best practice for CFD simulation rendering

[fraggle]

Thanks, I don't need the nightly for this test anyway.

What is the parameters to change to limit the flow into a value of X Liter per second?
Is the water viscosity included by default?

So I give it a try, but I didn't notice anything related to the incoming flow (except using the surface of the emitter to find out the volume, but it doesn't take into account the pressure to quantify the quantity per second).
As well for some reason whatever the timeframe, the sink never get filled up!
I also face a problem with particule poring trough my model. Does it have something to do with my model having 1.5mm thickness?

Thanks

[Svetlin.Nikolov]

Hey,

You can start off wither with Raven's scene, or you can check the liquid settings with the TapWater preset on the toolbar as well - it will automatically adjust the dynamics of the simulator to that of water. You can adjust the viscosity from the Dynamics rollout.

For the emitter flow, you need to set the Source into Surface Force mode and you can regulate the Outgoing Velocity parameter - it's in voxels/second in order to decouple its meaning from the shape and surface area of the emitter geometry. You can also use the Polygon ID param in the source to make just one set of polygons emit, instead of the entire emitter surface.

You must make sure your model is a watertight geometry - it must be shelled, must not have holes, inverted normals or any other kinds of broken topology. If this is okay, another possible reason is indeed that the grid resolution of the simulator is too low compared to the thickness of the geometry. Can you share a screenshot of how the leak looks? If there are leaking particles, this would be a reason why it never get filled. You also need to watch out for the unit scale of the scene - Phoenix takes into account the scene scale and you can check in the Grid or the Simulation panel how big the sim container is in world units - these must be accurate or at least in no less than half and no more than twice the scale of the real-world effect you simulate.

Cheers,

[fraggle]

it's in voxels/second

Can we convert voxels/second to Liter/second to be in a physical scale instead of a graphic scale?

[Raven]

My settings where all pretty much default. All I changed was steps per Frame to 3.
Didn't attach the scene last time, just the video, but here it is now. If it is any help

[Raven]

If I'm right just create a 10cm x 10cm x 10cm grid at the same cell setting. That should tell you how many voxels to a liter you have.

[fraggle]

Hello,
Here is a screenshot in max as well as the one in SW.
The sink get fill-up but no water are dropping from the overflow.

I will tried again with a basic box and see if it works better.
In anycase, if I can't regulate in Liter/second the incoming flow from the emmiter, the simulation will have no sense.

[fraggle]

If I'm right just create a 10cm x 10cm x 10cm grid at the same cell setting. That should tell you how many voxels to a liter you have.

Thanks for trying to help.
If the doc is right, voxel = cell. https://docs.chaosgroup.com/display/...+FD+in+3ds+Max


So, a 10x10x10cm grid with a cell size of 0.1mm give 1 000 000 (1M) voxels, that's a beginning.
Considering that the quick water setup give a value of velocity of 14.346 cell/second, we can conclude that this value correspond to 0.000014346 liter/second.
0.2 liter/second is then equal to a velocity of 20 0000 cell per second.
But None of the value change when the size of the emitter change, so this shortcut won't give the right result I guess.

Looking at the simulation with 20 000 Cell per second, it's definitively not the right approach!

[Svetlin.Nikolov]

Yup, you need to take into account the surface of the emitter geometry to get L/s.

[fraggle]

Yup, you need to take into account the surface of the emitter geometry to get L/s.

Sure, but I can't guess the relation between surface emitter and velocity. The default value of 14.346 cell/s correspond to what surface? Without this info I'm stacked.

[fraggle]

Yup, you need to take into account the surface of the emitter geometry to get L/s.

Sorry to insist, but without your help, the conversion into L/s can't be done.

[fraggle]

Sorry to insist, but without your help, the conversion into L/s can't be done.

sound like you don't have the matht o convert cell into physical scale

[Svetlin.Nikolov]

Not sure why you need this in real world units though, can you explain more about the purpose of your setup? If you are going to use it simply for visualization purposes, the discharge doesn't really matter - you can just set any value that makes it look good. The only thing you need is just the surface of your emitter - A * (voxels/sec) * voxel_size will give you the volume of emitted liquid which you can easily convert into liters...

[fraggle]

Svetlin, I just follow your advice.
If Phoenix is as good as a CFD, why should I make 2 simulation, one as a proof of concept and one as a marketing purpose when both can be done in one shot?
The proof of concept inside phoenix will be valid only if the incoming flow respect the requierement of 0.2l/s.
That's the purpose of the Velocity, but in cell/s it doesn't allow me to connect with the real world unit. Proof of concept can be made in voxel because there is no vexel in the physical world.

Thanks.

Well, Phoenix is a fluid dynamics solver and shader. It's oriented towards VFX, but that doesn't mean it entirely fakes its results and is inaccurate - like all CFD solvers you can choose how precise the results are, which is a tradeoff between simulation time and quality. The current state of scientific progress on CFD is pretty low in my personal opinion and the same numerical approximations are used in both Phoenix and the responsible industrial CFD solvers. Of course, you should not use Phoenix for solutions that would decide people's lives, or for supersonic fluids which go by different laws, but the reference video is something that is easily achievable with Phoenix - all the visible aspects of it at least. If you need it for visualization purposes, I think you should give it a shot

[fraggle]

World unit are necessary to make sure that the overflow is sized properly to drain out as much water as the incoming water.

Not sure why you need this in real world units though, can you explain more about the purpose of your setup? If you are going to use it simply for visualization purposes, the discharge doesn't really matter - you can just set any value that makes it look good. The only thing you need is just the surface of your emitter - A * (voxels/sec) * voxel_size will give you the volume of emitted liquid which you can easily convert into liters...

[Svetlin.Nikolov]

Indeed, keep us posted