Learning Forum

[Erling Gjeset]

Hello all,

I'm struggling with setting up a multiphase simulation of an impinging jet and I would like some help with troubleshooting. 

The case is this: I have a certain numer of segments arranged in a circular pattern, and a system supplying fluid from jets located on the outer radius. One such segment is shown in the screenshot below.

In this example, I have considered the segment on the immediate right of this one to be its mirror image, therefore a symmetry axis on the right hand side. The left side is "empty" (outlet BC) in this case, as I am currently trying to get the first piece to work before setting up the next one.

The model setup is (briefly summarized):

• steady state
• gravity (y=-9.81)
• volume of fluid with implicit vfp and body forces
• k-epsilon turbulent viscous model
• phases are air and water with a surface tension coefficient of 0.073 and wall adhesion turned on
• water enters the inlet at 2 m/s
• SIMPLE pressure velocity coupling

More details can be provided upon request as I'm at this moment uncertain which parameters are the most crucial to consider. I've kept most of the solver settings as default.

When I run this setup the right hand side of the jet "fills up" with water, as shown in the following screenshot. I would suspect some flow recirculation as a result of the jet on the other side of the symmetry axis, as has been shown experimentally, but it seems strange to me that the fountain created in my simulation is this large..

 

The velocity fields look as follows:

 

I've also tried to setup another model similar to this one, only expanded to include the next segment on the left of the first one, but this model (with more or less the same setup) is not solvable.

What can I do to improve the model?

[Rob]

If you compare the inlet to outlet area why wouldn't it flood? The impinging jet hits the far surface and splits, where does the trapped air go? With the left side being open the jet may bias in that direction, but on the right liquid will flow down due to gravity. Check the back flow conditions. 

[Erling Gjeset]

I see! That is a good point!

I've tried to increase the outlet width to be 40% larger than the inlet to see if this would affect the phase contour. However, now my solution diverges and I cannot get a result at all.

The outlet backflow condition is set to have 0 water volume fraction.

[Rob]

If the outlet is all gas why won't you have bubble rising from the surface? 

Sketch out the domain on paper, and then work through what is going on in reality. Add any information you know to the sketch (flow rates etc). 

[Rob]

To add, you also want to avoid high aspect ratio cells with VOF so I'd focus on a refined mesh throughout the domain rather than use any inflation options. 

[Erling Gjeset]

Thank you for your feedback. 

In truth, the fact that this "flooding" occurs is not a huge concern. The main issue with this model is that it is extremely sensitive to any small adjustments I make to it trying to implement your feedback. That was why I was questioning the recirculation zone in the first place.

I've been fiddling around with this model for a while now, testing to see if the mesh size/shape would affect it, whether or not I should use symmetry axes or expand the model to contain neighboring segments, as well as testing different under-relaxation factors. It seems like I rarely am able to replicate my previous results, even when I try to change the settings back to my previous ones. I also tried with your suggestion for mesh settings, but this does not seem to give a good convergence.

Do you have any idea or other suggestion on how to refine my model?

[Rob]

You may find you need a transient solution, it will depend on how stable the jet and recirculation regions are.