Meshing settings for skyscraper outdoor analysis


I went through the example file for outdoor analysis, everything works fine. I mostly figured out how to setup mesh settings to get faster result. But I have problems when it comes to a larger scale and more complicated geometry.
The wind tunnel is quite large (1.8x4.8x0.7 km) because of the surrounding area and building height, therefore there are a lot of cells in tunnel grading, even with a cell size of 10.

Snappy hex meshing took about 12 minutes with default settings from example file. Skyscraper’s geometry is curved and I’m not really sure how to set up snappy meshing properly.

Since my goal is to use Galapagos to minimize wind load on the structure, I guess that my probes should be face centers of meshed skyscraper? The problem is, in my opinion, that there are a lot of points, about 4k.

Bear in mind that for this part of the algorithm, i.e. wind load optimization, I only need Galapagos to be able to make difference between good and bad results (higher and lower cumulative wind pressure). So I suppose that low meshing settings should do well without taking too much time to get the results.

Sorry for the long introduction to the problem, hopefully someone got this far :slight_smile:

I guess my main questions are:

  1. If the wind tunnel is so large, would increasing the cell size even more cause some problems down the road? What would be the limit?
  2. How should I set snappy hex mesh dict, since I got lost with all of the inputs?
  3. In order to reduce the number of probes, can I use some kind of low poly mesh (currently it’s just brep to mesh) and how “low” can it be in terms of number of points/faces?

Hopefully all of this is not too computationally intensive and I don’t need a supercomputer. I’ve got 4 core/8 thread CPU (4 of them are used for virtual machine) and 16GB of RAM.

Kind regards,

skyscraper_cfd.3dm (310.0 KB) (85.1 KB)

1 Like

Please find the file attached. With the settings in the file I could get snappymesh like the following image. I believe this is a good start for your studies. (81.1 KB)

1 Like

Hi, devang, thanks for taking the time to respond.

While waiting for someone to reply, I more or less figured out how different parameters affect the time needed for snappyhex meshing.
I ran your file and the meshing took about 8 minutes. Since I need it to be faster for galapagos optimization, I managed to take the time down to about 3 minutes, with of course less detailed mesh. The solution still took more than half an hour.
So I guess the whole process can’t be sped up much more?

I often get some defects after meshing, no matter the quality.

Can these affect the results to some high degree? I know that the probes have to be outside the geometry, but anything else?

Since I might be able to use a computer lab at the faculty, do you know whether other computers have to have OpenFoam installed or not?

A question regarding your gh file. Why did you turn off the snap option? I noticed that it reduces the meshing time, but does it affect the results since the geometry is curved?

Not really sure if I have to create a new topic for this but I’ll ask here. After running the solution there were negative pressure values. Is this an error or it might be the suction forces that are represented as negative pressure?

Thanks again :slight_smile:

1 Like

Absolutely yes.

I mainly chose the option to reduce the meshing time. You should try to set it to true. It will require a finer mesh. Which will again increase the meshing time.

Negative pressure on the leeward side should be fine.

1 Like