Tension only member and Response Spectrum Analysis

[stanchan]

Since Response Spectrum Analysis is a linear analysis, it is unable to account for the tension only Braces in a braced bay.

What is the best way the model tension only braces while performing RSA?
I am trying to model my braces with a 0.5 stiffness modifier, so that the brace bay would act as if it was a tension only brace bay in RSA. Is the method generally valid? or would I get completely erroneous results?

Another thing that I notice is that RSA only show Maximum axial forces in elements. This is leading me to think that RSA theory doesn't distinguish between positive and negative force? Is there any way of finding compression forces?

Any advice would be much appreciated!

 

[JoshPlumSE]

Stanchan -

It is not possible to run an RSA analysis with tension only or compression only elements. So, if you have an X brace type configuration (or any configuration where there is an equal / symmetric arrangement of braces) then it makes sense to use a 0.5 multiplier on the axial stiffness of the brace. That should work pretty well.

When you get your analysis results the axial forces in the braces are equally likely to be tension or compression.... you don't know. So, in an X brace configuration you would add them together. The total sum is the maximum axial force in the tension brace. That works great if your assumption is that the movent of the structure is almost entirely lateral.

Note: I'm not a SAP guy at all (I actually work for a SAP competitor). But, I can't imagine that this basic concept would be different between programs.

 

[stressed]

RSA is by definition a frequency domain linear analysis, so nonlinear boundary conditions cannot be considered. If you wanted to go beyond your .5 multiplier approximation, you would need to analyze as a NL modal time history analysis using Ritz vectors (or NL Direct Integration). A NL time history analysis would consider tension/compression limits in the dynamic analysis (you can apply TH functions to acceleration loads or other assigned load patterns) and also provide + - directional output results

 

[johnbridge231]

A trick for the case of X bracing is to assign an axial force modifier equal to 0.5 to ALL braces and assume that both tension and compression member are active for horizontal loads. In this way, you "linearize" the problem and a linear response spectrum analysis is allowed. The global behaviour of the structure will be as the whole tensile brace worked and the compressive did not carry any force.

Regards.

Analysis and Design of arbitrary cross sections
Reinforcement design to all major codes
Moment Curvature analysis

http://www.engissol.com/cross-section-analysis-design.html

 

[stanchan]

Thank you so much for all your advice!!! Problem solved!

 

[bkal]

I think it is important to remember, in case you that are using a 0.5 factor, that forces in members to which the braces are connected might then be incorrect.

 

[johnbridge231]

bkal, for sure, but you will be able to perform a response spectrum analysis. Then you should multiply the axial force of the braces by 2.
Which one is easier: A multiplication by 2 or a Non-Linear Time History Analysis?

Analysis and Design of arbitrary cross sections
Reinforcement design to all major codes
Moment Curvature analysis

http://www.engissol.com/cross-section-analysis-design.html

 

[bkal]

Johnbridge,

I was not advocating use of N/L analysis; it was just a reminder. Also, use of 0.5 factor will affect not only the bracing members, but also the force in the members connected to the braces.

 

[johnbridge231]

Alright, thanks for the contribution.

Analysis and Design of arbitrary cross sections
Reinforcement design to all major codes
Moment Curvature analysis

http://www.engissol.com/cross-section-analysis-design.html

 

[dgerashki]

There is another approach. You can only leave one diagonal of the X-bracing and design only for the tension force. In that way you will have correct values for the members that are connected to the diagonal (column base, footing, etc.).
To design the column and its base on the other side of the bracing, you can make a second model with mirrored diagonals.

 

[johnbridge231]

But, since the quake load is taken as positive and negative in relevant seismic combination, compression will 99% occur in this case as well!


Analysis and Design of arbitrary cross sections
Reinforcement design to all major codes
Moment Curvature analysis

http://www.engissol.com/cross-section-analysis-design.html

 

[dgerashki]

Yes and only the tension force must be taken into account, as the compression member is assumed to be buckled.

 

[johnbridge231]

In Response Spectrum Analysis, since the earthquake is applied as + and - , tension will also occur in this case. You can try it.

Analysis and Design of arbitrary cross sections
Reinforcement design to all major codes
Moment Curvature analysis

http://www.engissol.com/cross-section-analysis-design.html