problems for unsteady calculation
 

Hi everybody, I want to do an unsteady calculation for a flow over a flate plate in horizontal duct (like a forward-facing and backward-facing step).

-I calculate a steady-state solution before.Is it the best solution ?

-I tried with an adaptive time stepping method with the default values. But which is the criteron to determine the best time step size for a fixed time stepping method.

-Why my unstedy model give me a steady solution when I visualize my animation? My model is maybe steady or there is a problem in the calculation ?

Thanks a lot !!!

Re: problems for unsteady calculation
 

Hello,

I think that if you are that your problem is unsteady, you don't need to calculate a steady-state solution before. You can directly calculate an unsatedy-state solution.

For the criterion to determine the best time step size, it is the CFL number. You initialize this criterion in solve -> controls -> solution.

Laure

Re: problems for unsteady calculation
 

Thanks Laure (merci d'avoir répondu et bon courage pour la suite)

Re: problems for unsteady calculation
 

Hi,

You didn't say whether you were using RANS or LES. With RANS, you might seem to be having a steady state solution at the onset. You'll have to allow it to run for a considerable period of time before you get rid of the steady state behaviour.

Hope this helps.

James

Re: problems for unsteady calculation
 

Hi James, I used a k-e model. I don't know which model is the most adapted for this kind of calculation. I'll try a RANS model. Maybe my unsteady calculation didn't work over a rather long periode of time to became unsteady? Thanks James.

Re: problems for unsteady calculation
 

Hi,

I suggest you do some computation of the calculation time it will have to take you for your information to travel through your computational domain...with this, you will then allow your RANS simulation to run for this period of time after which you see some unsteady effects. You need to know the length of the domain and the average axial velocity and the ratio of the two will give you an idea of the computational time needed. If it doesn't work, i suggest you switch your RASN steady state solution to LES...i mean from your RANS steady state solution, you will switch over to LES and see how it will be behave...maybe it will be able to capture all the unsteady effects associated with your flow.

Hope it helps.

James