[jins9158]

hello

I am trying to simulate the watertight test with cfd in the figure below.

The figure below is a watertight test of the ventilation cap.

Water is sprayed from above, and water is introduced into the ventilation cap by a fan.

The water flowing into the interior fills up through the drain tank.


The analysis conditions were set as follows.

1. Water/air phase (water is a dispersed fluid and average diameter is 0.1 cm)
2. Apply buoyancy
3. Apply inhomgeouts model
4. Application of surface tension
5. Applying drag (option is Grace model)
6. Non-drag force application
1) lift force (Ignoring the viscous effect around the particle, the coefficient of lift is applied as 0.5)
2) Apply Virtual Mass Force
(Ignore the viscous effect around the particles and set the coefficient to 0.5)
3) Apply Turbulent Dispersion Force
(option is Favre Averaged Drag Force)

* Ignore wall lubrication force
(Because there is no reaction by the wall surface)

I don't have much experience with multiphase flow, so I'm not sure if the values ​​I set are appropriate.

We would appreciate it if you could review the settings to make sure they are appropriate.

Also, I set the setup as an inhomgeous model (considering the speed separately for the phases), can I do it with a homgenous model when water is sprayed?

Thank you for your comments.

[Gert-Jan]

Really, you can get any answer you want using multiphase analysis like this. But I'm quite sure you'll never get it running this way. What is the purpose of your CFD analysis?

Nevertheless:
- Don't click on all models available and then cross fingers. This is not going to work. Start as simple as it can be. Then increase complexity bit by bit, while reading the documentation to understand what is important and what not.
- Start with Schiller-Naumann for drag or a constant value (as simple as possible)
- Stay away from surface tenseion. If you read the documentation, then you should understand that you don't need this here.
- Don't use virtual mass force. If you read the documentation, then you should understand that you don't need this here.
- Don't start with turbulent dispersion. Start as simple as possible. You can always add it later.
- Start homogeneous, restart with inhomogeneous.

[ghorrocks]

As Gert-Jan says, this is quite a tricky model and will take some preparation to get working. A critical thing, in addition to Gert-Jan's comments is you need to think about the detailed physics which is going on here - droplet wall impact, droplet relative slip being the most important I see here - and read the CFX documentation in detail to understand how these physical models are handled in CFX. There are many physical models in CFX, and you need to choose the appropriate ones for your application. For example the question about inhomogenous/homogenous is concerning, as this suggests to me you do not understand this at all as the answer should be obvious. You need to do more reading and understanding of your application - when you know the answer to this question yourself then you can choose the appropriate model with confidence.

And once you have chosen the appropriate models you should do a simple benchmark validation case to convince yourself you can get accurate results on a simple model. Applying complex physical models on complex geometries NEVER gives accurate results unless you can accurately model a simple case first.