different results for same optimization


i performed a parametric study including an optimization for each set of parameters. In the parameter study some sets are equivalent with respect to the model (that is due to the fact how i defined the parameter), but the results are not.
What may be the reasons that the same model produces different results? Even an optimization study should result in a unique solution unless the model or the starting point or the objective function or the optimization technique are different. But they are not, as all optimizations were performed within the same parametric study.
What other input parameters can be responsible for the different results?

(To make clearer why different parameter values result in the same model: I changed the coordinates (parameter 1) of points of different segements (parameter 2). If i change the coordinates by 0 mm, it is not important of which segment the coordinates were changed.)

Thanks for AnyIdea,


Hi Thom,

I think i get your point. My thought about this is that the sets of parameters you belive to be equivalents are in fact not, for some reason, probably something forgoten in their definition.

Of course it is hard to tell without the model, i can’t help you much more right now. You can upload the model to the debug section if you want us to look at it.

Best regards, Sylvain.

Dear Sylvain,

thanks for your offer to look at my model. But i would rather like to be able to find the problem than knowing the source of the problem in this specific case. Therefor i have some questions:

1.) Is there a possibility to debug the model, e.g. dump some values during the calculation or is it only possible to dump values after the analysis with the right mouse click?

2.) When a parameters min is -1 and max=+1 with nStep of the ParamStudy being 2, are then -1 and +1 used or values in between?

3.) What does the documentation about AnyDesVar mean: “Val is associated with a special evaluation moment for design variables”? What is the evaluation moment? Am i correct that i would get an error message when i would try to vary a parameter that would not be allowed to vary? I want to change node coordinates within a parameter study. Is this all right?

Thanks for your help,


Hi Thom,

  1. I think you can only dump the values in text file after the simulation is done.

  2. With nStep=2 in the parameter study then the two extrem values are used (max and min). The max and min are always used unless nstep is 1.

  3. Each object is evaluated at a certain moment during the loading. The design variables have a special status as they are updated during the simulation. It is perfectly all right to change the node coordinates (sRel). Lots of object can be used as design variable, but the system will complain if you try to use one not allowed.

Best regards, Sylvain.