Dynamic Data-Driven Smagorinky Closure for LES
Last updated
Last updated
Before reading this tutorial, you should know a bit of Python. If you would like to refresh your memory, take a look at the .
We are going to work on the reduced that is available on the . Make sure to download the data folder before starting.
This tutorial is aimed at a user who has a solid understanding of Computational Fluid Dynamics (CFD) for Combustion. A basic understanding of Machine learning is not required.
The tutorial could also be beneficial for those who do not have strong competencies in CFD but know Data Science; in fact, apart from some computations to compute interesting quantities in reacting flows, the basic operations that we are going to perform on the data are based on filtering of 3D fields.
The dataset is extracted from a DNS simulation of a . The variables saved comprise Temperature, Species Mass Fractions, 3 Velocity components, and Pressure. The chemical mechanism used comprises 9 species.
You will learn:
how to use aPriori to compute the Strain Rate,
how to filter the results obtained to resemble an LES field,
and build a data-driven closure based on Neural Networks to model the subgrid stresses.
