LES time step
 

Hi

If you are running a LES simulation with CFX which is a fully implicit code then how would you choose your time step. Thank you

Chris

Re: LES time step
 

Hi,

Either do a mesh independence check (this can be tricky with LES), estimate a timestep based on the flow timescales or just try the recommended 3-5 iterations per timestep and see if the final results are OK.

Glenn Horrocks

Re: LES time step
 

What do you mean by flow time scales. By iterations do you mean coefficient loops? I can only check final results once i have run my simulation for a week. This is not a good option. Is there something else that you can recommend.

Chris

Re: LES time step
 

Hi,

Flow time scales - LES simulations typically resolve the turbulent eddies down to the inertial sub-range. Therefore you need to resolve to this frequency and filter out the inertial sub-range.

Iterations/coefficient loops - yes, I mean the same thing.

Glenn Horrocks

Re: LES time step
 

Hi,

A common guideline for LES is to keep the CFL number below 1.0. You can find even more rigorous guidelines recommending CFL ~ 0.5. The CFX manual also recommends this range. In several papers, researchers claim to have used CFL between 0.4 and 0.8. If you cannot afford such small time steps, maybe you could consider switching to SAS, but then you have to carefully evaluate if your transient results quantitatively correspond to reality, because there may be a tendency to damp the turbulent effects.

Regards, Magnoli.

Re: LES time step
 

Thanks for your answer Magnoli. We can fix the max courant number to whatever value we like under simulation type form. However this still does not give us our time step. How can we calculate time step?

In my current simulation i have fixed the max courant number to 0.5 and am using adaptive time step, but still i had to put an initial value for time step. How do we find an initial time step then, if we use adaptive time stepping.

CFD newbie

Re: LES time step
 

If person who initially posted this message have understood how to calculate LES time step and could share his understanding with me, it will be very helpful.

CFD newbie

Re: LES time step
 

Hi,

Personally I don't like to use automatic or adaptive time steps, I prefer to specify the time step and take the decision. If you you want to base your simulation in the CFL number (following the guideline for LES), you can run the simulation for a few time steps, check the CFL number and set it to a new value, adequated to your simulation. About how to calculate the time step, I don't know any relation which would be valid for all cases (except this guideline for LES), or which is only dependent from your numeric setup. You should use your knowledge of your physical problem and the oscillation frequencies, you're expecting to observe, as an intial guess for your time step. I usually use 1/20 of the expected characteristic period as the time step. After a first simulation I would additionally vary the time step and check its influence on the solution, to be sure that you achieved a time step independent solution. If you have no idea of what time step to start with, I would suggest that you take a further look at the physical nature of your problem. You should also evaluate your real need for LES or if DES or SAS would be good enough.

Regards, Magnoli.