Web Stability and Point Loads 1

[mathlete7]

Hello folks,
So, we all know about the basic design practice of using stiffeners at all locations where point loads are introduced into beams, right? For instance, if you have an I-beam that is loaded in the middle you'd put a stiffener where the point load is introduced to stabilize the web locally.
Well, i'm stuck with some hardware that doesn't have this feature, we "don't have time to modify", and I need to figure out some hand-waving to write a margin against localized web buckling. Does anyone know of a methodology for this (Bruhn/Niu/etc)?
I should note that the upper and lower chords are stabilized in the out-of-plane directions so out-of-plane chord buckling isn't a concern. Only localized buckling of the web where the point load is applied.

Thanks for you input...

 

[edbgtr]

Mathlete7

There are few issues you should take note of. Firstly a vertical stiffener(VS) on the beam web is not only to break the panel, it serves a few other functions as well, one of which is to prevent the flanges from moving towards each other as per a truss-type beam. The VS introduces the load into the web via shear thereby eliminating the point load introduction effects you are proposing. If the load is in tension relative to the beam NA, then the web will experience some tension as well, which combined with the web shear, may delay the onset of elastic buckling. In compression this will advance the onset of elastic buckling. These issues may be insignificant however compared to the local stresses introduced into the web which is now under a very localised direct stress that will combine with the web shear stress at that point. Such a condition may not show immediate problems but may show up later in fatigue. This problem is similar to the "new practise" of running a VS out halfway down the web. Invariably cracks appear at the mid-web runout of the VS. A way of tackling the local direct stress in the web is either to build a small FE model of the flange, web and nearest VS members to your new applied load, or use some classic indeterminate structure analysis on the attachment flange (chord) to allocate the load being seen by the web.
Obtaining the localised stress in the web could then be calculated using a beam on an elastic foundation as per Roark's Flexure of Straight Beams. Note however that the bending moment formula will need to be differentiated twice to yield the load/unit-length on the web. Correct this load for the web thickness and you have the local direct stress. The flange will also bend and you need to know the local effect on the flange stresses. Assess these factors and decide whether you will be creating a possible future problem in that area. If the point loads being introduced here are large, the answer will probably be yes. The original beam structure will have been designed sans your new load introduction. If you are introducing this load via the overhang flanges, then make sure they are supported via a throat-plate (filler) or large stiff washers modified to tuck right into the radius at the vertical flange member.
What you have been asked to justify is a tough ask, which may come back to bite your company later. Make sure you cover all aspects, especially the long-term durability aspects of the design.
Ed.

 

[rb1957]

it sounds like you've got a fitting attached to the web of a beam, or maybe to the cap flange ?

if the web, i'd worry about the off-set loads ... you've got a bearing allowable in the web and a shear allowable of the fasteners, to get the load into the web. but the load is applied not in the same plane as the web, no? and the unstiffened web doesn't like this (more thoughts later if this is were you're at)

if the flange, well there's less reason to stiffen the web. you've probably got tension fasteners attaching the fitting to the flange. hopefully here the load is close to the plane of the web. the flange will distribute the load into the web, much less of a point load. (again, more thoughts if this is your configuration)

 

[mathlete7]

well, this is a static test fixture and its essentially a box section sitting on top of an i-beam. So in this application the upper beam will press into the web of the lower I-beam.
i've never actually known what the failure mode would look like with an unstiffened web in a configuration like this. Does Bruhn/Niu talk about this at all? Would the upper chord fold over?
Thanks...

 

[rb1957]

won't it press on the upper cap (of the lower beam)

at least the load is applied in the plane of the web.

you could consider the upper cap and web as a Tee-section in bending. I'd suggest 30t of web to be effective, span of the beam ? maybe the depth of the web, both sides of the load, maybe pinned ends (cantilevered seems more reasonable physically, but will give better results; so being conservative ...)

 

[mathlete7]

Correct, the lower cap of the upper beam is pressing into the upper cap of the I-beam. That's a good idea using the effective "tee-section" and checking for bending. That would certainly demonstrate capability at ultimate loads.
I was trying to think through whether the I-beam web would buckle before the effective "tee" goes into bending (I guess if would have to for the load to react this way) and whether this could cause some undesirable effects at limit loads...

 

[rb1957]

of course initially the entire web is effective, but at some stage it's going to start to buckle, let's say elastica buckling, which would allow the Tee section to deflect in bending. I'd say that by the time the Tee section cripples then the beam web is close enough to saying "game over". but then isn't your rig designed with a "huge" safety factor ?

 

[mathlete7]

i wish. i'm looking at some hardware that was designed by someone else and the analysis was done very poorly (essentially wasn't analyzed).
so now i'm stuck with a poor design that "we don't have time to modify" and i need to figure out a way to show it good. or at least that nothing catastrophic will happen when we load it.
thanks for your input rb!

 

[rb1957]

check your design guide. it should specify the safety factor required for test rigs. as an internal requirement it is changeable, but mgmt need to know and to make the decision to deviate "knowlingly". IMHO, it sound a 1/2 a$$'d design, to have one part of the rig bear up against another as a means of transmiting load ...

 

[mathlete7]

agreed. i would definitely classify my company as a "fly-by-night" organization. we're a case study in why having designers do their own analysis is a bad idea.

 

[FeX32]

What exactly does "we don't have time to modify" really mean?
For example, if your analysis shows a very simple plate welded to the side of the web of the "I" beam could increase the safety factor by "X" then could you not have a welder spend the 10 minutes ?
Or maybe I am not seeing something in your posts.


Fe