CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLUENT (http://www.cfd-online.com/Forums/fluent/)

 Swarup September 29, 2006 23:03

Hi All,

I wonder how large scale domains with dimensions like: 200X30X30 m^3 are handled in FLUENT. A common case can be airflow around a group of buildings.

Are they handled at their actual scale or there is some other way?

Any comment on this and any advice is highly appreciated.

Swarup.

 Ralf Schmidt October 1, 2006 07:29

Hi!

You can scale every grid in Fluent to the dimensions you want to. Use grid->scale.

So, if you build a grid in meters, just check the scale option, if fluent has it. Otherwise modify the unit, the grid was generated and scale again.

In this way, you can have every grid dimensions you want to!

Ralf

 Swarup October 2, 2006 07:31

Hi Ralf!

Thanx a lot for the reply. If I have not understood your post properly, bear with me.

When I scale the grid, domain extents are also scaled. So I will end up simulating a scale model instead of actual domain.

As far as I know, one should use scaled NS Equations so that Re is the only factor to be specified. Then results are applicable at all scales provided Re is maintained same.

So, will you still advise me to scale the grid and go ahead?

Sincerely.

Swarup.

 Ralf Schmidt October 2, 2006 08:34

Hi Swarup,

yes, you must scale your grid!

An example: you create a geometry (in gambit) that is 100 by 200 by 300 (in gambit, there are NO units!!).

Now, you import it into Fluent. The standard in Fluent is meters, so if your geometry is in meters (100m by 200m by 300m) you have nothing to change. But if it has any other dimensions, it has to be scaled (!).

Otherwise, Fluent returnes wrong results. The BC you enter (like mass flow inlet or density of the material) have real physical units (kg/s or kg/m3) and these have to be calculated with real dimension in m, mm, km, ft or whatever unit you want!

So, there will be a difference, if you pour 10 kg/s water in a shoe carton or in a canyon!

Another point in that topic are the reference values you set (Report->reference values). They are important for reporting results like drag, Nu-number, Tu-intensity.

Hope, it helps

Ralf

 akr October 2, 2006 19:56

Hi Ralf, I am a beginer of fluent. So could you teach me? I'm just wondering... If I create a geometry in gambit that is 100 by 200 by 300 and create meshes by 1 by 1 by 1 cube, the mesh size will depends on the scale. That is, when I scale the grid by metre, the mesh size will be 1m by 1m by 1m. When I scale that by inch, the mesh size will be 1inch by 1inch by 1inch. Is it no problem?

Sorry for terrible English.

Thanks, akr

 Ralf Schmidt October 3, 2006 02:12

hi akr,

you are absolutly correct! The grid size depends on the scale of the mesh in Fluent! If your grid is 100 by 200 by 300, and you have grid-cubes of 1 by 1 by 1, Fluent will scale the domain AND the grid in the unit you apply! So if you use mm, your geometry is 100mm by 200mm by 300mm and the cell zie is 1mm by 1mm by 1mm.

Take care, the scale function has to be used correctly: first, enter the dimensions, the grid was build in, then, click the scale button. Check, if the overall dimensions of your geometry are correctly.

BUT take care of HTC, Nu-number, TU-intensity, drag and others, they are related to the reference values.

Ralf

 akr October 3, 2006 04:15

Thank you, Ralf. Actually, I am going to create a grid of aeration tank. It is quite big (10m by 3m by 5m). I want to mesh the grid in small number as possible, because I don't have powerful computer. Anyways, I will try to make the number of mesh smaller. Now I want to know the detail of your comment. "BUT take care of HTC, Nu-number, TU-intensity, drag and others, they are related to the reference values." This comment seems to mean that I can get a good model by checking HTC, Nu-number,,,, Could you teach me what this is meaning and how I can do that?

Thanks, akr

 Ralf Schmidt October 3, 2006 04:18

no no no..

the model has to be ok anyway...

The reference values are important, ONLY for reporting the results!

So it might be happend, that you are scaling your grid in mm, but the reference length is 1000mm - that will give you totally different Turbulence intensitys...

Ralf

 akr October 3, 2006 20:15

Thanks, Ralf.

O.K. I understood about the scaling. But what I want to know is how to decide the mesh size. I think the smaller mesh size will give me good results. But I need much time to calculate. That is why I want to make the mesh size larger. And I can make the mesh number smaller.

Is there something like rule about mesh size?

Thanks,

akr

 Ralf Schmidt October 4, 2006 04:11

Hi,

ok... mesh size is a totally different topic. I would suggest to read fluent manual or general CFD book, there are some information on that.

In general, some ideas that may help:

- Your grid should be small, in the regions, where you will have high gradients of velocity, temp., ...

- You should have at least 5 cells in a gap between two walls

- Use sizing function and Boundary layer options to achieve the change in grid size over your domain.

For more questions, I suggest to open a new question in the forum.

Ralf

 akr October 4, 2006 23:09