Epsilon residual in FLUENT
 

Hi Dear Friends,

I have a problem for convergence of turbulence in fluent. In my model mesh quality and aspect ration are 0.27 and 16.1 respectively and use k-epsilon model for turbulence. Epsilon residual is not stable and after some iteration and adding different model like energy, radiation and etc for simulation, epsilon residual rise and i can not control it. Also i use first-order scheme for discretization of trubulaence model. Can anyone help me to set up fluent solver like Multigrid type, under relaxation factor and other way for solution epsilon equation and good simulation.

The residuals graph attached.

Thanks,,,

50 iterations is too few to make any judgement. Does it diverge? If not, keep cranking. When you change models, expect residuals to spike. Don't touch multigrid parameters. If it diverges you can try playing with urf's but you have to know why it diverged. If it doesn't diverge, leave it alone.

Hello,

I'm also facing the same situation. Epsilon oscillates too much even though i went through large number of iterations.
Any suggestions to get smooth residuals.

Thanks in Advance

Samm.

In your case, all your residuals are oscillating. It's not just the epsilon equation. Your solution can be improved everywhere.


There is a warning that the pdf enthalpy is too low, lower than your table, and this means your solution is being clipped. It's only 2 cells so maybe it's not important. Check if your solution makes sense (that you have a flame where you expect it to be). Probably, the solution is completely wrong. If it looks okay... then it's just a matter of better meshing.

Thanks for your reply

Yes,You are correct .Still I couldn't get the flame that i expected.
My problem is 2d axisymmetric pressure based k-epsilon model.
I'm using Non-premixed chemical equilibrium model with 12 species

Methane Velocity inlet 4.62 m/s with pressure 4.113
Oxygen Velocity inlet 59.89 m/s with pressure 4.89
temperature at both inlet 290 K.

Hybrid Initialization

My case seems simple but i don't know where i'm making mistake
please help with your valuable suggestions.

If you don't get the flame you expect, then your solution is just wrong. It's meaningless to look at or compare residuals. Your residuals look bounded asymptotic, so it's more-or-less reached iterative convergence. You have to think about why the flame is not where it is. What's wrong with the flame? Do you even have a flame or did it blow out? etc. It could be initialization and/or wrong BC's and/or settings. I see you chose hybrid initialization. Well, a lot can go wrong.

Thank you sir

My boundary conditions seems well as it was taken from the experimental work.so maybe no problem with that. Most probably initialization issue, if i use standard initialization,will it be good or something else i have to change.

Currently in this case residuals oscillating too much and flame seems completely not OK..no flame in the chamber,near inlet it was looking less temperature complete domain is same max temp say(3050 k). this may imbalance stuck in my domain or i don't know how to sort it out.

help me with some suggestions

Hybrid initialization is nice and automatic and in most cases but stupid.


Try either:
1) Standard initialization. You have to do some thinking to figure out the initial conditions.

2) Patch the progress variable in the combustion chamber. This forces there to be a flame and hopefully it anchors itself in the right place. You can also patch the temperature where it doesn't make sense. This is more-or-less the same as standard initialization, but you get to keep your presumably nice flow field that you've already achieved over 100k iterations.

Thanks again

I started the case again with standard initialization (atmospheric condition(pressure and temp) and other values zero). But i couldn't get the desired flame.

I attached residuals and temperature contour.

how can i tune this case in right way.Help me

Hi Jegan,

Use FMG initialization after standard initialization. Hopefully it will work. Give initial value of epsilon wisely in standard initialization. For achieve that value you have to do run model for 3-4 times. But, you will get experience from that.

Thanks.

Regards,
Sanket

Thank you for your kind response

I think FMG initialization is unavailable for Non-premixed combustion:confused:
In standard initialization i simply gave atmospheric condition (pressure and temperature) i have no idea about other values so i put zero for all...now the case achieved the same flame as i attached in my previous reply.
Completely helpless situation now ..don't know what to alter to get on right way.
Give your valuable suggestions

Thanks

Any help please..why the heavy imbalance occurs. Unable to achieve the desired flame.:confused: constant maximum temperature after some iterations how to sort out this issue. where and how to resolve this problem