Stochastic estimation of multipoint conditional averages and their spatio-temporal evolution

Abstract

A novel application of the stochastic estimation procedure is used to examine the three-dimensional structure of the turbulent wake of a cylinder atx/d=100 and Re _D =5000. The basis for this method is the measurements of the full 3-D correlation tensor for all components across the width of the wake. Using this large data base, events contributing most to the Reynolds stress are identified and the most likely ensemble-averaged structures corresponding to these events are reconstructed using the stochastic estimation procedure. This powerful technique offers the distinct advantage that multi-point conditional averages can readily be estimated. An extension of this multi-point conditional average estimation is presented in this paper in the form of a pseudo-dynamic reconstruction of the flow field. To that effect, measured time sequences of “conditions” at several points in the wake are used to estimate the three-dimensional structures most likely associated with these events, hence providing a useful tool to reconstruct the evolution of the large-scale flow structures.