Unusual Beam Support for Column Removal

[CTW]

I'm analyzing an existing beam which had a column removed about 12 years ago. See attached sketch. The beam was supported and extended in an unusual manner. A new beam was added parallel to the existing one with the bottom flange of the new beam welded to the top flange of the existing beam. The new beam length is 7'-5" with a 48" overlap with the existing beam. The connection on the other end of the new beam is two bolt double angle. The other end of the existing beam bears on top of a column.

I'm curious how others would approach the analysis. The total span is 44'-2". New beam size is W14x43 and existing beam size is W18x40.

 

[SlideRuleEra]

Oh, that's a good one. I would start with the basics, make assumptions of the supports (simple supports sounds reasonable). Then draw shear and moment diagrams. You did not mention the magnitude or location of the loads or if there are intermediate compression flange connections to limit LTB.

One thing for sure, with a span to depth ratio of about 30 to 1 (44'2" / 18") allowable bending stress will be maxed out for even modest UDL loads. Additionally, since the W14 is near the end of the simple span there will be even less bending stress at the joint no matter how the loads are applied. Shear could be another matter if you have heavy point loads near the joint.

What to do next really depends of the unstated details, loading and compression flange supports.

http://www.SlideRuleEra.net

http://www.VacuumTubeEra.net

 

[BAretired]

Calculate the moment and shear at each end of the overlap and apply statics.

Details of the existing welds affects the load transfer between flanges. The added beam has a wider flange than the original. Are they welded parallel to the beam axis at the outer edge of the narrower flange? Or are they welded across the flange at each end of the 4'-0" splice section? Are there stiffeners in the beams at the ends of the splice section?

BA

 

[msquared48]

Just analyze it as a haunched, built-up girder.

No stiffeners?

Mike McCann
MMC Engineering

 

[CTW]

No stiffeners. The weld is a 1/4" fillet all around.

 

[Teguci]

I don't like the idea of crossing the full flange with a fillet, but unless you have lots of load cycles, it shouldn't be a problem. I definitely recommend looking for any distress at these welds though.

 

[ishvaaag]

I'll be looking the joint itself.

This is a weld rectangle of throat width; assume such rectangle stays in the XY plane and the webs are in plane Z. Make a cut in the interface notional XY plane. All the analysis carried at service level as per Pilkey's.

The shear V at center of the rectangle is redistributable in stresses shearing the whole rectangle of value

fv=V/(a·(2·b_min+2·length_of_splice)), these anywhere in the weld.

Then the standing moment will produce normal stresses to the notional rectangle of width a, in accord of the elastic modulus of the notional rectangle section of throat a thickness. Let's call it fn

Then according to Pilkey, p.384, the worse points will be in the narrow edges of the rectangle weld, so you build your comparison stress as sqrt(fv^2+fn_narrowedges^2). This needs to be compared to the allowable shear stress, Table 8-9 in Pilkey.

From there we would look to introduction of the loads to the flanges, say, examining in what extent the localized introduction may cause stress concentrations that need be factored in the design. Normally must be scarce, since the complete perimeter weldment somewhat forestalls the development of prying effects.

 

[racookpe1978]

Then according to Pilkey, p.384, the worse points will be in the narrow edges of the rectangle weld, so you build your comparison stress as sqrt(fv^2+fn_narrowedges^2).

1/4" (fillet I'd assume) welds? Concur with your observation that the biggest threat is not the beam itself (flange or web failure) but at the tip of the welds tearing out at the point of max loading.

 

[racookpe1978]

To be more clear.

Assume some "downwards" on the main beam back over the left of the sketch. Assume that the two angles and the bolts can hold the load. Both beams are relatively stiff: assume for first checks that they don't bend very much.

If so, the lower (original) beam is being pried apart from the upper (new) beam: the pivot point is the left edge of the 4 foot overlap, the separating point (point of max tension on the weld) is the weld at the extreme right side of the 4 foot overlap. Worse, if any part of that fillet weld cracks or rusts, then the rest continues cracking and the failure accelerates, since the splitting force is increased as the crack (weak point) gets bigger.

 

[BAretired]

Yes, it seems that stiffeners would be in order at the right end of the overlap.

BA

 

[TXStructural]

Agree about needing stiffeners, although I'd think that the webs nearer the left end (nearer center span) would be in compression.

I'd be concerned that the load is getting from one beam to the next through the tips of the flanges.

 

[ishvaaag]

Innermost short sides' weld of the inner is not in "compression", but in an state of stress that is characterized to be checked against allowable stress in the weld through separate examination of notional shearing and bending stresses on the notional rectangle weld.

What precedes, of course if full contact in compression is not ensured. If it is, you substitute direct bearing where we have been considering action on weld; this way you may consider even a more favourable distribution of stress on the welds, where the neutral axis will migrate towards the innermost edge of the rectangle. Mechanical arm will stand around the same at 2/3 of the splice length, but since you have now more weld available for the tensile part it may turn out a more favorable "normal stress" component on the tensile side. For the shear stress component it would be rare that other thing than the whole rectangular weld (throat) perimeter be used, since it can make use of it to pass shear.

Even so, I prefer the first scheme that will result typically more conservative, is along the lines of what made many times for welds, and is even simpler numerically.

 

[msquared48]

Shore, cut off the flanges, weld the webs together and install stiffeners as needed to limit the web shear stress.

Mike McCann
MMC Engineering

 

[ishvaaag]

The 2/3 part of my last entry is somewhat inaccurate since the weld is a rectangle shape, not a rectangle. Anyway in general it is quite likely the final notional normal stresses at the tensile side will get lower when we accept compression between flanges at the other side and the comments stand.