Negative Cp for three straight blades VAWT

Joao Bomfim · December 11, 2018, 11:49

Hi friends,
This is my first post. I am trying to validate an simulation using ansys CFX with FORTRAN routines, but for this case, the routines don't interfere. The simulation is a case transient, of rotor-stator with sliding mesh. In the article, for the TSR = 2.5, cp is approximately 0.14. The mesh in the work don't have many elements (43 524), but have a good quality (the less is 0.67) and is composite by hexaedra structured elements. The cp obtained in the simulation for the same TSR is about -0.016. I tryed increase the number of elements, but I obtained a similar result. The torque in the blades are calculated by CEL, adding for each blade, the torque in extern and intern side. The wind velocity is 8.5 m/s. Anybody have an idea what problem is it? If need an archive, I could provide. Sorry for my bad english. Thank's for the attention!

ghorrocks · December 11, 2018, 16:55

I can see a few problems straight away:

* Your mesh is definitely way too coarse. You will need to refine it a lot. I can tell this by the blocky contour lines on your plots.
* You have a lot of user fortran. Why so many? Can't you replace these functions with CEL?
* How are you defining the time step? A very common error is to use too large a time step. I recommend using dynamic time stepping, homing in on 3-5 coefficient loops per iteration. Make sure the initial time step is close, and make sure the max and min time steps are wide enough you never hit them.
* Why are you doing this as a transient rotor stator simulation? This is very slow. Why can't you do it as a frozen rotor simulation?
* Your residual target is very loose. Are you sure this is adequately converged?
* Why do you have your expert parameter defined? Don't use expert parameters unless you need them and you know what they do.

Joao Bomfim · December 11, 2018, 17:37

Glenn, thanks for your observations. I am simulating now without Fortran routines, for simplify the problem. The routines are needed only when I am working with evolution of turbine's rotation. The timestep is defined for the expression dt=dtheta*(2pi/360 degree)/rotational velocity. I will try simulate the way you recommended me. The configuration was transient rotor-stator because my advisor request me to do. After solving with a mesh more refined, I post the results with new CCL, and ask for help in more anything

. Thanks so much!

ghorrocks · December 11, 2018, 17:45

This simulation will be much faster and easier if you use Frozen Rotor rather than Transient Rotor Stator. So unless you have a good reason to use TRS you should convince your advisor to use Frozen Rotor.

Even better, do a TRS run and compare it to some Frozen Rotor runs. Then you will see the difference for yourself and whether the frozen rotor assumption is appropriate for your case.

December 11, 2018, 16:55		#2
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,705 Rep Power: 143	I can see a few problems straight away: * Your mesh is definitely way too coarse. You will need to refine it a lot. I can tell this by the blocky contour lines on your plots. * You have a lot of user fortran. Why so many? Can't you replace these functions with CEL? * How are you defining the time step? A very common error is to use too large a time step. I recommend using dynamic time stepping, homing in on 3-5 coefficient loops per iteration. Make sure the initial time step is close, and make sure the max and min time steps are wide enough you never hit them. * Why are you doing this as a transient rotor stator simulation? This is very slow. Why can't you do it as a frozen rotor simulation? * Your residual target is very loose. Are you sure this is adequately converged? * Why do you have your expert parameter defined? Don't use expert parameters unless you need them and you know what they do. Joao Bomfim likes this. __________________ Note: I do not answer CFD questions by PM. CFD questions should be posted on the forum.

December 11, 2018, 17:45		#4
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,705 Rep Power: 143	This simulation will be much faster and easier if you use Frozen Rotor rather than Transient Rotor Stator. So unless you have a good reason to use TRS you should convince your advisor to use Frozen Rotor. Even better, do a TRS run and compare it to some Frozen Rotor runs. Then you will see the difference for yourself and whether the frozen rotor assumption is appropriate for your case. Joao Bomfim likes this. __________________ Note: I do not answer CFD questions by PM. CFD questions should be posted on the forum.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
[Other] mesh airfoil NACA0012	anand_30	OpenFOAM Meshing & Mesh Conversion	13	March 7, 2022 17:22
[blockMesh] non-orthogonal faces and incorrect orientation?	nennbs	OpenFOAM Meshing & Mesh Conversion	7	April 17, 2013 05:42
[blockMesh] error message with modeling a cube with a hold at the center	hsingtzu	OpenFOAM Meshing & Mesh Conversion	2	March 14, 2012 09:56
[blockMesh] BlockMesh FOAM warning	gaottino	OpenFOAM Meshing & Mesh Conversion	7	July 19, 2010 14:11
[blockMesh] Axisymmetrical mesh	Rasmus Gjesing (Gjesing)	OpenFOAM Meshing & Mesh Conversion	10	April 2, 2007 14:00

December 11, 2018, 17:37		#3
Joao Bomfim New Member João de Sousa Bomfim Neto Join Date: Oct 2017 Location: Maranguape, Ceará, Brasil Posts: 25 Rep Power: 8	Glenn, thanks for your observations. I am simulating now without Fortran routines, for simplify the problem. The routines are needed only when I am working with evolution of turbine's rotation. The timestep is defined for the expression dt=dtheta*(2pi/360 degree)/rotational velocity. I will try simulate the way you recommended me. The configuration was transient rotor-stator because my advisor request me to do. After solving with a mesh more refined, I post the results with new CCL, and ask for help in more anything . Thanks so much!