Topology optimization question

[ParabolicTet]

Once you run topology optimization, how do you interpret the element density plots to know how to size your design? I have been using anything higher than 0.3 density to include in the design as suggested in the Optistruct training.

 

[mloew]

ParabolicTet,

Good question. Part of the answer is that it depends on what your objective function is. If your goal was to minimize mass while constraining displacements, stresses, etc. then the density results should be those that meet that criteria. More often I think that the results of a topological optimization can only be used as a template for visualizing the required load paths.

Think of it this way, if you took the results of a particular density output, remeshed that, and performed the same FE analysis (without the optimization loop) and the design constraints were not met, then the density was too low.

I hope this makes sense. I am not opining from a position of great experience.

Best regards,

Matthew Ian Loew
"I don't grow up. In me is the small child of my early days" -- M.C. Escher

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips Fora.

 

[ParabolicTet]

I am trying to minimize mass subject to a displacement constraint. In other words I am saying cut away at this block of material till the tip displacement is no more than some number. Once it converges how do I know how much material I need to achieve the same result?

http://web.peoriadesignweb.com/eos

 

[mloew]

ParabolicTet,

Is there no way to have the post-processor step through the results? For each density iteration there is a corresponding displacement result calculated right? Pick the lowest mass structure that meets the constraint. I hope I am helping, I have very little experience with HyperMesh.


Best regards,

Matthew Ian Loew
"I don't grow up. In me is the small child of my early days" -- M.C. Escher

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips Fora.

 

[ParabolicTet]

There are element densities available at each iteration. However the densities are not discrete values, they range from 0 to 1. So some elements have 0.8 while others could be 0.1. OptiStruct says to use 0.3 above in your design. I am wondering what the reasoning is for that and if it is accurate.

Thanks!

 

[mloew]

ParabolicTet,

My guess is that experience has shown that structures derived from topological optimization runs encompassing densities > 0.3 give reasonable results. I would be looking for a way to set up the post-processor to determine what the greatest threshold density would be such that the resultant structure will still meet the design constraints.

Best regards,

Matthew Ian Loew
"I don't grow up. In me is the small child of my early days" -- M.C. Escher

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips Fora.

 

[feajob]

Parabolic Tet,

I am using Optistruct, I think that HyperMesh guys found 0.3 only by experience. I don't think that there is solid theoretical reason behind this number.

I use this number only to find out the load path on my structure, then I keep the material suggested with this number for the first run of shape optimization. I usually get my final design (optimal design) by doing some shape optimization study.

Well, most of the time manufacturing considerations plus suggested material (with 0.3) gives me only the first design. Final design is coming from shape optimization.

I hope that this helps you.


A.A.Y

http://www.geocities.com/fea_tek/asd/