Development of Boundary Vorticity Dynamics Based Closed Loop Flow Control

In this proposal closed loop control design tools will be developed using three key ingredients. These methods are (i) neural network based control methods, (ii) boundary vorticity based flow control strategies, and (iii) a reduced order numerical scheme that solves an asymptotic high Reynolds number approximation to the Navier-Stokes equations on coarse Euler type grids. Neural networks are chosen since they are universal approximators as well as being inherently non-linear. Further, neural networks lend themselves naturally to implementation in multiple input/multiple output control environments. The neural network will be used to control a combination of traditional actuation methods such as blowing/suction in conjunction with novel boundary vorticity flux methods based on tangential wall jets to optimally achieve the desired flow control. The reduced order numerical scheme will be used to make the effort computationally tractable. In Phase-I separation flow control will be demonstrated for a NACA 0012 airfoil undergoing dynamic stall. This effort will be collaboration between CFDRC and Dr. J.Z. Wu of the University of Tennessee. In Phase-II, the closed-loop control methodology will be tested on a three dimensional wing in a series of wind tunnel tests conducted at Virginia Tech University in collaboration with Prof. Demetri Telionis.