Uncovering the Nature of Large-Scale Flow Events in Turbulent Boundary Layers Using Data Driven Methodologies

Turbulent boundary layers (TBLs) are complicated flows characterized by a hierarchy of flow scales. Historical research has identified large-scale spatio-temporal events in such flows. From a control standpoint, it is important to model these larger flow scales so they can be used as inputs to flow actuation. To this end, the current work attempts to isolate spatio-temporal large-scale flow events in a zero-pressure-gradient TBL flow using data driven methodologies including proper orthogonal decomposition and a U-Net based autoencoder. Such methods can help extract spatio-temporal signatures of large-scale events in the flow utilizing datasets that are readily available from Particle Image Velocimetry and Direct Numerical Simulations. The results obtained will provide insights into the non-linear and multi-scalar nature of these large-scale events, thereby better informing modeling efforts for physics-based flow control.