Y+ question

Tom · June 8, 2009, 01:53

I am having a very weird problem. I have read that a Y+of 1 should bring my lift and drag values closer to experimental values. However, it is not in my case. As the Y+ decreases, the lift coefficient obtained is quite good and drag is higher than the values in the journal im studying. As the Y+ increases to about 30 the drag coefficient is quite good and lift coefficient suffers. So im trying values in the buffer region to minimise the error. But i am not getting any results at Angle of attack of 10. The solution does not converge. Same for Y+ values of 30.

Anyone has any tips for guiding me on adjusting the values in my SST model. The turbulence intensity im using is less than 0.1% so im using values between 0.03 to 0.08%. Length scale auto, not sure how small of a length scale to use.

Thanks.

ghorrocks · June 8, 2009, 06:45

Hi,

Is turbulent transition important? Often for accurate drag numbers it is. This model requires the use of a Y+=1 grid as you can't use wall functions with that model.

There is also a heap of other potential problems... Mesh sensitivity and quality, boundary proximity, convergence, transient flow features etc etc

Glenn Horrocks

Tom · June 8, 2009, 10:12

Transition is important. As i need that info. Yea i did it with a Y+1 Grid and it gives a high drag result at low angle of attack for instance 0 degree. But at high angle 10 degree its gives me a good result.

This makes me wonder if the information/experiment result in the journal is accurate.
Was wondering if getting drag coefficient values from wake surveys in the experiment data i have be the same as the data im getting for my drag in CFX, I obtain my drag coefficient using force_x function.

ghorrocks · June 8, 2009, 18:53

Hi,

The journal article should have some form of error estimate. My understanding is the wake survey method is not very accurate but people more experienced in the field may have more detailed comments.

If you are using the turbulence transition model then inlet turbulence levels and surface roughness become critical. Are you sure you have them correct?

Glenn Horrocks

Tom · June 8, 2009, 21:01

Yea tried playing around with the inlet turbulence level till it becomes quite low about 0.0001 with a small length scale. Still it has quite a high drag with SST. Guess i will use another model with scalable wall and compare my results Thanks alot.

ghorrocks · June 9, 2009, 07:57

Hi,

You are unlikely to do much better than SST with the turbulence transition model for aerofoil flows unless you start looking towards DES or LES. If a simple two-equation turbulence model does better than SST it is probably just luck.

I suggest you look at general simulation accuracy issues - mesh sensitivity, mesh quality, inlet and outlet proximity, surface roughness etc etc.

And of course are you sure the results you are comparing to are correct?

Glenn Horrocks

paulo · June 9, 2009, 09:00

Hi to all,

Since you are not sure of the paper's results, you could start validating you problem with the classical cylinder, which is very well documented (rotating cylinder for lift predictions).

At 10 degree AoA, you may have vortex shedding, what won't let your simulation to converge, and you will have to use a transient simulation.

Hope that helps anyway,

Best Regards,

Paulo Rocha

Sudharshani · October 11, 2009, 06:09

hai....
i am new to CFX.. modeling a supersonic CD nozzle...with an exit Mach number just under<2(keeping in mind to match nozzle exit outlet pressure to ambient at certain altitude....my case at 60,000ft Pambient=0.043atm)
maintain a NPR 8:1 to generate supersonic flow down stream

Q(1) if iam using SST what Y+ value should i use?

Q(2) thinking of two options to set up BC(stuck here)
option 1: if i set "inlet as static pressure"/0.344(NPR 8:1) and can i use outlet as 'static pressure" of 0.043 atm to match ambient?
option 2: if i set "inlet as static pressure/0.344"...and "outlet as opening"...then how can i match the exit presure to ambient?

some help would be appreciated
Sudhar

Quote:

Originally Posted by ghorrocks

Hi,

Is turbulent transition important? Often for accurate drag numbers it is. This model requires the use of a Y+=1 grid as you can't use wall functions with that model.

There is also a heap of other potential problems... Mesh sensitivity and quality, boundary proximity, convergence, transient flow features etc etc

Glenn Horrocks

ghorrocks · October 11, 2009, 06:24

I assume for supersonic flows you are not interested in turbulence transition and suck things. In fact I don't think the CFX turbulence transition model is calibrated for supersonic flows anyway so I would not recommend using it. That means you are probably OK with a more normal wall function grid. In this case you should aim to have y+>20 or so. Exactly what you need you will have to determine with a sensitivity analysis.

Set the reference pressure to the ambient pressure and use zero relative pressure at the outlet.

June 8, 2009, 01:53	Y+ question	#1
Tom New Member Join Date: Jun 2009 Posts: 14 Rep Power: 16	I am having a very weird problem. I have read that a Y+of 1 should bring my lift and drag values closer to experimental values. However, it is not in my case. As the Y+ decreases, the lift coefficient obtained is quite good and drag is higher than the values in the journal im studying. As the Y+ increases to about 30 the drag coefficient is quite good and lift coefficient suffers. So im trying values in the buffer region to minimise the error. But i am not getting any results at Angle of attack of 10. The solution does not converge. Same for Y+ values of 30. Anyone has any tips for guiding me on adjusting the values in my SST model. The turbulence intensity im using is less than 0.1% so im using values between 0.03 to 0.08%. Length scale auto, not sure how small of a length scale to use. Thanks. Last edited by Tom; June 8, 2009 at 02:28.

June 9, 2009, 09:00	My two cents	#7
paulo Member Paulo Alexandre Costa Rocha Join Date: Mar 2009 Posts: 71 Rep Power: 17	Hi to all, Since you are not sure of the paper's results, you could start validating you problem with the classical cylinder, which is very well documented (rotating cylinder for lift predictions). At 10 degree AoA, you may have vortex shedding, what won't let your simulation to converge, and you will have to use a transient simulation. Hope that helps anyway, Best Regards, Paulo Rocha

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
Unanswered question	niklas	OpenFOAM	2	July 31, 2013 16:03
Question concerning flow3D	Julian Gacek	Main CFD Forum	0	October 1, 2007 16:00
Poisson Solver question	Suresh	Main CFD Forum	3	August 12, 2005 04:37
CHANNEL FLOW: a question and a request	Carlos	Main CFD Forum	4	August 23, 2002 05:55
question	K.L.Huang	Siemens	1	March 29, 2000 04:57

June 8, 2009, 06:45		#2
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,700 Rep Power: 143	Hi, Is turbulent transition important? Often for accurate drag numbers it is. This model requires the use of a Y+=1 grid as you can't use wall functions with that model. There is also a heap of other potential problems... Mesh sensitivity and quality, boundary proximity, convergence, transient flow features etc etc Glenn Horrocks

June 8, 2009, 10:12		#3
Tom New Member Join Date: Jun 2009 Posts: 14 Rep Power: 16	Transition is important. As i need that info. Yea i did it with a Y+1 Grid and it gives a high drag result at low angle of attack for instance 0 degree. But at high angle 10 degree its gives me a good result. This makes me wonder if the information/experiment result in the journal is accurate. Was wondering if getting drag coefficient values from wake surveys in the experiment data i have be the same as the data im getting for my drag in CFX, I obtain my drag coefficient using force_x function.

June 8, 2009, 18:53		#4
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,700 Rep Power: 143	Hi, The journal article should have some form of error estimate. My understanding is the wake survey method is not very accurate but people more experienced in the field may have more detailed comments. If you are using the turbulence transition model then inlet turbulence levels and surface roughness become critical. Are you sure you have them correct? Glenn Horrocks

June 8, 2009, 21:01		#5
Tom New Member Join Date: Jun 2009 Posts: 14 Rep Power: 16	Yea tried playing around with the inlet turbulence level till it becomes quite low about 0.0001 with a small length scale. Still it has quite a high drag with SST. Guess i will use another model with scalable wall and compare my results Thanks alot.

June 9, 2009, 07:57		#6
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,700 Rep Power: 143	Hi, You are unlikely to do much better than SST with the turbulence transition model for aerofoil flows unless you start looking towards DES or LES. If a simple two-equation turbulence model does better than SST it is probably just luck. I suggest you look at general simulation accuracy issues - mesh sensitivity, mesh quality, inlet and outlet proximity, surface roughness etc etc. And of course are you sure the results you are comparing to are correct? Glenn Horrocks

October 11, 2009, 06:24		#9
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,700 Rep Power: 143	I assume for supersonic flows you are not interested in turbulence transition and suck things. In fact I don't think the CFX turbulence transition model is calibrated for supersonic flows anyway so I would not recommend using it. That means you are probably OK with a more normal wall function grid. In this case you should aim to have y+>20 or so. Exactly what you need you will have to determine with a sensitivity analysis. Set the reference pressure to the ambient pressure and use zero relative pressure at the outlet.