I decided to try something like Computational Fluid Dynamics in Houdini using the smoke solver. Obviously the output data will not be used for prototyping or anything, but it looks cool. I've got some draft screenshots up on flickr.
http://www.flickr.com/photos/heydabop/sets/72157622551197396/ [flickr.com]
And of course, I'm open to any comments, even criticisms.
CFD in Houdini
20062 9 6- heydabop
- Member
- 203 posts
- Joined:
- Offline
- phtj
- Member
- 224 posts
- Joined: 6月 2009
- Offline
- heydabop
- Member
- 203 posts
- Joined:
- Offline
phtj
I am very interested in this - using houdini for ‘rough’ CFD analysis. In our case, we are interested in ventilation in buildings. I am wondering what level of accuracy could be expected - i.e. could houdini's pyro be used to better understand airflow in buildings.
Well, I don't really know how accurate it is, probably not greatly accurate, as it's normally used for movie effects as opposed to technical prototyping. But perhaps someone else can shed some light on the subject. I just did it because I like how it looks.
- probbins
- Member
- 1145 posts
- Joined: 7月 2005
- Offline
Well houdini isn't an engineering tool, however companies are starting to use it for just this sort of simulation.
It provides a faithful representation of a fluid solution, and is excellent for rapid prototyping the setup you want to inject into your mainframe simulator.
It provides a faithful representation of a fluid solution, and is excellent for rapid prototyping the setup you want to inject into your mainframe simulator.
“gravity is not a force, it is a boundary layer”
“everything is coincident”
“Love; the state of suspended anticipation.”
“everything is coincident”
“Love; the state of suspended anticipation.”
- heydabop
- Member
- 203 posts
- Joined:
- Offline
probbins
Well houdini isn't an engineering tool, however companies are starting to use it for just this sort of simulation.
It provides a faithful representation of a fluid solution, and is excellent for rapid prototyping the setup you want to inject into your mainframe simulator.
Ah, okay cool. Thanks for shedding some light on the subject.
- vigneshkaushik
- Member
- 20 posts
- Joined: 9月 2010
- Offline
- old_school
- スタッフ
- 2540 posts
- Joined: 7月 2005
- Offline
Simple to set up a fairly accurate smoke sim these days. All you need is a smoke tank set up as a wind tunnel and evaluate just the velocity against your colliders.
For accuracy, it's best to use the PCG Projection method. It's slower than Multigrid but it is more accurate.
After that it is all about constructing your collision geometry. It's probably easiest to add an RBD passive object to the scene rather than pump in colliders.
By default when you add an empty smoke container to start the set-up, it sets the Smoke Object DOP's Velocity Sampling to Faces which is the more accurate method. Note that some of the preset pyro tools may move this off the default of Faces.
Because you are solving just for the velocity, you can strip out most of the fields. Just keep vel, pressure, confinement, collision and collisionvel in the Smoke Object DOP.
Now crank up the resolution and sim.
The default Smoke Solver has no shaping micro-solvers in it. No turbulence, shredding, etc. You might want to add the Gas Vortex Confinement DOP to add confinement to the simulation.
It would be trivial to add tools to inspect the velocity field to see what the speed is in various parts of the scene.
See the attached file.
For accuracy, it's best to use the PCG Projection method. It's slower than Multigrid but it is more accurate.
After that it is all about constructing your collision geometry. It's probably easiest to add an RBD passive object to the scene rather than pump in colliders.
By default when you add an empty smoke container to start the set-up, it sets the Smoke Object DOP's Velocity Sampling to Faces which is the more accurate method. Note that some of the preset pyro tools may move this off the default of Faces.
Because you are solving just for the velocity, you can strip out most of the fields. Just keep vel, pressure, confinement, collision and collisionvel in the Smoke Object DOP.
Now crank up the resolution and sim.
The default Smoke Solver has no shaping micro-solvers in it. No turbulence, shredding, etc. You might want to add the Gas Vortex Confinement DOP to add confinement to the simulation.
It would be trivial to add tools to inspect the velocity field to see what the speed is in various parts of the scene.
See the attached file.
There's at least one school like the old school!
- vigneshkaushik
- Member
- 20 posts
- Joined: 9月 2010
- Offline
- mogul4200
- Member
- 4 posts
- Joined: 5月 2016
- Offline
- boya_wangca34e169c5bc427a
- Member
- 1 posts
- Joined: 10月 2019
- Offline
-
- Quick Links