[abhi084]

I came to know tat there is a difference b/w unsteady simulation and transient simulation? If so, can i do an unsteady simulation in fluent and how? How will the BC be described?
Also the simulation i am planning to do is for supersonic flow over open cavities? How to go about it particularly the turbulence model to be chosen?
thanks in advance for any help.

[ghost82]

As I know there is no difference between unsteady and transient: solutions are time dependent.
What's the difference?

[abhi084]

Originally Posted by ghost82

As I know there is no difference between unsteady and transient: solutions are time dependent.
What's the difference?

thanks for Answering..but what i meant about unsteady and transient is mentioned in the post below:

http://www.cfd-online.com/Forums/mai...ady-state.html

Please kindly go through it and if possible please help since i understand the concept of what is being said, but am not able to understand on how to set this up in Fluent.

[ghost82]

In fluent you can setup steady (no time dependent solution) and transient (time dependent solution) problems; so, setup your simulation as transient.
Then, it is up to you to analyze the results (monitors for example) of your problem and understand if it reaches a steady state, a quasy steady state or if it is "unsteady".

[CeesH]

Hi,

In FLUENT, there is no discrimination.

Also, I think the distinction some of the people make in the post you mentioned is more an argument about language than really a technical point. Both transient and unsteady can be used for both types of flow that they name, in my opinion.

Anyway, the difference they indicate is as follows:

Transient: A flow that has yet to develop to it's steady state value, for example the startup of flow through a pipe, where the flow has to develop from 0 velocity to a certain steady velocity profile. But, after some time, it will reach a steady state.

Unsteady: A flow that is never going to reach a steady state; for example the flow of blood through a vein. The flow velocity continuously changes due to the pumping of the heart. This flow is periodic, so the same pattern repeats over and over, but never steady.

But as I said, it's mostly a difference in wording. In FLUENT, you'd tackle either problem by using the transient solver

[abhi084]

thank you....i will look into it........now one more thing like in a periodic problem as mentioned in the post (on unsteady vs transient) let us assume a flapping shear layer which oscillates at a particular frequency (we know the frequency from experiments) so how to go about setting up the problem?
And yeah i know this will be out of context here but how to give value for turbulence intensity if we don't know these experimentally....the formula for intensity (based on Re) can be used anywhere?

[ghost82]

Originally Posted by abhi084

thank you....i will look into it........now one more thing like in a periodic problem as mentioned in the post (on unsteady vs transient) let us assume a flapping shear layer which oscillates at a particular frequency (we know the frequency from experiments) so how to go about setting up the problem?
And yeah i know this will be out of context here but how to give value for turbulence intensity if we don't know these experimentally....the formula for intensity (based on Re) can be used anywhere?

If for example the shear layer oscillates because of flow separation/vortex shedding you can capture the oscillations with the transient solver.
The equation for turbulent intensity I think you refer is valid for fully developed turbulent flows inside pipes.

[CeesH]

setting up the problem: use the transient solver, and set a timestep sufficiently smaller than the characteristic time of the oscillations (I don't know exactly how much timesteps are required - you can do a sensitivity study by trying say, 20, 50, 100, 200 and 500 steps per oscillation and see how the solution changes).

Regarding turbulence intensity, a starting point would be to set it at 5%, but there are some better guidelines in the FLUENT manual regarding the turbulence models. So I'd suggest to look in there. (edit: as Daniele said, it might not be valid for the particular case you are looking at!)
edit: regarding turbulence intensity, do you have data of k & epsilon near the domain inlet? If so you can estimate the turbulent viscosity, and by that the viscosity ratio.

[abhi084]

Okay thanks will try to do that. yes that turbulent intensity equation was for fully developed turbulence. So how should we input the value for the turbulence model....and which model would be most suitable for a wall bounded shear flow with a small cavity....i have seen some papers where some ppl have used S-A model others have used K epsilon.....so how to go about it....plz help

[CeesH]

I have no experience modeling such geometries, so I can only advice you by saying literature is your friend. Look at what others did, what they found and what they recommend regarding turbulence models, and perhaps also regarding boundary conditions. It's not a quick solution, but well, that's how it is...

[ghost82]

Originally Posted by CeesH

I have no experience modeling such geometries, so I can only advice you by saying literature is your friend. Look at what others did, what they found and what they recommend regarding turbulence models, and perhaps also regarding boundary conditions. It's not a quick solution, but well, that's how it is...

I agree.
Each turbulence model has its pro and cons.
You should investigate not only one turbulence model but at least one of each "family" and see what are differences.

PS: need to add to my signature "literature is your friend also"

[abhi084]

Originally Posted by CeesH

I have no experience modeling such geometries, so I can only advice you by saying literature is your friend. Look at what others did, what they found and what they recommend regarding turbulence models, and perhaps also regarding boundary conditions. It's not a quick solution, but well, that's how it is...

Thank you very much....As suggested will look more into the literature. Till now was focussing more on the experimental part but now will definitely look into it.

Thanks Again..

[Logha94]

[QUOTE=CeesH;522967]setting up the problem: use the transient solver, and set a timestep sufficiently smaller than the characteristic time of the oscillations (I don't know exactly how much timesteps are required - you can do a sensitivity study by trying say, 20, 50, 100, 200 and 500 steps per oscillation and see how the solution changes).

Hello,

mind if I ask a question in here? because I'm struggling to find some answers to my problem.

I am carrying out an unsteady flow simulation based on my steady computation using spalart-allmaras and SST k-w model.
firstly, I step up the time step size(s) to 1e-4 with time step = 100 and max iter/time step = 20, with these numbers it started oscillating and once I started decreasing the value as 1e-3,1e-2,0.01,0.02 and so on, the graph no oscillating.

I hope youll could help me.
Thank you