HSS to HSS connection 1

[OffshoreWindStructures]

I'm looking at an accidental case in a 2d HSS frame that would result in load in the branch members for all 6 degrees of freedom (axial, torsion, major axis shear, minor axis shear, major axis bending, minor axis bending).

The members have the capacity for the combined load so the next question are the welded connections (k-joint and y-joint). Using matched sections (weld is a combination of flare bevel and fillet ) and beta=1 and b/t < 16 so connection in considered rigid enough to carry the moment.

Most of the examples i've seen focus on either axial only cases or axial and moment. I would like to compare some of my assumptions with examples that include torsion as well as axial, and in plane and out of plane shear and bending. Do you know of any such examples?

My assumptions are:

Axial load for branch = axial load for weld (considering the 1/sin(angle) transformation) = shear load in chord (with some axial)
in plane shear load for branch = in plane shear load for weld (considering the 1/sin(angle) transformation) = axial load in chord (with some shear)
out of plane shear load for branch = out of plane shear load for weld = out of plane shear load in chord (with some torsion)
in plane bending for branch = in plane bending for weld = in plane bending for chord
out of plane bending for branch = out of plane bending for weld = torsion for chord
torsion in branch = ?torsion ? for weld = out of plane bending for chord

My main question is : am i right in saying the torsion in the branch should also be transformed out of plane bending for the weld depending on the angle ?

My thinking is that If the angle was 90 degrees torsion in branch = torsion in weld and if the angle was approaching 0 degrees torsion in branch = out of plane bending for weld so it should be transformed.


do you have any examples of a fully loaded connection that involves a skewed member with a reference on the transformations for the weld?

 

[CJLCivilStruct]

Interested to here if there is a worked example for such a problem.

My guess is this is a bit too complex to 'calc out' reliably and would be best suited to a localised FE model (Ideastatica maybe?).

 

[OffshoreWindStructures]

If i had ideastatica to hand i'd give that a go, but i dont. Since I posted yesterday, i decided to use a beam element FE model (a 3 member, 4 node model : 2 members for chord and 1 for a y-joint branch) with 6 different loadcases, expecting that it should give me enough guidance to confirm or reject my assumptions.

the results of the tests are as follows:

torsion in branch =
- torsion in chord
- out of plane bending in chord
- out of plane shear in chord

out of plane bending in branch =
- torsion in chord
- out of plane bending in chord
- out of plane shear in chord

in plane bending in branch =
- in plane bending in chord
- in plane shear in chord

axial in branch =
- axial in chord
- in plane bending in chord
- in plane shear in chord

in plane shear in branch =
- axial in chord
- in plane bending in chord
- in plane shear in chord

out of plane shear in branch =
- torsion in chord
- out of plane bending in chord
- out of plane shear in chord

and for the weld (added fixed node at weld) rather than the chord

torsion in branch =
- torsion in weld
- out of plane bending in weld

out of plane bending in branch =
- torsion in weld
- out of plane bending in weld

in plane bending in branch =
- in plane bending in weld

axial in branch =
- axial in chord
- in plane shear in weld

in plane shear in branch =
- axial in weld
- in plane bending in weld
- in plane shear in weld

out of plane shear in branch =
- torsion in weld
- out of plane bending in weld
- out of plane shear in weld

i should be able to use this info capture the loading transformations and magnitude in the calcs reasonably well. There's some additional effects i hadn't considered initially

 

[ViaParadise]

Just a few scattered thoughts:

Packer and Henderson, as well as CIDECT and AISC DG 24 (heavily references them) have the formulas and examples for considering axial + in-plane and out-of-plane moment.

Beam element model is good for check reactions in the elements but not the connection itself.

To avoid stiffeners, I've thought of these connections like this. Since you're attaching to a single side, as you introduce forces and moments, you're locally introducing them on the face you're welded to and not the entire member (and therefore its' full capacity) at once. Balanced forces in your branches only have to pass through the space in between. The forces and moments unresolved actually have to transfer from branch -> weld -> connecting surface before distributing in your member. If you can break up your components into the way the connection accommodates them then you can evaluate each weld and face of the connection.

Flare-bevel welds are difficult to accommodate because of the curved corners of rolled HSS. If you make the depth of your branch sections narrower by 6*chord thickness + 2*weld size, you can weld the branches to the flat section of your chord.

I'd highly recommend an area element model if torsion exceeds 20% of branch or chord section util. (just a rough rule here). Otherwise, axial and moment can all be dealt with using DG24.

 

[EngDM]

Flare-bevel welds are difficult to accommodate because of the curved corners of rolled HSS. If you make the depth of your branch sections narrower by 6*chord thickness + 2*weld size, you can weld the branches to the flat section of your chord.

Even if you reduce the branch member as you've described, there are still limits to length of weld on the branch member face such that you cannot rely on the entire branch face that is perpendicular to chord span direction. See table K5.1 of AISC 15th ed manual.

Could also check out Table K4.2 of the 15th edition of AISC's steel construction manual, looks like it has a combined unity equation for in-plane+out-of-plane+axial connections. I'd imagine if you could determine a way to calculate a torsional resistance applied to the face of the chord member you could throw it into that unity equation. (Although it is only for 90° connections, maybe there is something you could look into for how to apply it to other angles, maybe Packer?).