Tension-only members in RISA 3D 2

[tonycampos88]

Hello,

I am modeling some X braced frames and I want to use 3/4" threaded rods as my braces. I am trying to set them as "tension-only" members in RISA 3D, but as I apply any gravity loads, the model just goes unstable. I already set their Self Weight to zero. Also I have tried the Euler buckling strength, but as they are just rods, this strength is almost zero. Is there any other way to model these braces??

Thanks

 

[EngineeringEric]

You are correct in modeling them as tension only members for that toggle. I am guessing you have something else incorrect in the model. Make sure your basic load cases are correct, i had a problem with not making my Dead load the actual DL category in the drop down menu. Also check the Boundary conditions. Other than that, have someone else in your office (hopefully possible) look over your shoulder once

 

[JoshPlumSE]

Tony -

Remember that RISA is not doing a true non-linear solution for those tension only members. Rather it is iterating a linear solution. By that I mean that the program first assumes all members are active, then solves the model. Based on the results, it decides which tension only members are in compression and turns them off for the next iteration. At the end of the next iteration, the program may turn some members off and other members back on. The program determines convergence when there are no load reversals in these tension only members.

If the program experiences an instability for an iteration, then it deals with it in the normal way (by locking certain degrees of freedom). What is likely happening for your gravity only solution is that all braces go into compression (and get turned off) then the model is re-solved.... only now it doesn't have any lateral force resisting system. Hence the program reports an instability.

There are a number of ways to deal with this. For rods or cables (like you have in your model) I would suggest adding in a negative thermal load to give the X bracing some pre-tension because they will likely have pre-tension in reality.

 

[msquared48]

Did you remember to lock at least one direction at every node, pinned or otherwise?

Mike McCann
MMC Engineering

 

[nutte]

Josh, thank you for that explanation. I take it this is why some simple models won't run under just dead load, but they will run when a lateral load is applied?

 

[JoshPlumSE]

Yep.... That's why lateral loads aren't a problem. Because it is totally clear during the analysis which tension only members are in tension and which need to be turned off. And, the program never goes through an iteration where the model becomes understable.

 

[EngineeringEric]

I wonder Josh, could one just apply a small 10# lateral DL to the system to overcome this issue?

---Sorry for hijacking!

 

[bookowski]

Add a very low stiffness x spring to one top node, say 10#/in, just enough to keep it stable when braces are removed.

 

[JoshPlumSE]

Both those suggestions could work.

Eric's suggestion (adding a small lateral load) is akin to a "notional" load such as those used by AISC for their direct analysis method. The only sticking point is that it has to be large enough to overcome the gravity loads tendency to cause compression in the braces. So, it should work most of the time. But, it won't work ALL of the time.

Bookowski's suggestion (adding a lateral spring) can work as well. It will work well for all gravity only load combinations because it truly adds lateral stability. Make the spring too stiff, though, and it will also affect your lateral load combinations. Make it too soft and you will still be prone to P-Delta type instabilities in your frame.

Any of these solutions are okay. I prefer the pre-tensioning because it feels more robust and more realistic to me. But, that's a personal preference based on my on personal engineering judgment. Sometimes, if pre-tensioning isn't enough, I will add in notional loads to the gravity LC's as well.

 

[bookowski]

For a frame as described I don't think the spring can be too soft and create a p-delta issue if there are no initial imperfections or notional loads.