How to setup initail conditions in steady solution

aja1345 · May 8, 2016, 04:59

Hi,

I am simulating a trbine and I have problem in running after stall...

In fact, Residuals have fluctuation after stall because solution is steady and steady solution fails after stall because there are massive separations.

But i want to have better convergence after stall for having better results. I think that this convergence is related to initial guess but i am new in CFX and i don't know that how to set up that. I am grateful that guide me about these cases.

Now, Have you any idea and suggestions?

Thanks.

Best.

Aja

ghorrocks · May 8, 2016, 06:36

This FAQ covers this topic: http://www.cfd-online.com/Wiki/Ansys...gence_criteria

But stall is usually characterised by large scale transient turbulent structures which cannot be modelled steady state (hence your steady state model is not converging). You might be forced to use a LES class model to handle this, and that means running transient. LES class model means LES, SAS, DES or one of that class of turbulent model.

aja1345 · May 8, 2016, 07:49

Quote:

Originally Posted by ghorrocks

This FAQ covers this topic: http://www.cfd-online.com/Wiki/Ansys...gence_criteria

But stall is usually characterised by large scale transient turbulent structures which cannot be modelled steady state (hence your steady state model is not converging). You might be forced to use a LES class model to handle this, and that means running transient. LES class model means LES, SAS, DES or one of that class of turbulent model.

Hi, thanks.

I reduce time scale and convergence is better......Is this logical?

I can't find the following point:

"Read the CFX documentation. Specifically section 15 ("Advice on Flow Modeling"), section 15.10.6 ("Problems with Convergence"), and and any best practises guide applicable to your simulation"

Can you give me link of these documentations please? i can't find them.

Thanks.

ghorrocks · May 8, 2016, 08:07

The FAQs are a few years old. Those sections in the documentation are probably a different section number now. But hopefully they should not be too far away so you should find them quickly enough.

If reducing time scale improves convergence this suggests you have significant numerical instability in your simulation, and the normal cause of that is mesh quality. So if you improve your mesh quality it is likely your convergence will significantly improve.

aja1345 · May 8, 2016, 08:40

Quote:

Originally Posted by ghorrocks

The FAQs are a few years old. Those sections in the documentation are probably a different section number now. But hopefully they should not be too far away so you should find them quickly enough.

If reducing time scale improves convergence this suggests you have significant numerical instability in your simulation, and the normal cause of that is mesh quality. So if you improve your mesh quality it is likely your convergence will significantly improve.

Mesh quality is good and also It doesn't improve more because geometry is complex....I think this is because of steady solution.....No?

ghorrocks · May 8, 2016, 20:07

How do you know the mesh quality is good? Different models have different requirements on mesh quality, and the comments by the CFX solver are just a guide.

As I stated in post #2, this probably is a LES class model and those models are inherently transient. So that means you will not be able to get good results running them steady state.

aja1345 · May 12, 2016, 08:26

Quote:

Originally Posted by ghorrocks

How do you know the mesh quality is good? Different models have different requirements on mesh quality, and the comments by the CFX solver are just a guide.

Skewness is lower than 0.61. Mesh quality is good. right?

-Maxim- · May 12, 2016, 10:53

Quote:

Originally Posted by aja1345

Skewness is lower than 0.61. Mesh quality is good. right?

skewness is not the only factor for a good mesh.
maybe you can post pictures of your mesh so that we can have a look.

aja1345 · May 12, 2016, 12:47

Quote:

Originally Posted by -Maxim-

skewness is not the only factor for a good mesh.
maybe you can post pictures of your mesh so that we can have a look.

aja1345 · May 12, 2016, 13:06

Why are removed pictures????!!!!!!!!!!!

aja1345 · May 12, 2016, 13:09

Mesh........

aja1345 · May 12, 2016, 13:10

Mesh........

aja1345 · May 12, 2016, 13:10

Mesh.........

aja1345 · May 12, 2016, 13:14

I used boundary layer option in Gambit. ...

I don't have set condition...that is default...Is this OK?

...

I am grateful that guide about these cases.

Thanks.

ghorrocks · May 12, 2016, 19:47

A little distance away from the airfoil your elements have far too big aspect ratio. At the foil surface you have a small aspect ratio (which is correct, you want to model the boundary layer), but further away your aspect ratio looks to be about 10 and that is far too large.

aja1345 · May 13, 2016, 05:33

Convergence became better in high angle of attack. I already had divergence in high angles of attack... thanks.

But the following convergence is not good. Why? Is there any problem?

Thanks.

aja1345 · May 13, 2016, 05:42

Torque monioter:

aja1345 · May 13, 2016, 05:45

-Maxim- · May 13, 2016, 05:52

looks good. What is your convergence criteria? You are well on the way to RMS < 10^-5

aja1345 · May 13, 2016, 06:05

Quote:

Originally Posted by -Maxim-

looks good. What is your convergence criteria? You are well on the way to RMS < 10^-5

Thanks.

My convergence criteria is 10^-5.

May 8, 2016, 04:59	How to setup initail conditions in steady solution	#1
aja1345 Senior Member Aja Join Date: Nov 2013 Posts: 496 Rep Power: 15	Hi, I am simulating a trbine and I have problem in running after stall... In fact, Residuals have fluctuation after stall because solution is steady and steady solution fails after stall because there are massive separations. But i want to have better convergence after stall for having better results. I think that this convergence is related to initial guess but i am new in CFX and i don't know that how to set up that. I am grateful that guide me about these cases. Now, Have you any idea and suggestions? Thanks. Best. Aja Last edited by aja1345; May 13, 2016 at 15:50.

May 8, 2016, 08:07		#4
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,854 Rep Power: 144	The FAQs are a few years old. Those sections in the documentation are probably a different section number now. But hopefully they should not be too far away so you should find them quickly enough. If reducing time scale improves convergence this suggests you have significant numerical instability in your simulation, and the normal cause of that is mesh quality. So if you improve your mesh quality it is likely your convergence will significantly improve. aja1345 likes this.

May 8, 2016, 20:07		#6
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,854 Rep Power: 144	How do you know the mesh quality is good? Different models have different requirements on mesh quality, and the comments by the CFX solver are just a guide. As I stated in post #2, this probably is a LES class model and those models are inherently transient. So that means you will not be able to get good results running them steady state. aja1345 likes this.

May 12, 2016, 13:09		#11
aja1345 Senior Member Aja Join Date: Nov 2013 Posts: 496 Rep Power: 15	Mesh........ Last edited by aja1345; May 15, 2016 at 13:43.

May 12, 2016, 13:10		#12
aja1345 Senior Member Aja Join Date: Nov 2013 Posts: 496 Rep Power: 15	Mesh........ Last edited by aja1345; May 15, 2016 at 13:42.

May 8, 2016, 06:36		#2
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,854 Rep Power: 144	This FAQ covers this topic: http://www.cfd-online.com/Wiki/Ansys...gence_criteria But stall is usually characterised by large scale transient turbulent structures which cannot be modelled steady state (hence your steady state model is not converging). You might be forced to use a LES class model to handle this, and that means running transient. LES class model means LES, SAS, DES or one of that class of turbulent model. aja1345 likes this.

May 12, 2016, 13:06		#10
aja1345 Senior Member Aja Join Date: Nov 2013 Posts: 496 Rep Power: 15	Why are removed pictures????!!!!!!!!!!!

May 12, 2016, 13:10		#13
aja1345 Senior Member Aja Join Date: Nov 2013 Posts: 496 Rep Power: 15	Mesh......... Last edited by aja1345; May 15, 2016 at 13:42.

May 12, 2016, 13:14		#14
aja1345 Senior Member Aja Join Date: Nov 2013 Posts: 496 Rep Power: 15	I used boundary layer option in Gambit. ... I don't have set condition...that is default...Is this OK? ... I am grateful that guide about these cases. Thanks. Last edited by aja1345; May 13, 2016 at 10:01.

May 12, 2016, 19:47		#15
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,854 Rep Power: 144	A little distance away from the airfoil your elements have far too big aspect ratio. At the foil surface you have a small aspect ratio (which is correct, you want to model the boundary layer), but further away your aspect ratio looks to be about 10 and that is far too large. aja1345 likes this.

May 13, 2016, 05:33		#16
aja1345 Senior Member Aja Join Date: Nov 2013 Posts: 496 Rep Power: 15	Convergence became better in high angle of attack. I already had divergence in high angles of attack... thanks. But the following convergence is not good. Why? Is there any problem? Thanks. Last edited by aja1345; June 4, 2016 at 09:32.

May 13, 2016, 05:42		#17
aja1345 Senior Member Aja Join Date: Nov 2013 Posts: 496 Rep Power: 15	Torque monioter:

May 13, 2016, 05:45		#18
aja1345 Senior Member Aja Join Date: Nov 2013 Posts: 496 Rep Power: 15

May 13, 2016, 05:52		#19
-Maxim- Senior Member Maxim Join Date: Aug 2015 Location: Germany Posts: 413 Rep Power: 13	looks good. What is your convergence criteria? You are well on the way to RMS < 10^-5 aja1345 likes this.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Unsteady simulation with steady solution as initial value	kiddmax	OpenFOAM Running, Solving & CFD	8	August 20, 2015 06:12
is it possible to predict how long it takes to reach steady state solution in unstead	Alimohamadi_nasr	CFX	4	November 11, 2013 07:11
Transient Solution looks like Steady State	ljwnow	FLUENT	0	March 26, 2012 02:54
Steady solution from Transient simulations	wawa	FLUENT	2	November 9, 2010 18:44
About the difference between steady and unsteady problems	Lisa	Main CFD Forum	11	July 5, 2000 15:37