CFD Online Discussion Forums

CFD Online Discussion Forums (
-   Siemens (
-   -   centroid problem (

eos December 4, 2006 10:55

centroid problem

I am trying to solve a 3d wing which is moving but when i use the check tool for the mesh i have the following message: *** WARNING #020 *** CELL CENTRE OUTSIDE OF THE CELL FOR CELL NO. 71 and this warning is displayed for many cells (about 200000) my mesh consists of approximately a million elements and created in a commercial mesh generator, Gridgen. does anyone know how to fix this problem? Thanks.. Eos

Venkatesh V December 4, 2006 16:18

Re: centroid problem
Hi, Check the quality of your mesh in GridGen, if the element quality is bad improve it. But if find the quality is very good in GridGen and you are getting warning in Star, please scale the geometry so that the dimensions are between 0.01-10000

eos December 5, 2006 03:21

Re: centroid problem
the problem comes from the scaling the chord of the wing is supposed to be 1 cm but i am meshing a geometry of chord 100 units (as far as i know Gridgen has no dimensions) i tried to scale the geometry in Gridgen, in Starcd by vscale and while writing the geom file (not at at the same time of course) which one do you recommend scaling during preprocessing in Starcd (vscale) or during thw writing of the geom file? by the way Starcd works in meters right? thanks for the response i will try again


Venkatesh V December 5, 2006 09:55

Re: centroid problem
Hi, Sorry for the confusion. ProStar or ProAm (not the solver)is written in such a way that it can handle the numbers only between 0.01 - 10000. You have to scale the geometry in prostar such that, all the dimensions lie between this range, eventhough the units doesn't make any sense dont worry. Now check the qulaity and do manipulation if needed. While writing the geometry file revert back to original dimension you need.

Thanks Venkatesh V

TG December 5, 2006 22:19

Re: centroid problem
The post from Venkatesh is absolute nonsense. STAR works in meters and you MUST scale the model to meters. You can do it before you write the geometry file with vscale or on output with the geom command. The result is the same. pro-STAR prior to 4.0 stored single precision coordinates and so there was a limit in the ratio between the biggest and smallest or you lost accuracy but in 4.0 its double precision. The idea that you have to work in units between .01 and 10000 is most ridiculous thing I have ever heard. That's why you should call user support instead of asking here. You never know how good the answer is.

Venkatesh V December 6, 2006 09:42

Re: centroid problem
Dear TG, Thank you very much for your feedback. I have asked him to resize the geometry( that is in Prostar and not in Star) to be between 0.01 to 10000. If dimensions are far away from this range Prostar will create problem. Even though it is not be strictly followed, it suggested by CD-Adapco to follow this as best practice. I have faced similar problem, even though the mesh look for visualy, pro-star used to give very bad qulaity. That is why on my first post I have told if Grid Gen is showing very good quality and ProStar is showing something wrong, i have ased to check dimensions.

If you have doubt about this please check the Star 3.26 user guide page number 3-27.

If you have any other suggestion please let me know, It would be a great learning oppertunity for me.

Thanks Venkatesh V

Tom December 6, 2006 13:13

Re: centroid problem
This is also a problem for us that work in English units. One gets the iges or other solid model in inches and work with that until the geometry write step. Unfortunately, when you try to build a subsurface that may be 0.01 inches, star doesn't know it is inches, only units. So the first step is use vscale to increase the size of the model to get around small units. By the way, this was recommended to me by adapco support.


TG December 7, 2006 17:45

Re: centroid problem
Dear Venkatesh I reread your post and I now realize I read it wrong the first time (sorry). I thought you said the solver required units between .01 and 10000. You didn't. In any event, I think that storing double precision for coordinates in prostar 4 should remove any restrictions. I also think even in prostar3 with single precision, its not the absolute value that matters but the ratio of the biggest coordinate to the smallest increment between adjacent vertices.

All times are GMT -4. The time now is 21:08.