CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   CFX (https://www.cfd-online.com/Forums/cfx/)
-   -   Enclosure in The Geometry (https://www.cfd-online.com/Forums/cfx/187276-enclosure-geometry.html)

AS_Aero May 3, 2017 03:48

Enclosure in The Geometry
 
Hi All

I am doing a multiphase simulation where Water from diffuser is sprayed on to the atmosphere. So here I was planning to use the option ''Enclosure'' and select it as fluid. And then I do the enclosure around the difusser volume. Does anyone have done like this or do you think it will work or not ? Or any suggestion how I can proceed ? The problem will be the interface between the water and air. Also comes the problem with the inlet and outlet Boundary. Inlet I can give the difuser inlet but outlet ?
Kindly help me on this.

ghorrocks May 3, 2017 08:44

Words like "spray", "diffuser" and "enclosure" mean very different things to different people. Indeed, different industries use these words to describe very different things. So can you please explain what you are trying to do a little more and include some images.

AS_Aero May 3, 2017 10:03

1 Attachment(s)
Hi Glenn

Thanks for your reply , Sorry for my misleading question...So to elaborate I want to do a multiphase simulation of a Difuser ejecting water out of it to the atmosphere. So in this basically I need to check the atomization of the water droplets, its velocity and size. To start with I thought I will create an enclosure around the geometry and keep it as air (in ansys designmodleer Tools/Enclosure ) but that seems not to be a good option. So now I created a new geometry at the exit surface of the difuser as you can see on the image. but I am not sure if this is the right way to proceed.

The other option I had was to do the single phase of the difuser and take the outlet data and feed to a rectangular channel. But then again its the same process I guess.
So I thought to try this present way. But from here I dont know how I can proceed. Multiple zone mesh ? and what about the interface etc etc.
Kindly give me some valuable advice.

What you see in the image : 1 is the fluid (water) and 2 is the Airatmosphere to which the water is sprayed.

Thanks

ghorrocks May 3, 2017 18:51

If you are looking to do direct modelling of the spray break up process (like this https://msu.edu/~iranneja/research%20projects.html) then you will need access to a big compute cluster and a good knowledge of CFD (multiphase, validation and verification). Is this what you are trying to do? You should not underestimate the difficulty of a model like this.

AS_Aero May 4, 2017 03:07

Hi Glenn

Well the project is almost like the one you have posted. But mine is more like a simple water spray where I dont have any chemical reaction or combustion etc. And I am interested in doing VOF, Eulerian and Lagrangian method to identify the droplet size and distribution of velocity etc. The problem is CFX dont have any tutorial that explains how to do this or similar one. So am bit confused. I could do just the difuser part and then take the outlet data from the difuser to the inlet of the rectangle box. This is one approach which came to my mind. but am not sure how to proceed.
And yeah computation power I have, i can run it on HPC cluster with several nodes.

ghorrocks May 4, 2017 07:07

A "simple" water spray with VOF/free surface model will still require a large compute cluster and careful validation of a multiphase model. The lagrangian droplet breakup and transport models are more tractable.

CFX has tutorials on all these options, but not specifically for spray breakup. It does not have a VOF spray break up tutorial as this is a seriously difficult simulation and definitely NOT suitable for a tutorial! It does however have free surface models for flow over a bump. So this is where to start for VOF. But if you are a CFD beginner then don't even consider VOF for spray breakup modelling, it really is a very difficult task.

The lagragian spray breakup model is more likely. There are lagrangian particle tracking tutorials. I cannot recall one for the spray breakup model however. ANSYS Support might have an example of this.

AS_Aero May 4, 2017 07:34

I thought VOF is more easy that Lagrangian or Eulerian !! In the case of VOF we dont have to give the particle size and so on. And rarely I see any paper for these. most of them are related to the IC engine, combustion stuffs.

ghorrocks May 4, 2017 17:48

VOF with surface tension is considerably more difficult. Surface tension is required for spray breakup. But the big issue is when you try to do accurate VOF with surface tension - that is the really difficult bit.

Getting the Eularian and Lagrangian models accurate is considerably easier.

AS_Aero May 5, 2017 02:22

Sorry !! I have never done it in CFX and I thought it should be similar to the settings what we do in LES everything will be there and I will have to play around to get the accurate results.
In my case its important to get the droplet size and velocity distribution and also the impact force at which the water hits the surface as its part of a cleaning nozzle.

ghorrocks May 5, 2017 02:28

As I said, if you intend to model the breakup process directly by VOF then this is a very difficult task. You will need to do a lot of work on the free surface settings as the defaults in CFX are not the best for that sort of work.

And I think you also find that Fluent is 10x or more faster at this sort of model than CFX. It has a much better free surface model for micro stuff like this compared to CFX. Again, you have to do many adjustments to get the simulation speed up, but I would expect an optimised Fluent model of spray breakup to be 10x or more faster than an optimised CFX model.

AS_Aero May 5, 2017 02:35

Ok then I will have a look on that too. But I also have a doubt about how we can see the cleaning process or dirt removal process on a plate using this spray ?

ghorrocks May 5, 2017 05:14

You can model just about anything you know the detailed physics for. It you know exactly how the dirt removal process works then you might be able to model it. If you don't know exactly how it works you have no hope (until you do a literature review and figure it out, at least).


All times are GMT -4. The time now is 07:49.