Need help with CFX simulation for airfoil

ghorrocks · May 20, 2009, 21:41

Quote:

Originally Posted by Arti

I did a mesh test for 50K 60K and 80K plus cells of my setup. Am now using the 80K cell as my choice. Am trying to do the lift and drag curves to confirm it with the experimental results i have. I think im getting it. Some settings were slightly off. Will try again. Thanks for the advice.

Hi,

I would be very surprised if you have achieved mesh convergence with a mesh of this size. The differences between your meshes is not enough to get meaningful results. Generally you half the element edge lengths, so you will get around 4 times as many elements per refinement in a 2D simulation or 8 times as many elements in a 3D simulation.

Until you have a mesh converged solution then there is no way your results are accurate.

Glenn Horrocks

Arti · May 21, 2009, 03:41

hmmm Not sure whats wrong with my software.. I can now get a better grid size with the required wall distance than those i got before. But just to ensure im going the right way.
I hope you can tell me that the way of mesh refinement that im doing is correct. The smallest number of element points i have on the airfoil is 60 top 70 bottom without any errors.
So i would have about 59 cells top and 69 cells bottom right? So refinement is done by halving the 59 cells which makes 118 and so on till i have a converged result. Is this what you mean?

ghorrocks · May 21, 2009, 20:23

Hi,

Yes, halving the mesh element edge length is the usual increment for mesh sensitivity studies. Refer to "Computational Fluid Dynamics" by Roache for a detailed discussion on this. It has also been implemented as an editorial policy by Journal of Fluids Engineering, see http://journaltool.asme.org/Template...umAccuracy.pdf

So if your mesh has 60 elements on the top face of the foil, ideally I would then use 120 and 240 and so on. As a minimum (as discussed in the JFE reference) you should increase by a factor of 1.3, so 60, then 80 then 100 but bigger steps are desireable.

Glenn Horrocks

Arti · May 22, 2009, 14:00

ah I get it now. Thanks for the information.

Was wondering does setting a specific boundary condition work for all situations for example in an airfoil simulation can setting the boundary condition for one angle of attack be used for the remaining ones.
It seems like im having problem with some angles even with the same boundary. Am trying to collect data to see if my mesh convergence has reached.

LSC · May 26, 2009, 11:52

Hi,

currently I have managed to get my CL within 7% error and 60% error for CD (more or less expected). This was done with SST k-omega and intermittency-ReTheta transition model at default values. I did a CF plot and found that transition was predicted further upstream when compared with experimental (therefore resulting in large drag error). Are there any best known methods on tuning the transition model coefficients?

ghorrocks · May 26, 2009, 20:02

Hi,

Before you tune the coefficients, are you sure you have the inlet turbulence level correct? Also what about surface roughness - does your model properly account for surface roughness?

Glenn Horrocks

LSC · May 27, 2009, 06:57

Hi, the turbulence intensity was calculated based on the width and height of the wind tunnel geometry given in the journals where the authors did an experiment on the airfoil which I am simulating as well. However, I did the calculations based on a rectangular cross section wheresa the actual wind tunnel cross section is octagonal. In fact, one of my main objective is to look into the effects of surface roughness on airfoil but currently I am in the midst of getting the clean airfoil data first and subsequently add roughness on the surface. Currently looking into tuning the transition model coefficient to reduce the CD error. Please advise

sivarama1 · May 28, 2009, 01:20

Hi all,
yes,i am also doing same type of problem,I create model in I Gambit and meshed,but that is 2d model,how to export to CFX-11.
IS IT POSSIBLE TO EXPORT TO CFX-11(2D GEOMETRY IN GAMBIT)

Arti · May 28, 2009, 04:35

Please read a few post back. About how to simulate a 2D effect for Airfoil. There is NO actual 2D simulation for CFX u will have to extrude it to 3D.

sivarama1 · May 28, 2009, 08:06

I am new in this field,
what i am analyzing
1) first i was taken one airfoil in that i was done 2D mesh in GAMBIT,That i should export to solver means CFX-11,but CFX-11 most i was used in 3D simulations only,2D mesh not imported in cfx-11.
if dont mind your MSN ID,I SHOULD CONTACT.
Thanking you.

Arti · May 28, 2009, 12:39

As i was saying. Extrude (or use the sweep function in Gambit) your 2D mesh into 3D and u would have solve your problem. Cant be simpler than that.

LSC · May 30, 2009, 02:11

Quote:

Originally Posted by ghorrocks

Hi,

Before you tune the coefficients, are you sure you have the inlet turbulence level correct? Also what about surface roughness - does your model properly account for surface roughness?

Glenn Horrocks

Hi Glenn,

you are right! I made a mistake in the inlet turbulence level..The CD and CL error are greatly reduced. Really appreciate your invaluable advice and many many thanks!!

meow55 · July 14, 2009, 14:23

hi there, currently i'm having a problem with my simulation. i had done a simulation with 0 angle of attack, but when my angle of attack increases, the error percentage gets larger n larger. But when i'm doing 0 angle of attack , my errors are less than 1% for both lift n drag. I'm using mesh transformation to perform the angle of attack. Can anyone advise on this issue? thanks

meow55 · July 14, 2009, 14:30

hi there, currently i'm having a problem with my simulation. i had done a simulation with 0 angle of attack, but when my angle of attack increases, the error percentage gets larger n larger. But when i'm doing 0 angle of attack , my errors are less than 1% for both lift n drag. I'm using mesh transformation to perform the angle of attack. Can anyone advise on this issue? thanks

May 21, 2009, 03:41		#42
Arti New Member Join Date: Apr 2009 Posts: 26 Rep Power: 17	hmmm Not sure whats wrong with my software.. I can now get a better grid size with the required wall distance than those i got before. But just to ensure im going the right way. I hope you can tell me that the way of mesh refinement that im doing is correct. The smallest number of element points i have on the airfoil is 60 top 70 bottom without any errors. So i would have about 59 cells top and 69 cells bottom right? So refinement is done by halving the 59 cells which makes 118 and so on till i have a converged result. Is this what you mean? Last edited by Arti; May 21, 2009 at 12:00.

May 28, 2009, 01:20	Cfd 3d airfoil	#48
sivarama1 Member sivaramakrishnaiah Join Date: Mar 2009 Location: india,pondicherry Posts: 76 Rep Power: 17	Hi all, yes,i am also doing same type of problem,I create model in I Gambit and meshed,but that is 2d model,how to export to CFX-11. IS IT POSSIBLE TO EXPORT TO CFX-11(2D GEOMETRY IN GAMBIT) __________________ sivaramakrihnaiah

May 28, 2009, 08:06	2d export to CFX-11	#50
sivarama1 Member sivaramakrishnaiah Join Date: Mar 2009 Location: india,pondicherry Posts: 76 Rep Power: 17	I am new in this field, what i am analyzing 1) first i was taken one airfoil in that i was done 2D mesh in GAMBIT,That i should export to solver means CFX-11,but CFX-11 most i was used in 3D simulations only,2D mesh not imported in cfx-11. if dont mind your MSN ID,I SHOULD CONTACT. Thanking you. __________________ sivaramakrihnaiah

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Pros and Cons for CFX, CFdesign, COMSOL	Val	Main CFD Forum	3	June 10, 2011 03:20
nucleate boiling simulation in CFX	Anil	CFX	3	August 25, 2010 15:18
PhD using CFX	Rui	CFX	9	May 28, 2007 06:59
2D simulation - ICEM meshing for CFX question	Ben Makhal	CFX	5	April 11, 2007 09:44
Simulation of turbine cascade in CFX.	Jonas Pedro Caumo	CFX	0	December 9, 2006 14:54

May 21, 2009, 20:23		#43
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,870 Rep Power: 144	Hi, Yes, halving the mesh element edge length is the usual increment for mesh sensitivity studies. Refer to "Computational Fluid Dynamics" by Roache for a detailed discussion on this. It has also been implemented as an editorial policy by Journal of Fluids Engineering, see http://journaltool.asme.org/Template...umAccuracy.pdf So if your mesh has 60 elements on the top face of the foil, ideally I would then use 120 and 240 and so on. As a minimum (as discussed in the JFE reference) you should increase by a factor of 1.3, so 60, then 80 then 100 but bigger steps are desireable. Glenn Horrocks

May 22, 2009, 14:00		#44
Arti New Member Join Date: Apr 2009 Posts: 26 Rep Power: 17	ah I get it now. Thanks for the information. Was wondering does setting a specific boundary condition work for all situations for example in an airfoil simulation can setting the boundary condition for one angle of attack be used for the remaining ones. It seems like im having problem with some angles even with the same boundary. Am trying to collect data to see if my mesh convergence has reached.

May 26, 2009, 11:52		#45
LSC Member LSC Join Date: May 2009 Posts: 58 Rep Power: 17	Hi, currently I have managed to get my CL within 7% error and 60% error for CD (more or less expected). This was done with SST k-omega and intermittency-ReTheta transition model at default values. I did a CF plot and found that transition was predicted further upstream when compared with experimental (therefore resulting in large drag error). Are there any best known methods on tuning the transition model coefficients?

May 26, 2009, 20:02		#46
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,870 Rep Power: 144	Hi, Before you tune the coefficients, are you sure you have the inlet turbulence level correct? Also what about surface roughness - does your model properly account for surface roughness? Glenn Horrocks

May 27, 2009, 06:57		#47
LSC Member LSC Join Date: May 2009 Posts: 58 Rep Power: 17	Hi, the turbulence intensity was calculated based on the width and height of the wind tunnel geometry given in the journals where the authors did an experiment on the airfoil which I am simulating as well. However, I did the calculations based on a rectangular cross section wheresa the actual wind tunnel cross section is octagonal. In fact, one of my main objective is to look into the effects of surface roughness on airfoil but currently I am in the midst of getting the clean airfoil data first and subsequently add roughness on the surface. Currently looking into tuning the transition model coefficient to reduce the CD error. Please advise

May 28, 2009, 04:35		#49
Arti New Member Join Date: Apr 2009 Posts: 26 Rep Power: 17	Please read a few post back. About how to simulate a 2D effect for Airfoil. There is NO actual 2D simulation for CFX u will have to extrude it to 3D.

May 28, 2009, 12:39		#51
Arti New Member Join Date: Apr 2009 Posts: 26 Rep Power: 17	As i was saying. Extrude (or use the sweep function in Gambit) your 2D mesh into 3D and u would have solve your problem. Cant be simpler than that.

July 14, 2009, 14:23		#53
meow55 New Member Join Date: Jun 2009 Posts: 2 Rep Power: 0	hi there, currently i'm having a problem with my simulation. i had done a simulation with 0 angle of attack, but when my angle of attack increases, the error percentage gets larger n larger. But when i'm doing 0 angle of attack , my errors are less than 1% for both lift n drag. I'm using mesh transformation to perform the angle of attack. Can anyone advise on this issue? thanks

July 14, 2009, 14:30		#54
meow55 New Member Join Date: Jun 2009 Posts: 2 Rep Power: 0	hi there, currently i'm having a problem with my simulation. i had done a simulation with 0 angle of attack, but when my angle of attack increases, the error percentage gets larger n larger. But when i'm doing 0 angle of attack , my errors are less than 1% for both lift n drag. I'm using mesh transformation to perform the angle of attack. Can anyone advise on this issue? thanks