Strange RSM Temperature Divergence

dvcauwe · March 7, 2012, 09:57

Hello,

I am currently working on my Master's Thesis and I'm trying to simulate turbulent air flow through finned tubes. For this I would like to use the Reynolds Stress Model with QUICK discretization scheme for all equations.

Unfortunately, whether I initialize from a fully converged ke-RNG (all QUICK) solution, or start from a fully converged RSM (all QUICK, except for 1st order upwind reynolds stresses), I can't achieve convergence.

For about 50 iterations, all appears normal, but then suddenly in 1 iteration, the temperature leaps out of bounds and all is lost.

I have also tried 2nd order upwind schemes and lowering the energy URF but nothing seems to work.

Thank you very much in advance for any advice you could give me!

ghost82 · March 7, 2012, 10:45

I don't know why, but in my experience sometimes I can't get convergence when I switch from 1st order to 2nd order scheme previous initialization from a 1st order converged solution...I got convergence by reinitializing the solution and starting the simulation directly with 2nd order scheme..
You can try to start your simulation directly with reynolds stress model and see if something changes...

Daniele

dvcauwe · March 8, 2012, 11:47

This actually seemed to work for about 100 iterations (furthest I got so far!) but after that, all comes crashing down again in a single iteration. Is this normal diverging behavior?

What I find strange also, is that it's always the temperature that diverges while there isn't even any heat transfer going on. If I turn off the energy equation and work with constant gas density, after 50 iterations, out of nowhere the stress-residuals jump 4 orders and everything diverges as well.

Any help or previous experiences with this issue would be greatly appreciated!

March 7, 2012, 09:57	Strange RSM Temperature Divergence	#1
dvcauwe New Member David Van Cauwenberge Join Date: Feb 2012 Location: Ghent, Belgium Posts: 19 Rep Power: 14	Hello, I am currently working on my Master's Thesis and I'm trying to simulate turbulent air flow through finned tubes. For this I would like to use the Reynolds Stress Model with QUICK discretization scheme for all equations. Unfortunately, whether I initialize from a fully converged ke-RNG (all QUICK) solution, or start from a fully converged RSM (all QUICK, except for 1st order upwind reynolds stresses), I can't achieve convergence. For about 50 iterations, all appears normal, but then suddenly in 1 iteration, the temperature leaps out of bounds and all is lost. I have also tried 2nd order upwind schemes and lowering the energy URF but nothing seems to work. Thank you very much in advance for any advice you could give me!

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
# Divergence detected in AMG solver: temperature -> Increasing relaxation sweeps!	pastarox	FLUENT	7	July 24, 2017 10:03
Calculation of the Governing Equations	Mihail	CFX	7	September 7, 2014 06:27
divergence detected in AMG solver: temperature	zambroa	FLUENT	4	September 11, 2009 10:06
" Divergence detected in AMG solver: temperature"	Suman	FLUENT	4	September 11, 2009 09:31
strange divergence when solving multiphase problem	tanghao	FLUENT	2	July 27, 2006 19:47

March 7, 2012, 10:45		#2
ghost82 Senior Member Rick Join Date: Oct 2010 Posts: 1,016 Rep Power: 26	I don't know why, but in my experience sometimes I can't get convergence when I switch from 1st order to 2nd order scheme previous initialization from a 1st order converged solution...I got convergence by reinitializing the solution and starting the simulation directly with 2nd order scheme.. You can try to start your simulation directly with reynolds stress model and see if something changes... Daniele

March 8, 2012, 11:47		#3
dvcauwe New Member David Van Cauwenberge Join Date: Feb 2012 Location: Ghent, Belgium Posts: 19 Rep Power: 14	This actually seemed to work for about 100 iterations (furthest I got so far!) but after that, all comes crashing down again in a single iteration. Is this normal diverging behavior? What I find strange also, is that it's always the temperature that diverges while there isn't even any heat transfer going on. If I turn off the energy equation and work with constant gas density, after 50 iterations, out of nowhere the stress-residuals jump 4 orders and everything diverges as well. Any help or previous experiences with this issue would be greatly appreciated!