Seismosoft | Earthquake Engineering Solutions

hello seismostruct !
I found an ucexpected error to report
i made 2 model similre to each other just with difference in defintion of diaphragms in story levels
in one model master node is in corner of storey
and in other model master node is in middle of story
resulting periods and other results are very different while any other difference does'nt exists in two model.
in help menu of seismostuct is mentiond that master node doesn't influence in model and results
so what's the problem?
models are available in below urls:

http://www.4shared.com/file/239581801/4 ... value.html
http://www.4shared.com/file/239581830/1 ... alue2.html

grateful for your attention so soon

Hi,

There is no error in the program; the behaviour you observed is expected and normal. Indeed, when modelling a rigid diaphragm, the master node should obviously correspond to the slab's centre of gravity.

The Help System comment to which you allude to refers to 'rigid links'; we will clarify this in future versions.

Seismosoft Support

Hello,

I also have problems with constraints - rigid links.
I am modelling an arch bridge. In the model arch is defined as one beam while the piers above arch are eccentrical to it, so I have to connect them somehow. When I use rigid link (arch node is master, and base nodes of two piers are slaves), I get imaginary solution with Warning: Could not re-orthogonalise all Lanczos vectors. On the other hand, when I connect the arch node to base nodes of piers with stiff beams, model works fine.

Could you tell me what I am doing wrong?

best regards!

Hi msavor. I'm not exactly clear on how this model is pieced together, but this could be a case - I have run into them myself with rigid links - where the Penalty Function parameters need to be tweaked for the situation in which you are using rigid links.

In Tools -> Project Settings -> Constraints, check to see if you are using Penalty Functions (the default) or Lagrange Multipliers.

You may need to gradually increase the Penaly Function exponent to arrive at a solution with rigid links. This is not always straightforward and I would recommend browsing over the "Constraints" Help topic, as well as the references listed there.

And I would also recommend a forum search on the word "penalty". You'll get a handful of hits.

Hopefully this will help you solve the problem, msavor. If not, at least you might be able to eliminate this as the source of the problem.

Best of luck.

Thank you very much for your answer. I will try to do what you recommended and if it still won't work properly, I will model connections as very stiff beams.
Best regards!

hello seismostruct,

is there any chance that I send you my model (very simple one) to check what is wrong with the constraints? I am very curious about it...

Hi Marta. In my experience, SeismoSoft is very good about a willingness to review models. So I am sure they will soon respond to your request.

In the meantime, if you wish, I would be happy to take a glance at your model and see if anything glaring leaps out to me. So feel free to send an e-mail with your model attached if you wish.

Marta, hi.

If you use Penalty Functions (e.g. with a value of 1.0E+12), rather than Lagrange Multipliers, you should get similar results between the two models (we tried it, and it worked).

As stated in the Help System, in this Forum, and in the technical literature in general, the use of numerical constraints is not always straightforward, hence experimentation is often required, as the model (or the analysis type) changes.

Seismosoft Support

Many thanks to Seismosoft support and also to Tim!
I was really curious why Lagrange Multipliers ("conceptually superior") lead to numerical difficulties in my model, but I will read some more literature to be clear about it.

Best regards!