Knife Plate Design

[jmurph36]

In the attached file, I am showing a bracing connection that I am attempting to design. The connection is failing according to RISA Connection because the software is considering bending in the knife plate as well as a combined tension/compression and flexure interaction per Chapter H in the AISC Spec. The only force that is passing through the knife plate is axial load of 69.1 kips T/C from the HSS brace. I have never seen any textbook or design example that considers building in this type of arrangement. Does anyone agree with me that this is a tension/shear failure only in the plates? I ran through the numbers and if RISA is correct then the knife plate would have to be 1 3/8" thick to make the connection work and this just seems way too thick when the tensile strength of a PL 6 5/8" x 3/8" is 80.5 kips (LRFD).

 

[CANPRO]

You're transferring an axial load from one plate to another, and the plates aren't aligned - there is absolutely bending in the plates. You could go brute force here and use a really thick plate, or you could make your connection symmetrical. You could keep your connection exactly as shown, but don't overlap the HSS gusset with the gusset on the beam - join the (2) gussets with (2) loose plates either side. Puts your bolts in double shear, so you should be able to reduce to (4) bolts, and makes your connection symmetrical, eliminating the bending in the gusset plates.

 

[TehMightyEngineer]

The knife plate and the gusset plate are not perfectly in line. There's a non-zero eccentricity here and despite it being small, the large axial forces involved will cause some bending of the flat plate along it's weak axis.

Break out the pen and paper and figure out a suitable stiffener shape and weld those stiffeners to the knife plate. Or you could use a C-channel as your knife plate and verify you have sufficient room for the welds to the HSS. Alternatively, slot the HSS and do a slotted HSS to gusset welded connection. Or just go with a 1-3/8" plate.

Ian Riley, PE, SE
Professional Engineer (ME, NH, VT, CT, MA, FL) Structural Engineer (IL, HI)

 

[jmurph36]

Then why does RISA not check the gusset for this same moment interaction? If there is moment, then it should transfer to the gusset as well. And the gusset is passing with a 3/8" thickness. Every design example I've seen for this type of situation only considers tension and shear.

 

[KootK]

Some Euro-info on this:

http://www.newsteelconstruction.com/wp/wp-content/uploads/TechPaper/1401NSCTech.pdf

 

[WinelandV]

jmurph,

Just like in shear connection design, you pick a zero moment point and measure eccentricity from there. I'm betting the RISA connection is picking the center of the main gusset as the zero point, forcing all the bending from the eccentricity into the knife plate. If you want to crank it by hand, you could pick the faying surface of the plates as the zero point, and dump the corresponding moment into each plate.

For this situation, I recommend the slotted hss brace with a single hole for an erection bolt, and field welding the hss to the gusset (as TME mentioned)

 

[jmurph36]

Ah, that makes sense. That was why I was questioning the analysis initially. We like to have knife plates because we want to eliminate the need for erection crews to field weld connections. I have taken this to my manager and he thinks that it is more economical and in our best interest to just "sandwich" the gusset with two knife plates. That will solve our problem.

 

[JoshPlumSE]

AISC Design Guide 24 for HSS connections. See chapter (tension and compression connections. Figure 5-5 Tee stem / gusset as an eccentrically loaded column. As well as Example 5.1.

 

[CANPRO]

Using the sandwiched knife plate solves the eccentricity problem, but can cause real erection problems - instead of just placing your brace beside the gusset and bolting, now you have to maneuver both ends (assuming both ends connection the similar)between the sandwhich plates - which depending the geometry of the brace bay might not be physically possible. I highly recommend drawing your braced bay and connections to scale and checking that the brace can be rotated in place.

Your proposed solution is similar to my suggest above, but I recommended keeping your connection more or less as is, and using two loose plates the make the sandwich.

 

[jmurph36]

One more question. I've reviewed the AISC DG 24. It seems as though the only time eccentricity is considered is when the brace goes into compression but I don't see the DG checking eccentricity for tension. I also see that the stem of the tee or knife plate is assumed to absorb all the buckling cause by the eccentrically loaded column situation. Is this accurate?

 

[JoshPlumSE]

The flange of Tee tends to be a lot thicker than the stem. So, you can usually assume that the flange is a "fixed" end compared to the stem.

Tension and moment is just as valid as compression and moment (IMO). Though probably less likely to cause failure at similar force levels.

 

[jmurph36]

Yes. I'm now understanding that. Maybe the example in the DG just assumes that the compression/flexure interaction is more critical than tension/flexure.

 

[CANPRO]

Maybe the example in the DG just assumes that the compression/flexure interaction is more critical than tension/flexure.

Well it is, isn't it? Considering buckling and all. Plus, I feel like the tension situation is almost self-correcting - the eccentricity will tend to roll the gusset away from the knife plate, but the tension will tend to straighten it back out. The compression scenario is the opposite where the eccentricity pulls the gusset towards the knife plate and the compression just keeps it going.