Horus |
February 26, 2013 09:31 |
Trouble with simple AMI case
Hello,
I have a case of a simple geometry on which I want to use AMI interpolation for rotation.
I construct it from two meshes. The first one is a large box and cylinder. The second one the same cylinder but mesh generation startpoint inside the cylinder and a simple non-axisymmetric wheel which is completely inside the cylinder. The box has slip condition at the walls and zeroGradient/fixedValue in/outflow. The meshes are generated using Numeca Spider and converted to OF using fluent3DMeshToFoam.
Turbulence is disabled.
The two meshes are merged using mergeMeshes. To construct a zone from the inside region I use topoSet:
Code:
actions
(
{
name tempBox;
type cellSet;
action new;
source boxToCell;
sourceInfo {box (-100 -100 -100 ) (100 100 100); }
}
{
name rotZone;
type cellSet;
action new;
source regionToCell;
sourceInfo { set tempBox; insidePoint (0.03 -0.02 0.05); } // A point inside the rotating mesh
}
{
name rotZone;
type cellZoneSet;
action new;
source setToCellZone;
sourceInfo { set rotZone; }
}
{
name tempBox;
type cellSet;
action remove;
}
A changeDictionary call modifies the AMI patches in the boundary dict.
I've taken my system dictory from the propeller tutorial cases, my solver is pimpleDyMFoam as well. Initial timestep is 1e-5 and maxCo 1. fvSchemes and fvSolvers are unmodified except pRefCell / pRefValue.
I'm not sure if AMI needs a overlap or gap between the two meshes. I've tried three methods: unchanged sized and the inside scaled to 99,9% and 100,1% resulting in a visible gap/overlap using transformPoints -scale. All cases crash with similiar behavior.
The case was decomposed using scotch for 128 processors.
The case starts just fine, but after a few timesteps Co rises / timestep shrinks until it crashes.
A timestep looks like that for example:
Code:
Courant Number mean: 1.4220815e-09 max: 1.986819
deltaT = 4.552641e-13
Time = 1.09198398e-05
solidBodyMotionFunctions::rotatingMotion::transformation(): Time = 1.091984e-05 transformation: ((0 0 0) (0.99999846 (0 0.0017549257 0)))
AMI: Creating addressing and weights between 75608 source faces and 128434 target faces
AMI: Patch source weights min/max/average = 0.91541718, 1.0695428, 0.99960354
AMI: Patch target weights min/max/average = 0.98438672, 1.1577481, 1.0012444
PIMPLE: iteration 1
smoothSolver: Solving for Ux, Initial residual = 8.2800706e-05, Final residual = 3.2848892e-06, No Iterations 1
smoothSolver: Solving for Uy, Initial residual = 5.247264e-05, Final residual = 2.5868548e-06, No Iterations 1
smoothSolver: Solving for Uz, Initial residual = 7.837248e-05, Final residual = 3.5215317e-06, No Iterations 1
GAMG: Solving for p, Initial residual = 0.2816471, Final residual = 0.0027095654, No Iterations 4
time step continuity errors : sum local = 3.7889088e-11, global = -1.9208336e-13, cumulative = 4.9011936e-10
PIMPLE: iteration 2
smoothSolver: Solving for Ux, Initial residual = 9.2696402e-06, Final residual = 7.1946491e-07, No Iterations 2
smoothSolver: Solving for Uy, Initial residual = 6.6339078e-06, Final residual = 5.1357597e-07, No Iterations 2
smoothSolver: Solving for Uz, Initial residual = 1.0457223e-05, Final residual = 7.8017092e-07, No Iterations 2
GAMG: Solving for p, Initial residual = 0.23159011, Final residual = 0.0020297622, No Iterations 4
time step continuity errors : sum local = 3.7371482e-11, global = 1.304155e-13, cumulative = 4.9024978e-10
PIMPLE: iteration 3
DILUPBiCG: Solving for Ux, Initial residual = 3.4757571e-05, Final residual = 1.0572902e-07, No Iterations 3
DILUPBiCG: Solving for Uy, Initial residual = 2.0913946e-05, Final residual = 5.5933975e-07, No Iterations 2
DILUPBiCG: Solving for Uz, Initial residual = 4.4476877e-05, Final residual = 3.8734961e-07, No Iterations 3
GAMG: Solving for p, Initial residual = 0.30341415, Final residual = 8.5899541e-07, No Iterations 16
time step continuity errors : sum local = 1.5030424e-14, global = 3.4608435e-16, cumulative = 4.9025012e-10
ExecutionTime = 104.46 s ClockTime = 105 s
I've uploaded the complete case to http://xgm.de/upload/OF/ (archive and web-browsable). The case I have uploaded uses a 99,9% scaled inner mesh. I can of course provide the other ones, too. Not all timesteps are included but the first three ones..
I would really appreciate if someone would have a look and an idea why these cases all diverge.
Thanks!
Florian
|