Two questions regarding solids vs shell and mesh generation for assemblies

[jsteffle]

I have two questions that I'm hoping to get help on:

Referring to attached image....

1. I have module that I need to obtain stresses for. The module consists of a bunch of components sitting on a larger component. These are created in an assembly. All of the components contain different materials so my understanding is I need to keep it as an assembly vs converting to a part. The components are basically "cube" in form. Currently, all of the components are defined as solids. I would like to know if I should be converting these to shells rather than keeping them as solids? The components themselves are solid. I'm just not sure if they should be processed as shells to make it easier for calculations? Any help is appreciated.

2. When I generate the mesh for the components I've indicated above the nodes don't line up where the components meet. I have specified with the continuous meshing option but it doesn't seem to have an impact. I have created the contacts and have refined the mesh along the applicable edge. The larger (bottom, GREEN) mesh stays larger and the upper (smaller) box gets refined. I'm concerned this is affecting the results. Can anyone explain how I can do this correctly? I've attached a screenshot for reference.

[KubliJ]

Hi @jsteffle,

What is the object that you are trying to solve for?  The solid blocks are too blocky to convert to shells, but if the plate they are attached to is the important part, then I don't see why the blocks cannot be replaced with a point mass.

What type of analysis are you trying to do?

Regarding your second question.  The continuous meshing is mostly for shell elements and connecting them.  Using contact is the correct way to bond them.  Mesh refinement will only affect the body that the edges/faces are connected to.  To refine the face of the green part, you will need to select that face and add a refinement.

Thanks,

James

[jsteffle]

Hi James,

Thanks very much for your help. What I'm trying to solve out of this is just to be able to highlight areas of stress due to thermal expansion at various temperatures. I'm using body temperature along with initial temperature for loads in a linear stress analysis. The blocks are fully attached to the main block (aka substrate). I have used automatic contacts for this. The smaller blocks are mainly ceramic and the substrate is glass reinforced plastic. The larger blocks on top of the substrate are silicon. Expansion rates for these are much different. We know there are risks with this combination. To add another level of complexity, eventually all of the blocks will be covered with an epoxy.  Screenshot attached. Currently, I'm just trying to find the areas of peak stress. I'm hoping that by determining this I will be able to know if a certain component was moved it may be possible to reduce some stress. Hopefully this is more clear. 

We know that component placement inside this epoxy has a major effect on the device reliability. We have seen these assemblies come apart (delaminate after temperature cycles). I am trying to highlight and reduce the associated risks, wherever possible. I understand that simply highlighting the risks may not solve my problem but I'm just trying to identify possible areas of higher stress.

I thought that the continuous meshing would help me align the node points where two components met. This doesn't seem to be the case. The main substrate for example where it meets the smaller components continues to have larger shapes that don't align to the vertices of the smaller component edges. Is it critical/important that these align? I thought this would be needed to get a clear understanding of how the substrate was interacting with the components which are attached. Do you think it would be a good idea to create a sketch on the main substrate where the components meet and then refine using this edge? Perhaps this isn't even important?

Thanks again,

Joe

[KubliJ · Accepted solution]

Hi @jsteffle,

I understand the intent.  I don't think mixed elements of shell and solid elements are going to offer any improvements. Solid alone should be fine.

Now regarding the mesh.  Having an unmatched mesh with contact is fine.  No concerns that I can think of other than an extreme difference in mesh size can cause issues.  But the simplest way to correct this is to split the face of the large plate to match that of the attached blocks.  This new face can then be refined to match the contact of the other face.  Honestly though, I don't think you need to be local refinement with this model.  Just give a global mesh refinement.

Thanks,

James