Design of HSS - Include moment from connection offset 1

[EngDM]

Hey all,

I'm looking to get some input on what the common practice is for giving moment to an HSS column that is loaded by a shear tab connection. Is it standard to offset the load by half the column width + ~1 1/2" to the first row of bolts? For cases where connections are delegated and you don't know the # of bolt columns, should you take a higher offset to account for the true centroid of bolts?

I know RISA Connection allows you to ignore moments in the connection if the offset isn't that large, but does this mean that the column doesn't experience any moment as well?

With HSS you rarely have a concentric loading case unless you're running a gerber beam system above or explicitly detail a bearing connection, but I know that columns are just take with the typical compression equation and not really any moment.

In my case, we are doing a 4 storey building so I've been including the connection offset and pattern loading the live loads on either side to get a moment envelope, but it's making my column sizes way larger than we originally anticipated.

 

[KootK]

This has the potential to be a massive can-o-worms. Briefly:

1) This has been discussed ad nausea here. @human909 has been very active on this stuff, as have I. If you search for his handle + mine + eccentricity, I've no doubt that you'll turn up a dizzying amount of pedantic info.

2) Sure, of course you should make a conservative estimate of the eccentric moment on the column. It's there, right? Just kidding. There are, however, codes around there world that specifically say that you have to consider eccentricity to the bolt line which, in truth, may not even be a truly conservative approach when accidental beam-column fixity is considered.

3) There has been at least one study (Ionnidies?) where the author concluded that, in many practical situations, The rotational restraint provided by the beams (K=0.7 ish) outweighs the connection eccentricity. But, as @human909 has pointed out, there have also been failures owing to this very concern.

4) For situations such as I suspect you are contemplating, my current approach is:

a) Design the column for pure axial (K=1) and keep the combined stress index under 0.8.

b) Ignore the eccentricity beyond [a], which is admittedly non-rigorous.

And I rely very much on my judgement to gauge the suitability of this approach. It would be impossible for me to provide a full list of the things that might put me off of this strategy but a partial list would include:

i) Has to be a situation where the moment transfer is incidental (no moment frames etc).

ii) Pause to think if the column is supporting a significant beam cantilever with a shallow beam.

iii) Ix_beam >> Ix_column. I love me a W18 shear tabbed into a HSS4.

 

[EngDM]

This has the potential to be a massive can-o-worms. Briefly:

1) This has been discussed ad nausea here. @human909 has been very active on this stuff, as have I. If you search for his handle + mine + eccentricity, I've no doubt that you'll turn up a dizzying amount of pedantic info.

2) Sure, of course you should make a conservative estimate of the eccentric moment on the column. It's there, right? Just kidding. There are, however, codes around there world that specifically say that you have to consider eccentricity to the bolt line which, in truth, may not even be a truly conservative approach when accidental beam-column fixity is considered.

3) There has been at least one study (Ionnidies?) where the author concluded that, in many practical situations, The rotational restraint provided by the beams (K=0.7 ish) outweighs the connection eccentricity. But, as @human909 has pointed out, there have also been failures owing to this very concern.

4) For situations such as I suspect you are contemplating, my current approach is:

a) Design the column for pure axial (K=1) and keep the combined stress index under 0.8.

b) Ignore the eccentricity beyond [a], which is admittedly non-rigorous.

And I rely very much on my judgement to gauge the suitability of this approach. It would be impossible for me to provide a full list of the things that might put me off of this strategy but a partial list would include:

i) Has to be a situation where the moment transfer is incidental (no moment frames etc).

ii) Pause to think if the column is supporting a significant beam cantilever with a shallow beam.

iii) Ix_beam >> Ix_column. I love me a W18 shear tabbed into a HSS4.

Thank you for the detailed response. The spans tying into each side of the column are relatively similar, so any pattern loading moment is quite minimal and the dead load moments essentially cancel out.

Columns in this case are also supporting hollowcore which we detail a clip angle directly from plank to column for even more rigidity, leaning more to your K=0.7 statement.

As a sanity check, what are your thoughts on this, too conservative?

-For multi-story columns that I do include the eccentricity on I have always pinned the base at each floor transition, even though it might be a multi-story continuous column. From running a couple cases the single curvature bending is quite conservative when compared to a fully fixed continuous column. Just another safety factor in there. Then I design higher than 0.8 utilization typically, not going above 0.9.

 

[KootK]

Columns in this case are also supporting hollowcore which we detail a clip angle directly from plank to column for even more rigidity, leaning more to your K=0.7 statement.

Maybe. I'd want to get a look at the detail before committing. Often, I deliberately avoid extraneous plank connections to make the connection behavior more predictable. The accidental rotational restraint of plank ends is generally not your friend, both in terms of local plank cracking and the potential for one way shear failure.

 

[W460x68]

Maybe. I'd want to get a look at the detail before committing. Often, I deliberately avoid extraneous plank connections to make the connection behavior more predictable. The accidental rotational restraint of plank ends is generally not your friend, both in terms of local plank cracking and the potential for one way shear failure.

For this particular case, I have been explicitly advised by precast designers (precast design being a delegated design item for most of my projects) that they require some sort of bearing connection where the plank meets a column continuous to the floor above. Typically we provide clip angles from plank to column as @EngDM has described. Are you saying this is something you try to avoid?

 

[KootK]

Are you saying this is something you try to avoid?

Nope. I'm just not entirely clear on whether or not it is that kind of clip angle connection that @EngDM is alluding to. It probably is but, given that he sees the connection as having rotational restraint value, it's conceivable that he may be referencing a topside connection. I just don't want to give my blessing to something that I don't understand.

 

[W460x68]

Nope. I'm just not entirely clear on whether or not it is that kind of clip angle connection that @EngDM is alluding to. It probably is but, given that he sees the connection as having rotational restraint value, it's conceivable that he may be referencing a topside connection. I just don't want to give my blessing to something that I don't understand.

Got it. Thanks for the clarification.

 

[human909]

KootK pretty much covered most things.

To be honest in my day to day engineering I don't fret too much about eccentricity. Lots of things work out in the wash especially if you aren't designing things to 98% of capacity.

To side track things slightly how is this detail for eccentricity! 300mm eccentricity off a 200wide column in its weak axis. But given low loads on the beam it does actually still "work" I just find it horrible detailing.

 

[Tomfh]

This is an interesting topic. I've seen many engineers just design for the axial load and assume the connection and other stiffnesses get the load into the middle of the column.

I check for the eccentricity - the edge of the column, or the bolts, depending on the circumstance.

How do you guys account for P-delta/moment magnification effect, and your interaction?

 

[human909]

This is an interesting topic. I've seen many engineers just design for the axial load and assume the connection and other stiffnesses get the load into the middle of the column.

A dangerous assumption that you will get away with 95% of the time.

It is a code requirement in AS4100 to account for a minimum eccentricity. So I should be always including it. But most of the time I don't explicitly include it in my design calcs. For my designs it might knock a few 2-6% off the column ultimate capacity. My columns aren't designed that close to the line. Also I design and review the connections so I know what I'm dealing with.

I check for the eccentricity - the edge of the column, or the bolts, depending on the circumstance.

Same, if I'm checking.

How do you guys account for P-delta/moment magnification effect, and your interaction?

Non-linear analysis.

 

[Tomfh]

A dangerous assumption that you will get away with 95% of the time.

Have you ever seen it actually go bad, like a column buckling because of it? I know the code says to do it, and I follow it, but I’ve never been quite sure how critical it really is, given there are a lot of engineers who say they're never done it.

Non-linear analysis.

Probably the best way.

Do you ever use the code moment amplification method?

 

[human909]

Have you ever seen it actually go bad, like a column buckling because of it? I know the code says to do it, and I follow it, but I’ve never been quite sure how critical it really is, given there are a lot of engineers who say they're never done it.

It is pretty rare to see actual failures. I've seen serviceability failures on the same job due to not allowing for eccentricity and not allowing for moment transfer between a beam and a column. This was on a job when I was working for the fabricator and I wasn't a yet qualified engineer.

The engineers provided relatively slender SHS I believe 100SHS to support some long span rafters. The connections where end plate connections which essentially functioned as moment connections. Some of the columns had a distinct bend to them when they attracted moment, in the engineering analysis they were a pinned connection, reality was different

Likewise on the same job a cleat plate coming off the 100SHS with a heavy transfer beam also put a significant bend to it due to the eccentricity. The engineers really ran around in a fit about that one. I'd suggest it would have failed well under design loads.

125SHS or 150SHS would have been more suitable. All the guys on site commented how skinny the columns were. Sure they might of calced out for strength but their slenderness caused other issues.

Probably the best way.

Do you ever use the code moment amplification method?

No.

I pretty much let SpaceGass do the heavy lifting.

 

[EngDM]

A dangerous assumption that you will get away with 95% of the time.

It is a code requirement in AS4100 to account for a minimum eccentricity. So I should be always including it. But most of the time I don't explicitly include it in my design calcs. For my designs it might knock a few 2-6% off the column ultimate capacity. My columns aren't designed that close to the line. Also I design and review the connections so I know what I'm dealing with.


Same, if I'm checking.



Non-linear analysis.

Is that 2-6% consistent when using HSS columns? They have a larger moment arm than your typical W-beam web connection.

 

[KootK]

How do you guys account for P-delta/moment magnification effect, and your interaction?

I don't, at least not explicitly or rigorously. It's tough to magnify a moment that you've chosen to pretend is zero to begin with.

One has to feel confident that the moment magnification, in addition to the original moment it would magnify, is covered by this:

Design the column for pure axial (K=1) and keep the combined stress index under 0.8.

As you know, moment magnification is nonlinear and only starts to really run away from you when you approach P_Euler. So a great reason to keep your axial load well away from P_Euler when using this approach.

I won't ever use the approach that I described earlier unless, in my heart, I really do feel that the connection moment would have to flip to a buckling restraint moment before the thing would come down K=0.7-ish. At the same time, that does cover a whole lot of practical gravity column cases.

I wouldn't normally do this for, say, and angle bearing seat unless it had some very specific detailing features.

 

[Brad805]

This is a good example where we as designers could consider a small brace eccentricity to improve the detailers job. Maybe it was not feasible here, but I suspect the detailer did this to avoid field welds. With the level of automation these days field welding has become virtually a swear word. These type of things can get missed in the delgated world when a junior engineer is responsible to review the shops.