Discussion Forum

Hi

Read carefully the doc about Union and Assembly (it has nothing to do with CAD assembly) Union means that all common geometric surfaces between two geometric volumes become two Domains with each their overlapping surface. But there is no FEM link between these surfaces, hence no "(force, fluid heat ... Flux" passing through these boundaries. For that you must define the pair and set manually all the flux corresponding to your physics.
In contrary in Union mode, the two volumes and their respective surfaces are transformed into 2 domains and ONE common boundary (except if you chooses remove internal boundaries, then you become 1 Domain and NO common boundary at all). In this case COMSOL assumes continuity of "flux" over this common boundary (if it exist)

In contact problems, if the surfaces in contact do not touch at geometry definition, either Union or Assembly works. If you have a contact and common boundary, then you need to ensure that this boundary is de-doubled and allow the "contact surface physics to apply to each boundary in contact.

A third case arrives if you models the fluid between the contacts, then you must ensure (if the contact are NOT closed at solver start) that the mesh of fluid remains over the contact boundary, else you make a topology change during the contact and the solver will crash. You can use the "contact offset" to perform this, and keep an artificial 1-2 um thick fluid layer, but you should not consider the fluid flow as realistic in this layer. Still the solver will complain if the layer mesh "inverts" mesh elements that turn inside out.

--
Good luck
Ivar

Dear Ivar,

Thanks for your prompt reply. After carefully check my model, I think my model applies to the case that it has a contact pair between two solid domains and two common boundaries between each solid domain and the only liquid domain. In this case, the contact pair and common boundary co-exist concurrently. So It make me confused about the setting for "Union" or "Assembly".

I intended to set the continuity state (extensive quantities and their gradients, i.e. heat flux, heat, etc) across the common boundaries between solid domain and the only liquid domain. For the contact pair, I hope to get compression and friction or slip in next step. In this sense, the setting of this case is complicated. By the way, my consideration is based on this support description: www.comsol.com/support/knowledgebase/1216/

The whole .mph file is attached below. It's set up in 5.0 version. I hope to get your specialized idea for this model.

Thanks very much.

Hi

First of all you have a complex 3 physics case with many couplings, then you two half sphere that touches each other at a point, for a 3D physics with *contact*, you have set-up a system defined with a singularity at the contact point.

To get this model running I suspect that you are still missing the ALE part (a forth physics) to handle the coupling between the deformed solid structural geometry and the fluid domain.

So for me you should definitively start "simple". Try first only the solid + contact, then the conjugated heat flow with solid, and fluid, but without contact, and finally the full system. There are so many setting to tweak to get these two cases running alone, that combining everything at once is a very serious challenge.

Even for the mesh, you have no boundary mesh for the fluid part, essential for flowing fluid, and even more important for HT + fluid interactions. Such a model was simply not possible to do a few years ago, at least not for a single person on a single program and PC, so today it's possible but the amount of physics tweaking and expressions to verify are and remain large.

But by studying physics interaction two by two, structural+ contact, structural + fluid, fluid& solid + HT, I'm sure you will get the required settings, and this would then allow you to assemble a full model. And it would be much clearer and simpler to identify the weak spots to be improved. With too many physics you have to many hares to follow at the same time ;)
--
Good luck
Ivar

Dear Ivar,

Thanks for your suggestion. In reality I started the one physics model (structure, temperature, fluid, contact pair) several months ago. I followed the philosophy to construct from simple to complicated. After that, the two-physics models was also modeled and extended to the three-physics model without contact pairs. All of these are successful. But the current model is when I add the contact pair, everything changed. The error always occurs. For me, the contact pairs is tricky and criminal.

For the setting of "Uniform" or "Assembly", do you have some idea to my specific model?

One thing to make up, the current model has three physics, and each of them are decoupled in order to analyze each of them particularly. I thought the cure would come out in this way. How about this idea, running each physics on such complicated model with respective steps?

Hi

my advice is to split it up first, then combine, and for your contact either start with an open contact and a "small gap" or have a true interaction so the contact surface is a "surface" and not a degenerated point or line contact. Furthermore, read carefully the doc, for contacts and the source destination mesh density etc, including for the fluidics about boundary mesh etc ...

--
Good luck
Ivar

Dear Ivar,

Thanks a lot. I am trying to do that.

Dear Ivar,

As a main step of the aforementioned intention, the three physics coupling is modeled (without contact pair). In this model, the study step 1 is HT + Fluid coupling. It works well alone. Then I activate the solid mechanics and FSI under Multiphysics button. the error always occurred. List below,

Failed to find a solution for the initial parameter.
The relative error (22) is greater than the relative tolerance.
Returned solution is not converged.

I searched the similar problem in this forum and those solution didn't work. By the way, If I disable the heat transfer in solid physics, and just consider FSI, It worked. but the three-physics coupling didn't work.
The file is attached below. I hope to get your response.