Hi Yaguang,

The issue is quite specific to the exact mesh you create, not only the geometry.

Did you manage to create a periodic mesh in Gmsh as Guglielmo suggested?

Tests you can check that:

N-elements of surf1 and surf2 composites are the same
[xyz] location of centroids/vertices should be the same if you try to transform the x,y,z coordinate system to x1,y1,z1 with the PI rotation. (This is important!)

What types of elements do you use in the mesh?
Once you make sure the previous conditions are passed, does PerAlign still gives you the issue with abstol and orient flags on?

If this is still the case, we can try to arrange a short chat at some point in the next week.

Kind Regards,

Kaloyan

PS: I am copying the email from the previous thread, so everyone is up to date as well.

---------------------------------Previous Thread - --------------------

Hi Yaguang,

Thank you for getting in touch.

Unfortunately, I cannot open the files due to an email firewall, but I should be able to help.

There are several possible reasons you might be experiencing this issue:

Surfaces/Composites - 5,6 need to be the matching exactly - number of faces, location [x,y,z] of the vertices and centroids of the faces.

- This is related to the way you generate your mesh and the periodicity usually needs to be specifically enforced.
Machine precision tolerances in some of the elements in the original :

- Recommend adding abstol=tolValue

NekMesh -m peralign:surf1=5:surf2=6:dir=z:rot=PI/6:abstol=1e-6 test.xml test_align.xml
Type of elements used - if you have prisms or tets in the mesh:

- Need to add the orient flag

NekMesh -m peralign:surf1=5:surf2=6:dir=z:rot=PI/6:abstol=1e-6:orient test.xml test_align.xml

I think this should help. If it does not, could you please let us a bit more details about how you generated the mesh, type of elements, etc?

Kind Regards,

Kaloyan Kirilov

Hi Yaguang,

Try with the abstol=1e-8 parameter for the paralign module:

-m peralign:surf1=2:surf2=3:rot=PI:dir=z:abstol=1e-8

For me this makes the difference:

You can find about the options for the module by running it with -p proc:<module name>. In the case of paralign:

NekMesh -p proc:peralign

I am not sure it is listed in the user guide?

Best regards,
Stanisław Gepner, Ph.D.

Faculty of Power and Aeronautical Engineering
Nowowiejska 21/25 Str.
00-665 Warsaw, Poland
phone +48 (22) 234 51 70

On 14.04.2024 13:54, 谢亚广 wrote:

Hello Kaloyan Kirilov,

I want to start by using GMSH to NEKTAR, and then use the peralign module in Nekmesh to align p1 and p2 for rotational periodicity. To pinpoint the problem, I have simplified the mesh to its most basic test state. However, no matter what I try, I am unable to achieve the aforementioned requirements using NekMesh -m peralign; it always results in errors.

Assuming thi .geo file corresponds to a 2D mesh named test.msh, my operations are as follows:

NekMesh test.msh test.xml:xml:uncompress

NekMesh -m peralign:surf1=5:surf2=6:dir=z:rot=PI/6 test.xml test_p.xml

Subsequently, an error occurs:

[ProcessPerAlign] WARNING: Could not find matching edge for surface element 3. Skipping periodic alignment.

I hope I have made myself clear.

Thank you for your assistance.

Best regards,

Yaguang.

-----原始邮件-----
发件人:"Vivarelli, Guglielmo A" <guglielmo.vivarelli12@imperial.ac.uk>
发送时间:2024-04-14 19:30:07 (星期日)
收件人: "谢亚广" <xieyaguang@iet.cn>, nektar-users <nektar-users@imperial.ac.uk>
抄送:
主题: Re: [Nektar-users] Questions about the plane of rotational periodic symmetry

Hi Yaguang,

I am not too sure of what you are trying to do, do you just want the two curves to match periodically and then convert the mesh from GMSH to Nektar?

If you give me a little more info I may be able to help you.

All the best,

Guglielmo

From: nektar-users-bounces@imperial.ac.uk <nektar-users-bounces@imperial.ac.uk> on behalf of 谢亚广 <xieyaguang@iet.cn>
Sent: 13 April 2024 05:00
To: nektar-users <nektar-users@imperial.ac.uk>
Subject: [Nektar-users] Questions about the plane of rotational periodic symmetry

This email from xieyaguang@iet.cn originates from outside Imperial. Do not click on links and attachments unless you recognise the sender. If you trust the sender, add them to your safe senders list to disable email stamping for this address.

Hello Kaloyan Kirilov,

Thank you very much for providing those solutions. Unfortunately, I am still encountering the same error. I generated the mesh using gmsh, and for testing purposes, I simplified the mesh to the most basic case. Here is the content of the .geo file:

SetFactory("OpenCASCADE");
Circle(1) = {0, 0, 0, 4, 0, Pi/6};
Circle(2) = {0, 0, 0, 14, 0, Pi/6};
Line(3) = {2, 4};
Line(4) = {1, 3};
Curve Loop(1) = {3, -2, -4, 1};
Plane Surface(1) = {1};
Transfinite Curve {3, 4} = 3 Using Progression 1.1;
Transfinite Curve {1, 2} = 3 Using Progression 1;
Transfinite Surface {1};
Recombine Surface {1};//+
Physical Curve("p1", 5) = {3};
Physical Curve("p2", 6) = {4};
Physical Curve("up", 7) = {1};
Physical Curve("up2", 8) = {2};
Physical Surface("plane", 9) = {1};

In this file, I want "p1" and "p2" to represent the periodic boundaries for rotation.

Thank you once again for your guidance.

Best regards,

Yaguang.
_______________________________________________
Nektar-users mailing list
Nektar-users@imperial.ac.uk
https://mailman.ic.ac.uk/mailman/listinfo/nektar-users