Re analysis of optimized model

hemanth kollipara_20292
hemanth kollipara_20292 Altair Community Member
edited April 15 in Community Q&A

Hi, I have performed topology optimization of a model and I would like to perform FEA for the optimized model. Here is the technique that I am using. 

1) First perform optimization

2) go to OSS smooth 

3) select OSS with FEA Reanalysis option 

4) select (.fem file ) and (.sh file) 

5) run the solver

But after performing this operation I can see that the model that was generated in optimization is different from the model that got abstracted into hypermesh. Also I can see some entities like voids are filled with solid material instead of remaing as voids.  

Tagged:

Answers

  • Adriano_Koga
    Adriano_Koga
    Altair Employee
    edited April 9

    hi,

    this is normal, as there is a step of smoothening going on in the process.

    You could play around with the threshold value in order to have a more aggresive or more conservative overall result. I.e: by increasing the threshold value, it would create larges void areas, but also might cause some discontinuities in your model. By decreasing the threshold, you might get a smoother final part, but some of the voids might be closed.

    But most of all, you should remember that Topology Optimization is a conceptual design technique, and the resultant geometry is not the final one, as adjustments will be needed, concerning design and manufacturing enhancements.

  • hemanth kollipara_20292
    hemanth kollipara_20292 Altair Community Member
    edited April 9
    Adriano A. Koga_21884 said:

    hi,

    this is normal, as there is a step of smoothening going on in the process.

    You could play around with the threshold value in order to have a more aggresive or more conservative overall result. I.e: by increasing the threshold value, it would create larges void areas, but also might cause some discontinuities in your model. By decreasing the threshold, you might get a smoother final part, but some of the voids might be closed.

    But most of all, you should remember that Topology Optimization is a conceptual design technique, and the resultant geometry is not the final one, as adjustments will be needed, concerning design and manufacturing enhancements.

    Hi, I am aware of the threshold but the covering of the void is too much. Moreover it changed the entire shape of the design space. Why is this happening ? Is it because of the method I am trying? I have attached the two images, where you can understand the problem. 

    Also I have seen your reply at "https://community.altair.com/communityid=community_question&sys_id=078fa0f21b68cd18507ca6442a4bcbdc"

    I am trying in the same that you explained but I don't know why I am getting this problem

  • hemanth kollipara_20292
    hemanth kollipara_20292 Altair Community Member
    edited April 9
    Adriano A. Koga_21884 said:

    hi,

    this is normal, as there is a step of smoothening going on in the process.

    You could play around with the threshold value in order to have a more aggresive or more conservative overall result. I.e: by increasing the threshold value, it would create larges void areas, but also might cause some discontinuities in your model. By decreasing the threshold, you might get a smoother final part, but some of the voids might be closed.

    But most of all, you should remember that Topology Optimization is a conceptual design technique, and the resultant geometry is not the final one, as adjustments will be needed, concerning design and manufacturing enhancements.

    Furthermore, I have studied this article from the help section 

    "https://help.altair.com/hwsolvers/os/topics/solvers/os/topology_optimization_of_control_arm_local_stress_constraint_os_example_r.htm?zoom_highlight=fem+control+arm#reference_xzb_vc4_2cb__Untitled4_fig_fjv_lb4_2cb"

    Is this method only applicable for constraint and objective function defined in the problem? or can it applied for any other design contraint?

  • Adriano_Koga
    Adriano_Koga
    Altair Employee
    edited April 9
    hemanth kollipara_20292 said:

    Hi, I am aware of the threshold but the covering of the void is too much. Moreover it changed the entire shape of the design space. Why is this happening ? Is it because of the method I am trying? I have attached the two images, where you can understand the problem. 

    Also I have seen your reply at "https://community.altair.com/communityid=community_question&sys_id=078fa0f21b68cd18507ca6442a4bcbdc"

    I am trying in the same that you explained but I don't know why I am getting this problem

    i'm sorry. The pictures are in different orientations, I can't see exactly what you're trying to show.

    But have you tried different threshold values an how those affect the results?

    In theory the threshold will just remove more or less elements from your final result, according to what you can see in the 'iso' plot in HyperView.

    So, usually, I go to HV, plot the iso for density, adjust the iso value threshold there, and then use the same in OSSMOOTH to get a reasonable conceptual design.

     

  • Adriano_Koga
    Adriano_Koga
    Altair Employee
    edited April 9
    hemanth kollipara_20292 said:

    Furthermore, I have studied this article from the help section 

    "https://help.altair.com/hwsolvers/os/topics/solvers/os/topology_optimization_of_control_arm_local_stress_constraint_os_example_r.htm?zoom_highlight=fem+control+arm#reference_xzb_vc4_2cb__Untitled4_fig_fjv_lb4_2cb"

    Is this method only applicable for constraint and objective function defined in the problem? or can it applied for any other design contraint?

    everytime that OS runs an optimization run it will indeed calcuate the loadcases for each design iteration, BUT considering all the elements there, even those with small densities. I'm not sure if I got what was your question, sorry.

  • hemanth kollipara_20292
    hemanth kollipara_20292 Altair Community Member
    edited April 9
    Adriano A. Koga_21884 said:

    i'm sorry. The pictures are in different orientations, I can't see exactly what you're trying to show.

    But have you tried different threshold values an how those affect the results?

    In theory the threshold will just remove more or less elements from your final result, according to what you can see in the 'iso' plot in HyperView.

    So, usually, I go to HV, plot the iso for density, adjust the iso value threshold there, and then use the same in OSSMOOTH to get a reasonable conceptual design.

     

    Yes, I saved the file with 0.5 ISO threshold, and then opened by performing OSSsmooth>>FEAReanalysis>> selecting (.fem file) that I got from optimization or selecting current model>> selecting(.sh file)>>applying threshold value as 0.5>>run.

    But still I am obtaining a different model after the run.

Categories