MMAE Seminar - Dr. John K. Eaton - Using Magnetic Resonance Imaging and Machine Learning to Understand and Model Turbulent Mixing

Armour College of Engineering's Mechanical, Materials & Aerospace Engineering Department will welcome Dr. John K. Eaton, Charles Lee Powell Foundation Professor and Martin Family University Fellow in the Department of Mechanical Engineering at Stanford University, to present his lecture, Using Magnetic Resonance Imaging and Machine Learning to Understand and Model Turbulent Mixing. The seminar is a part of Midwest Mechanics Seminar Series.

Abstract

Turbulent mixing is the controlling process in many systems including atmospheric dispersion of gaseous or particulate pollutants, fuel combustion, gas turbine film cooling, and myriad other processes in nature and technology. Mixing rates can be accurately predicted by fully-resolved, direct numerical simulations of the Navier-Stokes equations, but lower fidelity models meant for practical computations do not have predictive capability. Flow measurement techniques based on Magnetic Resonance Imaging (MRI) are enabling a paradigm shift in how we study turbulent mixing in complex geometries. MR Velocimetry (MRV) measures 3-D mean velocity fields without flow tracers or optical access, and an entire velocity field comprising millions of individual data points can be measured in a few hours. MRT, MRC, and MRP measure flow temperature, scalar concentration, and particle concentration fields respectively. Velocity and scalar concentration fields will be shown for gas turbine film cooling flows to demonstrate how the 3D measurements lead to new physical understanding. We are developing new scalar transport models for Reynolds-Averaged Navier-Stokes (RANS) computations using physics-informed, machine learning (ML) techniques. ML models are trained and evaluated using a combination of high fidelity simulations and the MRI data sets, and provide significantly better scalar transport predictions than conventional turbulence models. Interpretation of the ML models leads to new understanding of the flow features that cause failure of conventional models.

Biography

Dr. John K. Eaton is the Charles Lee Powell Foundation Professor of Engineering and the Martin Family University Fellow in Undergraduate Education at Stanford University where he has been on the faculty since 1980. He earned all his degrees in Mechanical Engineering at Stanford. He conducts research in turbulence, convective heat transfer, advanced measurement techniques, multiphase flow, and flow through random media. Recent emphasis has been on high-fidelity, rapid turnaround experiments in complex flows, measurement and modeling of turbulent mixing, particle-laden turbulent flows, and extreme sensitivity of certain high Reynolds number flows to small perturbations. Much of Professor Eaton’s work is motivated and funded by problems in the gas turbine industry. He has supervised 54 completed Ph.D. dissertations including those of 17 professors. Professor Eaton was a co-founder and long time chairman of the biennial International Symposia on Turbulence and Shear Flow Phenomena. He has won multiple Stanford awards for teaching excellence, the NSF Presidential Young Investigator Award, and the Senior Award from the International Society of Multiphase Flow. He is a Fellow of the American Society of Mechanical Engineers and the American Physical Society.