unbraced length question 1

[delagina]

i have a steel platform with grating with no horizontal brace. see attached for partial plan.

unbraced length of this i set as total L for new structure or i add horizontal brace.

but this is existing and i dont want to retrofit this if not necessary. any chance this can work out as L/8. i've read about 2% flange force and also grating might help.

looking for some clarification.

my biggest worry is original engineer (this is a very old structure) might have considered L/8 as unbraced length for the beam which will cause them to fail in my new analysis if i consider total L.

thanks,

 

[ishvaaag]

Normally if plates of the grating were welded at ends to the member that needs the bracing, I think for most situations most designers would consider it continuously braced. If your grating goes without weld to secondary members, typically if the secondary members were welded atop most would consider the distance between secondary members the braced length.

In doubt, model the thing with initial imperfections in say RISA 3D and P-Delta activated, with the sounder representation of the joint constraints. Apply your load-case and see the stability of the thing.

 

[slickdeals]

@Delaigna,
I had a similar situation. See attached thread thread507-277139

 

[frv]

What kind of members are the girders?

If they are WF, you could look at them to make sure they can support the bracing forces in weak-axis bending (provided the columns, in turn, form part of the LFRS.)

Look at AISC Appendix 6 for requirements.

You mentioned this is a very old structure.

Make sure you use the proper Yield Strength in LTB calculations; this can make a difference.

 

[delagina]

Existing steel is A242 weathering steel. I'll use 42 ksi since it's the lowest grade.

I'll check appendix 6. If i get confused or will take too much time, i'll just use total length and retrofit the girder. LOL

Dont like the idea of welding the grating. I have no experience doing that.

Thanks,

 

[XRS]

What are the transverse members at L/8 in your partial plan view? These might be helpful as bracing.

Grating is probably not much help since it has low strength and stiffness in the horizontal direction. We did a detailed analysis of grating to see if it helped as bracing on a steel stringer bridge and found it wasn't much help. Better than nothing but not close to providing fully effective bracing. Also, it had to be welded to the top of the stringers. If the grating is just resting on top as independant sections it really can't do much to prevent lateral movement of the beams.

 

[delagina]

W12X19 do you think it's good enough?

I read about the global lateral buckling happens only if all girders are loaded but if not, some of the girders may help as brace to the loaded girder.

I will not be able justify this though as we design for WORST CONDITION so all girders have to be assumed loaded.

 

[XRS]

The W12x19's will as a minimum have the effect of forcing all the beams to buckling together. The buckling capacity of the platform as a whole will be the total of the bucking capacity of all the beams. So if only one beam is loaded its buckling capacity is increased by the unused buckling capacity of the unloaded beams. But like you noted, you have to assume all the beams are loaded at or near capavity so there won't be much help from that mechanisim

The W12x19 may also help as a torsional brace depending on the relative stiffness and conection details with respect to the main beams. I've analysed this problem using handouts and notes from a class on structural stability taught by J.A. Yura. Yura published an article on beam bracing that covered the basics but I haven't bothered to look for it since I already had the info. The basic theory should also be in more recent textbooks. The main parameters are diaphram length, diaphragm moment of inertia, beam length, beam moment of inertia, and distance between brace points. I can scan Yura's handout for you if you're unable to find a published source.

Connection details are important. Ideally you'll have a stiffener at your brace points (full depth or nearly full depth) to control cross section distortion at the connection. Stiffeners would be common on older structures. You're probably looking at an adequate torsional brace if the W12x19 diaphragm depth is at least half the beam depth. That was the old rule of thumb guideline for diaphragms in steel bridges.

 

[BAretired]

If the grating is not welded to the beams, they are not laterally braced.

BA

 

[JoshPlumSE]

In my experience grating should never be considered to brace a wide flange member. We're talking about steel slats that are something like an inch deep resting on the top of the beam. Probably with clips rather than welds. That's not going to be very effect restraining the Lateral Torsional Buckling of the beams.... even if it's welded it won't be very effective as a brace.

However, that picture of yours seems to show some sort of joist or purlin spaced at L/8 that is much more likely to be provide lateral and twist restraint to your girder. So, L/8 might make sense for the girders..... just not for the joists / purlins.

 

[delagina]

what's that book again by yura? i'll try to find it.

thanks,

 

[XRS]

Yura, J. A. (2001). "Fundamentals of Beam Bracing." AISC Engineering Journal, 11-26.

Available by download for $10 at the AISC website.

I've attached stability class notes on bracing. Your problem is covered in Example 7 and the "Tension Flange Bracing" notes on page 5.

 

[a2mfk]

Wouldn't the grating more or less have to work as a diaphragm to act as a lateral brace, and take the buckling force to the far right and left side members? And like BA said, this would likely mean they'd need to be welded at all supports including the north-south members.

 

[XRS]

One more tip if you're going to use the simplified bracing formulas to determine if the W12x19's provide adequate bracing. First determine what unbraced length you require to make the braced beams adequate for the load. You have W12x19's at L/8 but you might only need brace points at L/4 or L/2 to make the main beams adequate. Bracing in older structures tends to be over done. Engineers back then followed very conservative guidelines made in recognition of limitations in theoretical knowledge. Beam buckling equations were theorectically crude and especially conservation. If say you only need an unbraced length of L/2, then use L/2 in the bracing formulas because the required torsional stiffness of the brace is greater the shorter your unbraced length. In other words ignore unnecessary bracing in your model in order to minimize the required stiffness of the torsional brace.

If that doesn't prove sufficiency of the bracing you could try more sophisticated computer modeling to get a direct solution of the full system. Some of the current software can model a 3-D system including effects of beam bracing and buckling.

 

[delagina]

i use staadpro and i specify unbraced length. dont know if it's possible not to specify this and let the computer compute it.

 

[XRS]

Computer programs don't do that sort of thinking for you. It sounds like you need to go back and hit the text books on the subject of beam buckling / lateral-torsional buckling / flexural-torsional buckling etc. Try solving the appropriate beam buckling formula in the AISC for L, Lb, Lp, or Lr depending on the design approach you are using.

Some of these parameters are listed in section property tables in the AISC Manual Beam Section for 36 and 50 ksi steel. For example if you have a W21 x 50 beam with a yield of 36 ksi, you can look up an Lr value of 16.2 ft which means you can develop first yield on the beam if it's braced at no more than 16.2 ft. Lp for this beam is 5.4 ft which means you need braces at 5.4 ft to develop full plastic bending capacity in the beam. For 50 ksi the numbers for this beam are 12.5 ft and 4.6 ft.

 

[delagina]

@xrs,

what are you talking about? i know how to put unt, unl, ly, lz, etc...

you were the one who said something like i will quote you "Some of the current software can model a 3-D system including effects of beam bracing and buckling."

i thought you were talking about automatic unt, unl, etc.. from computer similar to direct analysis method where you dont put K values anymore

 

[XRS]

Sorry I'm not familiar with the capabilites of staddpro with regard to direct design methods. Stability problems can be complex and personally I wouldn't trust a direct design method unless I could verify by independant means that the program is handling the stability issues in each case properly. Questions such as what constitutes an adequate brace for a given member might be beyond the program's capability even if can solve the k-value or column buckling problem directly.

The reason I suggested you may need more study is that the question of what unbraced length is required to carry a given load on a given beam is not a design program problem. You could use a design program to give you the answer by trail and error where you do repeated runs with different unbraced lengths until you converge on the required capacity. But I believe if you were more familiar with the theory of lateral torsional buckling and were familiar with the design equations in the AISC Steel Manual you could easily find the appropriate equation and solve it for unbraced length and by that method determine the unbraced required unbraced length for your beam.

If on the other hand you are only plugging values into design programs the design equations might be meaningless to you and in that case you are seriously in danger of falling victum to the dreaded "Black Box" problem. I've investigated two failures where the design engineer used a program without understanding its capabilities and limitation. The output said everything was fine but the bridge failed because the program didn't handle out-of-plane buckling and bracing issues. In one case the designer was very inexperienced and lacked sufficient knowledge of the issues. In the other the designer was quite experienced but didn't recognize a global buckling issue of girder pairs and unfortunately his program and the way he modeled the girder didn't handle the issue either.

 

[delagina]

i'll just use full length here for unbraced length instead of getting creative.

if this were a new structure i would have no problem as i could easily add horizontal brace.

anyway, thanks!

 

[a2mfk]

delagina- XRS makes some very good points here. Don't get too caught up in analysis software that you may miss a few finer points. This is such a simple design problem I would consult AISC and a run a few hand calcs and be done, nothing you need staadpro for... And you can never go wrong assuming unbraced, which with steel grates it sounds like that is how most of us would assume.