[teguhtf]

Dear All,
I'm running airfoil simulation. but it takes much time. i wanna make less time to simulate. I wanna make simpler meshing. To take less time simulation.
What should I do??
THX
teguhtf

[tstorm]

Can you post your mesh?

[teguhtf]

Hi tstorm
Could you give me your email??I'll send you soon.
THXS
teguhtf

[tstorm]

I'd rather not post it on the forum. I don't want to get onto any more spam lists than I'm already on. I'd be happy to email you if you post yours, or if you post a few pictures of your grid I can try to help.

[Philipov]

The only one effective way to reduce computational time is.... more computational power - more cores, domain partition.... the reducing the mesh may lead to not convergences... be careful if you are planning to reduce the number of cells...

[Figd84]

You can try another pressure velocity coupling that fits better your problem. Also try to use higher URF´s to achieve higher convergence speed, but be careful, since your problem can diverge.

As Philipov said, the only noticeable way to improve simulation time is more computer power.

[tstorm]

There's much more that can be done than just computing power. The turbulence model, UR factors, courant number, grid density, residual smoothing, timestep, multigrid controls, etc. can all affect convergence, and if we knew more about the case setup we could try to help you choose the right settings. Send me a private message with your email and I'll try to give some tips.

[Figd84]

But diferent turbulence models and grid density could affect the solution accuracy.

The mesh should be fine enough to avoid results variations between one mesh an a coarser one. And the turbulence model usually is chosen because the problem´s own nature (reading papers and experimental data).

[tstorm]

You're right. If he's already using the right turbulence model, there's nothing we can do to help in that case. But if it's a simple external aerodynamics problem, for example, and he's using an RSM model, we could cut processing time and use the SA model without sacrificing accuracy. Or, maybe the first gridpoint is in the laminar sublayer, which isn't necessary for the SA model. Or, maybe cell growth is so small that cell count is unnecessarily large. My point is we can help a new FLUENT user by giving real tips instead of vague pointers.

[Figd84]

That´s right. I´m just being pesimistic .

Anyway, what are multigrid controls?? I haven´t done anything with that.

thanks

[tstorm]

Ha, ok.

FLUENT uses an algebraic multigrid for implicit solutions, and a full-approximation multigrid is an available option. The most commonly changed option is the cycle type (V or W), but there's plenty of other things to play with. Theoretically, the FAS multigrid is better for non-linear problems.

[Figd84]

hmm... interesting.

Thanks

[teguhtf]

Originally Posted by tstorm

There's much more that can be done than just computing power. The turbulence model, UR factors, courant number, grid density, residual smoothing, timestep, multigrid controls, etc. can all affect convergence, and if we knew more about the case setup we could try to help you choose the right settings. Send me a private message with your email and I'll try to give some tips.

Hi Guys!!!
Thanks for your response!!
Here is my email
huget_teguh@ymail.com