unexpected behaviour nonlinear truss elements

[lukanar]

Dear SeismoSoft,
I'm writing you about unexpected behaviour of truss elements that I've found. I'm modelling a quite large existing r.c. building and I would insert dissipative bracings as seismic retrofit technique.
At first I've modelled such dissipative bracings using truss elements with nonlinear material behaviour (st_bl) and rectangular cross section discretized in 100 fibers (as default). After a pushover run, the internal axial force of such truss elements was equal to 0 for all step! But the capacity curve was completely different from the one without bracings.
I've tried several attempts and decreasing the number of fibers for single element the result changes but it was rather unstable (very large values alternated to very low values). I've also tried with a purely elastic material behaviour but with no significative improvements.
I've also tried varying the integration step, the integration method, the tollerances etc but they do not work.
Obviously with reduced size model (two floors, two bays) with the same properties, everything was reasonable and looked fine.
Thus I'm actually trying with another modelling solution: replace nonlinear truss element with a nonlinear beam-column member with two link elements at the ends (link = elastic behaviour with extremely high axial and shear stiffness and very low rotational stiffness).
I would ask if you have already found similar problems with truss elements and which can be the reasons for such unexpected behaviour.
Thank you very much for your kind attention and help.
Luca

[seismosoft]

Dear Luca,

No, we have not experienced such behaviour (truss elements malfunction in large models), and cannot find a reasonable explanation for it. If you send us both the reduced (working) and large (non-working) models, we may try having a look at it.

Thanks.

Seismosoft Support

[lukanar]

Dear SeismoSoft,
I've sent you the working and non-working models at the email address info@seismosoft.com.
Thank you very much for you collaboration and help.

Luca

[Andre]

Hi,

Actually there is a topic in this forum created by RICHARD named "truss output forces and displacements" where similar problems are reported.

I am also having problems with truss elements right now. In pushover analyses, braces in compression are recovering its elastic stiffness a while after yielding. This happens at the same time in all the storeys, even though braces at different storeys have different stiffness and strength.

I have no idea why this is happening.

Thank you in advance.

[seismosoft]

André,

The problems reported by Richard have been solved, and the new release of the program will not feature such issues.

The possible problems reported by Luca seem to be of a different nature, and we are currently looking into it.

Feel free to send us your problematic model, if you wish us to have a look at it.

Seismosoft Support

[lukanar]

Dear Seismosoft,
After last messages, I’ve tried other modelling strategies for dissipative bracing system. The possible modelling solutions that I’ve tested on reduced and complete models for dissipative bracings are:
a) nonlinear truss element;
b) nonlinear beam-column element (infrmFB) + two “hinge” (link elements) at the ends;
c) very stiff linear elastic truss element + nonlinear link element;
d) 1 “hinge” (link) element + very stiff linear elastic beam-column element + nonlinear link element;
e) infill panel element with zero shear stiffness and tension only struts (bl_asm costitutive law).
All these solutions provide with a good approximation the same results on simple portal frame + diagonal bracing element, but when adopted for the complete building model some of them cause some problems also in the permanent load step. In another topic “modelling repair braces” by Andre, you suggested to use such modelling strategy to avoid the bracings carrying permanent vertical loads. In all the modelling strategy from a) to d), the bracing elements should theoretically absorb some permament loads, but not in e) solution due to the peculiarity of infill panel element implemented in SeismoStruct. With the last solution e), I’ve obtained something that looks reasonable (but I’ve still to complete the check).
I guess if such aspect be one of the causes of unexpected behaviour of nonlinear truss elements in the complete model.
Thank you very much for your kind help.

Luca

[seismosoft]

Dear Luca,

We can confirm that there was indeed a bug in the program, which will be solved in the upcoming release of SeismoStruct.

Thank you for you patience and support,

Seismosoft Support