Learning Forum

[marc.goldbach]

Hi,

I am running a VOF simulation where the free surface is refined using automatic mesh refinement, but the simulation is suddenly diverging after about 1000 iterations.  When inspecting the results after divergence, I see no obious signs of high velocities, residuals, etc.  I have noticed that the original mesh had a minimum orthogonal quality of > 0.1, but when checking the mesh after divergence, the minimum orthogonal quality is ~0.03, eventhough I specify a minimum quality of 0.1 in the automatic mesh settings

Original mesh quality check using Fluent meshing:

----------------------------------------

Cell Quality [Measure : Orthogonal]

Minimum = 0.10208334, Maximum = 1, Average = 0.95636215.

 

Marking Cells in Range (0.10208334 0.15208334)...

Info: Total marked entities in the range is 252, is limited to 15 by max number of allowable cell clusters (10).

--------------- ----------

 

Mesh quality after automatic mesh adaption and divergence:

Minimum Orthogonal Quality =  3.18312e-02 cell 301147 on zone 19062 (ID: 14520292 on partition: 39) at location ( 5.99704e+00,  2.32079e+00,  9.54323e-01)

 

Maximum Aspect Ratio =  1.05687e+02 cell 31803 on zone 19062 (ID: 21173019 on partition: 58) at location (-5.25139e+00,  5.40401e-03, -1.74540e+00)

 

I checked the location of the poor quality cell, and it seems to be on the water surface in a very turbulent part of the domain.

Any suggestions on how to correct?

I have also tried with the "prismatic adaption" option selected, since my mesh does include solid boundaries with prism layers, but then I get an error after a while when the mesh is updating:

eversed flow on 9748 faces of pressure-inlet 169.
   419  3.2815e-02  1.0527e-05  1.1973e-05  1.0660e-05  5.2760e-03  2.2468e-03  1.3402e-03 16:27:52 1428
terminate called after throwing an instance of 'std::runtime_error'
  what():  
 ~~~~~~~~~~~~~~~~~ adapt::BaseMesh::initAnisotropyIndex ~~~~~~~~~~~~~~~~~ 
 Failed to find initial opposing face!
 Partition: 23
 Cell: 57430832 :: polyhedra
 n: 23 
 (
   15529381 :: [2.9153029462865616e+00 2.3716788569593583e+00 1.6824858095966990e+00] {0 0 0}
   80488568 :: [2.9106995584188318e+00 2.3745322681973411e+00 1.6805485417735193e+00] {1 1 0}
   80494791 :: [2.9061045061794402e+00 2.3658542722152580e+00 1.6786988298365557e+00] {1 1 0}
   80488567 :: [2.9152345834174480e+00 2.3657246637854894e+00 1.6824554668049947e+00] {1 1 0}
   15613312 :: [2.9159362300134335e+00 2.3716575584291846e+00 1.6810281321082341e+00] {0 0 0}
   80490989 :: [2.9113164176483997e+00 2.3745323495270014e+00 1.6790830094844111e+00] {1 1 0}
   15613311 :: [2.9066966052833663e+00 2.3774071406248183e+00 1.6771378868605882e+00] {0 0 0}
   15529382 :: [2.9060961705511019e+00 2.3773856794353234e+00 1.6786112739503398e+00] {0 0 0}
   80483965 :: [2.9158685012458170e+00 2.3657247741517073e+00 1.6809977345199560e+00] {1 1 0}
   80497194 :: [2.9067040720456472e+00 2.3658543707577904e+00 1.6772248331972825e+00] {1 1 0}
   80484233 :: [2.9015706020637895e+00 2.3746897130562958e+00 1.6768320751483961e+00] {1 1 0}
   15529378 :: [2.8970450335764766e+00 2.3719937466772683e+00 1.6750528763464525e+00] {0 0 0}
   80482367 :: [2.8973272284199449e+00 2.3719830136342814e+00 1.6743076450496202e+00] {1 0 1}
   15613309 :: [2.8976094232634138e+00 2.3719722805912951e+00 1.6735624137527880e+00] {0 0 0}
   80484232 :: [2.9021530142733898e+00 2.3746897106080569e+00 1.6753501503066881e+00] {1 1 0}
   80482368 :: [2.8969545548016358e+00 2.3660065670793218e+00 1.6750180659789646e+00] {1 1 0}
   80492263 :: [2.8972370967829035e+00 2.3660065940075339e+00 1.6742729815503528e+00] {1 1 1}
   80482366 :: [2.8975196387641713e+00 2.3660066209357464e+00 1.6735278971217413e+00] {1 1 0}
   15529384 :: [2.9151662205483344e+00 2.3597704706116205e+00 1.6824251240132906e+00] {0 0 0}
   80488570 :: [2.9106846554077928e+00 2.3570451476708385e+00 1.6805391448585016e+00] {1 1 0}
   15613313 :: [2.9158007724782009e+00 2.3597919898742297e+00 1.6809673369316780e+00] {0 0 0}
   91869905 :: [2.9109930734626674e+00 2.3570451804965211e+00 1.6798060974676075e+00] {1 1 1}
   80488838 :: [2.9113014915175417e+00 2.3570452133222037e+00 1.6790730500767130e+00] {1 1 0}
 ) {0 0 0} <0> [ Old Active Anisotropic ]
 f: 17 
 (...

I am running a steady state simulation using the following:

[Federico]

Hello Marc, 

can you also show your calcuation controls (from Solution tab > Run Calculation...)?

[marc.goldbach]

[Rob]

The prism refinement is for inflation layers and generally assumes you split the whole layer; depending on the version you'll get an error/failure/warning if the end of the prism split isn't terminated at the "end" of a wall. Not something you want to use in this scenario. 

Is the refinement/coarsening region being updated a lot? Ie is the adaption routine being triggered very frequently in that region? Can you post some images of the mesh?