Optimization constraint: material failure
I want to use OptiStruct to optimize thickness of an aluminium component made of PSHELL elements by using free-size (it is not a composite). I want to minimize the mass and constrained it to material failure (yield and ultimate stresses for compression, tension and shear).
Where should I add these material failure data? Should they be specified in the material section or in the optimization constraint for stress response?
Thank you in advance
Find more posts tagged with
Altair Employeeyou could do it 2 ways:
1) inside the free-size deisgn variable, you could add a global stress constraint, under 'parameters'>>'stress constraint'. It will give you some control over the average stresses, then specify your limit stress. It applyes to all your model and all static loadcases
2) create a response of type DRESP1, 'static stress', and then assign it to a opt constraint and define the upper bound and the desired loadcases. This one you can specify region and loadcases, but depending on your model it could take longer to compute, as sensitivities need to be calculated for all your design variables.
please, note that this is intended for yield/limit stresses.
i believe it is not possible to use NL material with topology and free-size. (also remember that these are conceptual design approaches, not intended for detailed design).
if it was size or shape i believe you could use plastic strain.
please, note that this is intended for yield/limit stresses.
i believe it is not possible to use NL material with topology and free-size. (also remember that these are conceptual design approaches, not intended for detailed design).
if it was size or shape i believe you could use plastic strain.
Thank you for your response.
I understand that I cannot set different limits for yield and ultimate stress as non-linear materials are not used in free-size optimzation.
And how can I set different yield limits for compression and tension stresses?
Altair EmployeeThank you for your response.
I understand that I cannot set different limits for yield and ultimate stress as non-linear materials are not used in free-size optimzation.
And how can I set different yield limits for compression and tension stresses?
for defining different allowables in tesnion and compression you would need to work with responses for maximum and/or minimum principal stress values or even xx, yy, ... and then in the constraints panel, add them accordingly as lower bound for compression (-) and upper bound for tension (+).
For vonMises it would not apply.
for defining different allowables in tesnion and compression you would need to work with responses for maximum and/or minimum principal stress values or even xx, yy, ... and then in the constraints panel, add them accordingly as lower bound for compression (-) and upper bound for tension (+).
For vonMises it would not apply.
Thank you.
Then if I want to set a response with type DRESP1, 'static stress', for stresses xx and yy, those are the types: normal x and normal y, respectively, right?
Altair EmployeeThank you.
Then if I want to set a response with type DRESP1, 'static stress', for stresses xx and yy, those are the types: normal x and normal y, respectively, right?
correct. DRESP1 with these directions.
Just take care with element/material orientation. Element stress values are related to the element orientation (in general node 1 to node 2). Depending if you need a specific direction it would be good to define the material orientation.
I would go for maximum/minimum principal stresses.
you could do it 2 ways:
1) inside the free-size deisgn variable, you could add a global stress constraint, under 'parameters'>>'stress constraint'. It will give you some control over the average stresses, then specify your limit stress. It applyes to all your model and all static loadcases
2) create a response of type DRESP1, 'static stress', and then assign it to a opt constraint and define the upper bound and the desired loadcases. This one you can specify region and loadcases, but depending on your model it could take longer to compute, as sensitivities need to be calculated for all your design variables.