Linear predictive filtering in a numerically simulated turbulent flow

A numerical investigation is made into the use of linear predictive (Wiener ) filters to predict flow quantities in the near-wall region of a turbulent boundary layer for use in active control algorithms. Optimal filters for t he prediction of Reynolds stress and fluctuating streamwise velocity compon ents using wall-shear stress information are developed and their dependence on the number and location of shear sensors is explored. It is found that a densely populated, wall-based, sensor system can predict the near-wall Re ynolds stress with good accuracy, and that 76% of the optimal performance c an be achieved with as few as four sensors whose locations coincide with th e strongest weights of the Wiener filter derived using a very dense network of input sensors. (C) 2000 American Institute of Physics. [S1070- 6631(00) 50112-3].