Wind load on building

I am attempting to simulate the wind load on a building. I would like to know the pressure on the cladding of the building. The project is here:

The values that I am getting do not seem reasonable. When I have the wind blowing from the direction that I have identified as north, the pressures are negative everywhere on the surface of the building, including the wall that is facing directly into the wind, which is counter intuitive. The values also have a much larger absolute value (more negative) than what I would expect based on calculating the wind load from the building code.

I have tried the simulation using the default numerics, and the numerics that were used in this wind load simulation for the Bank of China Tower:

The results using the two different sets of numerics have a similar pattern, but the absolute value is much greater with the default numerics.

The mesh that I am using is relatively simple, to save on CPU time. Perhaps the problem is just that the mesh of my enclosure is too simple, so I have generated a more detailed mesh, but I don’t want to waste the CPU time running a bunch of longer simulations if there is some basic mistake that I am making.

Thank you!

Hey @stu_crawford!

The discretization looks quite okay to me :thinking: Did you start a convergence study so far?

Tagging the @cfd_squad for more ideas here.

Cheers!

Jousef

Thanks @jousefm!

I just ran the simulation with a more detailed mesh.

When I ran the simulation with a simple mesh, the convergence plot residuals were much better using the numerics from the Bank of China Tower simulation compared to the default numerics, so I used those numerics for the simulation with the detailed mesh.

The results look similar with the more detailed mesh compared to the simpler mesh, so perhaps the mesh doesn’t need to be that detailed. But the results still seem wrong. The flat wall facing into a 100 km/h wind is experiencing relatively uniform pressures of -1000 Pa, when it seems more logical that it should be experiencing pressures of +500 Pa. The entire surface of the building is experiencing negative pressures, which seems unlikely.

The wind load from a south wind makes intuitive sense, with positive pressures on the wall facing the wind, and negative pressures in the lee of the building. The magnitude of the forces are similar to what would be predicted from the building codes. Pressures are shown below.


However, the results from a north wind don’t make sense. Even the vertical wall facing into the wind has substantial negative pressures. Every part of the building has negative pressure, and the magnitudes are significantly greater than what would be predicted from building codes. Pressures are shown below.

If the wind is from the northwest, the resulting pressures are quite different, and make more intuitive sense. Positive pressures facing into the wind, and negative pressures on the lee side, with reasonable magnitudes. Pressures are shown below.


I have also tried running the simulation with winds from two other directions, southeast and east/west. However, both of these simulations failed. One because the pressure field started diverging, the other because the velocity field started diverging. Not sure of the cause of that?

Picked by curiosity I did a particle trace on your Northwest simulation and it looks normal:

Perhaps you fixed already the simulation setup (like re-setting velocity inlet / pressure outlet)?

Take care,

Retsam

1 Like

Thanks @Retsam!

I’m new to simscale, and didn’t realize that the particle trace was an option. I played around with it a bit, it is very neat! Very useful to help understand what is happening. But what is the second screenshot that you provided? It also looks very useful, and I couldn’t figure out how to create it.

The results for the northwest simulation looked good, so it makes sense that the particle trace looks normal. It was the north simulation (which is different by 45 degrees) that was giving unreasonable results. But based on your suggestion, I ran a particle trace, and I think that explained the issue. In the north simulation, my velocity inlet was in the positive y direction, and I had my inlet velocity as a positive number. I think that this meant that it was sucking instead of blowing. I changed the inlet velocity to negative, and the results look much better.

My simulations for a west wind and a southeast wind both fail, and I haven’t been able to figure out why. One says that the velocity field started diverging, and the other that the pressure started diverging. Both give very unreasonable results before they fail. I looked at the position where it fails, but it isn’t on the surface of the building, and I can’t see any issues with the mesh. Are there other things I should look at for troubleshooting?

So far, so good.

For second screen shot, I suggest you use it systematically to spot wind directions in a plane.

  • Use ‘Cutting plane’ and move it to appropriate geometry height.
  • Use ‘Vectors’ and set it for ‘Velocity’.
  • Adjust arrow (vector) size tweaking it in ‘Results’.

Learning correct visualization takes time. And it is by doing, so few days are necessary to be at ease, alas.

Regarding mesh you make, for starting I say that back, curved face of the building is too fine (2 cm cell?). I suggest you use ‘Advanced setting’ while meshing and set ‘Small feature suppression’ to bigger value than 1e-5 m. This could help with building edges.

Bottom line is to make your mesh smaller while evaluating. You should be able to simulate in few minutes only.

Later on, you can check how finer mesh relates to your first and quick mesh settings: it is possible, that your are simply wasting your core-hours now, and these resources are limited.

Cheers,

Restam