[linnemann]

Hi.

Please share with dropbox/gDrive etc.

I'm sure there is a simple solution.

[arashfluid]

Thanks Friends,
Problem is solved.Now rotor moves properly.
My main problem is too much run time.This is why I got into the AMI.
If using the AMI ,can be used both of fixed topology and changing topology mesh manipulation models in dynamicMeshDict?
What is your opinion about reducing the run time?

[Mashiro5]

Dear All,


I'm working on the same topic (mainly a propeller rotating and interacting with a rudder) and I would like to share my knowledge and ask for some problems I'm having.

I succesfully run the propeller tutorial and I set up my case with a slightly complicated propeller geometry. The snap stage worked well up to snappyng. At that stage the AMI weights were abt. 0.8 1.2 1.0001 and if I run this case, eithout the addition of prism layers, everything works fine.

When I add prismlayers in snappyhexmesh (I found that the propeller tutorial case neglects prism layer addition) the resulting mesh, due to the layers addition itself, is slightly different also close to the interface and, there, shows some "deformations" (in particular for only one of the two interfaces patch) and "holes" (if seen using cutting planes). The resulting AMI min weight falls very close to 0 and the simulations stop after a while, when in correspondance of a particular time step, min AMI wheigh becomes 0.

Do you have any sugestions on how to prevent the modification of the mesh at the interfaces during the layer addition?

Many Thanks,

Stefano

[wyldckat]

Greetings Stefano,

Originally Posted by Mashiro5

Do you have any sugestions on how to prevent the modification of the mesh at the interfaces during the layer addition?

Without any more information (images and/or example case), the best I can do is suggest that you read+study the tutorials listed here: http://openfoamwiki.net/index.php/Sn...als_and_Guides

In addition, if you don't have experience using snappyHexMesh for adding layers, then you should use a small test case and test out the settings explained in the tutorials.

Best regards,
Bruno

[Mashiro5]

Dear Bruno, as usual you are right.. an appropriate image can explain better than hundreds of words.

I have a reasonable experience with snappyHexMesh (I already read the tutorial you suggested) and I usually use snappy to mesh propellers and ship hulls with satisfactory results, of course adding also boundary layers.

Now I'm trying to model the propeller operating behind the hull. The snap works well, as you can see in the image. The innerCylinder and the innerCylinder_slave have almost the same surface mesh, with well defined edges, even without an explicit refinement. If I start my simulation without the layer addition everything works fine: the AMI weights are well above 0 (abt. 0.8 - 1.2 - 1.0000x). During the simulation especially the min value oscillates but never comes below 0.5.

When instead I include, in the mesh generation, also the prism layer, something goes wrong with the mesh. As from the figure the layers are well added to the hull (and also to the propeller) surface even if the relativeSize option produces a variably prims layer thickness (but at thip point this is not important). The layer addition, unfortunately, changes the mesh arrangement and in particular change the mesh near the edges of the interface, only from the innerCylinder_slave point of view. This tooth-shape is responsible of very low values of the min AMI weight, abt 0.0006. The simulation starts but, in correspondence of certain angular positions the minimum weight becomes 0 and openFoam stops.

I tried different mesh arrangements, decreasing the size of the mesh in correspondance of the interface, clustering cells on its the edges (that are properly marked used eMesh files), changing a bit the parameters of the layer addition process. Everything is ok up to the snap stage. After the layer addition the AMI weight tremendously falls...

Now I'm wondering if it is more appropriate to create separately the two meshes and finally merge them, but after a week of "try and error" I'm looking for some feedback or, at least, consolations...

[wyldckat]

Hi Stefano,

For better or for worse, the image attachment system here on the forum will automatically rescale the images to a size that will (hopefully) be less than 100kB. This means that the image you provided doesn't have enough resolution to see all of the details

Although from what I can see, the mesh without layers isn't exactly perfect, or perhaps it was the JPG image compressor that added an artifact to the image...
Anyway, I don't have enough experience with adding layers in snappyHexMesh, so the best I can do is to suggest that you share an example case that reproduces this same problem. In addition, knowing which OpenFOAM version you're using would help.
A good meshing base case would be the tutorial "compressible/rhoPimpleDyMFoam/annularThermalMixer", which is available at least in OpenFOAM 2.3.

Best regards,
Bruno

[crixman]

Hi all,
did anyone tried to run an AMI simulation using cfMesh?
How would you go into that?
I'm not sure if I have to make two geometries for inner and stator, mesh them separately and combine them, or if I just need a single stl file with all patches.
How would I make the rotating cellZone then?

[Jetfire]

Dear all

Can someone please explain the difference between baffles and patches, for simulations using AMI do we need to create patches for the AMI interfaces or baffles i am confused.

[wyldckat]

Greetings to all!

@crixman:

Originally Posted by crixman

did anyone tried to run an AMI simulation using cfMesh?

Sorry for the very late reply. The answer is simple:

1. Create the meshes separately.
2. Merge meshes, no stitching required. Have a look at this post of mine: Problem using AMI post #184

Originally Posted by Jetfire

Can someone please explain the difference between baffles and patches, for simulations using AMI do we need to create patches for the AMI interfaces or baffles i am confused.

Again, have a look at this post of mine: Problem using AMI post #184

As for baffles vs patches... mmm, it's best to use the description given here: http://www.openfoam.org/version2.2.0/meshing.php

The createBaffles utility creates zero-thickness baffles by converting internal mesh faces into (pairs of co-located) boundary faces.

In other words, baffles can only be composed of faces that belong to 2 or more cells. Patches on the other hand are usually for faces that belong to only one cell.

Best regards,
Bruno

[Jetfire]

@wyldckat
Thanks for your reply.

Can you please help me with my simulation
My task is to simulate compressor stage of a turbocharger.
My first question:

Can we simulate these kind of turbomachinery simulations using OpenFOAM v2.3.0 with rhoPimpleDyMFoam using AMI Approach?

I have 3 separate meshes generated using ANSYS ICEM CFD
1.Inlet
2.Volute&Outlet
3.Compressor Domain

Should i import these mesh files separately to openfoam and then use mergeMeshes? or merge 1. and 2. making it a stationary mesh using ansys and 3. will be rotating mesh and import these two mesh files to openfoam? please suggest me with a strategy

And for performing rotation using AMI for this simulation how do i define the AMI interfaces. should i enclose my compressor domain with a cylinder similar to the propeller case? I'm stuck with this for months , please help me

Thank you

[openfoam_user]

Hi,

the following computation was done with ICEMCFD Hexa meshes (OpenFOAM 2.2.x).

https://www.youtube.com/watch?v=HJ2s...cn5Tt-X_CqxshQ

Take care at the interface between the 2 meshes. Same cell length is better.
Sometimes computations can crash with too fine grids. So it's better to begin with coarse grids as in the propeller tutorial.

Best regards,
Stéphane.

[vince_44]

Dear All

I test with success the propeller tutorial. Now I test with my own geometry, a rotor from a water jet. The case work well but as you can see in the pictures attached, the flow is not really impacted by the rotor rotation.

Any idea?

Best regards

[Alhasan]

Hey,
Vincent can you please be much more clear with your question. what do you mean by 'not really impacted' ? the propellor is not moving like how you expect it to move..? maybe the velocity is not fast enough..?

Regards,
Hasan K.J

[vince_44]

Hey Hasan

Sorry if I was not clear. In fact, the rotor move as I expected. But the flow seem no rotate. I attached two picture. The first is the OF calcul and the second it's with an another CFD code. The conditions are Uinlet=3.5m/s et rotation 2500rpm.

Regards
Vincent

[wyldckat]

Greetings to all!

@Vincent:

Originally Posted by vince_44

The conditions are Uinlet=3.5m/s et rotation 2500rpm.

Some years ago I learned something very important with OpenFOAM: Never assume things work the way we "think" they should.
In other words: please provide proof as to how you defined those values, namely provide the files where you configured those values.
Also, run checkMesh and confirm what dimensions your mesh has got (please provide the output from checkMesh, for us to see as well). And please indicate which OpenFOAM version you're using.

In addition, please follow the instructions given here: http://www.cfd-online.com/Forums/ope...-get-help.html

Best regards,
Bruno

[vince_44]

P { margin-bottom: 0.21cm; } Dear Bruno and Hasan


I use pimpleDyMFoam (OpenFOAM 2.3) to simulate the flow around a rotor. The final goal is to calculate the water jet thrust. For the moment, I test pimpleDyMFoam only with the rotor. I adapt the propeller tutorial to my problem.


I have some questions :


-When I create the patch for AMI, I feel I make a mistake (I join the log). Indeed, it read 0 faces from faceSet inletFaces and 0 faces from faceSet outletFaces


-After run the calculation, when I view the flow around the rotor, it's like if the flow don't swirl


-Finally, in my last test, I expected Fpx=-450N and I have Fpx=-1700N


I hope, I'm more clear with my problem now. In the next post, I join another files



Best regards

[vince_44]

Here the blockMest, SnappyHexMesh, topoSetDict and createPatchDict.

Best regards
Vince

[Mashiro5]

If you are not going to include a non-homogeneous inflow in teh stationary block or you're not interested in the initial transient (only steady state condition under investigation) a more computational efficient simulation using simpleFoam and appropriate fvOptions will be a better choice... :-)

[wyldckat]

Greetings to all!

@Vincent: Stefano might be right here.

But the problem I'm worried about the most is that you did not provide the most important file of all, which was the one I was sort-of requesting, namely the file "constant/dynamicMeshDict". This is where you are meant to properly define the value or rotation.

Best regards,
Bruno

[vince_44]

Hey Bruno and Stefano

Thanks for your answer.

Indeed, I forgot to join the dynamicMeshDict. Sorry. Now I join this file.

Best regards

Vincent