Seismosoft | Earthquake Engineering Solutions

Dear Seismosoft:
My model has the folowing characteristics and when I run the model for an Eigenvalue Analysis an error occurs with the message "Range Check Error"!!

No. of Materials: 159
No. of Sections: 219
No. of Element Classes: 220
No. of Nodes: 4346
No. of Elements: 5493
No. of Loads: 0

Can you help me please??

I'm really worried...
Thank you
Marta

Dear Marta,

Without seeing the model, there is not much we can do, we are afraid.

Seismosoft Support

Dear Seismosoft,
How can I send you the model?
Thanks,
Marta

Please use the same email address indicated in previous posts: support@seismosoft.com

Seismosoft Support

Dear Marta,

Typically, messages as the one you report occur when a model features an error that the program did not ever had to deal with before. Hence, we will need to look into the model and try to understand what is that might be wrong with it.

However, as you can imagine, it will be difficult for us to go through a 5000 elements model built by a third-party and find out what might be wrong with it. Hence, you might also try to have a go at its debugging yourself, perhaps by following the recommendations provided in a previous post regarding models that do not converge (do a Forum search on "non-convergence problems").

Some questions that immediately come to our mind, by looking very quickly at your model are e.g.:
- convergence criteria, concrete cover, gravity value, all seem to feature very small values
- structural members are being subdivided and still force-based frame elements with 5 integration points are being used (in such cases use either displacement-based frame elements or decrease the number of integration points, as described in the Help System and Forum)

Anyway, as we said above, please do check the past Forum post mentioned above, and work from there. In the meantime, we will also try to find the time to look into this problem.

Seismosoft Support

Dear Seismostruct
thank you so much for your attention to my problem.
I'll try to debug myself following your recomendations and searching all the previous posts about non-convergence problems as you advised me.
If meanwhile you have anymore recomendations or if you find anything wrong with my model please tell me. I'm very worried about the lack of time I have to solve the problem and to show some results...

Grateful for your help,
Marta

Dear Marta,

Unfortunately, and as feared, we found a very large number of errors in your model, which, when corrected, allowed the analysis to run without problems.

Some of these errors were:
- some foundation nodes (“N1k_scv_1” and “N2J_scv_1”) were unrestrained
- the “z” coordinate of node “PAR_B2_6” was conspicuously wrong
- the presence of elfrm “rigid” elements at the base of the walls was redundant
- the 'Project settings' featured a number of wrong values (e.g. concrete cover of .025 mm, etc etc)
- the rebars in many sections were either missing or wrongly placed (we replaced the sections with default ones, for simplicity)
- small infrmFB elements with too many integration points were being used (we replaced them with infrmDB elements)

We once again insist that users must check their models more carefully before reporting a problem, since we really cannot afford to provide this type of modelling assistance.

Seismosoft Support

Dear seismsoft,
I'm sorry the time I've taken you. I really appreciate all your help.

I was now defining de masses to assign at the floor levels - as it was one of the possible reasons.
I've already corrected points number 1,2,3 and 6 of your error list.

In point number 4, "the 'Project settings' featured a number of wrong values (e.g. concrete cover of .025 mm, etc etc)", how can I find that cover values or all those wrong values?

In point number 5, "- the rebars in many sections were either missing or wrongly placed (we replaced the sections with default ones, for simplicity)", I wanted to define walls without assigning the rebars, isn't that possible? Instead of defining a U-shaped wall section shall I define 3 rectangular wall sections?

Could you kindly send me the model you've corrected so I can compare with the one I'm correcting? My e-mail address is: mmateus@lnec.pt

Very grateful for all your attention,
Marta Mateus

Marta, hi.

We have now re-checked your corrected model, which you state to still be giving you 'Out-of-memory' or 'Range Check' errors, and we can tell you that the problem is that your model is extremely large.

Indeed, the size of its stiffness matrix skyline is close to 100 million, which means that you need approximately 1GB of continuous free memory, something that is unlikely to be obtained in systems with less that 4GB of RAM.

For models this large, ideally one should be using dedicated numerical servers, with plenty of processing power, memory capacity and storage space. In future releases, we will try to introduce a more user-friendly error message, so that users are better informed of what the problem is.

Thank you for signaling this issue to us.

Seismosoft Support