Hi all!

Well, I am simulating an external flow in order to calculate drag coefficient of a car. First, I am running simulations in 2D, with the side view of the car. The mesh has 120.000 tri elements and inlet flow is 10m/s. The fact is that I first make 500 iteratinos in order to get a first solution with the k-e standard model and standard wall-functions and then, I run 1500 more iterations with the RNG model and non-eq wall-functions until drag coeff. is gets stable. The problem is that in the iteration 2200 more or less, residuals start to diverge (with a linear slope, not chaotically).

I am starting to think that Fluent first make a solution and when residuals are little enaugh, then Fluents tries refine it, but if the mesh is not fine enough, solution diverges.

I've refined my mesh with 200000 tri. elem. and then I've run the simulation again, and when it arrives to iteration 2200 solution still converges (but slower). My reasoning with "Fuent's way to calculate" is correct or am I making some mistake with the grid, etc.?

(In both simulations I retain y+ in the 30-200 range, so I don't think It's an y+ issue)

Thanks for advance!

Hi Freeman,

Try checking your outlet boundary conditions. What turbulence parameters have you set at the pressure outlet? Yes, you can play with relaxation parameters as well, they are a help at times!

Please let me know if it was of any help.

Alan.

Thanks Alan!

For the inlet Flow I use Intensity turbulence of 0.5% and the hidraulic diameter. For the outlet flow I use the "Outflow" boundary condition, not the "pressure outlet": perhaps my mistake is there, but I don't know the difference between these two boundary conditions. If I try pressure outlet, which turbulence parameters should I use?

Thanks a lot!

Hi Freeman,

It depends on your application. If you are simulating a wind tunnel, I would prefer using a turbulence intensity of 0.24% and a Hydraulic Diameter of 0.1m. This is good for fully developed flows inside a wind tunnel.

As far as the pressure outlet is concerned, if you dont have any data for turbuelence intensity etc, use the outflow boundary condition which is correct in your case. It is suitable for situations where turbulence data is not available!

Hope this helps.

Thanks, Alan

OK! But the hydraulic diameter doesn't have to be according to tunnel's diameter? Because I am simulating a flow over a car and tunnel must have a diameter of about 4 meters (and I still consider this a low value)

Or this hydraulic diameter that you say has nothing to do with tunnels diameter?

Thanks!

Hi freeman

the cross section of the most wind tunnels is rectangular so when calculating hydraulic diameter you must use the formula : Dh=4*A/P where A=area of rectangular cross-section and P=wetted perimeter for further help if you email me I can send a document about this subject

best regards Erdem

[abhishek.mnit]

Originally Posted by erdem
;126605

Hi freeman

the cross section of the most wind tunnels is rectangular so when calculating hydraulic diameter you must use the formula : Dh=4*A/P where A=area of rectangular cross-section and P=wetted perimeter for further help if you email me I can send a document about this subject

best regards Erdem

hello guys
i am doing the same analysis in 2d but i really dont know the other parameter. i used the k-epsilon eq. to solve it & after 930 itration solution was converged. but i not sure i am right or wrong ..
can you tell me all other parameter?
other thing is in boundary condition we are using air density constant but in real world that doesnt happen......... if we select density as ideal gas it enables energy eq & in boundary condition as temp is constant it does not solve & stops the itrations.....
help me guys...

[RaiSiN]

hi abhishek...for the iteration, try to reduce the convergence criteria maybe from 0.0001 to 0.01 or 0.05.. then it may stops after few iterations..

Solve >Monitor>Residual> convergence criteria

[Albis]

I'm making a simulation about combination of savonius and darrieus wind turbine in 2d. I also used boundary condition on the mesh. I'm not sure if it's necessary or not. And when I run it on fluent it never get convergent. I don't understand, I run it on laminar and the velocity is 2 m/s. Also green-gauss node based. I'm wondering what's the difference between node based and the other. I need some advice, I wish someone could give me one