[terryley]

Hello,

I am trying to simulate a LES of flow around a car. I am not able to compute Turbulent Kinetic Energy. It shows the same value for every iteration. I am not sure what i am missing in setting up the simulation. Kindly help me.

[AliE]

Hi,
Which model are you using? Which software? "Standard" les model like smagorinky, wale or dynamic model do not use tke and does not solve for this quantity.

[LuckyTran]

There is no turbulent kinetic energy (a time-averaged variable) for a time-accurate simulation (LES is time-accurate). How do you even have any values to look at?

[AliE]

Same answer at the same time up to my knowledge there is a les model with an improvede eddy viscosity formulation which solves for tke, but I don't know if this is the case.

[terryley]

Im using Star CCM+. Im using WallE model.
Could i get an idea of how i should set up LES for flow around a car pls. The user guide and star ccm+ help have steps only for Aeroacoustics and transient.

[AliE]

Ok, wale model does not solve for tke. You can look at this model here in the form to convince yourself. It calculates an effective viscosity using the strain tensor and the flow gradients.

First of all, LES is always transient, are you trying to do a steady state simulation?

Apart this, a LES is not so easy to setup, grid requirements are much larger than a urans and you need some testing to have a nice mesh. Then you shuold specify the filter to use (usually cubic root of cell volume).

If you use star, than there is an ad hoc forum where you can post you question.

[terryley]

I am trying to setup an LES on Star CCM+. I am not sure how i will know when the simulation has converged or not, because Star CCM+ does show that the simulation has converged like Ansys. I will have to do some more indepth study on Star CCM+ before i could start doing LES. Thank you very much.

[AliE]

Ok, but please note that a LES is always an unsteady simulation. Thus, what do you mean with "convergence"? You can have a convergence only from a statistical point of view!

[terryley]

I'm aware LES is unsteady. When i say converged, i mean the difference in Mean averaged data for time steps.

[LuckyTran]

Originally Posted by AliE

Same answer at the same time up to my knowledge there is a les model with an improvede eddy viscosity formulation which solves for tke, but I don't know if this is the case.

Just wanted to add that these models solve for the subgrid-scale tke, not the full tke. Even with Smagorinsky models you can calculate the sgs tke by using a cute formulas (the square of turbulent viscosity divided by length scale).


To calculate the full tke, you need the time-accurate filtered velocity (the instantaneous x-velocity, y-velocity, z-velocity from Star), take the rms, and add to it the sgs tke calculated using the same formula. The rms of the velocities you can get from Star tools built-in. You have to do some work to get the sgs tke. Maybe you have to define a field function if it's not built-in (I'm not sure if it is built-in or not in Star-CCM, I assume probably not).

[terryley]

Thank you everyone. I have a a clearer idea of what i need to do now.