Neglecting eccentricity due to beam heights 2

[canadiancastor]

I've come across a simple (almost shameful) question that has me stumped. The initial question is whether we should "offset" all of the beams in our FEA models of steel structures so that the top of the beams all is level, which is how it's going to be in practice. I've never done this, but thinking about axial loads in collector beams, I'm now wondering if there's a hidden moment I should be considering (and if there is, where should it be considered?) The two practical cases I can think of are:
A - different height of beams at a column
B - different height of beams in a gerber system
For A, I figure this can be resolved by applying to point loads to the column and making sure it can withstand the induced moment. In this case, using FEA offsets would probably be a good idea to get the right moments in the columns.
For B, this one has me stumped. I don't think it's a good idea to use a full moment connection, that would remove any efficiency gained using the Gerber system. Maybe the connection needs to be full moment on one side only? I can't visualize what would happen if I kept regular pin connections there and was a huge axial load. Would the ends of the large beams "curve down" to better align with the applied load?

 

[CDLD]

For case A you get a moment on the column.

For case B, you do not need moment connections for the drop in.
The eccentric axial load will cause an end moment at the Gerber cantilevers, which is resolved by the supporting columns the same way as vertical loads.

Case B:
Edit: The blue line is the moment diagram.

 

[phamENG]

For Case A, I wouldn't be too worried as the effects will be localized. You need to account for that moment in connection design and considering local effects on the column, but it's not going to impact the global model appreciably. (A small amount of shear at the base of the column, but unless it's a moment or braced frame column designed to a 0.99999999999999999999994 utilization, it's probably not going to change the design.)

For Case B, yes. That eccentricity will result in variations throughout the model and it may be wise to consider it.

Be careful, though. Those offsets are usually done by using rigid links (sometimes behind the scenes), and those can screw up results. So be mindful of that and monitor the results closely. Might be best to run it without it first and then change the offsets and see how the results differ and whether or not they make sense.

 

[human909]

Be careful, though. Those offsets are usually done by using rigid links (sometimes behind the scenes), and those can screw up results. So be mindful of that and monitor the results closely. Might be best to run it without it first and then change the offsets and see how the results differ and whether or not they make sense.

Great advice here.

You can get some really odd and incorrect results (at least in terms of what your are trying to model) if you start offsetting beams/columns off their centreline. Sometimes you can get the exact results you are after, but is can also introduce oddities locally or even elsewhere. I occasionally do it, but sometimes there are other approaches.

Though other times you get handed design that is 100% detailed to sign off on and you have to redesign 40% of it because the original engineers (external consultant and won't be used again!) think that all loads magically go to a node. This is a current project. And the 400kN load (110kips) is inducing torsion on the base of a beam!

 

[lexpatrie]

I don't think this is quite on point, however.....

Transfer Forces in Steel Structures, Dowswell, AISC Engineering Journal, 3rd Quarter, 2010. (currently a free download).

I think what you're getting at is the effects of eccentricity on tension loads from collector forces..... that's not really what Dowswell is talking about, from what I see.

On the Right Path, Drucker, Criste, Modern Steel Construction, November 2014 at least shows a floor level transfer force but I don't think they get into it all that much.

I don't usually see Gerber girder systems on the exterior of a building, so they aren't normally a collector or diaphragm chord when I've seen them, there's usually a wall there.

Not sure I've seen Case A in the wild, except perhaps in a floor system. Again, that's not where I normally see a lateral force resisting system.

If you have a tension force going through the connection, you could see some additional localized bending moment in the column that is probably not addressed in a basic analysis, but there are ways around that load path, anyway, as drawn, say, by adding a plate on the top of the beams and welding it for the needed load.