Composite size optimization

Altair Forum User

Hi,

 

I am trying to do a size optimization of a composite plate. My plate is composed of 8 small plates. I have run 2 models with the same parameters except that one is based on PLY card: each plate has 4 plies (0/45/90/-45) with a STACK (SMEAR). The other model is based on PCOMP, one per plate with the same plies and the SMEAR card.

 

Can you explain me why the results of the size optimizations are different?

 

Thanks,

Ludov

<?xml version="1.0" encoding="UTF-8"?>

Altair Forum User

Hi Ludov,

 

can you share both the models?

Altair Forum User

The 2 models are attached.

 

Thank you.

Unable to find an attachment - read this blog

Altair Forum User

Hi,

 

Thanks. I will check and get back to you.

Altair Forum User

HI,

 

Even though the analysis models are the same, the optimization setups are quite different.  In the PCOMP model there are 4 design variables (1 per ply in the PCOMP), while in the PLY model there are 32 design variables (1 per PLY).  That’s why the results at the first iteration are identical, and the results diverge during the optimization process.

 

To obtain the same results, you would have to reference the same design variable for all 0deg plies, same design variable for all the 90deg plies, and so on.  Alternatively you can use design variable linking (DLINK) to achieve the same goal.

 

Hope that helps,

Altair Forum User

Hi Prakash,

 

In the PCOMP model, I have set up 8 PCOMP (one per small plate) so there are 32 design variables as in the PLY model.

So I'm not shure to understand why the results diverge during the optimization process if they are the same number of design variables with the same bounds. There is a difference in the setup of the DVPREL1 card : in the PLY model, each DVPREL1 has its property which is equivalent to the ply so there are 32 properties while in the PCOMP model there are only 8 properties (the 8 PCOMP). Could it be the reason of the divergence ?

 

Thanks for your help

Altair Forum User

Hi Ludov,

 

Yes, there are 8 PCOMP properties buy only one is applied to all the elements (property ID 444).

 

When I checked for empty properties, 7 out of 8 are empty. Am I missing something?

Altair Forum User

Hi,

 

In 'PCOMP Model.fem', there are 8 PCOMP (ID 111 - 888) and all are applied to a part of the CQUAD elements.

In 'Ply Model.fem', there is 1 PCOMPP (ID 444) and it is applied to all the elements, the plies are defined by the PLY card.

 

So I don't understand why in my PCOMP model 7 out of 8 are empty. Is it impossible to set up several PCOMP cards in an optimization ?

Altair Forum User

Hi Ludov,

 

 

I will try to model a PCOMP and PCOMPP with your approach and will share the same with you.

Altair Forum User

Ok thanks for your time.

 

I am continuing the optimization of this ply model and I am trying to set up a shuffling optimization.

I have several plies on each small plate and now I need to optimizise the stacking of the whole plate.

So I need to have a continuity of the orientation of the plies of each small plate at the same layup of the laminate.

Is it possible to do it easily with Optistruct ?

Altair Forum User

Hi Ludov,

 

I need to investigate on this, but I think this can be achieved using DLINK card.