modelling repair braces

[Andre]

Hi, I'm modelling braces used to retrofit an old structure with inelastic truss elements and, as I don't want them to absorb permanent vertical loading (I'm performing pushover analysis), I tried to use repair links on the upper brace-column connection. However, as the relink isn't activated during the permanent loading, the braces become mechanisms and the permanent loading phase doesn't converge. Is there a way of overcoming this problem? A way of keeping the braces from absorbing vertical permanent loading before the horizontal loading beggins? And, by the way, should this work, would the braces get vertical loading after the relinks were activated? In other words, is the permanent loading redistributed?

[ricardomonteiro]

Hi Andre,
I believe the most suitable way for you to do so is to model your truss elements as tension resisting only, assuming that you accept that sort of behaviour during the pushover analysis as well.
As the braces would get part of the vertical permanent loading as soon as they were 'activated', this should be the kind of solution that would take care of that issue as well.
The modelling of tension-only braces has been already discussed in this forum, in the 'Tension only bracing' topic. Maybe you can have a look at it.
Cheers,
Ricardo

[Andre]

Thank you, it's a good idea. However I need 'compression too bracing' so the idea doesn't apply.

[seismosoft]

André,

The best way to achieve what you describe would obviously be to have truss elements would an activation time, something that currently is not available in SeismoStruct (though it is in our to-do list for near-future implementation).

In absence of such a modelling capability, the ingenuous workaround that you describe would be the best option indeed, but, as you stated, numerical problems have arisen, which is unfortunate.

We have not thought deeply about this, but we wonder if you could think of using the infill panel element, without any shear strength and with its diagonal struts properly calibrated in order to make them reproduce the behaviour you require? The reason behind this suggestion is because the infill panel is activated only after the initial loads have been equilibrated, as you also require.

Finally, regarding your second question, no, the initial loads will not be redistributed to the struts when these are activated, since the columns will have absorbed them fully through axial deformation.

Seismosoft Support

[Andre]

Ok, I tried in a very simple example that I can send you but the results weren't good. I defined the strut parameters as in a normal elasto-plastic brace (k0 and Fy)and 0 shear stiffness. The rest of the parameters I have no idea of how to evaluate them. Can you give me a hint? I could send you my model. It has only 2 columns and 1 infill panel.

[seismosoft]

André,

The "other parameters" (width, thickness, etc) would need to be set in such a way that the infill struts have a similar axial capacity to the brace.

One other modelling workaround could perhaps be to introduce link elements between the trusses and the beam-column nodes, featuring a gap corresponding to the vertical settlement of the frame due to vertical loads. Since you are running pushover analysis only, it should work.

Good luck with your modelling.

Seismosoft Support

[TUC]

"Now" that the option of Activation-Deactivation Time/L.F. is implemented it is a very good solution in such case of bracing for retroffiting purposes...
May i ask if the value of activation time that should be assigned at the bracing elements can be obtained after making an initial analysis with Permanent loads (i.e gravity loads) in the model without braces and observing the stress state of an arbitrary element and then make a second analysis, concernig again the model without braces, with Permanent plus incremental loading in order to obtain the time step corresponding to the same stress state of the previously selected element?
Or is it maybe any other more proper/rational way of obtaining the correct value of activation time..?

Thanks

[Stelios_Antoniou]

If you want to activate the braces right after the application of the initial gravity loads, any time between zero and the L.F./time of the second step of the analysis (i.e. the first after the initial loading) will be fine. I would suggest 1e-8, i.e. an extremely small value larger than zero

Stelios Antoniou
SeismoSoft Developer

[TUC]

Do you mean that at the 1st step of an incremental analysis, the equilibrium state due to permanent (I.e gravity) loads has been already achieved?

[Stelios_Antoniou]

Yes, at the first step of the incremental (pushover) analysis no incremental loading is applied yet, and the qequilibrated state reflects the initial (i.e. gravity) loading