You're invited to attend

Scale-Resolving Simulations and Data-Driven
Closure Modeling of Turbulent Flows

by

Aviral Prakash
Ph.D. candidate | Ann and H.J. Smead Department of Aerospace Engineering Sciences
University of Colorado at Boulder.

Monday, April 3
2 - 3 p.m.
Weber Lecture Hall 2

About the Seminar: 
Turbulence simulations are of immense importance to the aerospace industry, with applications in the design of rotors, gas turbine or scramjet engines, and aerodynamic bodies of aircraft. With advancements in computational hardware, more complex scale-resolving turbulent flow simulations of these aerospace applications are becoming increasingly possible. Despite significant progress in numerical algorithms and highly scalable parallel codes for tackling such problems, physical modeling of these complex flows and knowledge extraction from these simulations are identified as areas that need key improvements.

This presentation discusses efforts toward the direct numerical simulation (DNS) of a turbulent boundary layer flow and data-driven subgrid stress modeling for large eddy simulations (LES). Insights from a DNS of a turbulent boundary layer that exhibits a series of pressure gradients and curvature effects resulting in flow separation are discussed. Furthermore, efforts towards a symmetry-preserving data-driven subgrid stress model form for anisotropic grids are presented. The model generalizes well to anisotropies, Reynolds numbers and flow physics outside the training dataset. Lastly, ongoing efforts for in-situ/online model training will be briefly discussed. This approach addresses the data I/O bottleneck of using big data to train a data-driven model by enabling model learning using in-memory databases populated by an ongoing DNS.

About the Speaker: 
Aviral Prakash is a Ph.D. candidate in the Ann and H.J. Smead Department of Aerospace Engineering Sciences at the University of Colorado at Boulder. He received his bachelor’s degree in mechanical engineering from MNNIT Allahabad, India, and his master's degree in Aerospace Engineering from Polytechnique Montreal, Canada. His research focuses on the intersection of finite element methods, turbulence modeling and simulation, and scientific machine learning. He received Dean’s Fellowship and a Departmental Fellowship for his Ph.D. work at CU Boulder focusing on scale-resolving simulations of turbulent boundary layers and data-driven subgrid stress closure modeling.