Specifying Periodic Vertices Causes Strange Problem
 

Hi all -

I'm encountering a strange problem when I try to create periodic sides on a 3D airfoil.

I have a working, complete 2D mesh that ran successfully (shown below).

http://a.imageshack.us/img440/5615/periodic1.th.jpg

I want to create a 3D mesh with periodic sides. The span length is 100.85531 mm. First, I went into the Global Mesh Setup and defined a translational periodicity with the z (span) offset as 100.85531. I then extruded my 2D blocking scheme to 3D using a z translation of 100.85531.

http://a.imageshack.us/img338/7800/periodic2.th.jpg

So far, so good. The problem arises when I begin specifying my periodic vertices. The vertices in the outer boundaries (inlet, farfield, outlet) don't seem to cause problems. The inner vertices (around the airfoil), however, seem to cause the mesh to extrude itself in the opposite direction desired (see below). The first picture shows the proper extrusion.

http://a.imageshack.us/img819/1349/periodic3.th.jpg

The next pictures show the problem areas.

http://a.imageshack.us/img440/4381/periodic4.th.jpg

http://a.imageshack.us/img715/7899/periodic5.th.jpg

http://a.imageshack.us/img38/8537/periodic6.th.jpg

http://a.imageshack.us/img84/5229/periodic7.th.jpg

http://a.imageshack.us/img340/5215/periodic8.th.jpg

http://a.imageshack.us/img843/3928/periodic9.th.jpg

Is my strategy incorrect? Any suggestions?

Thanks!

Josh

Are you specifying the airfoil vertices as periodic? If so, don't do that. The airfoil surface is not periodic, only outer boundaries.

Grant

Order...
 

When making verts periodic, you must be consistent with the order of selection. Go back and make sure that when you do it again, you always pick the lower k index first and then the higher K index next (or vice versa), but never change the oder during the selection of periodic pairs.

I think this hassle has been improved in more recent versions you could alternate without the problem.

Thanks for the suggestions, guys. I'm swamped this week with a different problem but will return to this topic at a later time with updates.

It's been a long time, but I've finally returned to this problem (and, unsurprisingly, I haven't yet solved it!).

As I mentioned previously, I have a working 2D mesh. Let me just ask this to start: What's the best way to proceed from there in order to make a 0.1c (10% chord-lengths, which for a 1 meter chord is 0.1 m) span mesh with 11 nodes (10 elements) in the spanwise direction? As far as I know, there are 2 methods:

1) Mesh > Part Mesh Setup > Define Translational Periodicity > [0 0 -0.1];
Blocking > Create Block > 2D to 3D > Method: Translate > Define new part: Solid > Distance [0 0 -0.1];
Blocking > Edit Block > Periodic Vertices > (set periodic vertices);
Blocking > Pre-Mesh Params > Edge Params > (define spanwise edge and parallel edges as 11 nodes deep)

2) Mesh > Part Mesh Setup > Define Translational Periodicity > [0 0 -0.1];
Edit Mesh > Extrude Mesh > (select all elements, 10 layers, 0.01 spacing);
Edit Mesh > Repair Mesh > Make/Remove Periodic > Make Periodic > (choose individual nodes)

Method 1 fails at step 2. When I extrude the block to 3D and observe the new pre-mesh, only certain parts are extruded.

Method 2 produces the correct spanwise mesh, but now I have 78,000 nodes. I can't specify the periodicity of each node pair.

How would you suggest I go about this?

I believe I've solved my problem. I used method 2 from the above post but neglected the periodic steps. I simply extruded my 2D mesh to 3D, created new parts for the 3D boundaries, and exported the mesh to CFX. It seems that specifying periodicity in ICEM is only required when the periodic interfaces are not directly aligned 1:1, e.g., in a turbine cascade. In my case, the periodic interfaces (the side boundaries) are directly aligned.

Hope this helps others.