Skip to content

WHOI GFD summer school project on exploring the stochasticity of turbulence closures.

Notifications You must be signed in to change notification settings

iuryt/stochastic_closures

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

68 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Stochasticity of Turbulence Closures

This is a repository for the code from my WHOI GFD summer school project.

We use the Dedalus package to solve the system of differential equations on a doubly periodic square domain with 4096 points on each side. Then we apply a stochastic Gaussian forcing centered at k=32, with width of 2 and random phase to the model. The viscosity and linear drag were chosen to fit the power spectrum within the wave numbers solved by the model. We use adaptive time stepping for the simulations, always keeping the CFL condition stable.

The computational resources used by this project were a separate challenge. We ran each pair of simulations on MIT's Supercloud machine using 16 nodes and 768 cores. This added up to a total of 200,000 CPU hours, including simulations and the analyzes.

Figure: Spin-up time series of kinetic energy (A) and enstrophy (B). Colored vertical dashed lines mark the snapshots used to present the energy spectrum in C. D and E depict vorticity snapshots for t=0.2 and 10, respectively.

The main scientific questions explored are:

  • How different are subgrid stresses from different nearby turbulent solutions?
  • How fast do nearby coarse-grained solutions diverge?
  • How quickly does the subgrid stresses decorrelate compared to the LES timestep?

Check out the video of some of those simulations.

About

WHOI GFD summer school project on exploring the stochasticity of turbulence closures.

Resources

Stars

Watchers

Forks

Packages

No packages published