Mechanical and Civil Engineering Seminar
Experimental Fluid Mechanics with Machine Learning
Young Investigators Lecture Series
The ability to understand unsteady fluid flows is foundational to advancing technologies in energy, health, transportation, and defense. We use cutting edge data-driven methods (i.e. machine learning) to interpret and control unsteady fluid flows through experiments in the following three cases: 1. We use robust principal component analysis (RPCA) to improve flow-field data by leveraging global coherent structures to identify and replace spurious data points. In all cases, both simulated and experimental, we find that RPCA filtering extracts dominant coherent structures and identifies and fills in incorrect or missing measurements; 2. We optimize a two cross-flow turbine (i.e. vertical-axis in wind) array using a hardware-in-the-loop approach and find that arrays with well-considered geometries and control strategies can outperform isolated turbines by up to 30%; 3. Using similar turbines, we create an experimental framework to more efficiently explore arrays' high-dimensional parameter space. Our data-driven approach allows us to model parameter spaces using sparse data. As a result, we are able to map turbine system dynamics with orders of magnitude fewer data points.
Contact: Sonya Lincoln at (626) 395-3385 email@example.com