Cross Bracing Deflection 5

[TxGraeme]

My office was having a discussion about the design of rod bracing as a lateral resising system. The modeling software we are using (RAM Advanse) checks for bending + axial load stresses, which indicate the member is overstressed. The deflection in the model is over 10 inches, which I'm assuming is the cause of the high bending stresses. However, I have heard that braces are typically pretensioned to eliminate sag (and tensioned even further when lateral load is applied) and the brace straightens out so in reality there is no bending.

It is a light seismic zone and the rods are assumed to be tension only. The building is single story. Does anyone have any experience with this or have a reference we can check? I posed the question to AISC steel solution center but that didn't get me anywhere.

Thank you,
Graeme Sharpe

 

[Lion06]

I am picturing that you are using rods for diagonal bracing from top of one column to bottom of the other. This being the case, you should make sure the node at the junction of the rod and column is pinned, not fixed. You shouldn't be getting any moment in there as the braced frame system you are talking about is designed to carry everything in axial load (as far as the lateral forces go anyway).
If you apply any bending to a thin rod it will be overstressed in bending because the Sx is so small. As mentioned above, however, there shouldn't be any bending in these diagonal braces.

 

[kslee1000]

Have you release the joint when the rod is in compression?

 

[JAE]

TxGraeme,

A common problem with analysis programs with straight matrix solutions is that they do not resolve tension straightening as it is a second order effect.

What happens is the program thinks that the rod is a simple spanning member (pins at the ends of course) and there is some small, but significant, self-weight to the rod that, over long distances of an X-brace, create simple, positive moment due to that self-weight.

What you should do, since the rod weight is insignificant to the analysis, is assign to that rod a special material property whereby it has very small, or zero, density. This eliminates the bending in the rod.

Also - I agree with kslee1000, that you should ensure that the rod is a tension-only member in the analysis.

 

[sundale]

It is very easy to calculate the PL/AE elongation of a tension rod and then compute via simple trigonmetry the lateral drift that results.

There will de some additional drift due to overturning tension or compression axial deformations in the columns also, but the additional drift contribution of these will be minimal in a one story structure. A simple virtual work analysis can determine the column's contribution to drift if you are curious.

Give your computer results the "ho ho test" by HAND...

 

[csd72]

StructuralEIT,

I assume he means a high moment in the columns due to the 10" of drift. Could happen if the beams have fixed connections into the columns.

TxGraeme,

The rods do get tensioned up but it is nothing like the tension on a pretensioned bolt, I wouldnt really count on it. If the deflection is a problem, then I would suggest using anle bracing instead. 10" sounds like a lot of drift to me.

csd

 

[TxGraeme]

Sorry guys, I might have led you astray. Let me clarify. The member under consideration is the rod brace itself. The bending is from dead load / self weight of steel. Drift is not an issue, nor are moments in the columns. The brace member is pinned at both ends (relative brace from top of one column to base of adjacent column). There is no nodal connection at the intersection of X-braces, so no interaction exists there.

Basically the brace is trying to resist self-weight by going into bending, but in reality this will resisted by catenary action. I agree that a second-order analysis would be required to model this correctly, but I was hoping for some rules of thumb to (hopefully) disregard bending in rod braces.

Thanks for all the replies so far,
Graeme Sharpe

 

[JAE]

...which is pretty much what I said above. Simply assign the rods zero density and the sag goes away.

 

[hokie66]

Not for analysis, but for detailing, if your rods are roof bracing, sag rods or straps from the purlins should be used to control the sag.

 

[sundale]

I was thinking you must have a very tall and flexible structure for 10" of drift!

One firm I worked at would not detail "tension only" rod braces because on some previous project they had gotten noise complaints from the occupants when the compression brace buckled and "whapped" the drywall studs under just average type wind events.

 

[DaveAtkins]

Do what JAE said, or set the selfweight to zero (e.g., on RISA 3D, don't put a -1 in the Y Gravity column in the Basic Load Cases spreadsheet).

DaveAtkins

 

[swearingen]

For grins, break the rod into 20 pieces and see what happens. You should see it straighten out...



If you "heard" it on the internet, it's guilty until proven innocent. - DCS

 

[TxGraeme]

It seems that the consensus opinion on this issue is that sag won't be a problem, so it can be disregarded in the analysis by either setting the density to zero or ignoring the self-weight deflection. Sounds good to me.

The comments about the braces "whacking" the wall seems important, has anyone else had problems like this when using tension-only braces?

The AISC manual's chapter on braces says that sag problems can be avoided by "drawing" or pretensioning the brace and lists the length each brace should be shortened by. Has anyone ever specified this or had problems reported by installers?

Graeme Sharpe
Indianapolis, IN

Thanks for all the help!

 

[csd72]

TxGraeme,

If you simply ignored the bending results from the output and took the axial load then this would have negligible difference from reanalysing as stated above(I would expect much less than 1%).

Regarding pretensionining to take out the sag, as this is mostly a visual requirement then visual inspection and common sense should be sufficient. If you shake it and it moves around then it is too loose, if you can easily see the sag then a bit more tension should be applied.

Be careful using rods with light fraing as the pretensioning can sometimes prematurely fail the surrounding light members (although this doesnt seem to apply to this case).

csd

 

[Lion06]

I would you need to account for the pretensioning in the design of the rods? Is this correct?

 

[DaveAtkins]

I recently learned a neat trick--you can induce a pretensioning in a rod, in RISA 3D, by giving that member a negative temperature load (this contracts the member, putting it into tension).

But I believe it doesn't make a difference with total load--you will get the same tension in the rod under total load whether or not the rod is pretensioned.

Someone correct me if I am wrong.

DaveAtkins

 

[UcfSE]

I would account for the pretension in the rods.

 

[KCRatnayake]

JAE,
I have seen some engineers using circular hollow sections instead of cable bracing (X-bracing). Any pros and cons?

KC

 

[Lion06]

I would think that (depending on size) the hollow circular sections are less likely to buckle. That being the case, care should be taken to not necessarily treat the bracing as tension only and account for the compression in the column attached to the compression brace.

 

[TxGraeme]

swearingen,

I tried breaking up the rod and it does straighten out if it is under enough tension and a second order analysis is performed - a very good suggestion.



Graeme Sharpe
Indianapolis, IN

Thanks for all the help!