HSS Through Plate Moment Connection Weld Design Question

[Kstructuralguy]

I'm designing moment connections for 2 bay moment frame, with an HSS8x8x5/8 column in the middle that connects to W21x68 beams, each with matching 239.2 K-Ft end moments at the column. I have been following AISC Design Guide 24, example 4.2 to do this. This was pretty straight forward until I got to the required fillet weld thickness. Following their procedure, I ended up with a required fillet weld thickness of over 2". Intuitively this seems wrong to me. I have done my frame analysis in Risa 3d. When I look at the member forces, the beam end moments match and there is almost no moment in the column itself. Is this correct?

For those without the design guide, their equation for the "Required resistance per unit length of weld" = [Pa /(Aw/t throat]) + [Ma*(H/2)/(I/t throat)]. Where Pa = axial load, Ma = moment, H = Column width parallel to the moment force. I = the moment of inertia and Aw = Area of the weld, which is the perimeter of the column times the weld throat. They say use part 8 AISC for weld size required and set Rn/omega = 0.928*D*l (for ASD), where D = n/16" of an inch for fillet weld. Set the weld strength equal to the required strength and the l is divided out (since the required resistance is per unit length) and you solve for D.

I'm not understanding the "Required resistance per unit length of weld" equation. If the Pa (axial load) is all gravity (compression), why is any weld required beyond attachment for stability? More importantly, why is the Ma multiplied by half of the column width and then divided by the welds moment of inertia? When designing the plates & bolts, the moment force was divided by the beam depth to give the tension and compression in each chord. If I design the welds for the chord force at the base they do not need to be nearly this large. Is this an accurate way to analyze this? If not, I believe I would need a CJP weld to keep in manageable. what is the effective throat of a complete joint penetration weld for this case?

 

[CANPRO]

I believe you're using the wrong Ma. You're using the beam moment, but you should be using the column moment. You divide the moment by half the column depth because that you gives you your tension/compression component of your weld.

 

[dhengr]

Kstructuralguy:
Why not run the W21x68 beam continuous over the 8x8x5/8” HSS col? Then, you don’t have to design and fab. the heavy moment connections on the beam ends, except to add web stiffeners over the column. You will also have to pay some attention to proper bracing of the cont. beam and the frame in general. But, it seems that you are looking at a rather idealized condition, and you better pay some attention to other possibilities too. It appears that you are looking at a uniform or at least symmetrical loading on the beams to arrive at two 239.2 ft.-kip beam end moments at the col. While it is true that this arrangement will generate almost no moment in the col., it is hardly the only possible load arrangement. In every other load case, the moment distribution at this center support will be a function of the relative stiffnesses of the two beam spans and of the column. So, in any non-uniform, unsymmetrical or lateral loading condition, all very possible load conditions, there will be some sizable col. moments at this connection. But, they will still likely be smaller than the beam moments which you still needn’t do much design work on, probably a couple sets of web stiffeners to match the col. sides. Then, design the col. end moment connection as analysis dictates.

 

[Kstructuralguy]

CANPRO you are correct, the moment I was sizing the weld for was not the right one. The moments I were using were equal because the controlling load case for max moment was DL + LL. In that case the column has no moment because the beams are equal and opposite. Correct on the weld into the column too - its easy to see now that its treated the same way as the beam - divide the member depth by the moment to get chord forces. I believe the moment of inertia was used above to account for the weld along the column side walls as well. I don't believe its contributes a significant amount (similar to how a wide flange beams web is discounted in moment transfer) but will still use it.
As dhengr noted, The column has its maximum end moment in a wind load case OR an uneven live load case. Luckily, most the design work I did (plate thickness & width, bolt size and number, beam flange rupture check, etc.) was not wasted because those checks are still needed for the max beam moment case.

dhengr, the reason I haven't used a continuous beam is because the column serves as a moment frame in both orthogonal directions. To further complicate it, the perpendicular beams are 2.5" higher to match joist T.O.S. so a single through plate can't be used. Luckily the perpendicular moment is much lower, so beam direct weld to the column is sufficient . I considered using partially restrained moment connections, but that is a bit more complicated, and the limited height at the roof made using a top angle unfeasible. The advantage is that the beams are fully restrained which allowed for lighter/shallower members to be used (deflection controlled in the DL + LL simple span case).

Thank you both for your advice.

 

[dhengr]

Kstructuralguy:
And, why couldn’t you run the largest beam, with the largest/worst moment connections continuous and save the mess of these two larger moment connections, their design, detailing and fab? Ten bring the column up under this largest beam, and deal with its smaller moments, in whatever direction, at this lower level. Then, run the smaller beam, with its smaller moments over and through the larger beam, and not have to mess with everything going on and going through the top of the col. Now you have something of a grid-work of crossing beams running over the top of the col. and still acting of rigid frames (“moment frames”) if you wish. The only thing I worry about is that in your next post you will reveal some more new important/critical engineering design info. that changes the whole picture again. So many of you seem not to know how to conceptually define a structural engineering problem without leaving out half the important design info. You’ll have to show us your final detail when you are done, and a good sketch with some dimensions, load cases, member sizes, etc. would have really helped in the first place.

 

[CANPRO]

A sketch would definitely be helpful.

The weld in question is likely governed by your unbalanced load case where you will develop some moment in the column.

For the beam framing in perpendicular - you can drop the beam 2.5” and use a common top flange plate with the other direction, just have to add a 2.5” spacer on top of the beam to catch the deck.

 

[Kstructuralguy]

dhengr: Not sure if the first part or your reply got cut off. But attached are the details I produced. The wind load case controlled for column moment (not uneven live load) and that moment was much more reasonable. Outside of the center bay, plated moment connections work. Bolting the beams to the plates (welded in shop) will make erecting the steel much faster.

I appreciate all the advice on alternative ways to frame this, but as a reminder I created the post to ask specifically about the weld between through plate and column, not how to frame the connection. It was not my intent to leave out important info and next time I'll clarify more initially. I had considered what you are proposing but was uneasy about the perpendicular beams' moment transfer to the column through the (proposed) continuous beam. Since their top and bottom of steel elevations do not match there would need to be additional steel between them to transfer this load to the column. I'd be very interested in seeing a sketch on how this could be accomplished.

 

[Shu Jiang PEng SE]

Aw/t is the length of the weld, Pa/(Aw/t) is the force per unit length required; Ma*(H/2)/(I/t) can be rewritten as (Ma/(I/(H/2)))*t, I/(H/2) is weld section modulus, Ma/(I/(H/2)) is the stress of the weld at far end, stress times weld thickness t is the unit length weld strength required.

 

[CANPRO]

Running the beam continuous over the column and framing in the perpendicular beam wouldn’t look too different from what you have sketched. Over the column, put split HSS stiffeners on each side of the beam (same section as the column, in line with column) then frame the other direction as you had intended.

Have you considered lowering your higher beam per my earlier suggestion? It cleans up your connection quite a bit, especially if beams in both directions are the same depth. Putting some 2.5” spacers along the beam length is much easier and cheaper than having a messy moment connection with beams at different elevations.