Jumpy Wall Shear Behavior

Leifheit · October 27, 2014, 08:58

Hello,

when I took a look at some of my results today I noticed a rather strange wall shear behavior. A screenshot is attached.
Has anyone else encountered such 'spiky' wall shear behavior?
Velocity and other stuff is fine.

Leif

ghorrocks · October 27, 2014, 16:23

I do not know what you are graphing and I do not know what spike you are concerned about. Please explain what you are doing.

Leifheit · October 29, 2014, 07:25

Sorry about that. I am graphing the wall shear in streamwise direction over the span of an airfoil.
The spikes I am talking about are those in the graph - as far as I know it should be a smooth curve.
The case has reached convergence and other stuff, for example the velocity field around the airfoil, look just fine.

Turbulence Model used was SST Gamma Theta for Transition.

Leifheit · October 29, 2014, 08:43

I also made some more pictures ... maybe those help understanding the problem ... I am still kinda new to cfd stuff so I might give too little / wrong information - if thats the case please tell me what you need!

ghorrocks · October 29, 2014, 17:44

Which spike don't you like?

Don't forget that you will have very sharp pressure gradients near the leading and trailing edges. Also at a separation or a laminar to turbulent transition you will also get big pressure gradients. So sharp pressure gradients can be real.

Leifheit · October 30, 2014, 05:03

Hey,

thanks for the effort you put in this!
Its not a single spike I dont like - it is the „jumpy behavior“ ... sorry dont know how to put this in better words ... the graphs should be more smooth I think.
I found a graph of the drag coefficient over the span of an airfoil at multiple RE numbers - the behavior of the graphs should be the same as wall shear in x direction.
Apparently the calculations were also made with CFX so it should be possible to get results that are of similar quality.
The magnitude of my results is just fine (in terms of drag and lift coefficient etc) - it just would be nice to have good looking results since I want to show the jump from laminar to turbulent behavior.

ghorrocks · October 30, 2014, 05:29

This is sounding like a FAQ now: http://www.cfd-online.com/Wiki/Ansys..._inaccurate.3F

Pay particular attention to mesh resolution. Jumpy curves like what you see are commonly caused by too coarse meshes.

Leifheit · October 30, 2014, 08:16

Unfortunately I do no think this is the case ... I did a mesh study on this considering the parameters

Y+
Streamwise grid refinement
Expansion Rate of Cells

with no change in the behavior.

I also tried doing multiple transition iterations - thats when I ran out of ideas

Just looked into it again ... the velocity gradient (perpendicular to airfoil surface) shows the same behaviour ... which makes sense since

... cant really figure out why though ... I will try to run the simulation with a different turbulence model and see what happens.

ghorrocks · October 30, 2014, 16:53

Another issue to look into is how well the curvature is resolved. The solid modelling may have facetted the face.

Leifheit · October 31, 2014, 05:38

Sorry dont really understand your advice....
how well curvature is resolved = how many samples cfxpost takes ?

ghorrocks · November 1, 2014, 04:38

No. The solid modelling package (DesignModeller in particular) approximates curved surfaces with facets. For most applications the approximation is good and no problems occur. But for simulations which rely on very accurate definition of curvature (airfoil modelling is a good example) this does cause problems sometimes.

If you zoom into your geometry on designmodeller you will see it is facetted. There is a setting in DM to set the geometry tolerance to reduce this.

October 29, 2014, 07:25		#3
Leifheit New Member Join Date: Jul 2013 Posts: 21 Rep Power: 12	Sorry about that. I am graphing the wall shear in streamwise direction over the span of an airfoil. The spikes I am talking about are those in the graph - as far as I know it should be a smooth curve. The case has reached convergence and other stuff, for example the velocity field around the airfoil, look just fine. Turbulence Model used was SST Gamma Theta for Transition. Last edited by Leifheit; October 29, 2014 at 08:41.

Thread Tools	Search this Thread
Show Printable Version Email this Page	Search this Thread: Advanced Search
Display Modes
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
What does “Mean Wall Shear Stress” mean?	dfytanid	FLUENT	4	November 6, 2016 20:19
how to resolve high gradients of wall shear?	Agad15	FLUENT	0	June 9, 2014 23:59
Wall shear stress distributions	en51nm	STAR-CD	2	January 28, 2010 04:42
Problems with wall shear stress.	xiexiehezuo	FLUENT	1	September 22, 2009 12:26
Macro to set Wall Shear Stress	Satish	FLUENT	4	November 26, 2003 14:46

October 27, 2014, 16:23		#2
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,690 Rep Power: 143	I do not know what you are graphing and I do not know what spike you are concerned about. Please explain what you are doing.

October 29, 2014, 17:44		#5
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,690 Rep Power: 143	Which spike don't you like? Don't forget that you will have very sharp pressure gradients near the leading and trailing edges. Also at a separation or a laminar to turbulent transition you will also get big pressure gradients. So sharp pressure gradients can be real.

October 30, 2014, 05:29		#7
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,690 Rep Power: 143	This is sounding like a FAQ now: http://www.cfd-online.com/Wiki/Ansys..._inaccurate.3F Pay particular attention to mesh resolution. Jumpy curves like what you see are commonly caused by too coarse meshes.

October 30, 2014, 08:16		#8
Leifheit New Member Join Date: Jul 2013 Posts: 21 Rep Power: 12	Unfortunately I do no think this is the case ... I did a mesh study on this considering the parameters Y+ Streamwise grid refinement Expansion Rate of Cells with no change in the behavior. I also tried doing multiple transition iterations - thats when I ran out of ideas Just looked into it again ... the velocity gradient (perpendicular to airfoil surface) shows the same behaviour ... which makes sense since ... cant really figure out why though ... I will try to run the simulation with a different turbulence model and see what happens.

October 30, 2014, 16:53		#9
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,690 Rep Power: 143	Another issue to look into is how well the curvature is resolved. The solid modelling may have facetted the face.

October 31, 2014, 05:38		#10
Leifheit New Member Join Date: Jul 2013 Posts: 21 Rep Power: 12	Sorry dont really understand your advice.... how well curvature is resolved = how many samples cfxpost takes ?

November 1, 2014, 04:38		#11
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,690 Rep Power: 143	No. The solid modelling package (DesignModeller in particular) approximates curved surfaces with facets. For most applications the approximation is good and no problems occur. But for simulations which rely on very accurate definition of curvature (airfoil modelling is a good example) this does cause problems sometimes. If you zoom into your geometry on designmodeller you will see it is facetted. There is a setting in DM to set the geometry tolerance to reduce this.