Fixed-Pinned Column Connections

[BridgEI]

Hi, I'm designing a lobby addition to an existing building in a non-seismic region using structural steel framing. Within our design group, we have been having a debate. Each of the columns is designed such that it is fixed at the base. Since the lobby is only one story, we are seeking to minimize the moments transferred to the column by "pinning" the tops of the columns to the beams/girders and moment connecting the beams/girder as a roof support structure so that it acts as a rigid diaphragm.

The structure has quite a large roof cantilever (10.5 ft) considering it's size, and the design is done such that the girders frame over the tops of the columns and extend to support the canopy. My question is, theoretically speaking, we have modeled the structure such that the columns are acting as cantilevers to support a rigid diaphragm roof structure, so the forces are transferred via shear at the top of the column and counteracted by the flexural resistance in the column which is transferred to the foundations, etc., but practically speaking, how can this connection be accomplished? If a cap plate is welded to the top of the HSS column and the bolts are spaced sufficiently far apart due to the column size and bolt gage of the girder itself, wouldn't the cap plate, if sized such that it transfers the compression from the roof loading to all 4 sides of the HSS, be sufficiently rigid that it would tend to transfer moment, despite not being modeled as such? I.e. The connection is modeled as a pin, but it is sufficiently stiff such that it prevents rotation in the girder to such a degree that the connection behaves as fixed.

The way that I've been thinking about this problem up until now is similar to how I would approach a base plate and anchor design: If the plate is sufficiently thick that prying action is eliminated, then the column and plate will act rigidly and transfer moment. Therefore, if the cap plate is sized such that the the thickness exceeds t.np per equation 9-17 in AISC 15th edition, then the connection cannot help but to transfer moment because the column and cap plate behave rigidly. If this is the case, would this same behavior occur at the lower threshold of t.min per equation 9-19?

Am I thinking about this correctly? If not, what are we missing that needs to be considered in the design and/or modeling of this structure? If so, can a connection be designed such that moment is not transferred from the girder to the column? Does the moment transfer need to be considered in this connection?

Any help would be appreciated in considering this problem.

 

[sdz]

It depends how you design your connections. A sketch would help enormously.
Do you have to transfer the load evenly to all four sides of the HSS? There is no such thing as a perfect pin joint so you should be designing for at lease a nominal eccentricity in the connection.
If necessary simply design and detail it as a rigid connection.

 

[BridgEI]

sdz, here is the connection in question:

On one side of the connection, the beam cantilevers. On the other side, the beam spans over to another column and connects with a similar cap plate connection.

 

[hardbutmild]

I think that this would be closer to a pinned connection than the drawing you provided (in the frame plane of course). I don't know if this impacts constructability since I'm more of a concrete guy.

 

[BridgEI]

hardbutmild, the only way that could be accomplished would be to clamp the plate onto the bottom flange. The HSS tube section is 6 inches wide, and the bottom flange is 8 inches wide. That only leaves an inch of plate between the edge of the tube and the edge of the flange, which would not satisfy edge distance requirements for a bolt of any reasonable size.

 

[hardbutmild]

I thought that was the situation, but since it wasn't clear I figured I'd suggest it in case you haven't considered it.

Well, I have one wild idea that might work or might be a complete miss. I won't explain it in words, since I think the picture says it better. The weld seems short in picture, but obviously you need to calculate it.

 

[JAE]

Another idea is to use an elastomeric pad under the beam to allow a bit of flex.

 

[KootK]

@Design Engineering: You should check out this related thread, in progress right now: Link. Timely.

we have modeled the structure such that the columns are acting as cantilevers to support a rigid diaphragm roof structure, so the forces are transferred via shear at the top of the column and counteracted by the flexural resistance in the column which is transferred to the foundations, etc., but practically speaking, how can this connection be accomplished?

For a moment, let's set aside the question of how you would accomplish a pinned connection at the top of your columns and, instead, ponder whether or not that's even something that you should be seeking to achieve. I'm not sure that it is.

When your column to beam connection attracts moment, that's going to put your column into double curvature under lateral load cases such that:

1) Your peak column moments will be reduced relative to your pure, cantilevered column model and;

2) Your roof drifts will be reduced relative to your pure, cantilevered column model.

Those are both good things in my book. I think that you should leave your model as it is and just accept that these incidental moments are surely to your, and the project's, benefit. You'll want detail your beam to column connection to provide rotational restraint to the beams at those joints but, then, you should be doing that regardless, even if you did manage a pin at the tops of the columns.

Going back to the days of stonehenge and adobe huts, a fundmental tenet of structural engineering has been this: unless there's a good reason to separate things, tie them the heck together. And sometimes there are good reasons like thermal, acoustic, fire... I just don't see that there's a good reason here.