[badboyz31]

Hello all CFX subforum members,

I'd like to simulate transient effect of moving a car's sunroof back and forth while the car is in motion. The model has been simplified as a cavity which has a lid cover, and the velocity of the lid is constant with time.
I've tried to look into the ball valve example in the CFX-Pre documentation, but I think that my case is different as the lid 's mesh is not contained within the domain, but it's connected to the external boundary.

I wonder how to setup the mesh and CFX-Pre in this case. Thanks in advance.

[ghorrocks]

This sounds straight forward to set up. This simple drawing shows one setup which should work.

Example.png

[badboyz31]

Originally Posted by ghorrocks

This sounds straight forward to set up. This simple drawing shows one setup which should work.

Attachment 44175

Yep, this is exactly what I was looking for. The only problem is, I am still looking for a basic tutorial about how to setup a general mesh deformation case in CFX.

I'm kind of lost looking for proper tutorial.

[ghorrocks]

If the tutorials you can find don't help much:
1) Download more tutorials from the ANSYS community webpage (www.ansys.com)
2) Talk to ANSYS support and get some tutorials off them

If that does not work give it a go and post what you get here. Post an image of what you are doing and your CCL and output file (showing any error messages). We will do what we can to help.

[badboyz31]

Originally Posted by ghorrocks

If the tutorials you can find don't help much:
1) Download more tutorials from the ANSYS community webpage (www.ansys.com)
2) Talk to ANSYS support and get some tutorials off them

If that does not work give it a go and post what you get here. Post an image of what you are doing and your CCL and output file (showing any error messages). We will do what we can to help.

Okay, thank you. I will try and see if I can get something out of them.

But uh, a few questions first:
1. If I use mesh deformation option, then when the lid moves across the moving domain, would the length-wise spacing of the moving domain become stretched/compressed or would the number of grid will be added/reduced to match surrounding vertices?

2. Does CFX needs to re-mesh the domain for every timestep? (I used structured hexa-meshing in ICEM, which was then converted to unstructured .cfx5 mesh)

[ghorrocks]

You can handle quite a large range of motion with moving mesh only, just stretching the existing mesh. Also note you will need transient rotor-stator interfaces on the interfaces with the stationary domains so the interface can be recalculated.

No need for remeshing. This is a single mesh which is deformed. (Except if you want to run the lid to full closure, then you will have to remesh to handle this).

A final comment:
Are you sure the opening of the lid is important? If the lid opens on a slower timescale than the flow, then this can be modelled as a series of steady state simulations with meshes at various stages of opening. This is MUCH simpler to do, so if it is appropriate then I would recommend you do it this way.

[badboyz31]

Originally Posted by ghorrocks

You can handle quite a large range of motion with moving mesh only, just stretching the existing mesh. Also note you will need transient rotor-stator interfaces on the interfaces with the stationary domains so the interface can be recalculated.

No need for remeshing. This is a single mesh which is deformed. (Except if you want to run the lid to full closure, then you will have to remesh to handle this).

A final comment:
Are you sure the opening of the lid is important? If the lid opens on a slower timescale than the flow, then this can be modelled as a series of steady state simulations with meshes at various stages of opening. This is MUCH simpler to do, so if it is appropriate then I would recommend you do it this way.

Your final comment is surely interesting for me, thanks to that. The lid traveling velocity is going to be 0.1 m/s vs supersonic 400 m/s of freestream flow.

However, I'm unsure of getting steadystate result as cavity flow is naturally unsteady. I'm afraid that the result would be inaccurate.

[ghorrocks]

Yes, the flow is likely to be inherently unsteady. If the free stream velocity is supersonic then you really need to separate the time scales - which means do runs at fixed opening positions. So these are likely to be transient runs, but with a fixed mesh and that makes things easier.