Discussion Forum

Hi Matej,

If your geometric entities are not touching then Union and Assembly result in the same model, except that Assembly can automatically create the Contact or Identify Pairs for you. If they are touching or even overlapping then you need Assembly to let COMSOL know that they are disjoint.

Nagi Elabbasi
Veryst Engineering

Hi Nagi,

thanks for the reply. By overlapping you mean when two elements (geometric entities) are in interference, right?

Based on the tutorials, it seems you are right for "snap hook penalty" and for "tube connection" models, whereas the first one has two elements which aren't in contact an uses form union and the second one has touching geometry entinies and has fomr assembly and contact boundary conditions defined.
So when I search for structural mechanics analyses everything is fine and within your replied instructions, however when I search for thermal stress analyses (forexample tutorial "turbine stator") which are assembled from different elements and materials, the selected geometry finalization is form union and for all the rest of thermal stress analyses it's the same setting.

Do you maybe know why?

Thanks in advance,


--
Kind regards,

Matej

Hi,

When you do a Union, you essentially assume that all (overlapping or contacting) parts are 'welded' or 'glued' together. This means that there is no jump in displacement or temperature at an internal boundary. This assumption may or may not be a good approximation depending on your physics. It may even be so that, for the same connection, it is OK to assume continuity in displacements but not in temperature.

You can add conditions like 'Thin Elastic Layer' (in Solid Mechanics) and 'Thin Layer' (in Heat Transfer in Solids) on internal boundaries, even if you have used Union. These conditions will create an internal discontinuity based on a spring constant or a thermal resistance.

In the 'Turbine Stator' model, the geometry consists of one solid only, so there is no difference between using Union or Assembly.

Regards,
Henrik

Hi Henrik,

Thanks for your reply. if I understand correctly then forming a union is not appropriate for computing models which are assembled from different parts and materials. Let's say we have an interference fit and a bolted fit of two elements which are from different materials (in the second case the bolt is from the third material). In this case structural analysis is only possible if the geometry was finalized with "form assembly" and later identity or contact pairs were specified with contact boundary conditions?

Based on your post:
-
You can add conditions like 'Thin Elastic Layer' (in Solid Mechanics) and 'Thin Layer' (in Heat Transfer in Solids) on internal boundaries, even if you have used Union. These conditions will create an internal discontinuity based on a spring constant or a thermal resistance.
-
union form can also be used in some cases, however internal boundary conditions should be defined in order to implement internal discontinuity, right?

Yes you are right 'Turbine Stator' model is one solid element, but what about tutorial model "viscoplastic solder joints." This one has different elements with specified materials (4) and the thermal analysis was also made with "Form union," and ther is no additional internal boundary condition as you specified in your previous post.

My CAD assembly is formed out of different connections (interference fits and bolted connections) and I am curious how to define it in Comsol. If I try to use contact pairs and define contacts inside contact boundary conditions the computation tim extends extensively and ends up with error such as "out of memory during LU factorization". My WS has 64Gb of RAM and 6+6 cores, so I don't think it should be a problem with HW.

Am I using wrong contacts for my application? Should I use Identity pairs and later some other contact boundary conditions?

Thanks in advance,

--
Kind regards,

Matej

Hi Matej,

For an assembly with an interference fit it is often appropriate to use Union, when studied from the structural mechanics point of view. But if sliding could occur, then an Assembly should be used together with contact modeling. Whether the materials are the same or not does not really make a difference here.

If you study the heat transfer problem, then there will often be a significant thermal resistance at the interface, even if there is a large contact pressure between the surfaces. So in that case you need to use either an Assembly, or a Union with a 'Thin Layer' condition and the boundary.

In the Viscoplastic Solder Joints model, the connections between the materials are assumed to be 'perfect' which is reasonable for e.g. soldered or welded joints.

Using contact conditions will significantly increase your analysis time, since it adds a severe nonlinearity to your model. But it should not increase the memory requirements significantly.

Regards,
Henrik