wind machine mesh
 

Hello everybody

I´m getting lot of trouble to mesh a three-bladed wind machine, without nacelle, using gambit

I started to mesh each blade, giving 100 nodes at radial direction and 30 nodes at tip and hub profiles. Then i mesh each blade face using "quad" elements, "map" type. The far field boundary is made of a cylindrical surface. When i try to mesh the 3d volume a message appears to me, saying " initialization failed, pertutrb boundary nodes and try again...check the skewnesses of your face meshes...". I guess the mistake could be based upon as r/R value increases, distance between nodes decreases.

I also try to mesh blades, using "quad/tri" elements, "pave" type, and equal distance between nodes at tip and hub profiles, but the outcome surface was a non-smoothed surface, so it´s not valid for me.

Actually, i don´t know what more can i do, can you please help me giving some tips/ advices, to solve this problem.

Thanks in advance

Are you trying to mesh your domain with hexa or tetra?
Usually that kind of message comes from a small edge, small angle etc...
Once you have meshed your surfaces, check the 2d-mesh quality and see worst elements (Those cells may give you some trouble)

Thanks for your help Max. I have already done some meshes made of tetra, hybrid, hexa. But i guess that the mistake is on meshed faces at blades. What can i do to mesh blade faces on a pave distribution, and get a smooth meshed surface all along the blade? Because i think it´s not possible to mesh that kind of surfaces using a map distribution, it would be good to keep distance between nodes at tip and hub.

Thanks a lot

Post a picture

Picture please.
If u have the surface meshed and fail to mesh to volume, i think u should give up the surface mesh.

[IMG]file:///C:/Documents%20and%20Settings/miguel%20angel%20soto/Escritorio/Dibujo.JPG[/IMG]

In your case you only have one volume. And you won't be able to mesh it with hexa on the fly.
That means, that your surface mesh with pave or hexa won't bring you anything.
I would advise you to mesh your blades surfaces with tetra, AND a element size at least 10 times smaller that the actual one. (you aren't able to mesh your volume because of a too coarse surface mesh).
Then apply a size function (type meshed) and choose your blade surfaces as source.

Max was right. its the surface meshing limited ur volume meshing.
If u don't need different volume mesh size in your volume, you should give up meshing the surface. I guess if u directly goes for volume mesh, ur meshing should be done easily.

I have just finished trying to mesh the volume by using smaller element sizes, about three times smaller than the previous one, but still the same message appears, and it doesn´t looks like to be the right direction. If i do it, meshing with ten times smaller elements i will have some trouble to solve the case in fluent, due to iteration time.

The airfoil profile is made up of three curves, upper surface, lower surface, both join by a third line, in the same way as a gourney flap,

your mesh is still too coarse.
Your trailing edge enforces you to refine your mesh.
It should be the smaller edge from your model, and it is quite important (you cannot merge it).
Thus, I assume you want to resolve the wake from the airfoil.
Check the length from the trailing edge and give it as reference element size of your surface mesh. But it will be still too coarse, because it means you will resolve the wake with one element.
Anyway it should solve your mesh problem.
Then iterate, and try to refine with 8 element along the trailing edge.
Solving CFD will always ask time to iterate (don't expect to get your result in 5min, excepts if you have a CRAY or someting like that).
Your model seems to be periodic, you can compute only a slice and therefore refine your mesh

Hi,
looks like an Enercon turbine. We made some windturbine meshes with Gambit.

I think that will never work at that way. You have to cut your big Volume in much more smaller volumes and you should try to mesh only one 120 degree part and copy it.

thanks for your advice, i will try to do it meshing only 120 degrees. In this way the iteration time shall be at least less than computing the full volume.

hi,
no, for resonable results you have to do the full 360 degree, but you can copy and connect the mesh in gambit.

And you should try to use hexaeder, because with tetraeder you will get really a lots of elements.

Here is an older paper, there we used tetraeder meshing in gambit and then convert to polygon-cells in Fluent:
http://www.iop.org/EJ/abstract/1742-6596/75/1/012033/

But I think hexaeder are the best way.

What kind of job are you doing there with the windturbine?