Stress base Topology Optimization

Reza Karshenas

hi everyone,

 

I'm a bit confused about the stress response for topology optimization with Optistruct. I have a linear static analysis which stress must be controlled and the values do not go through the yield strength. Then I want to define yield strength of the material as restriction but in the model there is a high value of stress in some small points and this has influence on the optimization. How can I remove this stress peak from the design variables? Is it enough if I define different properties or do I have to do something extra? 

so I would also ask what kind of stress restrictions should I choose? normal static stress? Norm-based approach? Augmented Lagrange Method?  Global von Mises Stress Response? Grid Point Stress Responses? Failsafe Topology Optimization?

thanks

GTT Johan

Hi Reza
I would have started with normal static stress and as you suggest put the "hot spot regions" in a seperate part/property. BUT only if they are in close to a RBE2 or SPC there you normally get very high stress peaks!
So be careful, so you dont put a real "hot spot" in an non-design material!

Often it is recommended to have non-design material at load application points and at constraint! 

OS-T: 2010 Design Concept for an Automotive Control Arm (altair.com)

Hope this helps!
/johan

GTT Johan

Hi Reza
I would have started with normal static stress and as you suggest put the "hot spot regions" in a seperate part/property. BUT only if they are in close to a RBE2 or SPC there you normally get very high stress peaks!
So be careful, so you dont put a real "hot spot" in an non-design material!

Often it is recommended to have non-design material at load application points and at constraint! 

OS-T: 2010 Design Concept for an Automotive Control Arm (altair.com)

Hope this helps!
/johan

Reza Karshenas

thanks Johan