Different displacement results between optimization and analysis

Hello everybody,

I wanted to ask a question about the results in optistruct. This is a general question as i tried with different problems and the results are the same.

I have a problem with a constraint on the displacement of some points and the mass as objective function.
I run a free-size optimization and the result is a feasible design with no errors. The resulting displacements are below the requested constraint, so all it is ok.

Now I load into Hypermesh the free-size output model (last iteration) and I measure the mass: it is different from the one in output from the free-size optimization. This is the first problem.
Then I run an analysis and the resulting displacements are different from the ones in output from the free-size optimization. This is the second problem.

Could you explain me why I have these problems?

Thanks,

SPK

Find more posts tagged with

English

OptiStruct

Accepted answers

All comments

tinh

Hello everybody,

I wanted to ask a question about the results in optistruct. This is a general question as i tried with different problems and the results are the same.
I have a problem with a constraint on the displacement of some points and the mass as objective function.
I run a free-size optimization and the result is a feasible design with no errors. The resulting displacements are below the requested constraint, so all it is ok.
Now I load into Hypermesh the free-size output model (last iteration) and I measure the mass: it is different from the one in output from the free-size optimization. This is the first problem.
Then I run an analysis and the resulting displacements are different from the ones in output from the free-size optimization. This is the second problem.
Could you explain me why I have these problems?
Thanks,
SPK

what type is your model, shell or solid?

If it is shell (2D) please check if you did not run the output .cmf file to classify thickness values, mass and disp will not correct

to see last iteration results you can turn to page2 on hv, i think running it again is not necessary

Altair Forum User

Thank you for the answer.
The model is of shell type. How can I check if I run the .cmf file or not?
I know that if I turn to page 2 on hv I will see the results, but I think that if I run an analysis on the optimized model I hope to obtain the same results given by the optimization output.

Regards,

SPK

tinh

Thank you for the answer.
The model is of shell type. How can I check if I run the .cmf file or not?
I know that if I turn to page 2 on hv I will see the results, but I think that if I run an analysis on the optimized model I hope to obtain the same results given by the optimization output.
Regards,
SPK

Hi, did you see any file .cmf in same folder with .h3d file

after load model file of last iteration into hypermesh, you run the .cmf file by menu Files>Run>Command File

Altair Forum User

Hi Tinh,

I have that file in my directory and I did what you said but nothing has changed. I mean, I have new sets with the various densities, but I have the same results when I run an analysis of the optimized model.

To understand my problem, you can try running a simple problem like a plate simply supported at the four edges: you will probably obtain different results between free-size optimization and analysis of the optimized model (both mass and displacements will be different).

Thanks for your time,

SPK

tinh

Hi Tinh,
I have that file in my directory and I did what you said but nothing has changed. I mean, I have new sets with the various densities, but I have the same results when I run an analysis of the optimized model.
To understand my problem, you can try running a simple problem like a plate simply supported at the four edges: you will probably obtain different results between free-size optimization and analysis of the optimized model (both mass and displacements will be different).
Thanks for your time,
SPK

You ran '.ent.cmf' file

Please try running '.comp.cmf' file

do not load last iteration FEM into hypermesh. Just open original '.fem' file and run '.comp.cmf' file, it will organize elems into difference thickness

after that you have to make properties and assign thickness values to them

If you model is big, a tcl macro is neccessary to assign thickness values, so looking for such macro in Script Exchange

Altair Forum User

I do not have the .comp.cmf file but only the .ent.cmf file. Do you know why?

I tried to activate the output card CMF but nothing changed.

Regards,

Spk

tinh

I do not have the .comp.cmf file but only the .ent.cmf file. Do you know why?
I tried to activate the output card CMF but nothing changed.
Regards,
Spk

Oh, I am not sure about the cause

You can try running optimization again

Altair Forum User

Hi SPK,

Usually after the completion of topology optimization a .comp.cmf file will be written in the working directory. Can you re-run the optimization again to cross check?

If no .comp.cmf is written, please let us know what is the version of HyperWorks you are using and along with the updates.

Please use the link below for information on how to check for updates:

http://forum.altairh...n-your-machine/

Altair Forum User

Hi Prakash Pagadala,

I checked and my version is the latest available,

Probably this can help you: I am not doing a topological optimization. I know that this file is written only with topological optimization, isn't it?
I read and did the skateboard tutorial (http://training.altairuniversity.com/wp-content/uploads/2013/07/Skateboard_tutorial_Prakash_MGJuly1.pdf) and the results are the same.

Regards,

SPK

PabloCabezas

I have the same problem; when i run the free-size optimization phase, minimizing the mass of the model, an putting tsai-hill as a constrain with an upper bound of 0.8; i obtain a feasible design with any constrain violating. In the _s#.h3d, i see a value of tsai hill of 0.78 after the optimization, but; when i run an analysis of my optimized model (sizing.#.fem), with exactly the same load cases, i obtain a different value of tsai hill failure (actually 2.55); I don´t know why is this happening; as you say, it should be the same value; furthermore, this is a problem if you want to perform a sizing optimization (2nd Phase), as you start from a 'false' value