Turbulent Kinetic Energy


I've noticed in the changelog of version DualSPHysics Package v5.0 (26-11-2020) :

- MeasureTool: Computes Turbulent Kinetic Energy (TKE).

And in the help section, it is said that:

Turbulent Kinetic Energy is calculated using the average velocity of selected PARTS

I'm a bit confused. I was under the impression that DSPH was not able to model turbulence... Is the calculation done here have any sense, is it based on any studies ? Why average parts (time)?

Have I missed some new section in the pdfs that explain this option ?



  • Hello TPouzol!

    DSPH is able to model turbulence when you use SPS-LES, since this method will actively resolve eddies larger than a specific size, and then model (i.e. approximate) turbulence below a sub-scale. Artificial Viscosity is not similar to a turbulence model, since it acts more as a way to smoothen shocks, than actually trying to resolve eddies.

    When the turbulent kinetic energy is calculated, then remember its definition:

    Where u,v,w are velocities in x, y and z direction, while the prime denotes fluctuating components. Typically this definition is used when using a RANS average flow, where you decompose your velocities into average and fluctuating parts.

    Without knowing exactly how DualSPHysics are implementing it; if you can split your velocities into fluctuating and average parts - you are able to do this calculation.

    Please don't take this as a definitive answer, just something to keep one thinking, until the DualSPHysics team hopefully explains it a bit more, since to my understanding this new option should only work for SPS-LES.

    Kind regards

  • @Asalih3d Thanks for your input !

    Kind regards

  • @TPouzol I am not into the SPS-LES theory. I do not quite rule out that the LES in SPS-LES is a different flavour than the Large Eddy Simulation proper.

    Nonetheless, when you use Large Eddy Simulations, the turbulent oscillations are those with respect to the resolved flow, which is as unsteady as the resolved large eddies are. True to the name behind LES. So the TKE calculated as deviations from a time-averaged velocity, as @Asalih3d drafted out, sounds like an operation for Reynolds turbulence, that is a time average of the full kind which averages out anything, including the large eddies.

    So I concur that a line in the change log makes it premature to see the full picture. More information in due course on what this functionality actually entails will be welcome.

  • edited December 2020


    Thanks for sharing your ideas.

    I have submitted this issue in the main sticky topics regarding "List of parameters or options that need better explanation"

Sign In or Register to comment.