5 Aug 2019

Date: Thursday, August 15, 2019
Time: 11:00 a.m. – noon (EDT)
Room: NIA, Rm 137
Speaker: Vishal Srivastava

Speakers Bio: 

Vishal Srivastava received his B. Tech degree in Aerospace Engineering from the Indian Institute of Technology, Kanpur and is currently a Ph.D. Candidate in the Department of Aerospace Engineering at the University of Michigan, Ann Arbor. With Dr. Karthik Duraisamy as his advisor, he is studying data-driven techniques to improve turbulence modeling for use in predictive simulations.

Abstract: 

Reynolds-Averaged Navier-Stokes (RANS) simulations have been the go-to choice among computational techniques for preliminary design and optimization of flow configurations, owing to their speed and computational cost-effectiveness; but, they suffer from significant inaccuracies when compared to their higher-fidelity counterparts which can be attributed to inadequacies in the turbulence models being used. Data-driven approaches seem to be a lucrative solution to address this problem. While the data available from experiments and high-fidelity simulations cannot be directly used to improve these models, the relationship between flow features and the intended augmentation can be inferred and captured in a functional form using Field Inversion and Machine Learning. The presentation discusses the mathematical setting, framework, variations and applications of the methodology with results for both simple and moderately complex problems, and, finally, generalization strategies being pursued to broaden the predictive applicability of this technique beyond the class of problems used for augmentation, along with related preliminary results.