Learning Forum

[refetali]

Hi,

I have digitized a metallic planar particle distribution from an SEM image. The particles are randomly distributed. Whole structure is so large to simulate so I want to simulate a smaller portion and use a periodic BC. As we can see from the attached image, boundaries (have to) intersect some of the particles (spheres). I know that these sliced pieces will create anomalies but as the area of the section grows I believe it will yield a realistic result. When I use FDTD I can use periodic BC which intersects with particles; however with DGTD I recieve "Failed to match periodic face" error. Is it possible to use a periodic boundary condition which is intersecting a particle (whose other half does not match at the other boundary)?

Kind regards,

Refet

intersecting boundary.jpg

[Guilin Sun]

I think it is not due to periodic, it is due the meshing of spheres. It can be challenge meshing curved surfaces when two spheres are cut or very close. You may need to use very small edge min. You can also try to first disable the smaller spheres.
A side note: this current setting of periodicity does not qualify the periodicity, as a standard periodic structure duplicates its unit cells by the periodic BCs. When you duplicate the current settings, the geometry will not duplicate its self as expected since the spheres are not cut at the symmetry. Please modify the periodicity in order to get correct result.

[refetali]

Thanks for the quick reply.
I am aware of the problem you mentioned in the side note. I know that it is a periodic BC not a symmetric BC. That's why I said "I know that these sliced pieces will create anomalies but as the area of the section grows I believe it will yield a realistic result." in my previous message

Did you mean high number of edges per wavelength while saying "You may need to use very small edge min." If so, I tried from 1 to 1000
adddgtdsolver;
set("simulation time",100e-15);
set("auto shutoff min",shutoff);
set("edges per wavelength",1000);
But I always recieve the following error;
The program terminated due to an error: Failed to match periodic face.
However the same geometry runs on FDTD.

We also have problems while deactivating our licence (Deactivation failed with error code 6) We need to move it to a new PC. We sent an email (from @univ-poitiers.fr mail) Could you also check it?

Thanks!
Kind regards Refet

[Guilin Sun]

ÔÇ£Failed to match periodic faceÔÇØ most likely due to the periodic BCs do not see the periodicity, eg, left and right are not the same. So before fixing other issues, please make sure it is periodic as suggested previously. I also would suggest that you disable all the spheres, and then enable them one by one, and just do meshing without actual simulation and see if the mesh can be generated successfully and which sphere causes this issue. Once identified then we can try to find a solution.