Iteration stops and optistruct error termination report printed to file- showing

Tanzila

I am doing a simple topological optimization on a cantilever beam with point load at it's end. It is a stress constraint topological optimization. I created one response mass, objective to minimize mass and described the stress constraint within topology panel. I am doing the simulation in HPC. Now upto mesh 3d element number like 160000, the result converged. But when I increased the mesh element to 185000, it is being stopped at certain iteration and showng this- optistruct error termination report printed to the file 'Ele_den_21.stat'. I have attached two of the output file. In the first one iteration stopped at 2nd iteration, then I run it again and then the same file stopped at 18th iteration. 

What can be the reason, how can I resolve it and get the converged result? From another conversation of the forum I came to know this might happen due to internal error. I don't know what does that mean. I have designed the model same way i had designed the other which were having a converged result.

 

In HPC optistruct is using in core or min core processing. And I am running the solver by writing command in command line.  

Unable to find an attachment - read this blog

Adriano_Koga

I can't tell exactly what is going on, but there's a warning in both .out files.

 

 *** INFORMATION # 6356
 MAXDIM constraint is activated on DTPL card 1. MINGAP constraint of
 the same value as MAXDIM is automatically turned on. MAXDIM and MINGAP
 constraints introduce significant restrictions to the design problem.

 

You have a minimum member size of 0.01 and a maximum member size of 10.0.

This usually is not recommended.

MINMEMB in general is 2~3x average element size.

MAXMEMB should not be too large as well. Not quite sure if it is 12x average element size...something like this..

So unless you really need these minmember and maxmember, maybe you should try reviewing them.

 

 -------------------------------
 Topology Optimization Summary :
 -------------------------------

 -------- ----------- ----------- -----------
 DTPL ID    Minimum     Maximum     Minimum  
          Member Size Member Size  Gap Size  
 -------- ----------- ----------- -----------
        1  0.1000E-02  0.1000E+02  0.1000E+02
 -------- ----------- ----------- -----------

 

 

 

As a minor warning, check where is your model running. If the scratch folder is in a different machine this will really slow computations.

*** WARNING # 9113
 It appears that the directory used for scratch files:
    ./
 is located on remote filesystem. 
 This could result in very slow solver performance.
   Number of CPU processors        :          1