CFD domain size + position

romanos · Posts: 1

Hello,

I am trying to do a very basic external CFD simulation with only one building (usually only 1 building block). The goal is to see how different forms and dimensions under exact same wind conditions influence the wind flow.

1. domain size
I have read in the forum that theoretically the domain size should not influence the wind behaviour, BUT trying out different domain sizes in the same file, have reached the conclusion that IT ACTUALLY DOES!... something which I think is quite worrying considering the domain does not exist in real life...

Am I doing something wrong?

2. domain position
I am actually interested in what happens after the wind hits my building, but the domain is always centered, resulting in a huge area in front of it being calculated although it is of no relevance to me.

Can I shift or move the domain, so its center is behind my building, concentrating on the part which I am interested in?

3. calculation times
Even with a very basic geometry, a 1m default grid spacing and Uniform grid type (as suggested by the help files), I am still facing very long calculations times (30 minutes do not reach 1000 iterations)..

Is this normal? Are there any ways to speed this up, without compromising accuracy?

Thank you for your time and ideas!
greetings
R.

Steve Potter · Posts: 572 Location: newcastle

Hi,

For external analyses, if the domain extent were to be continuously increased in the X, Y & Z dimensions, there would come a point where the domain extent would have no effect on the flow regime. However, if the reverse was applied, and the extent was reduced until it's dimensions were comparable with those of the overall building extent, this would have a significant effect. Consequently, it would be wrong to say that the domain extent has no effect on the flow regime (it would be appreciated if you could point out where this assertion has been made on the forum so that we can correct it).

Determining the correct domain size becomes a balance between arriving at a relatively economic solution (i.e. not too large a grid) against allowing the flow to develop fully around the building. Bear in mind that the flow must be allowed to develop fully on the windward side of the building as well as the leeward side (a pressure gradient is established on the windward side due to increasing static pressure) - this is why the grid extends equally on both windward and leeward sides. You cannot currently move the position of the building in relation to the calculation domain.

Although the domain is automatically generated in order to provide a reasonable balance between flow development and grid economy in most cases, there will be cases where the domain may need to be increased in size to allow the flow to develop fully and this can be established through experiment.

CFD calculations are by nature a lengthy process, however we are planning to look into the development of solution accelerators in the future. You can of course experiment with increasing the default grid spacing, you may find that you can obtain acceptable accuracy with very coarse grids.

Hope this is of help.

Regards,

Steve

Shortcuts

Login