All tips on Tie contraint 1

[mdama]

When I use Tie constraint to attach two parts, It's hard to get convergence. I would like to know tips on using Tie constraint. Should the mesh size/element type exactly be the same on both size. if any other tips are important, It would be great to know
Thanks
Mohsen

 

[FEA way]

Tie constraint when applied properly shouldn't cause any convergence issues since it's a lineae contact and It simply merges nodes together. But you should carefully select master and slave surfaces (there are some rules like in case of contact) and what is probably the most important - make sure that either there's no gap between the surfaces or the tolerance for adjustment is set properly. It shouldn't be too high because the constraint may try to merge nodes that are far away and shouldn't be merged because their elements will become highly distorted (common "zero volume" error). But if the tolerance is too low, tie constraint may not be able to merge nodes that should be merge. When you use ties always make sure that the constraints worked ans all nodes that should be merged are actually merged. There are several ways to do it - visual inspection, datacheck analysis with unconnected_regions=on, examination of warnings and eigenfrequency extraction analysis which will show the parts not tied together as they should be.

Mesh size and type doesn't have to be the same on both sides. Tie constraint is meant to be used for dissimilar meshes.

 

[mdama]

Thanks
Can you tell how I can deal with "datacheck analysis with unconnected regions" and "eigen frequency extraction analysis" and how should the latter be?

 

[FEA way]

For the former simply run datacheck job but add unconnected_regions=on command to the remaining instructions (for example use abq2019 datacheck job=... cpus=... interactive unconnected_regions=yes). You will find new sets in the results file that represent disconnected regions (keep in mind that some of them may be disconnected on purpose).

For the latter copy your model and replace static general step with frequency one. Choose about 10 eigenmodes and run the analysis. Then inspect obtained mode shapes. You will see which parts fly off. Everything that's properly connected with tie should stay fixed.