Web Sidesway Buckling from Column Reactions on Continous Beam 1

[kar108]

I am designing a 30 ft long (3) span continuous steel beam for a residential basement. The floor joist system provides continuous support for the top flange. The beam is bolted to the top bearing plates of (2) adjustable steel columns spaced approximately 10’ apart. The column reaction creates a concentrated load acting on the bottom flange of the beam. This load must be analyzed for concentrated loads acting on the Flanges and Web as described in section J10 of the AISC Steel Construction Manual (13th Edition). Using equation (J10-7), I want to design the lightest possible beam for web sidesway buckling without using a web stiffener.

My question is what should I use for “l“(the largest unbraced length along either flange at the point of load)---
What is the proper choice for the unbraced length???

(1) The entire length of the beam – ~30ft? (I am assuming that the columns do not provide lateral bracing), or
(2) The maximum span between columns and/or foundation walls --- ~10ft?,
or
(3) The distance along the bottom flange where the moment is negative ( measured from each column location to the point of zero moment)--- ~5’ for the fully loaded condition; ~12.5 when the live load is removed from the center span? (Since the floor system fully braces the top flange which has a positive moment, we only need to consider the unbraced portion of the bottom flange subjected to a negative moment)

 

[KootK]

This is a great question. One I've been meaning to ask myself for some time.

As for the value to use for "l", my interpretation -- based on diagram C-J10.2 given in the commentary - is that you should use the full length of the beam (~30).

Based on the commentary info, it sounds as though adding stiffeners wouldn't improve matters anyhow.

I have a few related questions of my own:

1) In kar108's case, there are two point loads (reactions) causing sway buckling withing one length of unbraced bottom flange. Does anything additional need to be done to account for the fact that two loads may now contribute to the sway buckling?

2) It has always seemed to me that, if one provided full height stiffeners and a stiff cap plate connection, you could consider the stiffener / web combination to simply be an extension of the column below. In this way you could claim to have braced the beam using the column. Is this rational? Does anyone know of a procedure for checking the cap plate connection in this scenario?

Usually, when I ask this question around the office I get a big fat "Just brace the damn flange".

Adam

 

[JAE]

I may be a bit of an old fogie, but I always put stiffeners on beams when they go over the top of a column.

I know AISC provides a nice check for sidesway web buckling for these conditions but over the years, the main structural collapse mechanism that I've seen is simply beams without stiffners over the columns.

Due to tolerances in sweep and web plumbness, installation variability in column plumbness, and unbalanced loads from one side of the beam, you can get a pretty unstable condition without them.

Now to answer the first question - l is defined in section J10.4 as the "largest laterally unbraced length along either flange at the point of load."

So you would go essentially the full length of the beam in your case. One unstiffened column point doesn't brace another.

 

[Lion06]

I believe that if you provide stiffeners - which AISC requires at points of support - that it would be 10' (if the top flange is braced like you say). The top flange being braced along with the stiffener will brace the bottom flange. The is the same as bracing the bottom flange at a seat support (maybe on a concrete wall) and providing stiffeners or an end plate. That provides rotational restraint, and braces the section. If you provide a stiffener and the top flange is braced then you have the same condition.

 

[Lion06]

I also wanted to say that it's also the only way your column will be braced in both directions.

 

[KootK]

Where does AISC say that stiffeners are required at all points of support?

 

[Lion06]

I apologize. It doesn't require stiffeners, but it does require that the section be restrained against rotation at points of support.

 

[BAretired]

I agree with SEIT up to a point. You must use stiffeners or other form of rotational restraint over each column. If you do, the unbraced length is from point of inflection to point of inflection. It is not the 10' span of the beam, no matter what the so-called experts say, but if you want to use 10' as your unbraced length, you will be on the safe side unless you get negative moment all the way across the span.

In that case, the unbraced length is equal to the span plus the distance from the column to the point of inflection on each side of the span. If the point of inflection is at 2' each side of the column, the unbraced length is 10 + 2 + 2 = 14'.

Such a condition is not likely to occur in residential loading, but it is important to understand the principle involved.

Most codes recognize that, when the moment is variable across the unbraced length, another factor comes into play which acts in your favor. I believe AISC calls it Cb.

BA

 

[BAretired]

Sorry, I spoke without thinking on the matter of unbraced length. If the beam is continuous and torsionally restrained at all supports, the unbraced length would be 10' max.

If the beam was a 10' span with a 2' cantilever each end, each loaded with a concentrated load at the tip, then the unbraced length of the bottom flange would be 14'

BA

 

[csd72]

This is a classic failure mechanism that is often overlooked - put in the stiffener!

I disagree with BA, the deflection points are not points of restraint and your effective length is between points of restraint no points of zero moment.

 

[ishvaaag]

A traditional detail to prevent lateral buckling of deep beams in the bottom (compressed) flange is to provide struts at 45 deg upwards to the floor. I think rarely the floor is checked for these rotation prevention forces (not seen in the books, or don't remember). The point here is that technically is assumed that the rotation prevention provided by the floor, maybe even unchecked, has been accepted in practice. Using a stiffener may, depending the detail, signify as well the same kind of introduction of rotational forces to the floor than by the struts, only that bigger. Hence, seeing the cost of adding details to simple straight members it looks natural some try to use simply the beam, without stiffeners. And in the books there are formulae for concentrated loads without stiffeners for a number of cases, maybe one as the described as well.

So, technically, since it is the rotation prevention by the floor what is accepted in many cases as the stabilizing mechanism against the lateral buckling of the referred bottom flange, and seeing the books contemplate the case, one may think it is acceptable practice to go for the beam without stiffeners, since, once the stabilizing rotational loads are passed to the floor, there must not be much difference between a rigdized and not rigidized support point, as long its capacity is proven.

And so one might start to think in what width of the web could be taken for such check. 18 times -9 times each side- the thickness of the beam web plus the width of some quite stiff plate supporting the beam?

Of course, all forces, vertical compression and bending stressses in plane and out of plane (bracing required) force need be taken in to account for the accepted collaborating segment of web. If the loads are small, it might work.

There is also the question on how the working notional panel is working. Given the small dimensions, maybe just a stresses approach might work; you may see the implications and assumptions that need to be taken to analyze one of such panels in whatever the way.

And as well, beyond the science of construction, the crystallized mode of it, code and pathology of construction experience. So one might elect not enter somewhat uncharted territory and put stiffeners.

 

[Teguci]

Without the stiffeners you have a double pin connections at the top. Rotation is allowed at the floor to beam connection and at the beam web to beam bottom flange. If one of these pins is not elliminated, then the column reactions are variable and subject to redistribution due to high deflections.

 

[CentrePACE]

In regards to the unbraced length, "l", and beam rotation restraint at supports: If an adjustable steel support column is embedded in a 4" concrete slab at the bottom and the top bearing plate is bolted or welded to the bottom flange of the beam, could the beam be assumed to be braced or "restrained against rotation" at that point (with continuous bracing of the top flange from the floor system)?

 

[Lion06]

CenterPace-

I would tend to say NO. You can certainly check to see if the cantilevered column is adequate (for both strength and stiffness) for bracing the bottom flange of the beam, but that detail, in and of itself, doesn't brace the bottom flange. I would just provide the stiffener and call it a day.

 

[KootK]

Technically, you have to provide a stiffener AND a brace though right? According to the AISC provisions, the stiffener by itself doesn't do anything for you in this situation.

 

[Lion06]

What brace? If the top flange is braced as you say, and the bottom flange is positively connected to the column, then the stiffener itself does brace the bottom flange of the beam AND it provides a brace point for the column in a direction that is perpendicular to the longitudinal axis of the beam. The only way for the beam to buckle in a LTB fashion would be for the stiffeners to fail.

 

[CentrePACE]

I am referring to a beam without stiffeners at the support. Could the positively connected column, along with the top flange bracing, together act as a restraint against rotation without the need for a stiffener?

 

[Lion06]

No. At that point you have to rely on bending of the web and that won't cut it. miecz recently posted a calc showing that it is possible to get that work, but there is no point in going through that mess.

The basic problem is that without the stiffener, you are counting on the beam to brace the column, so you can't count on the column to brace the beam. Unless, like I said, it is a cantilevered column (not a pinned base) and you actually check the strength and stiffness of the column to brace the beam.

Providing stiffeners takes care of both issues.

 

[kar108]

Is anyone aware of any documented steel beam failures in residential basements due to unreinforced point loads?

I ask this question, because my own observations (and also responses from several steel suppliers) indicate that residential steel basement beams are typically not reinforced at column supports in our area (PA – low seismic risk).

Is this possibly because basement beams are subject to gravity loads only (no wind loads to create drift in members above and eccentric loading situations, and no seismic in the area). The beams are supported by relatively short steel columns (4”x11Ga), only 7’ to 8’ long that are embedded in 4” of concrete and either bolted or welded to the beam.?

 

[Lion06]

I'm not aware of any, but I only have 3 years of experience. I would tend to think that there are several reasons for a lack of a ton of failures. The design loads are rarely seen - especially in residential construction. When is the last time someone had 40 psf LL in their house? The connection to the column does offer some benefit (though difficult to quantify and not reliable). You start eating into your factor of safety.

kar-
Where in PA are you?