Concentric bracing connection design

[csd72]

I have a connection designer that wants to use a concentric knife plate detail for the gussets at the end of tube braces.

The detail involves a knife plate and a gusset in line with each other and a cover plate each side with bolts into each.

I am not sure how to check this connection for buckling over the considerable 1'6" or so length. Does anyone have any good references for this type of connection?

 

[rb1957]

sounds like you've got tubes becoming double splice plates (welded?, swaged tube??) acting like a clevis lug, with a single joint splice plate joining the two tubes together. (or maybe the tubes become a single web and double splice plates join them together).

i'd be very carefull about the column allowable. to sounds like you've got a stepped column, which is critical about the weak axis of the joint.

there are "book" solutions, or eigenvalue solutions from your FEM of choice (i've recently used SOL105 from NASTRAN).

it'll probably be way too conservative to apply the weak inertia to the whole length of the column. i think it is unconservative to consider the end as a column in isolation with pinned/fixed supports.

 

[nutte]

The AISC Volume II (ASD/LRFD) Connections manual (green and blue book) has connections similar to this. Subsequent versions probably do, too.

Because the plates don't stiffen the gusset any beyond their weak axis stiffness, you should take your buckling length from the end of the HSS to the end of the gusset. This is much longer than a similar connection with say claw angles, where the angles themselves do stiffen the gusset, and the buckling length is taken from the end of the angles.

 

[rb1957]

i guess you could always use flanged splice plates (2 angles, or 1 I-section) which would greatly improve the weak axis column strength.

 

[csd72]

It is actually a horizontal tube brace into a vertical column. The gusset and knife plate are both horizontal and these are connected to each other with an additional plate top and bottom and 6 bolts into each section. There are 4 plates involved.

I have attached a really crude sketch.

 

[csd72]

Sorry this is the correct link

 

[rb1957]

couldn't get your file to open (no extension)

 

[csd72]

I did it the old way and browsed for it this time - third time lucky?

Thanks for the responses so far.

 

[nutte]

The black HSS connection manual is the reference I was thinking of. Look at page 6-8 for a condition similar to the one you're describing.

 

[csd72]

Thanks nutte,

Yes I found that bit, but I am concerned with the buckling behaviour of the connection as a whole. It seems to me that buckling effects would be far more exaggerated in a connection like this.

 

[nutte]

I'd use k=1.2, L=length from end of tube to flange of column, and check buckling on the thinnest part in the group, be it the single knife plate, pair of splice plates, or single gusset plate. Depending on how much load you have, it might not be a problem at all.

Globally, I don't see any problems with the configuration. The bolts will hold the plates tight; they're not going to be flopping around.

 

[rb1957]

could you turn the knife-plate 90deg ?

right now (the way i see it) the weak axis of the splice Could be seeing bending from the horizontal tube. you know if the tube loads are vertical or horizontal

 

[csd72]

nutte,

Thanks, that makes sense. All bolts are slip critical so it should be held tightly in position.

rb1957,

There are no major bending loads other than wind on the tube and self weight. The effect of the self weight on the buckling of the horizontal gusset is a bit of a concern though. I have asked them to put the gussets vertical where possible but it is not always possible.

Thank you both for your help, I now feel I understand it sufficiently to examine their calculations and make some intelligent comments.

 

[trainguy]

csd72,

Does anything prevent your brace from rotation, such that the connection plate effective length factor may need to be taken closer to 2.0? This is what I've done in the past, and checked the whole assy as a stepped column with Roark.

Also, you do not show column flange stiffeners in line with the brace conn. plates, you should have some.

Do you need to worry about seismic loads as well?

tg

 

[Dinosaur]

I would consider using two angles on the column to sandwich your knife plate rather that this approach. The connection will become more compact which should aleviate your concern of buckling. Also, the two angles could be detailed as a bolted connection that would eliminate a lot of critical welding this detail may require. I also would seek to turn the connection the 90 degrees mentioned wherever possible. Good Luck.

 

[hokie66]

I hate this connection, and would never detail it that way. Why is someone else detailing your connections?

Does your connection have to be concentric or can single plates suffice, with the strut sized considering the small eccentricity?

Other than the unnecessarily long connection, I suggest the worst thing about the connection is the way the knife plate is connected to the strut. I would always slot this into the hollow section so the connection is developed by fillet welding on both sides. The end plate is then just a closer.

Agree with others that the plates are best oriented vertically. That way, the plate will align with the column web and avoid need for stiffening the column.

 

[csd72]

Trainguy/Dinosaur,

Thanks for the suggestions, I think I will get them to incorporate one of those 2 options.

hokie,

This is the way it is done in East coast USA and in the UK, connections are designateed to the steel contractor and they employ an engineer to design their preferred connections to our design loads.

In some cases, such as this, I think it is more hassle than just designing them. Someone else designs them, but we as engineers of record must approve the design.

 

[hokie66]

Well, then, I would just reject it. The connection designer may be trying to save on length of tube by using this connection. What he has given you is in fact two connections when you only asked for one. Any slight misalignment will mean the connection is no longer concentric, and the plates will have to work in bending.