CFD Online Discussion Forums

CFD Online Discussion Forums (
-   ANSYS Meshing & Geometry (
-   -   3D s shaped pipe (

Prakash.Paudel November 1, 2010 15:25

3D s shaped pipe
Hello all ,

I created a s-shaped solid cylindrical pipe in Design modeler by sketching a circle in yz plane and sweeping it along a s-shaped curve which i sketched in xy plane. The s-shaped curve in xy plane used fillet feature to smooth the corners. The problem with my geometry now is that surfaces are not smooth. When I import this geometry in ICEMCFD i see some unnecessary edges along and across the surface of this geometry. How can i get rid of these edges. I dont have my geometry with me right now but i am willing to upload it later if that makes things easier for anyone who wants to help out.

Also, what kind of mesh would be appropriate for such a geometry to model incompressible, viscous, laminar fluid flow within this model. Is ICEMCFD or Workbench better for this ??

Thanks for any help


PSYMN November 2, 2010 13:46

I don't know what geometry issues you are having (a picture would help), but either tool could do a good job.

Workbench could sweep it.

ICEM CFD hexa could block it and put in an OGrid...

Workbench would be easier, ICEM CFD would be better quality and with more control (more control automatically means there is a learning curve).

Prakash.Paudel November 5, 2010 18:15

2 Attachment(s)
Thanks simon for answering my question. I hope things are very well on your side. Toronto is getting colder and soon the winter will hit us.

My problem was with the geometry and i got rid of it by using splines instead of using fillet when i created the sweep path for the circle. The 3D S- shaped pipe is smooth now.

For the mesh i used automatic mesh method in workbench and then i inserted an inflation for my boundary closeto the walls of the cylinder. I have attached a picture. I am modelling a visous laminar flow with reynolds number regime of 10 to 70.

Do you think i should better mesh it in ICEMCFD or this mesh that i have is enough??

Also, is there any way to find the equations of spline that i created using random points in DesignModeler?

Thanks for all the help.

PSYMN November 6, 2010 16:29

Not bad at all.
That is pretty good for a triangular mesh...

But what about a swept hexahedral or prismatic mesh? Workbench has a sweep method or the MultiZone method (ICEM CFD Hexa in ANSYS Meshing). Either of those should work nicely.

For the sweep method you will need to inflate the edges of the source face (weird quirk) instead of selecting the side walls of the solid as with other methods.

Either way, Workbench is definitely good enough for this sort of model.

I don't know of a way to extract the equation from WB, but you could extract locations along it and then send those to another tool to extract the equation.

All times are GMT -4. The time now is 16:17.