code_saturne User's Forum

Hello,

I have been trying to adapt the VIV ALE tutorial for use on an airfoil cross section. I am working in version 3.1.3. I have attached my xml file, my mesh and an abridged listings file. I have tried to simply replace the cylinder in the tutorial with an airfoil cross section tied to a spring system which only allows movement in the z direction (up and down).

Tutorial is here: http://code-saturne.org/cms/sites/defau ... action.pdf

I have issues with my grid being too refined for a k-e run and too coarse for a laminar run and I have tried using SOLU as suggested in another forum post (http://code-saturne.org/forum/viewtopic.php?f=2&t=1466) but my simulation keeps failing at the same point with this error:

Incoming flow detained for 105 outlet faces on 400
Warning:
--------
_iterative_scalar_gradient; variable: 1; sweeps: 100
normed residual: 9.9885e+55; norm: 1.4648e+01
Warning:
--------
_iterative_scalar_gradient; variable: 1; sweeps: 100
normed residual: 9.9885e+55; norm: 1.4648e+01
Warning:
--------
_iterative_scalar_gradient; variable: 1; sweeps: 100
normed residual: 9.9885e+55; norm: 1.4648e+01


Conjugate gradient [Pressure:0000]: divergence after 1 iterations:
initial residual: 2.1393e+02; current residual: 1.2431e+07



cs_multigrid.c:1230: Fatal error.

algebraic multigrid [Pressure]: divergence after 4 cycles:
initial residual: -1.0000e+00; current residual: 1.2431e+07


Call stack:
1: 0x7fd476403092 <bft_error+0x92> (libsaturne.so.0)
2: 0x7fd476083846 <+0x319846> (libsaturne.so.0)
3: 0x7fd476081da5 <cs_multigrid_solve+0x715> (libsaturne.so.0)
4: 0x7fd47608166f <resmgr_+0x16f> (libsaturne.so.0)
5: 0x7fd4760c13bf <invers_+0x10f> (libsaturne.so.0)
6: 0x7fd475f7d719 <resopv_+0x8289> (libsaturne.so.0)
7: 0x7fd475f0b7a1 <navstv_+0x2f61> (libsaturne.so.0)
8: 0x7fd475f9ff5c <tridim_+0x2a2c> (libsaturne.so.0)
9: 0x7fd475de6d04 <caltri_+0x3fe4> (libsaturne.so.0)
10: 0x7fd475dba142 <cs_run+0x8e2> (libsaturne.so.0)
11: 0x7fd475db983c <main+0x14c> (libsaturne.so.0)
12: 0x7fd471238d1d <__libc_start_main+0xfd> (libc.so.6)
13: 0x4007e9 <> (cs_solver)
End of stack


I am guessing it is something big I am doing wrong. IF anyone could point me in the right direction I would be very greatful!

Thanks in advance!

S

Hello,

Version 3.1 is not maintained anymore ,so I suggest you upgrade to 3.2 or 3.2 (or at least the latest 3.1, 3.1.4 in case you do not want to change your data setup).

In the listing, we can see the "right hand side" norm for the mesh velocity increases progressively before the crash. I do not know if it is related.

More important, at the iteration which crashes, the gradient reconstruction for pressure (the first variable solved for) fails (listing line 6096, then 6134, ...). This probably means your mesh has reached a "bad" shape (such as degenerate or tangled cells).
Postprocessing the mesh at the iteration just before failure (5009) could be interesting.

Your time step seems OK (by the CFL), but due to ALE, you might still try reducing it. Possibly better, you could increase the mesh viscosity, to better spread the deformation.

I'll let those more familiar than me with ALE complete these recommendations it they want.

Regards,

Yvan