code_saturne User's Forum

Hello,

I have been trying to join two meshes with the transient turbomachinery module for the past week, but I can't get past 10 iterations before there is a mesh joining error. I made sure the interface of the two meshes matched and the mesh was equally distributed in the theta direction. In the radial direction, the mesh is grouped together near the wall, but the largest aspect ratio is no greater than 10. I have tried changing joining tolerance from 0.05 to 0.5, but see no improvement. I have also changed the mesh distribution in the radial direction to be matching and non-matching along with the overall distribution. Would changing any settings with cs_join_set_advanced_param effect the joining? Attached is the error mesh file and images of the interface.

Note, at the beginning of the computation, I copy, rotate, and join the mesh to make the full compressor. I have no problem with the mesh interface between the impeller and diffuser, only the interface between impeller and inlet. I can run my computation with the impeller and inlet domains static with respect to one another. For this case, I use a tolerance of 0.2 and an angle of 25 degrees.

Regards,
Andrew

Hello Andrew,

What is the aspect ratio betwwen the surface and "depth/thickness" of the cells ? Double curvatures with thin boundary cells (such as those built for boundary layers) are the worst situation for mesh joining, due to the tolerance being based on the given multiplier times the shortest adjacent edge at each vertex (to avoid "entangling" cells) ?

Did you visualize where the joining error occurs (not actually sure joining visualization is not deactivated after the first time step, but if it is available, it may provide more info).

I would recommend keeping the parameters you use. If the gap is always positive, one of the advanced joining parameters (allowing basing the tolerance just on the surface and not boundary) may be useful.

Cheers,

Yvan

Hi Yvan,

The mesh upstream of the interface is completely normal to the interface. Downstream of the interface, towards the impeller, the angle of the mesh differs slightly from 90 to 75 degrees. With 90 degrees being normal.

Cell size in the normal direction to the interface is 0.0004. The aspect ratio varies between 3.2 to 8.0 for normal/radial direction and 1.41 to 0.54 for normal/tangential direction. Before this, I tried sizes of 0.00015 and 0.0003 while maintaining approximately the same densities in the radial and tangential direction. From the usage guide and previous posts, I think it's best to have an aspect ratio (normal to radial/tangential) greater than one, right?

I tried to visual the interface by setting the visualization parameter in the mesh joining function to 3, but I didn't see any extra information. I can only visualize the mesh joining if I statically join the two meshes. I will try to rotate the mesh by 10 time steps and then statically join them to see what happens.

Thank you,
Andrew