Connection Design

[strainstress]

Hello,

I have what seems like a simple question, but it has been bothering me for sometime.

The question, is it a always a safe approach, if I design a beam assuming it is simply supported on both ends, but then provide an end connection which is stronger(thicker plate/larger weld size/weld the flange) then what is required per AISC calculation for simple supports. I understand that the end connection will try to prevent rotation and develop moments. Can that end connection fail because of this additional moment loading which it was not specifically designed for ?

Thanks,

Vick Nand

 

[JedClampett]

A connection shouldn't fail due to moments if the beam is adequately designed for its moments. All connections, except for maybe rollers at bridge supports or sliding connections, supply some moment restraint. So they take some moment load. But before they can rotate enough to fail, the beam takes the moment.
A beam is usually a section made to support moments. It has flanges which have the majority of its cross sectional area. A connection is typically two thin plates (angles) oriented in a way that their moment stiffness and capacity are limited. They're made to be stiff in shear. Now you still have to worry about a small moment if the beam is coped excessively, but in a normal simple supported beam, the moment at the connection is ignored.

 

[ukengineer58]

My thought to these kind of things as they come to my mind is' whatever I write on my paper will the actual distribution of stresses be as I describe'. I think the answer is a resounding no! Look up wootens law a very good article describing this concept

 

[strainstress]

I just found this paper titled "Tests of Miscellaneous Welded Building Connections" from Lehigh University. (

http://digital.lib.lehigh.edu/fritz/pdf/194_2.pdf).

They performed a series tests on the welded beam connections. They state that (Pg 3) : "Many connections designed only for vertical reaction are inherently semi-rigid, thereby developing end moments which may not have been considered in the design. Such connections may not be strong enough to carry the end moments imposed by the normal tendency for the ends of the beam to rotate when the beam is loaded, in which case the design is faulty. In the design of welded connections it cannot be too strongly emphasized that end restraint and flexibility, as well as shear or reaction, must be given due consideration. The mere assumption of simple-beam end conditions is not adequate".


In Table - II Pg 21, they have summarized the results. What is worrisome is the fact that the shear carrying capacity in the connection fall below their design values in presence of the additional bending moments.

In light of these results, what are your thoughts on beam connection design? Does is it mean that making the joints stronger (by adding more weld) on the simply supported beam ends is not safe ? This just sounds so counter-intuitive.

Thanks

Vick Nand

PS : @ukengineer: Just read the paper, it is interesting, but he makes everything sound so simple.

 

[bpstruct]

In my opinion, you should try to avoid "overwelding" because it results in a condition you didn't design for. Failure, I don't know for sure. Logic would tell me that the connection would yield to the point that the designed condition (simple connection) is achieved and the beam would carry the loads as intended. But the AISC cautions about providing too much rigidity under their Shear section (looking in AISC LRFD 3rd Edition).

Not very likely to produce a true simple connection, but I wouldn't want to provide excessive rotational restraint when it wasn't designed for that.

 

[grantstructure]

Tabulated AISC connections are intended to provide minimal rotational restraint at simple shear connections--for instance, in a welded double angle connection, the angles which connect to the support are not welded along their top edges, to allow the angle to flex (yes, there is a short return on top to avoid a stress riser in the weld.)

I haven't read the paper, but does the shear capacity fall below its design value, or is it just inadequate to develop the moments that the connection could attract? That is, fixity at the beam end could result in higher shear at that end, and while the connection might have the bending capacity it might not have the shear capacity. But I'm having a hard time understanding how the shear capacity could actually drop below the intended, simple-span load that it was designed for. If you feel the need to create fixity at a beam end, you should analyze it with that fixity to see what it does to your reactions. I'm guessing that's what the report means, but I have been wrong before.

 

[JedClampett]

I've seen a lot of steel connections. And if we're doing it wrong, we're doing it wrong in every steel structure constructed. Every steel building has simple connections and they range from the relatively rigid (shear plates) to relatively flexible (double angles) to very flexible (beam seats).
I agree with grant. Are these problems with real life capacities or when tested to limits, do the moments start to contribute to failure?
I don't think we've been lucky all these years. Connections are recognized as the "weak link" in steel design and are conservatively designed.

 

[bookowski]

strainstress - What type of connection in particular are you talking about, you didn't give specifics - is it double angles, plate..? In your 'what ifs' you have 'Weld the Flange' - Do you mean that you are welding the flange of the beam, as in a moment connection? or the legs of a connection angle? It might help to clarify your concern if it was clear which connection types you are talking about.

 

[connectegr]

Very few connection details are a truly "pinned" condition. But in general connection design is not a case where "more is better". Conservative adjustments can be detrimental to the performance of the connections. For example increasing an endplate thickness or connection angle thickness, will reduce flexibility of the connection material. Welding beam flanges to a connection endplate will resist beam end rotation. Increasing the plate thickness of single plate or extended plate connections will reduce ductility. Providing excess weld on the welded outstanding legs can result in weld cracks as stresses are concentrated at the beginning of the weld. Research has shown the actual beam end rotations are much less than theoretically expected. Connection design are already conservative, as allowable strengths are used and concervative connection design forces are given. Adding additional conservativism, for a "warm and fuzzy" reason is not a good practice.

http://www.FerrellEngineering.com

Providing fabrication and erection efficient structural design of connections. Consulting services for structural welding and bolting.

 

[strainstress]

@ bookowski : Mine was a generic question, not specific to any particular application. I work in a custom engineering company, we design and manufacture Air Handling Units, Air Condition Units and Pressure vessels, where seismic loading is the governing load condition.

In some cases, when I design my beams (typically a structural angle, or square tubing) as fixed beams, I get very high end moments and it requires lot of welding hours to weld the end connection. Steel being cheaper, I thought, would it possible, to design my beams as simple supported, calculate the end reaction loads, and then find out the minimum weld size required to carry those loads and then add some extra, just for conservatism. The flexibility of end connection for the beams that I deal with, is not documented in AISC. So, I do not know, how to provide flexibility at the end connection, so as not to overly constrain it and result in an end connection failure.

Thanks

Vick Nand