Simulating Wind on Procedural Terrain Using Lattice Boltzmann Method | Hacker News Reader

Simulating Wind on Procedural Terrain Using Lattice Boltzmann Method

nickmcd.me

194 pointsdmit21 comments

There's a guy who tested his own LBM solver on a space shuttle model https://www.youtube.com/watch?v=5AzxwQpng0M at 60x higher resolution than the largest Space Shuttle CFD simulation ever done by NASA. Took 6 hours to calculate on an academia GPU setup.

Edit: He also says it's full engineering accuracy, not cheap CGI CFD that only has to look plausible. You could extract pressures, forces from the surfaces and use it for analysis. He's done extensive validation for his academic work.

That is the most amazing thing I have seen in months.

https://www.youtube.com/watch?v=pD8JWAZ2f8o

That would deserve its own submission

Project code here: https://github.com/ProjectPhysX/FluidX3D

I hope advances in this field would help to reduce turbulence in passangers flights.

One of my favourite wind implementations (both visually and from a game design standpoint) has been the winds in Ghost of Tsushima, there is a interesting talk/video with the developer who worked on it. https://www.gamedeveloper.com/api/redirects?to=/gdc2021

The way they decided to use the wind to guide players to the next point of interest is also an interesting concept in the minimap-plagued open world genre.

Slides of talk being referenced[0], not managed to find the video yet.

-[0]: https://www.gdcvault.com/play/1027124/Blowing-from-the-West-...

Damn, I saved the wrong link. There was a video tho.

Still not found it, if you do, please link! I'd love to watch it :)...

There is a youtube series with a coder doing this sort of thing that is absolutely amazing: https://m.youtube.com/watch?v=eaXk97ujbPQ

Shameless plug: I wrote a game that lets you play around with this sort of thing, it's on steam and it's free: https://store.steampowered.com/app/1482770/Terra_Firma/

WAsP CFD [1] is the industry-standard (at least it was when I worked in wind engineering 10 years' ago) for modelling flow over complex terrain, and is applied to wind farm modelling. [1] https://www.wasp.dk/waspcfd/flow-model

Not sure if they changed things, but Blender's fluid simulator was[1] based on the Lattice-Bolzmann method[2].

[1]: https://docs.blender.org/manual/en/2.79/physics/fluid/types/...

[2]: https://elbeem.sourceforge.net/

I'd like to see validation of these results with a traditional numerical meteorological model like WRF at microscales, or relative to actual measured winds from a wind farm. If accurate, it could be dramatically faster than the typical NS solvers.

Has anyone done this for fabric sails on a sailboat?

This is interesting to me but I have a question on the definition of a fluid that is presented here. It says it is a medium that cannot resist any stress applied to it —- is that not what surface tension does? Anyone smarter than me got an explanation?

Your question actually suggests that you are smarter than me, but anyway: the surface tension of an ideal fluid is indeed zero.

Real fluids are just approximations of the ideal. :) Their fundamental flaw is not being truly continuous, but rather consisting of sticky molecules. Surface tension is caused by the imbalance of forces acting on molecules not surrounded by other molecules on all sides:

https://en.wikipedia.org/wiki/Surface_tension

LBM is fascinating, an informatively different route to a fluid solver

So cool. Wish I have read this two years ago when I started to work with Lattice Boltzmann simulations using OpenLB (which is incredibly complex and rather underdocumented).

Does the page randomly scroll from time to time for anyone else?

This is extremely well written

Reply on news.ycombinator.com