Strange Result

Tremore · March 3, 2013, 12:47

I performed I simulation of a VAWT, using ANSYS CFX with the following data:

BLADES
NACA0018
0,5m coord
Set angle 0°

ROTOR blades
Diameter: 5m

SIMULATION TYPE
quasi 2D (5mm thin regions)
TSR: 3,8
INLET wind speed: 11 m/S
INTERFACE: rotating-static: STAGE.

MESH
300 000 elements: fitted near the blades(expecially in TE) and near the interface.

Everything look go in the right way , the solution goes at convergence in about 1500 interations but the results seems be completely wrongs.
if you Look at the velocity vectors:
Velocity vector.jpg
https://docs.google.com/file/d/0B81t...it?usp=sharing

and the pressure have a big step in the trasition beetween static and rotating domain:
pressure.jpg
https://docs.google.com/file/d/0B81t...it?usp=sharing

Someone have some idea why I have this kind of results?

Thank you in advance

ghorrocks · March 3, 2013, 18:05

Two initial comments:
* Please try to avoid jargon and acronyms. I have been on this forum a while so I know VAWT means vertical axis wind turbine, but many others may not.
* This is a general FAQ on accuracy: http://www.cfd-online.com/Wiki/Ansys..._inaccurate.3F

In this case I would be looking at mesh quality, and resolution.

Tremore · March 4, 2013, 07:10

Sorry for the acronimous.

I think is not a mesh problem.

mesh1.jpg mesh2.jpg mesh3.jpg mesh4.png mesh5.jpg

I solve the problem using frozen rotor instead stage in the interface option
Now the pressure distribution look like reasonable and also the totale torque of the three blade of 500Nm/m (for blade meter) very near at the hand calculated value.
The streamline are always completely wrong but probably this one is the way of CFX to illustre the streamlines on an rotating domain.

https://docs.google.com/file/d/0B81t...it?usp=sharing

https://docs.google.com/file/d/0B81t...it?usp=sharing

After the first simulation I tried to carry our a 360° simulation using a transition simulation:
0,75s totale time
0,02083 step time
I change the interface option again from rotor frozen to stage and use for initial condition the result of the previus simulation, the solver start normally but after 10 interation I have a overflow error.

Tremore · March 4, 2013, 09:19

I think I solved the problem, now is working:
I reduce the time step do 0,0075s (every step is solved in one o two interations and then the total time is almost the same) and now I'm at step 60 without any problem
I have forgotten to reduce the recorded timestep and then the solution file dimesion will be very big. It is possible reduce it?

ghorrocks · March 4, 2013, 17:37

Use adaptive timestepping, homing in on 3-5 coeff loops per iteration, then you know the tiem step is set correctly. And set transient result files every so often (maybe 1s or whatever is appropriate) so you get a manageable result file size.

March 3, 2013, 12:47	VAWT - Strange Result	#1
Tremore New Member Daniel Join Date: Mar 2013 Posts: 5 Rep Power: 13	I performed I simulation of a VAWT, using ANSYS CFX with the following data: BLADES NACA0018 0,5m coord Set angle 0° ROTOR blades Diameter: 5m SIMULATION TYPE quasi 2D (5mm thin regions) TSR: 3,8 INLET wind speed: 11 m/S INTERFACE: rotating-static: STAGE. MESH 300 000 elements: fitted near the blades(expecially in TE) and near the interface. Everything look go in the right way , the solution goes at convergence in about 1500 interations but the results seems be completely wrongs. if you Look at the velocity vectors: Velocity vector.jpg https://docs.google.com/file/d/0B81t...it?usp=sharing and the pressure have a big step in the trasition beetween static and rotating domain: pressure.jpg https://docs.google.com/file/d/0B81t...it?usp=sharing Someone have some idea why I have this kind of results? Thank you in advance Last edited by Tremore; March 3, 2013 at 13:35.

March 4, 2013, 07:10		#3
Tremore New Member Daniel Join Date: Mar 2013 Posts: 5 Rep Power: 13	Sorry for the acronimous. I think is not a mesh problem. mesh1.jpg mesh2.jpg mesh3.jpg mesh4.png mesh5.jpg I solve the problem using frozen rotor instead stage in the interface option Now the pressure distribution look like reasonable and also the totale torque of the three blade of 500Nm/m (for blade meter) very near at the hand calculated value. The streamline are always completely wrong but probably this one is the way of CFX to illustre the streamlines on an rotating domain. https://docs.google.com/file/d/0B81t...it?usp=sharing https://docs.google.com/file/d/0B81t...it?usp=sharing After the first simulation I tried to carry our a 360° simulation using a transition simulation: 0,75s totale time 0,02083 step time I change the interface option again from rotor frozen to stage and use for initial condition the result of the previus simulation, the solver start normally but after 10 interation I have a overflow error. Last edited by Tremore; March 4, 2013 at 08:33.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Strange Nut behaviour with K-OmegaSST	nicolarre	OpenFOAM Running, Solving & CFD	12	March 19, 2019 21:35
the result of the velocity and pressure drop is a little strange.	yuhehuan	FLUENT	0	December 18, 2012 20:48
How to display the whole result of a periodic model via one period result?	dixylo	FLUENT	2	February 23, 2012 05:11
Strange Result for Versteeg Testcase (2D, convection only, steady state)	caramelo	OpenFOAM Running, Solving & CFD	4	August 17, 2011 18:55
Strange Result of turboPassageRotating tutorial.	Ohbuchi	OpenFOAM Running, Solving & CFD	0	July 27, 2011 22:06

March 3, 2013, 18:05		#2
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,871 Rep Power: 144	Two initial comments: * Please try to avoid jargon and acronyms. I have been on this forum a while so I know VAWT means vertical axis wind turbine, but many others may not. * This is a general FAQ on accuracy: http://www.cfd-online.com/Wiki/Ansys..._inaccurate.3F In this case I would be looking at mesh quality, and resolution.

March 4, 2013, 09:19		#4
Tremore New Member Daniel Join Date: Mar 2013 Posts: 5 Rep Power: 13	I think I solved the problem, now is working: I reduce the time step do 0,0075s (every step is solved in one o two interations and then the total time is almost the same) and now I'm at step 60 without any problem I have forgotten to reduce the recorded timestep and then the solution file dimesion will be very big. It is possible reduce it?

March 4, 2013, 17:37		#5
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,871 Rep Power: 144	Use adaptive timestepping, homing in on 3-5 coeff loops per iteration, then you know the tiem step is set correctly. And set transient result files every so often (maybe 1s or whatever is appropriate) so you get a manageable result file size.