2 questions about fluids, velocity and collisions

[simonj]

Hey,

I'm working on an experiment with fluids, and well, I've made some progress in gaining a better understanding of fluid simulations in Houdini, but some things I just can't figure out.

My setup is simple, I'm trying to control the velocity of one smoke field using a particle field as a velocity "guide".

1: Collision velocity

I'm using a "gas particle to field"-node connected to the pyro solvers "velocity adjustment" input (#3), the "gas particle to field"-node is set to multiply values in the DOP field "pyro/vel" from the particle field, thus a particle value of 0, 0, 0 would create no motion in the smoke, and a value of 1, 1, 1 would allow full speed in all directions. However, when I first tried using geometry to collide with the smoke I still got velocity from the collision, I fixed this by also entering the DOP field "pyro/collisionvel" in the "gas particle to field"-node. And this resolved some of the motion, but not all, the velocity field seems to be smoothed out affecting the whole simulation, as visualized in picture 1.

2: Stuck smoke

When I run my sphere through the smoke I seem to getting some smoke trapped inside the sphere, this is not desired, the sphere is a simple polygon divided static object, what can I do to prevent this? Picture 2.

Once, I had a third question, but now I can't remember it

Anyway, any tips and tricks about this are super welcome!

Thanks,

Simon

[Overload]

I adjust collision velocity in the SOP level with a multiplier of the velocity(through either a point or trail sop). Then in your pyro solver, turn on use point velocity for collisions. So far I've found its the easiest way to adjust how things collide. You can just promote parameters for the velocity so your not diving in and out of networks.

[simonj]

I adjust collision velocity in the SOP level with a multiplier of the velocity(through either a point or trail sop). Then in your pyro solver, turn on use point velocity for collisions. So far I've found its the easiest way to adjust how things collide. You can just promote parameters for the velocity so your not diving in and out of networks.

Hi, thanks for the answer! I tried that way of "pushing" the smoke a bit further from the ball, and I got a little bit better results actually, still don't understand why the smoke would break through the geometry in the first place, but well, this will work

Question #1 still stands though, does anyone have an idea how to avoid the velocity from collisions to "spread", even after adjusting the velocity with a point field? Or does this involve developing a custom solver?

Thanks-

Simon J