Non linear Cosmos Assembly Problem

[SMarch]

I am simulating a rubber seal compressing on a shaft. The inner bore of the seal is 3 mm larger than the cable dia so there is displacement involved. I am using COSMOS Professional Advanced. The study is set to Nonlinear and Static. No penetration is set up also between the models.

There are 25481 nodes in the model.

When I run the simulation it is taking 2.5 hr to produce 0.04 seconds of results. That works out at 2.6 days to get 1 sec of data.

Is this due to me having my settings wrong or the shear amount of processing involved? I expect it to take a long time but I can’t even check that the simulation is working as it needs to be left for a day. (This has been done by the way and it had crashed after the first 1hr but it didn’t tell me.)

Am I better off reducing the mesh size? Concentrating it around the seal and increase it outwards? Any help will be appreciated.

Computer is Intel Core 2, CPU 2.4, 2GB RAM.

 

[rstupplebeen]

It might be that there is too much initial penetration for the model. You could scale the seal up for the start of the analysis and artificially shrink it by cooling the part with and appropriate CTE (coefficient of thermal expansion). I have much more experience with Abaqus and believe that there would be no problems with this methodology there. You could also do an axisymmetric analysis from your description to greatly reduce the solve time. Hope this helps.

Rob Stupplebeen

 

[SMarch]

Thanks for you reply.

By “initial penetration” do you mean that the fact that the assembly is set to no penetration on all parts and this may be slowing the simulation down?

The problem is that I do not know how the seal will react as it is not a simple o ring. Trying to design a seal that compresses and forms a seal but if it is over tightened it will still retain the seal but collapse so that the seal is not damaged. The ranged of the cables that will be used vary from 3 – 8 mm is dia so there will be a lot of displacement as the seal must work with both types of cable.

Have just googled Abaqus and it seems like a good bit of software will have a look on the web site.

I have already applied axisymmetric analysis to my assembly. Used 90 deg to simulate the seal but this can probably be reduced down to 2-3 degrees.

 

[rstupplebeen]

If the no penetration requires the nodes to move a large distance with respect to the element size the elements might become inside out or have poor aspect ratios. This could negatively affect solve time or make it unsolvable. I hope this helps.

Rob Stupplebeen