CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLUENT (http://www.cfd-online.com/Forums/fluent/)
-   -   Dynamic mesh + grid adapt = Crash! (Files included (http://www.cfd-online.com/Forums/fluent/45450-dynamic-mesh-grid-adapt-crash-files-included.html)

BillH July 20, 2007 07:43

Dynamic mesh + grid adapt = Crash! (Files included
 
Hi all,

I am having no luck trying to get dynamic mesh and grid adaptation working in even a trivial example.

I have set up a simple four sided 2D block with an inlet, outlet and two walls. Air enters the inlet at 1 m/s.
http://img181.imageshack.us/img181/4509/meshdv8.jpg

I have included a small square inclusion in the domain called "rod". Rod is driven to rotationally oscillate with a trivial DEFINE_CG_MOTION macro.

The idea of this test case is to test adaptation in conjunction with a dynamic mesh.

The case (as included in the zip file below) runs fine without adaptation turned on. I have tried to include adaptation in two ways:

a) Run for say 50 timesteps. Do a manual gradient adaptation on say the pressure gradient and then restart the run for another 50 timesteps. This immediately crashes the run with the following generic error message:

Error:
FLUENT received fatal signal (ACCESS_VIOLATION)
1. Note exact events leading to error.
2. Save case/data under new name.
3. Exit program and restart to continue.
4. Report error to your distributor.
Error Object: ()

(b) Turn on dynamic adaptation before stating the solve on say the pressure gradient. This runs fine until the first adaptation step, seems to complete the fist post-adaptation timestep and then crashes again with the above error.
http://img76.imageshack.us/img76/696...ergencebc8.jpg

I have tried a myriad of gradient based adaptation combinations on various variables but they all fail in exactly the same way.

This leads me to suspect that dynamic mesh and grid adaptation cannot be used concurrently but other fluent users (e.g. red lemon) have hinted that this should in fact be possible.

I would be grateful if anyone can give my any input on this problem. I have included a tiny zip file with the case files if anyone would like to re-run my problem. As I mentioned before, the case runs fine as is because adaptation is off. However, if adaptation is turned on in any form, dynamic or once off, the solve crashes.

Case files zip (350 kB): http://rapidshare.com/files/43981168...adapt.zip.html

Thanks!

red lemon July 20, 2007 16:31

Re: Dynamic mesh + grid adapt = Crash! (Files incl
 
Hi Adaption requires that hanging node grids are created. I dont think on reflection it is possible to remesh a hanging node grid. In same way you cant adapt a polyhedral or hexcore mesh. I'd need to check limitations on that in UG. cheers

BillH July 22, 2007 08:57

Re: Dynamic mesh + grid adapt = Crash! (Files incl
 
I thought about what you said about hanging node meshes and dynamic meshing being incompatible and it made sense.

I therefore switched from hanging node mesh adaptation to conformal mesh adaptation. Now it works perfectly!

So it seems that dynamic mesh and mesh adaptation can be used simultaneously but only if you use conformal adaptation, which is fine for my aplication.

BillH July 22, 2007 09:17

Re: Dynamic mesh + grid adapt = Crash! (Files incl
 
Strange. Repeated manual adaptation works fine, but automatic adaptation using the exact same adaptation settings hangs like in my first message.

Strange ....

BillH July 24, 2007 15:31

Re: Dynamic mesh + grid adapt = Crash! (Files incl
 
OK. I have got automatic mesh adaptation + dynamic remeshing working together. For the automatic mesh adaptation settings ONLY mark deforming tet/tri fluid zones as being adaptable under "controls".


All times are GMT -4. The time now is 04:50.