CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   OpenFOAM Running, Solving & CFD (http://www.cfd-online.com/Forums/openfoam-solving/)
-   -   Calculation of k and epsilon freezes (http://www.cfd-online.com/Forums/openfoam-solving/109232-calculation-k-epsilon-freezes.html)

Nigirim November 13, 2012 05:13

Calculation of k and epsilon freezes
 
Hi all,

I've got a problem with my calculation von k and epsilon.

I'm trying to simulate a rotating geometry with a modified version of interFoam. The geometry is rotating in different directions and my fluid is
sloshing, but the values for k and epsilon just dont chance anymore after some iterations.

At the beginning there are some Iterations for k and epsilon
Code:

MULES: Solving for alpha1
Liquid phase volume fraction = 0.979575 Min(alpha1) = -2.49977e-20 Max(alpha1) = 1
MULES: Solving for alpha1
Liquid phase volume fraction = 0.979575 Min(alpha1) = -7.02031e-21 Max(alpha1) = 1
MULES: Solving for alpha1
Liquid phase volume fraction = 0.979575 Min(alpha1) = -1.31893e-20 Max(alpha1) = 1
MULES: Solving for alpha1
Liquid phase volume fraction = 0.979575 Min(alpha1) = -4.39865e-20 Max(alpha1) = 1
DICPCG: Solving for p_rgh, Initial residual = 0.00109002, Final residual = 8.85396e-08, No Iterations 124
time step continuity errors : sum local = 1.83749e-09, global = 2.10194e-11, cumulative = -6.24155e-11
DICPCG: Solving for p_rgh, Initial residual = 0.000154468, Final residual = 9.29858e-08, No Iterations 114
time step continuity errors : sum local = 1.93264e-09, global = -1.87438e-10, cumulative = -2.49854e-10
DICPCG: Solving for p_rgh, Initial residual = 2.61918e-05, Final residual = 9.23886e-08, No Iterations 79
time step continuity errors : sum local = 1.9189e-09, global = 2.37918e-10, cumulative = -1.19357e-11
DILUPBiCG: Solving for epsilon, Initial residual = 0.000340262, Final residual = 6.24431e-09, No Iterations 2
bounding epsilon, min: -4.02028e-08 max: 0.107857 average: 5.80718e-05
DILUPBiCG: Solving for k, Initial residual = 0.000539149, Final residual = 1.33527e-11, No Iterations 4
bounding k, min: -7.24082e-12 max: 0.00100971 average: 1.00796e-05
ExecutionTime = 17.68 s ClockTime = 18 s

but later on, at about 0.2s (first directionchange is at 0.5s) it pretty much looks like this:
Code:

MULES: Solving for alpha1
Liquid phase volume fraction = 0.979575 Min(alpha1) = -2.15831e-20 Max(alpha1) = 1.00001
MULES: Solving for alpha1
Liquid phase volume fraction = 0.979575 Min(alpha1) = -2.96242e-20 Max(alpha1) = 1.00001
MULES: Solving for alpha1
Liquid phase volume fraction = 0.979575 Min(alpha1) = -1.20443e-20 Max(alpha1) = 1.00001
MULES: Solving for alpha1
Liquid phase volume fraction = 0.979575 Min(alpha1) = -9.61596e-21 Max(alpha1) = 1.00001
DICPCG: Solving for p_rgh, Initial residual = 0.000269819, Final residual = 9.59801e-08, No Iterations 26
time step continuity errors : sum local = 1.39057e-09, global = 1.24122e-10, cumulative = -4.05415e-08
DICPCG: Solving for p_rgh, Initial residual = 1.5512e-05, Final residual = 9.55703e-08, No Iterations 21
time step continuity errors : sum local = 1.3846e-09, global = 1.14579e-10, cumulative = -4.04269e-08
DICPCG: Solving for p_rgh, Initial residual = 1.98201e-06, Final residual = 9.75346e-08, No Iterations 5
time step continuity errors : sum local = 1.41305e-09, global = 1.08762e-10, cumulative = -4.03181e-08
DILUPBiCG: Solving for epsilon, Initial residual = 0.00010125, Final residual = 7.4232e-09, No Iterations 1
bounding epsilon, min: 1.98379e-18 max: 16.2816 average: 0.00581568
DILUPBiCG: Solving for k, Initial residual = 9.96483e-11, Final residual = 9.96483e-11, No Iterations 0
ExecutionTime = 936.41 s ClockTime = 977 s

Right now I'm using realizableKE as turbulence model, but I also tried to run it wie kOmegaSST with the same result.


I tried several things I found on the internet including:
  • Changing mesh quality (changed it between 20.000 and 100.000 cells)
  • Decreasing and increasing the timesteps
  • Changing my BCs for k, epsilon and nut
  • Used other schemes
  • Added some extra layers to my mesh
  • Changing the tolerance for k and epsilon
  • Changing my relaxation factors
But still my simulation looks pretty much the same, just the change of my tolerance resulted in an increase of the time that k and epsilon are calculated but still only about 1s out of 2s.

Here are my current BCs for k, epsilon and nut
k
Code:

boundaryField
{
outlet
{
type inletOutlet;
inletValue uniform 0.001;
value uniform 0.01;
}
symmetry
{
type symmetryPlane;
}
wall
{
type kqRWallFunction;
value uniform 1e-06;
}

epsilon
Code:

boundaryField
{
outlet
{
type inletOutlet;
inletValue uniform 0.0001;
value uniform 0;
}
symmetry
{
type symmetryPlane;
}
wall
{
type epsilonWallFunction;
value uniform 0.001;
}

nut
Code:

boundaryField
{
outlet
{
type calculated;
value uniform 0;
}
symmetry
{
type symmetryPlane;
}
wall
{
type nutkWallFunction;
value uniform 0;
}

I still think there is something with my BCs, but I dont know whats wrong or
causing the problem.

If anyone has an idea what I could do, please let me know.

My englisch isnt the best so if you dont understand something just ask me :D

Greets
Nigirim

hfs November 14, 2012 08:52

I got a similar problem!

here:

http://www.cfd-online.com/Forums/ope...alcontrol.html

Let me know when you have a solution... Thanks!


All times are GMT -4. The time now is 14:00.