I am using simpleFoam with kEpsilon model. After few iterations it looks like it stops solving equations of k and epsilon as I don't see error on my terminal. Does this mean that these equations diverged? How to make it visible all the time. I have experimented with "tolerences" as well as "relaxation factors" no improvement.

Regards
GS

[liu]

Maybe the initial residual is less than tolerences, so these equations don't need to be solved. Steady state?

Yes, steady state, I was guessing the same, but I still want to solve. How do I do that.

Regards

[henry]

Do you see the initial residual fall? Does it drop below your convergence tolerance? Have you tried making the convergence tolerance very tight?

>Do you see the initial residual fall?
Yes, initially there is reduction in residual.
>Does it drop below your convergence tolerance?
Yes
>Have you tried making the convergence tolerance very tight?
Yes

Currently I am compiling the code by making following change in "tests.C"

//converged_ = true;
converged_ = false;

Will this help?

Regards
GS

[henry]

I would be able to help you more if I had the answers to my questions or better still a section of the log file from just before k-epsilon stop solving to just after.

[henry]

Sorry, I missed the answers buried there. Anyway, I would still like to see the section of log file. Have you looked the the field valeus for k-epsilon? Are they VERY large?

BICCG: Solving for Ux, Initial residual = 0.0114222, Final residual = 0.000931753, No Iterations 1
BICCG: Solving for Uy, Initial residual = 0.0227955, Final residual = 0.00190204, No Iterations 1
BICCG: Solving for Uz, Initial residual = 0.00691902, Final residual = 0.000593082, No Iterations 1
ICCG: Solving for p, Initial residual = 0.0216008, Final residual = 0.000214944, No Iterations 116
time step continuity errors : sum local = 0.00871451, global = 0.000437168, cumulative = 0.0171293
BICCG: Solving for epsilon, Initial residual = 0.021943, Final residual = 2.2619e-08, No Iterations 1
BICCG: Solving for k, Initial residual = 1.78372e-05, Final residual = 1.1457e-06, No Iterations 1

ExecutionTime = 201 s


Time = 257
BICCG: Solving for Ux, Initial residual = 0.0114393, Final residual = 0.000938528, No Iterations 1
BICCG: Solving for Uy, Initial residual = 0.0228175, Final residual = 0.0019204, No Iterations 1
BICCG: Solving for Uz, Initial residual = 0.00691351, Final residual = 0.00059707, No Iterations 1
ICCG: Solving for p, Initial residual = 0.0229688, Final residual = 0.000222948, No Iterations 115
time step continuity errors : sum local = 0.00902571, global = 0.000449829, cumulative = 0.0175791
BICCG: Solving for epsilon, Initial residual = 0.0225105, Final residual = 2.18334e-08, No Iterations 1
bounding epsilon, min: -3.5104e-12 max: 822122 average: 4651.92

ExecutionTime = 205.23 s


Time = 258
BICCG: Solving for Ux, Initial residual = 0.0114469, Final residual = 0.00095594, No Iterations 1
BICCG: Solving for Uy, Initial residual = 0.0226974, Final residual = 0.00194632, No Iterations 1
BICCG: Solving for Uz, Initial residual = 0.00688176, Final residual = 0.000596608, No Iterations 1
ICCG: Solving for p, Initial residual = 0.0220571, Final residual = 0.00021926, No Iterations 118
time step continuity errors : sum local = 0.00883914, global = 0.000396233, cumulative = 0.0179754

Will this help to clarify myself? You can see after Time(iteration) # 257, k and epsilon disappears forever.

Regards
GS

[henry]

What convergence tolerance have you set for k and epsilon?

For both 1e-5(Tolerance), 0.1(RelativeTolerance)

[henry]

For cases will wall-functions I use 1e-8.

Now I am running with 1e-15 and 0.1 for both. It appears again. This way what is the lower tolerence limit?

Thanks and regards
GS

[henry]

Check the k and epsilon fields, it sounds like they are becoming VERY large which affecting the residual normalisation.

Max epsilon is 676404.2 m^2/s^3 and k is 72 m^/s^2, which is not too large value.

It has disappeared again even after giving tolerance 1e-15 for both k and epsilon.

Regards
GS

[henry]

It appears k and epsilon have converged for your case.

But I do not want it to stop solving k and epsilon, as flow field is still developing. How do I do that?

Regards
GS

[henry]

For some reason, probably due to something in your setup/BCs, the k-epsilon system has converged and nothing you do to the tolerances will make it converge further. You need to find out what is wrong with your setup.

We would be happy to check and correct your case as part of your support contract with OpenCFD.