigations are compared with other methods, in particular the Hough transformation from the preliminary work, Proper Orthogonal Decomposition (POD), Dynamic Mode Decomposition (DMD) and Artificial Neural Networks (KNN). These investigations provide basic knowledge for the identification of flow field structures via wall signals in a turbulent flat-plate boundary layer.In the end, the underlying bionic “predator-prey principle” will be demonstrated in a direct numerical simulation of a turbulent flat-plate boundary layer containing a sensor grid and a simple actuator. According to current understanding, low- and high-speed streaks can be detected in the boundary layer, which a fixed actuator can reduce either by suction or by blowing. The task of the coincidence detector is then to identify those streaks that are approaching the actuator in such a way that the actuator has a chance to reduce them.

• DFG Programme: Research Grants