fan vs experimental data

ijk · June 1, 2010, 08:03

Hi all,

I am trying to generate a fan curve from experimental data using CFX. I have found that it does not converge steady state so I run it transient.

The fan is an axial fan but I am running in the stalled mixed flow region. I have found that my course mesh using wall functions (yplus100 on blade) is about 15% low in mass flow for a given loading. The steady state is way off.

I have yplus 30 and 1 cases running but they take forever. Perhaps a PHD using CFD was a bad idea.

Are these normal results for CFX? I have some friends who have done similar work with Star CD and they have got transient results within 5%.

I am not 100% that the experimental results are correct as the airflow was measured using anemometers and not a proper venturi setup.

Any advice anyone could give would be much appreciated.

Thanks

ghorrocks · June 1, 2010, 19:46

Getting accurate results in the stalled region is difficult, so you have a difficult project. That means you can expect long run times.

It also means you may not be able to use a turbulence model to get good results, you may be forced to use DES/SAS to get improvement - but that will just increase your run time more.

There are no simple answers in CFD, especially with difficult work like stall modelling.

I would be very surprised that the difference between your results and Star-CD is purely the solver. I would suggest either the user knew what he was doing better or was lucky.

ijk · June 7, 2010, 05:42

Thanks Glenn

June 1, 2010, 08:03	fan vs experimental data	#1
ijk Member Join Date: Apr 2010 Posts: 33 Rep Power: 16	Hi all, I am trying to generate a fan curve from experimental data using CFX. I have found that it does not converge steady state so I run it transient. The fan is an axial fan but I am running in the stalled mixed flow region. I have found that my course mesh using wall functions (yplus100 on blade) is about 15% low in mass flow for a given loading. The steady state is way off. I have yplus 30 and 1 cases running but they take forever. Perhaps a PHD using CFD was a bad idea. Are these normal results for CFX? I have some friends who have done similar work with Star CD and they have got transient results within 5%. I am not 100% that the experimental results are correct as the airflow was measured using anemometers and not a proper venturi setup. Any advice anyone could give would be much appreciated. Thanks

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Tecplot. add experimental data into a mesh	Alejandro	FLUENT	0	December 10, 2008 08:08
How to update polyPatchbs localPoints	liu	OpenFOAM Running, Solving & CFD	6	December 30, 2005 18:27
Looking for the experimental data	COLOPOLO	Main CFD Forum	2	January 29, 2004 00:38
Simulated data vs. experimental data	moiami	FLUENT	0	December 19, 2002 02:03
Experimental data for fence at low Reynolds number	Kim Bindesboll Andersen	Main CFD Forum	1	September 26, 2002 14:54

June 1, 2010, 19:46		#2
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,871 Rep Power: 144	Getting accurate results in the stalled region is difficult, so you have a difficult project. That means you can expect long run times. It also means you may not be able to use a turbulence model to get good results, you may be forced to use DES/SAS to get improvement - but that will just increase your run time more. There are no simple answers in CFD, especially with difficult work like stall modelling. I would be very surprised that the difference between your results and Star-CD is purely the solver. I would suggest either the user knew what he was doing better or was lucky.

June 7, 2010, 05:42		#3
ijk Member Join Date: Apr 2010 Posts: 33 Rep Power: 16	Thanks Glenn