Wind & adaptive grid

[LarsSonparsson]

I'm having problems how the Max wind affects to my simulation.

I need to do a very slow large scale smoke, where the smoke is emitted from the right side of the image and during 1000-1500 frames wind blows it to left filling the whole shot. Check the attachment.




Some how the left edge of the smoke keeps raising all the time and I think that the reason is adaptive grid. Because the grid keeps growing to left following the smoke, there isn't actually any "wind movement"/velocity yet in the new area of the grid. Right? How to avoid this?

Simulation is so long, that I really don't have time to simulate it with out adaptive grid. Otherwise I would be simulating 2500 frames using 80-100 million cell grid...

[Svetlin.Nikolov]

Can't open the attachment for some reason. Are you sure you don't adapt by speed? That's the only case when the grid should really grow if you put a wind force, because the velocity on the far walls points out of the grid and it decides it should expand every frame.

[LarsSonparsson]

The way how the grid is growing is just fine. It follows the smoke in thresholds like it should.

Check this video. There is few moments from two different simulations showing the problem.

https://www.dropbox.com/s/55e1d1809j...FD001.mp4?dl=1

In both simulations, the left side of the grid is growing exactly how it should, staying quite close to the edge of the smoke.
So the grid isn't the problem, but the shape of the smoke is.

[Svetlin.Nikolov]

Btw, are you using a recent nightly build? If these videos are from the GPU preview, the shaking grid was fixed a while ago.

[Svetlin.Nikolov]

You could use a bit of Extra Margin so the smoke will have more place to grow to the left and not smash against the wall like that, or you might try to lower the smoke threshold further.

[LarsSonparsson]

Not the latest. Preview is done in every 3rd level of detail, so that might cause the shaking.