Divergence from pressure outlet

Ananthakrishnan · August 9, 2012, 09:58

Hi,
I am analysing a structured mesh with spalart alamaras model with 1% turbulence level. The quality in ICEM is >0.45 but the residuals in fluent start diverging beginning from the pressure outlet..

Can anyone share some thoughts upon the reasons!!!!

RodriguezFatz · August 9, 2012, 10:16

Transient or steady state? What kind of solution methods (discretization schemes) do you use? Did you try to switch all of them to standard / (1st order upwind if possible)?

Ananthakrishnan · August 9, 2012, 14:10

Its steady state..everything is first order.. no change.. It goes on till some 500 iterations.. I also need to run it in second order scheme.. so if i try this solution with second order from 501th iteration it diverges even more faster (which is quite logical)...

RodriguezFatz · August 9, 2012, 16:16

OK, give some more information: 2d or 3d? Compressible gas? One pressure outlet and one velocity inlet? Any walls? How do you initialize? What happens in the first 500 iterations with your residuals? Any additional models, like energy, multiphase? Do you use SIMPLE for solving?

... you could lower the under-relaxation factors.

Ananthakrishnan · August 9, 2012, 17:05

i have attached the picture of the residuals.. Its a 3d scalloped flipper wing

http://www.cfd-online.com/Forums/ans...-geometry.html
If you wish you can see the geometry of the wing in the above link.. but the mesh is not the one in the link.. the latest one is a completely structured mesh..

The mach number is around 0.2.. So i am using incompressible flow..out of the six sides of the flow domain, the two on left and right side have been set as symmetry. front side, top and bottom are velocity inlet..no additional models and its just SIMPLE..

hope you get some idea from these info and thanks a lot for your help..

Far · August 10, 2012, 08:32

can you please show the residuals with latest mesh? Did you try the solution streeing for your case (if it is available for pressure based solver)?

Far · August 11, 2012, 06:09

Could you please show us the comparison of old and new mesh on convergence ...

Ananthakrishnan · August 11, 2012, 06:24

This is the residuals of the latest mesh.. I had started the transient simulation, thats why you see the huge variations near the end. pls ignore that.

In the second order, the residuals initally decrease, later they neither decrease nor increase!!

Far · August 11, 2012, 06:37

did you give some special consideration to mesh for using DES? or same mesh as used in RANS?

Ananthakrishnan · August 11, 2012, 06:39

its the same mesh. I run it in RANS initially. Later when it has converged, i switch to URANS. If you are concerned about the peaks occurring after 800 iterations, then dont be.. because its normal in transient simulation.

Far · August 11, 2012, 07:00

do you think the quality of mesh (min angle in this case) has the detrimental effect in convergence?

Ananthakrishnan · August 11, 2012, 07:30

After you increased the min angle to 24 the convergence improved hugely. before at 17 degree as you see even the first order was not converging..
Is it possible to increase the min angle further.. i am trying from last night but i am not getting any big difference..

Far · August 11, 2012, 07:33

it is restricted by the angle at the sharp trailing.

Ananthakrishnan · August 11, 2012, 07:40

yeah, i saw it.. but since it is in the flow region (adjacent to the wing) i guess it should not be a serious problem.. so i am ignoring it and trying to improve in the far field region, in the boundary. do you think something like this is possible or is it interrelated!!!

August 9, 2012, 09:58	Divergence from pressure outlet	#1
Ananthakrishnan Senior Member Ananthakrishnan.A.S Join Date: Feb 2012 Location: Mumbai (Bombay), India Posts: 157 Rep Power: 16	Hi, I am analysing a structured mesh with spalart alamaras model with 1% turbulence level. The quality in ICEM is >0.45 but the residuals in fluent start diverging beginning from the pressure outlet.. Can anyone share some thoughts upon the reasons!!!!

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
pressure oscillation near pressure outlet boundary	kino	Main CFD Forum	5	April 13, 2011 12:03
Pressure Outlet setting	CoG	STAR-CCM+	4	June 9, 2010 22:47
Backflow occuring at a pressure outlet?	Dave	FLUENT	1	August 12, 2004 18:39
Gas pressure question	Dan Moskal	Main CFD Forum	0	October 24, 2002 23:02
Hydrostatic pressure in 2-phase flow modeling (long)	DS & HB	Main CFD Forum	0	January 8, 2000 16:00

August 9, 2012, 10:16		#2
RodriguezFatz Senior Member Philipp Join Date: Jun 2011 Location: Germany Posts: 1,297 Rep Power: 27	Transient or steady state? What kind of solution methods (discretization schemes) do you use? Did you try to switch all of them to standard / (1st order upwind if possible)?

August 9, 2012, 14:10		#3
Ananthakrishnan Senior Member Ananthakrishnan.A.S Join Date: Feb 2012 Location: Mumbai (Bombay), India Posts: 157 Rep Power: 16	Its steady state..everything is first order.. no change.. It goes on till some 500 iterations.. I also need to run it in second order scheme.. so if i try this solution with second order from 501th iteration it diverges even more faster (which is quite logical)...

August 9, 2012, 16:16		#4
RodriguezFatz Senior Member Philipp Join Date: Jun 2011 Location: Germany Posts: 1,297 Rep Power: 27	OK, give some more information: 2d or 3d? Compressible gas? One pressure outlet and one velocity inlet? Any walls? How do you initialize? What happens in the first 500 iterations with your residuals? Any additional models, like energy, multiphase? Do you use SIMPLE for solving? ... you could lower the under-relaxation factors.

August 9, 2012, 17:05		#5
Ananthakrishnan Senior Member Ananthakrishnan.A.S Join Date: Feb 2012 Location: Mumbai (Bombay), India Posts: 157 Rep Power: 16	i have attached the picture of the residuals.. Its a 3d scalloped flipper wing http://www.cfd-online.com/Forums/ans...-geometry.html If you wish you can see the geometry of the wing in the above link.. but the mesh is not the one in the link.. the latest one is a completely structured mesh.. The mach number is around 0.2.. So i am using incompressible flow..out of the six sides of the flow domain, the two on left and right side have been set as symmetry. front side, top and bottom are velocity inlet..no additional models and its just SIMPLE.. hope you get some idea from these info and thanks a lot for your help..

August 10, 2012, 08:32		#6
Far Senior Member Sijal Join Date: Mar 2009 Location: Islamabad Posts: 4,558 Blog Entries: 6 Rep Power: 54	can you please show the residuals with latest mesh? Did you try the solution streeing for your case (if it is available for pressure based solver)?

August 11, 2012, 06:09		#7
Far Senior Member Sijal Join Date: Mar 2009 Location: Islamabad Posts: 4,558 Blog Entries: 6 Rep Power: 54	Could you please show us the comparison of old and new mesh on convergence ...

August 11, 2012, 06:24		#8
Ananthakrishnan Senior Member Ananthakrishnan.A.S Join Date: Feb 2012 Location: Mumbai (Bombay), India Posts: 157 Rep Power: 16	This is the residuals of the latest mesh.. I had started the transient simulation, thats why you see the huge variations near the end. pls ignore that. In the second order, the residuals initally decrease, later they neither decrease nor increase!!

August 11, 2012, 06:37		#9
Far Senior Member Sijal Join Date: Mar 2009 Location: Islamabad Posts: 4,558 Blog Entries: 6 Rep Power: 54	did you give some special consideration to mesh for using DES? or same mesh as used in RANS?

August 11, 2012, 06:39		#10
Ananthakrishnan Senior Member Ananthakrishnan.A.S Join Date: Feb 2012 Location: Mumbai (Bombay), India Posts: 157 Rep Power: 16	its the same mesh. I run it in RANS initially. Later when it has converged, i switch to URANS. If you are concerned about the peaks occurring after 800 iterations, then dont be.. because its normal in transient simulation.

August 11, 2012, 07:00		#11
Far Senior Member Sijal Join Date: Mar 2009 Location: Islamabad Posts: 4,558 Blog Entries: 6 Rep Power: 54	do you think the quality of mesh (min angle in this case) has the detrimental effect in convergence?

August 11, 2012, 07:30		#12
Ananthakrishnan Senior Member Ananthakrishnan.A.S Join Date: Feb 2012 Location: Mumbai (Bombay), India Posts: 157 Rep Power: 16	After you increased the min angle to 24 the convergence improved hugely. before at 17 degree as you see even the first order was not converging.. Is it possible to increase the min angle further.. i am trying from last night but i am not getting any big difference..

August 11, 2012, 07:33		#13
Far Senior Member Sijal Join Date: Mar 2009 Location: Islamabad Posts: 4,558 Blog Entries: 6 Rep Power: 54	it is restricted by the angle at the sharp trailing.

August 11, 2012, 07:40		#14
Ananthakrishnan Senior Member Ananthakrishnan.A.S Join Date: Feb 2012 Location: Mumbai (Bombay), India Posts: 157 Rep Power: 16	yeah, i saw it.. but since it is in the flow region (adjacent to the wing) i guess it should not be a serious problem.. so i am ignoring it and trying to improve in the far field region, in the boundary. do you think something like this is possible or is it interrelated!!!