CFD Online Discussion Forums

CFD Online Discussion Forums (
-   Main CFD Forum (
-   -   How to reduce total number of elements? (

Pouya October 23, 2012 04:32

How to reduce total number of elements?
1 Attachment(s)
Hi every one!

I've a couple of questions! I'm using ANSYS Mechanical APDL to create geometry and mesh. The geometry is simple (flow around a circular cylinder!). I've used "Flotran 142" as fluid element.

I want to reduce total number of elements specially in the z direction (assuming xy on the cross section). And of course I have to reduce Y+ in z direction to a value of below 1.

I decided to use finer elements around the riser (similar to the picture). I'm using structured mesh. In this picture, 2 elements are connected to one. I should merge nodes and keypoints. Is it possible? (I know it is! I've extracted the picture from an OMAE paper!)
But how? Is it possible in Mechanical APDL? Is any special command necessary? or maybe I've to use another meshing software?!!!

I was wondering if you'd mind explaining it! Thank you.

cdegroot October 23, 2012 15:38

A structured mesh has the elements ordered in rows and columns. The picture you are showing is a block structured mesh, meaning there are certain blocks in which the mesh is structured (with a row/column format). I'm not sure about creating the mesh in the software you mentioned; never tried it. I would be more concerned about whether or not the CFD solver you are going to use will accept meshes with faces that aren't connected one-to-one. What solver are you going to use?

Pouya October 23, 2012 16:47

Thanks for your answer. I'm using CFX & I must use LES or DES turbulence model. As you mentioned faces of elements aren't connected.

cdegroot October 23, 2012 16:59

Okay. So while I can't comment on how to construct the mesh, you shouldn't have a problem in CFX with faces that do not line up properly. CFX is very good at handling this using the General Grid Interface (GGI) connections. Just make each block a separate fluid domain and define a conservative domain interface between them. Good luck.

Pouya October 24, 2012 16:54

Thanks Chris. Today I studied about GGI! I think you are right. I'm going to try it!

All times are GMT -4. The time now is 03:42.