Tied vs Partition inorder to have 2 layers connected

[cdixonm]

I am building a model where I have two layers of material. The outer layer is growing faster then the inner layer in my model. When I use partition to separate my two areas everything works great and I get the results I expect. However when I use the tie function, iterations or merged assembly I have not been able to get the results I expect like i am getting from partitioning my two layers apart. Does anybody know what setting I may be missing that would make these layers interact the same way they do when I partition them apart? The only reason this is a problem is that the next step is to apply this to a more complicated geometry and I will not be able to partition the sections apart in ABAQUS.


Thanks Campbell

 

[Mustaine3]

Do the parts deform together? What is the difference? Just from your explanation it is not possible to say anything specific.

 

[cdixonm]

Yes my two parts are deforming together. My outer layer is slightly stiffer then my inner layer. I am using abaqus explicit my densities are the same. And I expect relatively large deformation. My test case is on a half sphere where the outer layer has thickness of 2 and inner layer has a radius of 48. I expect the see the surface buckle and fold against its self. My step type is dynamic explicit and I am increasing the temperature of my system to simulate growth.

 

[cdixonm]

It appears that my problem is that the only way that I can get my parts to deform together is by using partition in all of the other methods I have tried my base layer does not deform with my surface. I am wondering if my problem is that some how my surfaces can slide across each other. The models where I use methods other then partition I get spikes that grow out of my surface.

Thank you Campbell

P.S. A list of operations i have tried:
merge
tied constraints(default settings) :no contact properties, contact friction less and normal tangential, and rough
contact properties only: contact friction less and normal tangential, and rough
maybe missing some other setups i have run that i just can't think of at the moment

 

[Mustaine3]

Just use a Tie constraint in the tie interface, not a contact with special properties.

 

[cdixonm]

That is one of the setups that did not work. I tried running just tie constraint again and got a very similar result to this one. In this you can see that it is forming spikes instead of building up pressure and then bucking

This is what I get when i partition the two layers out of a single object. This is the result i expect to get but am not

I don't understand what is going wrong. The most promise I have seen from any of these models is the rough contact with no separation set on the normal behavior but it still looked like my failed case where I had my outer layer forming spikes instead of buckling in on itself. I have found that the correct model takes much longer for Abaqus to process and it normally takes about 2 hours to run while all of the ones that form the spikes take about 10 minutes. With the tied condition I am having some of the surface move with the spikes that are forming but I it is still very far from being the bucking seen when using the partition method. I think the problem starts early in running the model analysis. When I use the partition method the model has uniform stress distribution all across the system however when I use any other method I do not have this uniform stress distribution instead the top of my model has lower stress. I expect that the problem I am getting is from this and the spikes reliving the built up stress instead of the system bucking very rapidly when a maximum stress is reached. Is there a setting in the tied condition that I am missing, I have tried both analytical default and surface to surface? or maybe a way to influence the mesh to help it have the uniform stress early on in the simulation?

 

[Mustaine3]

Can you upload the two models?

 

[cdixonm]

Yes Thank you Mustaine3 here is a link to a Google folder I made with 3 different case files. One file is of my geometry partitioned apart. One is of the two sections merged together and one is of them not merged together. They should all already be meshed and setup to run in one form or another. All of them should have a predefined Field that has a 700 temperature step. the outer layer should expand with .001 and have a c10 value for the neo-hooken step of 2000. the inner layer should have an expansion of 0 and a c10 value of 1000. If you have any questions please ask I should be quick to respond again now that i am back in town.

[URL unfurl="true"]https://drive.google.com/folderview?id=0B5Owd1gTUKIjM2NRYW0wWkZvbkU&usp=sharing[/url]

Thanks very much Campbell

 

[cdixonm]

Could the problem I am having be caused by density? I have been playing with the density and it seems like it is causing some what closer results but nothing I can be positive about?
Does anybody know if Density is treated differently when an object is partitioned vs tied?
I am wondering if for some reason it stops using a quasi-static analysis. Is that possible?

 

[cdixonm]

I ran a simulation that took a while but dropping the density in the tied model did solve my problem. Does anybody know why this worked I would really appreciate any insight you may have into this phenomenon. But for anybody who may work on a project and has a problem with tied vs partitioned sections in a dynamic explicit step the problem that it seems that i ran into was that is stopped calculating it as a dynamic explicit step and lowering my density seemed to give some promising results I lowered my density from 640 to 0.01 in this case.

Thanks, Campbell