Restraining the top of a top column when using SAFE

[U4ENIK]

Hi Everybody,

What would be the appropriate restrain for the top of the column above slab when using SAFE? and Why?
Why is that the column below the slab is always fixed at the bottom, but we are not fixing the top of the column above slab.
Some of my colleagues only restrain the column in Rx and Ry. Some restrain the column in Ux and Uy as well.

I would appreciate if somebody can give me some explanation or advise of how to model the columns above slab, particularly what would be the appropriate way of restraining the top of the column.

Thank you for your time.

 

[bookowski]

How you model column end fixity isn't unique to safe. It depends on your particular conditions. As a general rule I like to model 1/2 column height above and below with the far ends pinned (but upper column not taking axial). The default in safe does not make sense in my opinion so I change it to the above, but again it would depend on your condition.

 

[KootK]

I'm assuming that your slab isn't part of your designated lateral system.

The direct design method and modelling recommendations in ACI tend to steer one towards using full height columns above and below your slab with the far ends fixed for both. That's the most common approach from what I've seen. Bookowski's method will yield pretty much identical results.

My main concern with pinning the far ends of full height columns is that it may lead to an underestimation of punching shear moments.

The greatest trick that bond stress ever pulled was convincing the world it didn't exist.

 

[bookowski]

Agree with KootK. I typically do 1/2 above and below with pinned far ends because it is a relatively reasonable assumption for most cases and has the added benefit that if you are taking reactions from safe for your column loads you do not double count columns self weight. Depending on building height this may or may not be a big deal. Full length with fixed ends is also reasonable for the slab design.

 

[U4ENIK]

Thank you guys.
So would that be a mistake if the column above is made weightless (still accounting for stiffness) and is full height and is only restrained in Rx and Ry ( rotation) directions. Bottom column is fully fixed.
Do I have to always fix the columns or pin them at the top for the top columns?

 

[KootK]

@Bookowski: I'm starting to question our recommendation. With full height columns fixed at the far ends, the stiffness contribution of each column would be 4EI/Lc. With half height columns pinned at each end, it would be 3EI/(Lc/2) = 6EI/Lc. That's a 50% difference.

The greatest trick that bond stress ever pulled was convincing the world it didn't exist.

 

[KootK]

The primary purposes in modelling the columns is to check punching shear and provide a realistic amount of rotational restraint to the joint. To those ends, I'd model the columns full height and fixed at their ends, both for rotation and lateral translation. Obviously, at least one column end will need to fixed for vertical translation to keep your model stable. Column weight is irrelevant.

If you're interested in obtaining average column reactions per floor, you can play some accounting tricks as Bookowski has suggested. Take care to ensure that those tricks don't muck up your other results however.

The greatest trick that bond stress ever pulled was convincing the world it didn't exist.

 

[bookowski]

@KootK - You're right about the stiffness. If you wanted the far end fixed to be accurate though you'd also have to impose the rotation/moment from the floor above which would crank the column back a bit to get it in double curvature w/inflection at midpoint. The 1/2 height pinned should be accurate. This all assumes that you're at some mid story of the building with repetitive conditions above and below. The only completely accurate way would be to model the whole thing. At the end of the day either one is probably justifiable - punching shear results do not seem that accurate to me anyway, go from one software to another and you get very different results (try running identical models in safe v8 and then safe v12, big differences).

 

[KootK]

I see what you're saying Bookowski. I default to full length columns with the ends fixed simply because ACI seems to point one in that direction. Your method actually seems more rational to me. It would tend to overestimate punching shear moments and top steel requirements while underestimating bottom steel. The reverse with my method. Top steel and punching shear seem more critical than bottom steel to me. I usually throw 10-15% extra bottom steel in anyhow. Thanks for your response.

The greatest trick that bond stress ever pulled was convincing the world it didn't exist.