High Residuals

alw7 · September 14, 2018, 14:01

Good Afternoon,

I am currently working with a k-omega turbulence model and my results converge but my residuals are very high, specifically the TKE as shown. I already tried reducing the under-relaxation factors but it does not seem to be helping. Could someone please help me?

alw

Baseline_Case_Surface_Control_VC-50perc_VC-70perc_URF_VCfix_KOT-.4@24001_Residuals.png

fluid23 · September 14, 2018, 17:03

Have you tried locating cells with high residuals?

rasidual in each cell, location of the cells with maximum residuals

alw7 · September 15, 2018, 20:52

I have located the cells with high residuals but I am unsure what to do with this information. Do I need to adjust the cells to a better quality or do I just need to know where they are giving me trouble?

fluid23 · September 17, 2018, 12:13

It depends. If all of your high residual cells are in an area that you feel is not important or is not impacting your extracted results then you can decide to ignore them. I would typically only apply this approach to turbulence residuals.

If you think it is impacting your results then you can apply a local mesh refinement and make the cells in that area smaller. In some cases that will be enough.

The last resort would be to resort to an unsteady analysis. Most real flows are unsteady and there could be enough unsteady behavior that a time averaged RANS analysis is not sufficient to capture the phenomena that are driving your flow.

That all being said, your residuals do not look sufficiently converged. If you could provide more info on what you are modeling, how well your mesh is suited (for example what are your wall y+ values), and so on it would help in diagnosing your issues.

ashokac7 · September 18, 2018, 03:35

Yes. Need more info to comment something better, Like why K-Omega. And what fluid23 is asking. Most of the times if other residuals are converged then high Tke and Tdr don't create an issue. If you turn off the normalization for Tke they will fall near the other residual range. This happens because during normalization, initial guess is so small that dividing the later residual by small quantity produce high normalized Tke residual monitor.

September 14, 2018, 14:01	High Residuals	#1
alw7 New Member Indiana (IN) Join Date: Apr 2018 Posts: 6 Rep Power: 8	Good Afternoon, I am currently working with a k-omega turbulence model and my results converge but my residuals are very high, specifically the TKE as shown. I already tried reducing the under-relaxation factors but it does not seem to be helping. Could someone please help me? alw Baseline_Case_Surface_Control_VC-50perc_VC-70perc_URF_VCfix_KOT-.4@24001_Residuals.png

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
how to understand high resolution scheme and high order scheme	iilw1314	Main CFD Forum	7	April 12, 2022 13:29
Fluctuating residuals at high angle of attack for NACA0018	Anand Thamban	FLUENT	1	June 20, 2016 08:58
Continuity Residual high but total mass convergence achieved?	MichaelStark331	FLUENT	1	February 8, 2016 11:24
High Lift Airfoils At High Angles Of Attack	Luiz Pancini	FLUENT	2	April 9, 2015 09:01
plot residuals	pigna	OpenFOAM Running, Solving & CFD	2	December 19, 2014 03:32

September 14, 2018, 17:03		#2
fluid23 Senior Member Matt Join Date: Aug 2014 Posts: 947 Rep Power: 18	Have you tried locating cells with high residuals? rasidual in each cell, location of the cells with maximum residuals

September 15, 2018, 20:52		#3
alw7 New Member Indiana (IN) Join Date: Apr 2018 Posts: 6 Rep Power: 8	I have located the cells with high residuals but I am unsure what to do with this information. Do I need to adjust the cells to a better quality or do I just need to know where they are giving me trouble?

September 17, 2018, 12:13		#4
fluid23 Senior Member Matt Join Date: Aug 2014 Posts: 947 Rep Power: 18	It depends. If all of your high residual cells are in an area that you feel is not important or is not impacting your extracted results then you can decide to ignore them. I would typically only apply this approach to turbulence residuals. If you think it is impacting your results then you can apply a local mesh refinement and make the cells in that area smaller. In some cases that will be enough. The last resort would be to resort to an unsteady analysis. Most real flows are unsteady and there could be enough unsteady behavior that a time averaged RANS analysis is not sufficient to capture the phenomena that are driving your flow. That all being said, your residuals do not look sufficiently converged. If you could provide more info on what you are modeling, how well your mesh is suited (for example what are your wall y+ values), and so on it would help in diagnosing your issues.

September 18, 2018, 03:35		#5
ashokac7 Senior Member Ashok Chaudhari Join Date: Aug 2016 Location: Pune, India Posts: 260 Rep Power: 11	Yes. Need more info to comment something better, Like why K-Omega. And what fluid23 is asking. Most of the times if other residuals are converged then high Tke and Tdr don't create an issue. If you turn off the normalization for Tke they will fall near the other residual range. This happens because during normalization, initial guess is so small that dividing the later residual by small quantity produce high normalized Tke residual monitor.