[Lapidus]

Good Day everybody,

At the moment I am supposed to run a simulation that is the same as the one in the video below. Water droplets impact a wall. Some of them will flow out through an opening at the lower part of the wall, others will be dragged by the gas to another exit. Basically what I want to simulate is the same that can be seen in the video, but adding a small slith next to the corner.

I am having trouble setting this up, haven´t been able to chose the right physics. Neither have I been capable to figure out how the simulation is introducing the droplets in the volume (as a Lagrangian injector?).

Do you have experience with something similar? Do you know what physics are used in the video? Then just let me know, it would be incredibly helpful.

Thank you!!

VIDEO: https://www.youtube.com/watch?v=yvqL5v0KhUQ

[pbrady2013]

Hi,


What package are you using? I've done similar simulations in the past with CFD-ACE+ if its a help. Just looking at OpenFOAM now.



Broadly though:

1. I modelled a nozzle to provide a jet that then broke up
2. It is very mesh dependent
3. Start without surface tension but you will most likely need it depending on drop sizes
4. If in doubt drop the time step - counterintiuitvely low CFLs run faster that higher as the numerical work per time step is lower
5. Get your gaseous pressure boundaries right - depending on scale I tend to enforce a barometric pressure across the domain. With mass flows relatively small compared to the domain small pressure gradients in the gas can drive large circulations.

Hope that helps,
-pete

[Lapidus]

Hey Pete,

This is being done with StarCCM+. The particles are brought in with an surface injector, since what is modeled is an aerosol.

The idea is to run a similar simulation than this, but in 3D. The container where the flow has to pass through is nor rectangular, thus all slices wouldn´t look the same and the interactions in the perpendicular plane would be lost. If the computational power required for this endeavour is insurmountable, 2D would cut it.

The droplets (or at least the water that builds on the wall) must be modeled as VoF. All tutorials I have seen have the fluid(s) already in the space and only see how they behave. Is it possible to allow new fluid into the space, maybe through a function? Preferably not as a volume fraction of the new incoming fluid.

At the moment everything is laminar, without surface tension and so forth, to keep the model as simple as possible. Yet, what physiscs to choose is a tricky question. The problem seems to be solved as implicit unsteady with some Lagrangian Particles for the injector, Wall Impingement, Fluid Film on the walls and VOF for when that film grows thick enough.

In my experience, the Lagrangian particles can only be traced when solving steady. I am a bit stuck on how to have all the physics working with each other and what multiphase interactions to use. The simulation is possible, because it has been done in the video of the OP, as well as can be read in this article:
http://mdx2.plm.automation.siemens.c...hans/node/7661#

Any infos in this regard are welcome. Truly thankful!

[pbrady2013]

OK, so I'm super familiar with StarCCM but I think you are confusing to separate physics.


In the video link you posted it appears to be a 100% VOF simulation, which is an Eulerian grid system where the VOF volume fraction is advected with the flow and pressure fields. And yes, this will be fully time domain and, most likely, not trackable.


Lagrangian particles can be tracked but are time domain. Conceptually these are very different though - they are generally assumed to be discrete particles, significantly smaller than the mesh so they do not affect the flow field.


With respect to your question about inlets: for sure, simply define a small region of your inlet as where the liquid phase will come in. I suspect that is exactly what is happening in first image on your linked article.


My thoughts then are:

1. if you want to simulate the video you should only need VOF. No lagrangian discrete particles
2. beyond that, add the physics one step at a time until you get where you want to be.

Cheers,
-pete