[HiddenSquid]

Hello everyone,

Has anyone had any success creating a rotating AMI zone that also undergoes translation?

I am trying to simulate a two dimensional case that involves a patch both translating and rotating. I have previously been using the dynamic mesh motion solvers to warp the mesh to my moving geometry but have had issues with poor mesh quality at high deformations.

For rotational only problems I have had better results with the use of AMI patches as this allows the mesh close to the rotating body to be preserved. I am now trying to set up a case where the zone close to the moving patch (i.e. the "AMI zone") both rotates and translates while the cells outside of the AMI zone only have to deal with the translations thanks to the nature of the AMI separating these two zones.

I have read a number of papers and other sources from Prof Jasak explaining how a capsizing ship was modelled in this way (http://www.sciencedirect.com/science...92678216303879) but I am yet to determine how to go about doing this for my case. I have tried to specify the motion within the 0/pointDisplacement file and define the points within the AMI zone as frozen within the constant/dynamicMeshDict but this means that these cells do not move at all. If the points are not specified as frozen then the patch with motion prescribed at the centre of the AMI zone moves while the AMI boundaries do not with the mesh within the zone deforming - i.e. how it would be without the use of AMI at all.

Something similar that I have come across is this work by Tobias Holzmann https://www.youtube.com/watch?v=XC-_F42CZjg.

I am not sure if a better approach would be to try to apply non uniform mesh diffusivity (i.e. set it to zero in the rotating zone) and then apply the AMI rotation to the cell zone separetely but I am not sure if this would require two different types of dynamic mesh solver. As far as I can tell pointDisplacement can only specify rotation to a patch while specifying the motion within dynamicMeshDict (type solidBodyMotionFvMesh) can apply motion to cellZones. If mesh diffusion is used then the diffusion is specified within the dynamicMeshDict (type dynamicMotionSolverFvMesh) hence the need to move the motion definition to the pointDisplacement file but then losing the ability to apply to cellzones rather than patches...

The motion that I am looking to apply is predetermined and read in from an external file.

I hope that makes sense! If anyone has heard of this being done before or has any ideas of how to achieve something like this then it would be interesting to discuss it with you.

Thanks,
Nick

[louisgag]

Hi Nick,
were you able to get such a case working?
-Louis