Mode Tracking in GSO

Torben0159

Hello,

I am currently doing size optimization with Optistruct. I want to optimize material parameters (5 design variables), so that the frequencies of some desired modes match the frequencies obtained form experimental data. I have done a modalanalysis and checked which mode number my desired shapes have (e.g. shape 1 = mode_number 7; shape 5 = mode_number 32 and so on). As responses I defined the frequencies at the mentioned mode numbers. I activated mode tracking, because the mode numbers of my desired shapes are changing throughout the optimization process. I performed an optimization, but the converged solution wasnt ideal. So I tried the Global Search Option with the hope of finding other local minima. So far this worked as planned, I am getting some better designs.

But there is one problem: the GSO uses different starting point afaik. With that, the mode numbers may also change (e.g. shape 5 isnt mode_number 32 anymore, but mode_number 36). This completely ruins the idea of the optimization, because now different modes are optimized which even have the wrong desired frequency. This means that like half my starting points are wrong and it would take a lot of time to find out which are. Can I somehow tell Optistruct to always use the originally defined mode number (and change it through mode tracking if necessary) for each starting point? Or is there another work around?

Adriano Koga_20259

wow...this is a hard problem to solve

 

your modes will naturally change as your starting points change.

I can't see an alternative for you in this case, unfortunately. But maybe someone has a good tip for you.

 

In your initial problem, are the constraints violated from the beginning? or they are respected? it might help start from a good initial point, with no violations (if possible).

 

 

Torben0159

Ty for your answer, I expected that this issue will be tough. I dont even have any constraints, just want to minimize difference between measured and simulated frequencies. I guess i will try some startpoints manually. Maybe you can consider this feature in future releases, all that needs to be done (in my eyes) is to identify the modes for the initial modalanalysis and then use mode tracking to adapt the mode number for other starting points.

Dominik T.

Hello together, I have the same problem right now.
Is there any progress in the meantime?

My case: "In an optistruct optimization there is the option of activating modetrak so that the correct modes are still optimized despite the shifting of modes.
Now I am optimizing several load cases with hyperstudy. How can I best ensure that the correct modes are optimized when the modes are shifted. So I need a kind of integration of modetrak in hyperstudy."