[hirota]

Hi Foamers,

Currently, I am trying to reproduce the rotation of a propeller using pimpleDyMFoam.
Specifically, I want to simulate a rotating region (propeller rotation region) inside a stationary region.

First, I assume that I will be using pointwise to generate the grid.
Next, I am considering using the AMI function in consideration of the case where propeller rotation is derived in the future.

As shown in the Fig1, the computational domain is a hexahedral region with a cylinder rotation region inside.

This analysis was output from pointwise as polyMesh to openFoam specification, and after that I had a hard time setting the AMI boundary, and tried to find a solution, but still could not reach a solution.

The procedure is
1, Create a rotated region and a stationary region by pointwise, and output them by polyMesh.
　The boundry files at this time are shown in Figure 2 and Figure 3.
2, Perform renumberMesh.
3, Create a cellzone as a rotation area by topoSet (topoSetDict is shown in Figure 4)
4, MergeMesh to combine two meshes.
5, Use changeDictionary to change the type of AMI1 and AMI2 of the boundary from patch to cyclicAMI (changeDictionaryDict is attached in Figure 5)
6, Describe the name of the cellZone that will be the rotation region in the dynamicMeshDict, and execute the simpleDyMFoam.

If you follow these steps, you will see that the calculation turns, but the velocity and pressure distributions do not change even though you are just rotating an empty mesh inside a rectangular body.
However, the velocity distribution fluctuates and the flow is disturbed. (Figure 6; the circle near the center is the cross section of the cylinder in the rotating region)
After investigating the cause, we found that when step 5 was executed in changeDictionary, an impossible velocity distribution was given to the cellZone, the rotating region, even though it was the initial condition (Fig. 6) Since it is the boundary of AMI, it should be zero.
The pressure and other distributions are still zero, but for some reason only the velocity distribution is like this, and this seems to be the cause of the flow disturbance during the calculation.
I am at a loss and am having a very difficult time. I would be grateful if someone could help me.

[hirota]

Here are some additional images.
Fig.6 &Fig.7

[Swift]

Hi Hirota,


I don't work with pointwise meshes nor have I used changeDictionaryDict.


However I can make some recommendations:


Have you tried the tutorials in pimpleFoam. There are two that handle mesh rotation using AMI. They are the propeller and rotatingFanInRoom. The use snappyHexMesh, but they might shed some insight and help you get started. It will also help the the BS definition.

When you define your rotating and stationary mesh, the 'AMI' patch must be named differently eg: AMI_stationary and AMI_rotating. When you merge the meshes using mergeMesh, you must have two patches in your boundary dict file. Then you can convert them into cyclic AMI. I normally use createPatch as per the tutorials for this.
You can also test if your mesh is created properly with moveDynamicMesh. This will move the mesh and run checkMesh. If your mesh quality is changing then you haven't set up your AMI properly.


You mentioned simplyeDyMFoam. What version of OpenFoam are you using?


Is there a specific reason for wanting to use a pointwise mesh? If not, I would adapt the tutorials to use your geometry and work from there.


Regards,
Thomas

[hirota]

Hi Thomas,
I'm glad you got a reply.

I had already read the tutorial and thought I understood the process.
I'm thinking that the only rough difference between snappyHexMesh and snappyHexMesh is the creation of the rotation boundary, and that the rest of the process is the same, but I'm having trouble.

The OpenFOAM version used is OpenFOAM-4.x.

『When you define your rotating and stationary mesh, the 'AMI' patch must be named differently eg: AMI_stationary and AMI_rotating. 』

Is this correct in the sense that my approach of AMI1 on the rotating boundary side and AMI2 on the fixed boundary side is correct and they are not grouped by "AMI"?

And the procedure for using createPatch is
topoSet, createPatch, createBaffles, mergeOrSplitBaffles
How do you do this?
I couldn't get this to work either...

The reason for using pointwise is that we plan to analyze hulls with propellers in the future, and we believe that a finer grid layout will be necessary.

I look forward to your continued advice.

Regards,
Hirota

[Swift]

Hi Hirota,


Are you able to share your case. I can then take a look.

Regards,
Thomas