[kandelabr]

Hello!

How do I set up an ACMI patches for a pump? I have an impeller that's always touching the volute boundary, but volute is not always touching the impeller.

The incompressible/pimpleDyMFoam/oscillatingInlet2D tutorial case is very complicated and I can't tell what's what in there. But I can't just 'copy' it because in my case impeller has no 'nonOverlappingPatch'. Or does it? Do I need to run createBaffles?

Thanks!

[kandelabr]

Maybe I should ask this question first: is it even necessary to use ACMI or should AMI suffice? The thing is that I'm starting to believe the way I split the mesh for AMI is ruining my results. Here's my impeller that includes a 'rim' of fluid that actually belongs to volute:

https://www.cfd-online.com/Forums/at...h-impeller.jpg

Am I right?

[Dipsomaniac]

You should use an AMI boundary for a system like yours. Have a look a this tutorial which sets up the case for a rotor using AMI boundaries.

Moving Rotor (AMI) (Holzmann CFD)

[kandelabr]

according to the tutorial, my case is set up exactly the same.
my theory was that the ring that is 'attached' to impeller (but is not rotating in reality, see the attached image) messes up the results.

if that works for mr. Holzmann, i guess it should work for me too , but it should work for dynamic mesh only, not for MRF. in MRF, centrifugal and coriolis force is added in the ring as well. and that's wrong. am I right?

thanks!

[Dipsomaniac]

I would assume those forces would have to be added in the MRF case since it is not actually a moving/dynamic mesh, but rather a moving reference frame, though I have only been working with AMI patches for my work. Good luck!

[Tobi]

Hi all,

first thanks for mentioning my tutorial. I appreciate this and I am happy that it will help the community. I just want to mention a few things related to my case and other cases using ACMI and MRF.

• The case provided above is build in a way that the whole rotor and also the static stuff is within the cell zone that is moving. In a numerical point of view it is not the best choice because in the outer regions which are normally static, we induce mesh fluxes which will influence the numerical results. Please do not ask me how much. I never made an error analyze for that case
• The same case could be done with ACMI and AMI which might be a much better set-up. The connection between the centerline pipe (static) and moving part could be done using AMI while the rotating propeller could be modelled using an ACMI (the sliding parts; like this here: https://www.youtube.com/watch?list=P...&v=ZsdoAQ9hQUM)
• Using MRF can be done - of course and even without any arbitrary interface. You have to provide the mesh zone for the MRFProperties thats all. Please keep in mind that this will be correct if the speed is high. For low speed rotations I would expect complete different results with AMI / ACMI because of the impact of the velocity field based on the moving impeller. But I am not so familiar with MRF and that topic.
• MRF can be used with AMI too but it is not needed. AMI is only the interface between two different meshes. It maps and iterpolates the values from one side to the other one. If both are static, does not matter, but if it is so, there is actually no need for AMI (slows down the simulation).
• The centrifugal forces should be implemented in the MRF while in dynamic meshes is is natural based on the natural rotation and the mesh fluxes.

To be honest, I am not sure if everything is 100% correct here. Please correct me if I am wrong or missed something. I am actually not an expert in that field of MRF nor which forces occur in addition for moving meshes etc.

[kandelabr]

so my guess is the closest-to-actual model would be AMI on inlet and ACMI on the outlet.

i can't set up ACMI because in all tutorials there are four combinations of patches:

• moving part: matching
• moving part: blocking
• static part: matching
• static part: blocking

but in my case, moving part (impeller) is never blocked so i have no idea where to start. i hate it when tutorials use some crazy shit voodoo to set up something very simple and never explain anything.

[Tobi]

Hi,

you should provide both blockage and coupled patches. Both are just some duplications. Actually, you should have both of them. If for any reason you will never have any blocking on one patch, who cares? Imaging the oscillating tutorial and reduce the oscillation. Then you would probably only have a blockage on one side while the other one is always coupled. Could you provide a picture/draft of your geometric problem?

By the way. What you explain fits 100% to my other case
Simple ACMI (Holzmann CFD)

[kandelabr]

if that is so, let me check your tutorial first. i'll bother you later if i can't get it working

thanks!