Size optimization don't minimize the objective function

Campo
Campo Altair Community Member
edited September 2023 in Community Q&A

I have a problem with size optimization of a composite shell. In a nutshell, I notice that the global mass (objective function to minimise) increases even when all constraints are satisfied, so the optimization ends with non optimal solution (see pictures). Why?

I want specify that I also performed the same optimization with less stringent convergence criterion and in that case the problem doesn't occur. In my opinion it is very strange. What am I doing wrong? 

The model has only two loadcases and three constraints. The objective is the minimization of mass. 

imageimage

 

Tagged:

Answers

  • Adriano_Koga
    Adriano_Koga
    Altair Employee
    edited September 2023

    could you double check your problem formulation for the design variables?

    From your images i can see that the manufacturing constraints are active and they're at the maximum (upper bound). This means that the optimizer is probably trying to add more thickness to your your part, but it has no place to go further. Maybe this is causing some troubles.

  • Campo
    Campo Altair Community Member
    edited September 2023

    Thank you so much for your suggestion.

    I double checked the problem formulation but I didn't find wrong things.

    I understand that the activation of upper bound of manufacturing constraints might cause some problems but I don't understand why the solver tries to add mass when the constraints are largely satisfied. 

    In my opinion, It should try to reduce the thickness in other areas rather than adding mass by increasing thickness. Moreover, I think it should try to reduce thickness in order to reach the lower mass limit, even at cost to finding some configuration where the constraints are violated

    I performed several tests to different models to understand the behaviour of size optimization but the process is still not clear for me. For example, some times it adds mass even when the constraints are satisfied and then it stops the optimization at iterations that are worse than previous ones. I'm really confused about its behaviour.

    Could you help me to understand what it does?
    Is there any output file that can give me more information about optimization process or optimization parametesr that could avoid this problem?

    I've already tried to use DGLOBAL card but problems still occur in the optimization with some starting points. 

Categories