J-Type topology,shifted and non shifted periodicity
 

Hi Folks

Previously I tried several times to create the J-type mesh for turbomachinery using ICEM. with no success. http://www.cfd-online.com/Forums/ans...ine-blade.html

Finally I found a method to do so and wanted to share with you. Geometry is simple with leading edge type shape.

Picture of final mesh

http://img860.imageshack.us/img860/3653/jgridicem.jpg

Steps to create this topology(1-5):
http://img849.imageshack.us/img849/9365/24717448.png
http://img804.imageshack.us/img804/5121/70606230.png
http://img714.imageshack.us/img714/6333/95684120.png
http://img546.imageshack.us/img546/4121/55460009.png
http://img607.imageshack.us/img607/2798/94096378.png

steps 6-9

http://img835.imageshack.us/img835/8546/35112974.png
http://img52.imageshack.us/img52/622/16411932.png
http://img846.imageshack.us/img846/6188/76018423.png
http://img94.imageshack.us/img94/9779/55164288.png

Shifted periodic
 

Use of shifted periodic to define the periodic boundary conditions

Steps : 10-13

http://img715.imageshack.us/img715/4347/j10.png
http://img210.imageshack.us/img210/9020/j11g.png
http://img820.imageshack.us/img820/4686/j12f.png
http://img853.imageshack.us/img853/3508/j13v.png

Nonconformal periodicity treatment
 

Since latest version of Fluent has option for non conformal periodic boundary conditions. Therefore it is not always required to define the periodicity in ICEM and also there is no need to use the shifted periodic boundary condition. We shall create the non-conformal periodic boundary in Fluent. In ICEM define per1 and per2 as interface boundaries.

PS : Fluent V 12 and ICEM V 12 are used for all this work

steps 14-16

http://img443.imageshack.us/img443/3748/j14u.png
http://img833.imageshack.us/img833/2874/j15c.png
http://img713.imageshack.us/img713/9084/j16ai.png

Why not just apply a CGrid or Quarter Ogrid on this...

Maybe I am not understanding your goal, but I think you could have got all the hard part done in one quick top down procedure. Oh well.

I dont think that this is possible with quarter O-grid

The first 9 steps can be replaced by 2 simple splits, a quarter ogrid and one delete block. You will get to the same blocking, but the index will be much cleaner.

I didn't understand what you were trying to accomplish with steps 10 thru 13... I guess this is a periodic model and you were trying to create a shifted periodic blocking? Just put a quarter Ogrid the other way, then apply periodic verts and be done.

hi FAR.

thanks for your post - I think the Numeca mesher uses J-grids quite extensively.

Just a quick question though, were you able to make conformal periodics with this mesh?

I am trying to stay away from quarter o-grids since I often end up with a block of dense cells in the quarter o-grid and which then gives me mesh related issues with variables which are highly sensitive to mesh density!

Again thanks for posting this -

Best regards
Jonathan