Cyclic pattern grouping leads to infeasible design

Altair Forum User

Hello,

 

I have a question concerning the application of a cyclic pattern to my optimization. I have a cylindric shaped work space for my optimization and apply a moment at the end of this pipe. The second end is completely restrained. Now, when I run the simulation without the cyclic pattern it runs through in about 18 iterations and finds a design, which makes sense. However, in the final product, I will have moment loads in multiple directions, so I wanted to apply the cyclic pattern to find a design, which will function in every direction.

I started with 4 Ucycs and got an infeasible design. After that, I reduced the symmetry to 3, then to 2, and now I have just 1 Ucyc. Still, my optimization runs 80 iterations without finding a feasible design... How can this be? Is there an additional setting I forgot to apply?

 

Thank you very much for your ideas!

Altair Forum User

Hi @Kensei

 

Can you share the .out file?

 

Meanwhile can you increase the number of iteration limit to say 200 and check if this runs beyond 80 iterations?

Altair Forum User

I just sent you the out file of my latest optimization run.

 

Okay, that is a good idea, I will increase the run limitation to 200 and see where this leads.

Thanks for your help!

Altair Forum User

Hi,

 

The limit of number iterations reached the maximum limit. You can increase the limit and check if you are able to see any progress

http://www.altairhyperworks.com/pdfs/tips/1088.pdf

Altair Forum User

Hi,

 

you were right. It was just the iteration limit. After a bit of trial and error I found, that 220 iterations are enough to find a feasible design.

However, the resulting design does not really make sense. The element ensities are way too low (around 0.015). This can't be normal behaviour...

Could you have a look on my complete model to make sure?

I sent my hm, and h3d files to your FileTransferLink. Thank you very much for your help!