[NIU]

I want to calculate the heat transfer and pressure drop through a packed bed, so firstly , i create a random packed bed using DEM method,then i got the spheres coordinates. Secondly, i transfer the geometry to ICEM and mesh.
There are some problems i can't continue:
1:the contact point distort the mesh;
2:i use tetra+prism to mesh ,but the quality is too low to calculate.
3:there are so many cells for only 100 spheres, which took too much time to mesh.

So i want to know if anyone can share a more convenient method to mesh the spheres and if i have as up to 10000 spheres, how to mesh using ICEM?

Thanks!

[saidc.]

Hi NIU,

Were you able to solve your problem? I'm facing same difficulties right now. Lot of spheres, 50 million cells but still too coarse to solve. The best result I get with Salome (NETGEN Algorithm) for now is 5000 non-orthogonal > 70 faces (avg 20, max 88). Same as you I have a lot of spheres (300,900,2000 and so on). I know that if I use much lower max size parametre for NETGEN mesh will be fine but it is already has more than 50 million cells and this is not even the largest case I have. I wonder how did you approached this situation because I'm using OpenFOAM and SALOME unlike you. Please share your experiences with me.

Kind regards,
Said.

[Gert-Jan]

Quality is mainly determined by the contact angle between the spheres. If you use point contact, then you can imagine that none of all existing mesh packages is able to create good quality mesh. Therefore you need a compromise between real geometry with impossible mesh or a simplification (with overlapping spheres) with reasonable mesh.

Moreover, try CFX as a solver which is very forgiven for low quality (=bad) mesh.

[saidc.]

Originally Posted by Gert-Jan

Quality is mainly determined by the contact angle between the spheres. If you use point contact, then you can imagine that none of all existing mesh packages is able to create good quality mesh. Therefore you need a compromise between real geometry with impossible mesh or a simplification (with overlapping spheres) with reasonable mesh.

Moreover, try CFX as a solver which is very forgiven for low quality (=bad) mesh.

Hi Gert-Jan,

Thanks for your kind reply. I already did overlapping for ensure that the mesh can be done but that approach still not cover the problem. If I do much more overlapping (decreasing the distance) my needed porosity will gone and my accurisy too. I'm still thinking what should I do and my time running out. Umarım çözümü en kısa sürede bulurum.

Kind regards,
Said.

[Gert-Jan]

Around the the contact point you can create a small cilinder as part of the spheres. This will at least remove the very small angle.