[Jonathan]

Hi,

I have been working on generating some turbine geometry in Gambit which is part of a larger batch automated routine, and therefore have used vertex input files for the blade / volume geometry and a gambit journal to create the model ...

My problem is, based on tests, it seems i need to use a hex mesh to capture the flow dynamics correctly, and so need to decompose the flow volume. was wondering if anyone had any ideas of the best way to try to decompose the volume i.e. o / c / h grids as i have tried a heap of methods but dont seem to be able to get a decent mesh! part of the problem seems to be that the blade is highly twisted and coopering from hub to tip ends up with a heap of inverted cells ...

any ideas the best way to decompose the volume??
thanks a lot!!

ps attached are a couple screen shots of the geometry at the moment...

[PSYMN]

Try shifting the periodic... Add a quarter ogrid on one side and another on the other side to match everything out... I have added a couple example sketches. Please excuse that the first one isn't quite right because the lines don't cross the airfoil. The point is just that you need the faces on the periodic sides to be periodic, but you don't need the mesh to cross directly between periodic nodes. You can shift them. In the second image, I used two quarter Ogrids on the bottom right, one for the shift on the upper left and the other to make up for the collapsed trailing edge...

I have other posts about this. Search for "shifted periodic".

Simon

[la7low]

Wow, nice geometry! Probably the blade sections at higher radii do not "absorb" much work, right?
I think it is hopeless to do the mesh in Gambit, use Icem and consider topoligies like the shifted peridic or J-topology, which connects the blade with the inlet/outlet (see pictures).
Also you could use TurboGrid help to get inspiration and see other topologies...

Could you send pictures if you have managed to generate the mesh in GAMBIT? I would see it with my eyes if that was possible...

[Jonathan]

hi guys,

thanks a lot for your replies ... had begun to give up on this!!

Simon, havent come across the shifted periodic idea before - that seems like a really good idea - thanks a lot, much appreciated. will give it a try and see.

la7low - yup, have found gambit pretty tough to deal with this geometry but its all i have at the moment, plus i am using as part of an automated routine so need the scripting functionality ... blade twist is just to accomodate the change in onset flow angle due to the rotation of the blade. can't remember off hand what the work distribution along the span is, can look it up though - will post a couple pics if i manage to mesh it using a structured hex mesh ...

thanks
Jon