Part Insert Meshing Options

[Lorena.Rivas]

Hi Community!

Im working with a client that use models with Part Insert and we tried different options for the meshing.
In theory, we could meshed with Ignore Contact (unless we notice that some results are "weird").

So, we try 3 options:
1. Meshing with Ignore Contact

2. Meshing with Precise Match

3. "manual Meshing", basically create the inserts based on the plastic part mesh (using the Ignore Contact mesh as starting point)

The test was made just using FLOW analysis with same material, process settings, etc. and the majority of the results gave us a very small difference (as you could see in the image with Pressure V/P)

But there is 2 results that are very different (in the case of the client, one more important than the other one)

- Extension Rate (less important)

- Time to reach ejection temperature (high importance)

Specially with Time to reach, we notice a difference (Between Ignore Contact and Precise match)

We think that the main reason of this difference is related to the mesh generation through the thickness, but i will like to have your opinion regarding the case.

Our idea is to create a confidence that with Ignore contact the results will be good without having doubts so we could reduce the model preparation time.

As always, thanks for your feedback!

[raalteh]

@Lorena.Rivas , probably worth looking at the mesh in the local area. It's not clear where the inserts are but just looking at things like local aspect ratios (squished models) or how the two models are or are (not) matching... 

[Lorena.Rivas]

Hi @raalteh !

Let me be more specific.

The model that I’m showing is a “dummy” version of the real models that the client use (for confidential restrictions we can not use their real parts)

Saying this, normally the models have small pins (as I showed in the picture below) and the tolerances are small. This tolerance promotes a lot of interceptions/overlaps in the model that require manual manipulation.

To avoid this additional model preparation time, we are making exercises with the difference/impact on using ignore contact and precise match.

If I understand correctly your point, I have now this doubt:

If the meshes are different just in the contact areas between the inserts and the part, but the rest of the controls are the same… Could the mesh have a big difference (general or local)?

Because if that is the case, which results could be more reliable? I imagine that are the ones with Precise Match.

[raalteh]

Hi Lorena,

From the 'accuracy perspective' the precise match should be the best, but ... getting a mesh with precise match is not easy because in many cases the CAD models are not 'perfectly' aligned which causes problems in meshing. The low guarantee for getting a mesh caused us to take a different approach and map the data on the nodes. The recommendation is to have the mesh as similar as possible. The nodes will look (within a range) for the closes node on the other side to match with. There is obviously going to be a source of error; the mapping of nodes will not get better than perfect matching nodes. Its a matter of if going through the meshing 'pain' is worth extra accuracy. 

One consideration ... the different solvers all have different this internal node-to-node mapping of contact. The flow solvers uses the contact to transfer heat and pressure (for example). Then the warp solver uses the node mapping to transfer load from one part to the next. The core shift option is a warp solver embedded into the flow solver ,.. if some nodes (out of many) do not have a connection, the impact may be negligible, but I can imagine that for the temperature solution it could have a locally significant effect.

Regariding my comment to look at the meshing; precise matching will force nodes on on the surface to be identical, but the interior mesh could be different (possibly skewed tets that could have a numerical effect) ... I'm speculating a bit though. Something I would look at..