Advanced Techniques to Analyze Large Datasets of Turbulent Flow
Thursday, January 25, 2018, 10:30 a.m. — Room H-111
by Datta V. Gaitonde
Mechanical and Aerospace Engineering
The Ohio State University, Columbus, OH 43210
Abstract: Experiments and simulations can measure or simulate a wide range of spatio-temporal scales. Successfully resolved studies thus invariably generate large datasets, whose analysis poses a significant challenge and may be considered as a major bottleneck in educing phenomena of interest. In this talk, we consider analysis of data obtained from Large-Eddy Simulations (LES). In particular, we will discuss a number of decomposition techniques, including data-driven and operator-based variants, to educe phenomena of interest. Data-driven techniques are essentially agnostic about the underlying governing equations: of the several available, we focus on Dynamic Mode Decomposition (DMD) and Empirical Mode Decomposition (EMD). Operator-based approaches on the other hand exploit the properties of the Navier-Stokes equations – in this category we will consider Doak’s Momentum Potential Theory (MPT). The techniques are illustrated in the context of several specific problems, both separately as well as in conjunction with each other. The use of DMD to analyze separated flows, as well as to deduce global modes in conjunction with perturbation methods, will be discussed in the context of stall control analysis and cavity flows. The power of EMD is highlighted by examining its ability to automatically identify intermittent events, which are often a crucial element of turbulent flows, but are difficult to detect with traditional Fourier based methods. When combined with the Hilbert transform, EMD provides a time-local frequency for the primary phenomena. The effectiveness of the method is demonstrated in the context of noise generated by turbulent jets. Finally, the talk will discuss the use of MPT to separate the signatures of acoustic, hydrodynamic (or vortical) and thermal processes in arbitrary turbulent flows. The results highlight the striking ability of the method to analyze complex turbulent phenomena in a Kovasznay-type manner, facilitating a detailed understanding of wavepackets in both jets as well as hypersonic transitional boundary layers.
Brief Bio: Datta V. Gaitonde is the John Glenn Chair and Ohio Research Scholar in the Mechanical and Aerospace Engineering Department. He has diverse research interests, including shock/turbulent interactions, jet noise, hypersonic transition, flow control, bluff body flows, scramjet flowpaths and algorithm development. He has over 200 publications in journals, conferences and books and has delivered numerous national and international presentations. He has served on various advisory panels for national agencies and is a Deputy Editor of the AIAA Journal. He is a recipient of several awards and is a Fellow of the Air Force Research Laboratory, ASME and AIAA.
All AE/ME 595 students are encouraged to attend, but not required. Everyone is invited and welcome to attend.