CFX vs. Fluent Results

n.sharifi · October 31, 2010, 08:31

hello everybody.
i have a problem with CFX results compared with the same model Fluent results.
this case is a mixed regime of supersonic and subsonic compressible flow, but the mass flux of each flow doesn't equal to the same flux of alternative solver
it may be useful to know the main model solved by Density-Based Fluent Solver, the maximum mach number is about 4.5 and the mesh size is equal for both solvers.
is this problem because of pressure-based CFX Solver?

ghorrocks · October 31, 2010, 18:11

Both CFX and Fluent should be able to do Mach 4.5 flows accurately. There is no inherent accuracy problem with the pressure based CFX solver.

n.sharifi · November 1, 2010, 06:03

1) please clarify for me is CFX a pressure based solver or not?
2) if your answer is true then the difference is acceptable or not?
3) i check all of my problem setup in both solver and i have still non-similar results, what is your recommandation?

ghorrocks · November 1, 2010, 06:17

1) CFX is pressure based - read the documentation.
2) That is up to you. How accurate do you want your results? If the level of accuracy you already have is OK then why worry? If you need better accuracy then you had better do something about it.
3) There is no inherent problem with modelling this sort of flow in CFX and I dare say Fluent either. The difference is caused by your models. This FAQ may be of assistance http://www.cfd-online.com/Wiki/Ansys..._inaccurate.3F

michael_owen · November 1, 2010, 10:31

How different are the results?

n.sharifi · November 3, 2010, 02:04

Hi, I appreciate You for your attention. The results are in the attachment. It maybe useful to declare that the suction flow is subsonic and has a great importance in my study. this difference of suction mass flux is not acceptable at all (Fluent is about 4.7 kg/s but CFX is about 5.6 kg/s for models: 2,3,4). if you want ican tell you the other properties of flow field such as: Mach, Pr, T, Gamma and so on in both solver results. Thanks

ghorrocks · November 3, 2010, 05:35

The results don't look mesh independant to me yet. Have a look here: http://www.cfd-online.com/Wiki/Ansys..._inaccurate.3F and here: http//:journaltool.asme.org/Templates/JFENumAccuracy.pdf for some ideas about mesh refinement and other accuracy issues.

CFX can do 2D models. That might run a bit quicker for you.

Look in the documentation of Fluent and CFX about how they implement the boundary conditions. Especially for high speed compressible flows like this some subtle differences in approach can lead to significant differences.

n.sharifi · November 3, 2010, 07:45

thanks again for your quick and persuasive answers. I'd investigated my models again and here is some points: 1) I have adapted the 2D-Mesh two and three times, but the results didn't change significantly. so it must be enough to deduce that this model is mesh independent in the first level of adaption. (for 2D-Axisymmetric FLUENT) OK? 2) if I adapt the 3D meshes again, the problem would be very hard to solve by my system configuration. so i thought the first level of adaption is enough. but I agree with you that this matter doesn't guarantee the mesh in-dependency of my problem in this level. 3)please clarify "the implementation of the boundary conditions".Do You mean Turbulence Model,Pressure,Temperature and Type of inlet or opening boundaries or the other word that i didn't concern.

ghorrocks · November 3, 2010, 17:40

Mesh sensitivity - your 2D models will be good for this. If you have run Fluent to what looks like mesh independance then I would just use these settings in 3D. Likewise I would check that CFX is mesh independant on a 2D model (or at least a thin slice/wedge) and just use that result in 3D. But based on your comments it looks like mesh sensitivity should be OK.

BC implementation - read the theory documentation about inlets, outlets and pressure boundaries. You will find there is assumptions about variable gradients, initial values etc etc which will probably be different. You will have to look up the exact BC you are using to find out how it is implemented. If you want the two of them to be the same you will probably have to choose some non-default boundary options.

October 31, 2010, 08:31	CFX vs. Fluent Results	#1
n.sharifi New Member Navid Sharifi Join Date: Oct 2010 Location: Iran, Tehran Posts: 10 Rep Power: 16	hello everybody. i have a problem with CFX results compared with the same model Fluent results. this case is a mixed regime of supersonic and subsonic compressible flow, but the mass flux of each flow doesn't equal to the same flux of alternative solver it may be useful to know the main model solved by Density-Based Fluent Solver, the maximum mach number is about 4.5 and the mesh size is equal for both solvers. is this problem because of pressure-based CFX Solver?

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
CFX vs. FLUENT in multiphase models	Hossam	CFX	1	November 3, 2010 07:57
Gradient Discretization CFX vs Fluent	Scott Nordsen	CFX	1	December 2, 2009 17:46
Reading CFX-CFD results without Ansys CFX	JAY	ANSYS	2	July 7, 2009 17:48
Import results to CFX post	MatjazR	CFX	1	October 17, 2005 10:07
Switching from Fluent to CFX	Terri Novotny	CFX	1	December 7, 2004 22:59

October 31, 2010, 18:11		#2
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,854 Rep Power: 144	Both CFX and Fluent should be able to do Mach 4.5 flows accurately. There is no inherent accuracy problem with the pressure based CFX solver.

November 1, 2010, 06:03		#3
n.sharifi New Member Navid Sharifi Join Date: Oct 2010 Location: Iran, Tehran Posts: 10 Rep Power: 16	1) please clarify for me is CFX a pressure based solver or not? 2) if your answer is true then the difference is acceptable or not? 3) i check all of my problem setup in both solver and i have still non-similar results, what is your recommandation?

November 1, 2010, 06:17		#4
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,854 Rep Power: 144	1) CFX is pressure based - read the documentation. 2) That is up to you. How accurate do you want your results? If the level of accuracy you already have is OK then why worry? If you need better accuracy then you had better do something about it. 3) There is no inherent problem with modelling this sort of flow in CFX and I dare say Fluent either. The difference is caused by your models. This FAQ may be of assistance http://www.cfd-online.com/Wiki/Ansys..._inaccurate.3F

November 1, 2010, 10:31		#5
michael_owen Senior Member Michael P. Owen Join Date: Mar 2009 Posts: 196 Rep Power: 17	How different are the results?

November 3, 2010, 05:35		#7
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,854 Rep Power: 144	The results don't look mesh independant to me yet. Have a look here: http://www.cfd-online.com/Wiki/Ansys..._inaccurate.3F and here: http//:journaltool.asme.org/Templates/JFENumAccuracy.pdf for some ideas about mesh refinement and other accuracy issues. CFX can do 2D models. That might run a bit quicker for you. Look in the documentation of Fluent and CFX about how they implement the boundary conditions. Especially for high speed compressible flows like this some subtle differences in approach can lead to significant differences.

November 3, 2010, 07:45		#8
n.sharifi New Member Navid Sharifi Join Date: Oct 2010 Location: Iran, Tehran Posts: 10 Rep Power: 16	thanks again for your quick and persuasive answers. I'd investigated my models again and here is some points: 1) I have adapted the 2D-Mesh two and three times, but the results didn't change significantly. so it must be enough to deduce that this model is mesh independent in the first level of adaption. (for 2D-Axisymmetric FLUENT) OK? 2) if I adapt the 3D meshes again, the problem would be very hard to solve by my system configuration. so i thought the first level of adaption is enough. but I agree with you that this matter doesn't guarantee the mesh in-dependency of my problem in this level. 3)please clarify "the implementation of the boundary conditions".Do You mean Turbulence Model,Pressure,Temperature and Type of inlet or opening boundaries or the other word that i didn't concern.

November 3, 2010, 17:40		#9
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,854 Rep Power: 144	Mesh sensitivity - your 2D models will be good for this. If you have run Fluent to what looks like mesh independance then I would just use these settings in 3D. Likewise I would check that CFX is mesh independant on a 2D model (or at least a thin slice/wedge) and just use that result in 3D. But based on your comments it looks like mesh sensitivity should be OK. BC implementation - read the theory documentation about inlets, outlets and pressure boundaries. You will find there is assumptions about variable gradients, initial values etc etc which will probably be different. You will have to look up the exact BC you are using to find out how it is implemented. If you want the two of them to be the same you will probably have to choose some non-default boundary options.