Discussion Forum

Hi

I'm not by my WS so I acannot open your model just now. But I suspect that you have misunderstood something in the way to combine meshes, and domains for different physics.

basically you have
a) one complex geometry, defined in "union mode" wich means that all domains have single common boundaries between each other (when adjacent) in this way when you mesh one domain, it generates mesh nodes on the boundary for the adjacent mesh and you MUST use these as seed nodes for each domain, hence over the continuity boundary (by default common boundaries have continuity of flux defined across them) you need the SAME mesh density as you share mesh nodes on the boundary.
b) one complex geometry (could be partially grouped by union commands) in "assembly" mode. Here, each common boundary between domains assuming "assembly" (i.e. which have NOT previously been grouped by a geometry union node) will have two adjacent boundaries, one attached to each domain, here each domain can be meshed differently and there are no "common nodes" (requirement, still it's also possible to copy over the exact node pattern, but it must be done manually) and you can have differnt mesh densities on each domain for local adjacent dedoubled boundaries. Neverthelesss, COMSOL does NOT assume any continuity over these boundaries, so its YOU the user who have to define these "identity" boundaries, group them by pairs (or packets of pairs) define the physics for EACH dependent variable ... Then COMSOL will look after the mesh mapping from the different meshes over the identity pair. Whith the possibility to have a poor flux transfer if the two meshes are too different.

I agree this is not that obvious from the doc, it took me quite some time with tests, but if you start simple, with two adjacent blocks and try out the different cominations, i.e. in HT as then you have 1 single dependent variable "T" temperature and the analytical solutions are rather well known from the litterature, then you can test this out and understand better how it works

--
Good luck
Ivar

The way I understand you, you want to specify different mesh densities for your components. What you did, though, is you created three distinct meshes. What you should do instead is make all those mesh nodes part of the same mesh sequence, see file attached.